The efficiency of non-bank financial intermediaries: Empirical evidence from Malaysia The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 149 THE EFFICIENCY OF NON-BANK FINANCIAL INTERMEDIARIES: EMPIRICAL EVIDENCE FROM MALAYSIA Fadzlan Sufian The University of Malaysia and CIMB Bank Berhad Abstract This paper investigates the performance of Malaysian non-bank financial institutions during the period of 2000-2004. Several efficiency estimates of individual NBFIs are evaluated using the non-parametric Data Envelopment Analysis (DEA) method. The findings suggest that during the period of study, scale inefficiency outweighs pure technical inefficiency in the Malaysian NBFI sector. We find that the merchant banks have exhibited a higher, technical efficiency compared to their peers. The empirical findings suggest that scale efficiency tends to be more sensitive to the exclusion of risk factors, implying that potential economies of scale may be overestimated when risk factors are excluded. Keywords: Non-Bank financial intermediaries, Data Envelopment Analysis (DEA), Risk JEL Classification: G21, G28 1. Introduction Non-Bank Financial Institutions (NBFIs) play important dual roles in a financial system. They complement the role of commercial banks by filling in financial intermediation gaps by offering a range of products and services. They also compete with commercial banks, forcing the latter to be more efficient and responsive to their customers needs. NBFIs’ state of development is usually a good indicator to the state of development of a country’s financial system as a whole. The importance of investigating the efficiency and productivity of Malaysian NBFIs could be best justified by the fact they play important roles in complementing the facilities offered by the commercial banks, as well as being key players in the development of the capital markets. As sophisticated and well-developed as capital markets are considered to be as the hallmark for a market-based economy worldwide, such a study of this nature is particularly important as the health and development of the capital market relies largely upon the performance of NBFIs. Hence, efficient and productive NBFIs are expected to enhance the Malaysian capital markets in its pursuit to move towards a full market based economy. IJBF 150 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 Despite the significant, economic developments of the NBFI sector, studies that attempt to investigate this issue are relatively scarce. Over the years, while there have been extensive literature examining the productivity and efficiency of banking industries in various countries, empirical works on NBFIs’ productivity & efficiency are still in its infancy. To the best of our knowledge, there has been no microeconomic study performed with respect to NBFIs. The study therefore aims to fill a demanding gap in that case. Nevertheless, the study will also be the first to investigate the sources of NBFIs’ productivity changes in developing economies. Section 2 will provide a brief overview of the Malaysian financial system with reviews of related studies. Section 3 will outline the approaches to the measurement and estimation of efficiency change, while Section 4 will discuss the results. Naturally, Section 5 will conclude the paper. 2. Background and Related Literature The Malaysian financial system can be broadly divided into the banking system and non-bank financial intermediaries. The banking system is the largest component, accounting for approximately 70 percent of the financial system’s total assets. The banking system can be further divided into three main groups, namely the commercial banks, financial companies, and merchant banks. The commercial banks are the main players in the banking system. They are the largest and most significant providers of funds in the banking system, enjoying the widest scope of permissible activities, those of which are able to engage in a full range of banking services. Financial companies formed the second largest group of deposit taking institutions in Malaysia. Traditionally, financial companies specialize in consumption credit, comprising mainly of hire purchase financing, leasing, housing loans, block discounting, and secured personal loans. Merchant banks emerged in the Malaysian banking scene in 1970, marking an important milestone in the development of the financial system, alongside Malaysian corporate development. They play a role in the short-term money market and capital raising activities such as financing, syndicating, corporate financing, and management advisory services that arrange for the issue and listing of shares, as well as managing portfolios. The Malaysian financial system’s assets and liabilities continued to be highly concentrated at the commercial banking sector with total assets and liabilities amounting to RM 761,254.8 billion (or 3.05 times the national GDP at the end of 2004). Prior to the Asian Financial Crisis in 1997/98, financial companies’ assets and liabilities were seen increasing from only RM531 million (or 0.05 times the national GDP in 1970) to a high of RM 152.4 billion (or 0.77 times in 1997). The ratio however, has gradually declined to RM 123.6 billion (or 0.60 times in 1998) to RM 109,409.8 billion (or 0.52 times the GDP in 2000), before increasing again in year 2001, to reach a post crisis high of RM 141,911.0 billion (or 0.61 times the GDP in 2003). Due to further consolidation in the Malaysian financial sector, financial companies’ assets as a ratio of the national GDP declined again to reach a low of The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 151 0.27 times in 2004. As for the merchant banks, a similar trend is observed where their assets and liabilities (as a ratio of the national GDP) have been increasing since 1971, reaching a peak of RM 44.3 billion or 0.23 times GDP in 1997 (before the Asian financial crisis). During the post crisis period, the merchant banks’ assets and liabilities continued to remain stable at 0.17 to 0.22 times the national GDP. A combination of both financial companies and merchant banks’ total assets reveal that the non-bank financial sector commanded approximately 22.8 percent of the banking system’s total assets and liabilities.1 Table 1: Assets of the Financial System, 1960 – 2004 Year Commercial Banks Finance Companies Merchant Banks RM million As a Ratio of GDP RM million As a Ratio of GDP RM million As a Ratio of GDP 1960 1,231.9 0.21 n.a. n.a. n.a. n.a. 1970 4,460.2 0.38 531.0 0.05 19.6* 0.002 1980 32,186.1 0.63 5,635.4 0.13 2,228.7 0.05 1990 129,284.9 1.23 39,448.0 0.50 11,063.2 0.14 1995 295,460.0 1.77 91,892.0 0.55 27,062.0 0.16 1996 360,126.8 1.98 119,768.8 0.65 34,072.8 0.19 1997 480,248.1 2.46 152,386.8 0.77 44,300.0 0.23 1998 453,492.0 2.52 123,596.9 0.68 39,227.8 0.22 1999 482,738.3 2.50 116,438.0 0.60 39,184.0 0.20 2000 512,714.7 2.44 109,409.8 0.52 36,876.0 0.18 2001 529,735.5 2.51 121,811.1 0.58 41,025.2 0.19 2002 563,254.1 2.56 130,520.0 0.59 41,415.5 0.19 2003 629,975.3 2.71 141,911.0 0.61 44,103.6 0.19 2004 761,254.8 3.05 68,421.1 0.27 42,691.0 0.17 Source: Bank Negara Malaysia. *As at end 1971. The Malaysian financial sector is currently facing a number of challenges such as frequent changes in technology required for modern banking, increasing competition, rising customer expectations, etc. Hence, the efficiency and productivity issues have become a major area of concern for the banks’ management. In fact, productivity is an important criterion to measure the performance of banks in addition to profitability, financial, and operational efficiency. An efficient management of banking operations aimed at increasing the efficiency and productivity of the financial sector requires up to date knowledge. 1 The figure is at end-2003, prior to the consolidation of financial companies into their respective commercial banking parents. 152 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 A lot of research work has so far taken place concerning the views about the role of financial & banking developments in economic growth [McKinnon (1973); Shaw (1973); Rajan & Zingales (1998); Levine (2004); Singh (2005)], as well as banking efficiency and productivity [(Das & Ghosh (2006); Sufian (2007); and Weill (2007)].2 Similarly, some studies have been undertaken for measuring the productivity and efficiency of banks in Malaysia [most notably, Katib & Matthews (2000) and Okuda & Hashimoto (2004)]. Concerning our information, despite NFBIs’ significance towards economic development, studies that attempt to investigate this issue are relatively scarce. Over the years, while there have been extensive literature examining the productivity & efficiency of banking industries in various countries, empirical works on NBFIs’ productivity & efficiency are still in its infancy. 3. Methodology and Data A non-parametric Data Envelopment Analysis (DEA) is employed with a variable return to scale assumption, measuring Malaysian NBFIs’ input-oriented technical efficiencies. DEA involves constructing a non-parametric production frontier based on the actual input-output observations in the sample, relative to the measured efficiency of each firm in the sample (Coelli, 1996). Let us give a short description of the Data Envelopment Analysis3. Assume that there is data on K inputs and M outputs for each N NBFI. For the ith NBFI, these are represented by the vectors x i and y i , respectively. Let us introduce the K x N input matrix, X, and the M x N output matrix, Y. To measure the efficiency for each NBFI, we calculate a ratio of all inputs, such as (u’y i /v’x i ), where u is an M x 1 vector of output weights, and v is a K x 1 vector of input weights. To select optimal weights, we specify the following mathematical programming problem: min (u’y i /v’x i ), u,v u’y i /v’x i ≤1, j = 1, 2,…, N, u,v ≥ 0 (1) The above formulation has a problem of infinite solutions; therefore we impose the constraint v’x i = 1, which leads to: min (μ’y i ), μ,ϕ ϕ’x i = 1 μ’y i – ϕ’x j ≤0 j = 1, 2,…, N, μ,ϕ ≥ 0 (2) 2 See Berger & Humphrey (1997) for an excellent review. 3 Good reference books on efficiency measures are Coelli et al. (1998), Cooper et al. (2000), and Thanassoulis (2001). The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 153 where we change notation from u & v to μ & ϕ, respectively, in order to reflect transformations. Using the duality in linear programming, an equivalent envelopment form of this problem can be derived: min θ, θ, l y i + Yλ > 0 θx i - Xλ > 0 λ > 0 (3) where θ is a scalar representing the value of the efficiency score for the ith decision-making unit, which will range between 0 and 1. λ is a vector of N x 1 constants. The linear programming has to be solved N times, once for each NBFI in the sample. In order to calculate efficiency under the assumption of variable returns to scale, the convexity constraint (N1'λ=1) will be added to ensure that an inefficient NBFI is only compared against NBFIs of similar size; thus providing the basis for measuring economies of scale within the DEA concept. For the empirical analysis, all Malaysian NBFIs would be incorporated. The annual balance sheets and income statements used to construct the variables for the empirical analysis are sourced from published balance sheet information in the annual reports. Due to scarce data from M & A activity, the final sample was an unbalanced panel sample of 92 NBFI observations. There are two main approaches that exist in banking theory literature to define the banking function: the production and intermediation approaches [Sealey & Lindley (1977)]. Under the production approach, which was pioneered by Benston (1965), a financial institution is defined as a producer of services for account holders. That is, they perform transactions on deposit accounts and process documents such as loans. The intermediation approach on the other hand, assumes that financial firms act as an intermediary between savers and borrowers, hypothesizing total loans and securities as outputs; whereas deposits with labor and physical capital are defined as inputs. For the purpose of this study, a variation of the intermediation approach or asset approach originally developed by Sealey and Lindley (1977) will be adopted in the definition of inputs and outputs used. Given the sensitivity of efficiency estimates to the specification of outputs and inputs, we have estimated two alternative models. In DEA Model A, we model Malaysian NBFIs as multi-product firms, producing two outputs by employing two inputs. Accordingly, Total Deposits (x1) include deposits from customers and other banks. Fixed Assets (x2) are used as input vectors to produce Total Loans (y1), which include loans to customers and other banks. Investments (y2) include investment securities held for trading, investment securities available for sale (AFS), and investment securities held to maturity. To assess the importance of risk and lending quality problems in explaining the efficiency of Malaysian NBFIs, following the approach by the likes of Drake and Hall (2003) and Charnes et al. (1990), Loan Loss Provisions (x3) is incorporated as an input variable in DEA Model B. 15 4 T h e In te rn a ti o n a l Jo u rn a l o f B a n ki n g a n d F in a n ce , 2 00 7/ 08 V ol . 5 . N um be r 2: 2 00 8: 1 49 -1 67 Table 2: Descriptive Statistics for Inputs and Outputs The table presents summary statistics of the variables used to construct the efficiency frontier for both DEA Model A and DEA Model B over the period 2000- 2004. The sample is divided into peer groups (i.e., merchant banks and financial companies). MB denotes merchant banks and FC denotes finance companies. 2000 (RMb) 2001 (RMb) 2002 (RMb) 2003 (RMb) 2004 (RMb) Outputs MB FC MB FC MB FC MB FC MB FC Total Loans Min 172.05 1,927.44 135.04 887.41 136.73 1,116.10 89.77 1,363.46 136.55 1408.4 Mean 1,784.70 6,832.02 1,549.44 6,904.33 1,336.28 7,383.17 1,173.53 9,773.36 1,045.39 9,454.49 Max 7,677.01 15,743.03 7,571.63 15,765.02 6,906.83 16,732.43 5,582.32 25,160.44 5,274.91 26,048.86 S.D 2,426.46 4,537.73 2,192.84 4,929.12 2,014.58 5,095.30 1,706.99 7,690.041 1,628.510 8,241.46 Investments Min 61.79 180.97 74.8 40.55 57.82 41.69 99.51 75.77 98.67 69.91 Mean 1,710.31 1,473.56 1,530.41 1,116.91 1,655.44 818.38 2,085.44 966.80 2,058.65 797.97 Max 5,525.08 3,416.52 4,985.66 2,800.68 5,999.55 1,730.22 8,023.00 2,454.12 6,558.26 2,317.31 S.D 1,945.90 1,220.12 1,858.97 1,063.46 2,035.57 599.90 2,503.18 991.46 1,974.16 910.38 Inputs Fixed Assets Min 0.84 21.68 0.25 5.31 0.32 6.71 0.10 6.54 0.06 2.02 Mean 4.46 55.66 7.90 49.63 11.51 88.86 16.21 87.12 14.56 95.84 Max 11.71 186.94 39.69 205.86 45.00 425.22 53.69 439.35 54.17 424.60 S.D 4.07 54.42 12.45 58.76 16.62 130.90 19.61 134.07 19.93 141.93 Total De- posits Min 58.30 1,480.76 88.86 913.12 20.23 1,164.16 63.78 1,226.55 74.62 1,084.00 Mean 2,331.99 7,145.82 1,906.52 6,514.09 1,555.06 7,445.49 1,660.76 8,306.54 2,003.80 7,903.50 Max 8,110.02 14,546.27 8,853.50 13,928.60 5,356.46 16,025.89 5,302.27 19,609.19 5,929.86 20,411.79 S.D 2,543.59 4,183.57 2,596.72 4,757.44 1,676.94 5,590.04 1,676.48 6,506.15 1,796.95 7,057.08 Loan Loss Provisions Min 1.18 0.60 10.00 35.18 0.08 17.78 7.47 33.79 17.70 59.72 Mean 55.7925 136.40 71.09 119.96 32.78 107.03 47.45 108.10 58.43 155.74 Max 160.28 519.78 229.44 384.38 154.35 311.64 253.24 378.73 159.31 347.76 S.D 58.02 168.09 68.35 106.56 47.14 84.65 79.14 101.04 40.40 96.39 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 155 Table 2 presents the summary statistics of the input and output variables used to construct the efficiency frontier. During the period of study, it is apparent that the financial companies were almost three times larger (in terms of asset size) and commanded higher market share in terms of loans & deposits, compared with their merchant bank peers. On the other hand, although the merchant banks were smaller, they seem to have produced a higher amount of investments with lower amounts of defaulted loans. The differences are further confirmed by a series of parametric (t-test) and non-parametric (Kruskal-Wallis and Mann-Whitney [Wilcoxon Rank- Sum] tests), which suggest that the differences in the mean are significant for all variables at the 1 per cent level of significance4. 4. Results In this section, we will discuss the technical efficiency change (TE) of the Malaysian NBFI sector, measured by the Data Envelopment Analysis (DEA) method, along with its decomposition into pure technical efficiency (PTE) and scale efficiency (SE) components. With the existence of scale inefficiency, we will attempt to provide evidence on the nature of returns to scale of Malaysian NBFI. The efficiency of Malaysian NBFIs was first examined by applying the DEA method for each year under investigation by employing the traditional input-output variables. We extend the analysis to examine the merchant banks and financial companies’ efficiency results derived from an alternative model, which incorporates a non-discretionary, input variable. 4.1 Efficieny of the Malaysian NBFI Sector Table 3 presents the mean efficiency scores of the merchant banks for the years 2000 (Panel A), 2001 (Panel B), 2002 (Panel C), 2003 (Panel D), 2004 (Panel E), and All Years (Panel F). The results from DEA Model A seems to suggest that the merchant banks’ mean technical efficiency has been on a declining trend during the earlier part of the studies, before increasing again during the latter years. The decomposition of overall efficiency into its pure technical and scale efficiency components suggest that the merchant banks have exhibited higher scale efficiency during 2000 and 2002. Overall, the results imply that during the period of study, the merchant banks have been operating at the wrong scale of operations. During the period of study, the results from Panel F of Table 3 seem to suggest that the merchant banks have exhibited a mean technical efficiency of 69.6 percent, suggesting a mean input waste of 30.4 percent. In other words, the merchant banks could have produced the same amount of outputs by only using 69.6 percent of the amount of inputs it uses. From Table 3 (Panel F), it is also clear that scale inefficiency outweighs pure, technical inefficiency in determining the total technical inefficiency of the merchant banks. 4 Investment is not significant in the case of the Mann-Whitney [Wilcoxon Rank-Sum] and Kruskal-Wallis tests at any conventional levels. To conserve space, we do not report the results here. They are available from the authors upon request. 156 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 Table 3: Summary Statistics of Efficiency Measures – Merchant Banks (DEA Model A) The table presents mean, minimum, maximum, and standard deviation of Malaysian NBFIs’ technical efficiency (TE), its mutually exhaustive, pure technical efficiency (PTE), and scale efficiency (SE) components derived from DEA Model A (excluding the risk factor). Panel A, B, C, D, and E shows the mean, minimum, maximum, and standard deviation of TE, PTE, and SE of the merchant banks for the years 2000, 2001, 2002, 2003, and 2004, respectively. Panel F presents the merchant banks mean, minimum, maximum, and standard deviation of TE, PTE, and SE scores, respectively. The TE, PTE, and SE scores are bounded between a minimum of 0 and a maximum of 1. Efficiency Measures Mean Minimum Maximum Std. Dev. Panel A: 2000 Technical Efficiency 0.908 0.443 1.000 0.193 Pure Technical Efficiency 0.925 0.527 1.000 0.167 Scale Efficiency 0.974 0.841 1.000 0.056 Panel B: 2001 Technical Efficiency 0.745 0.342 1.000 0.271 Pure Technical Efficiency 0.897 0.547 1.000 0.180 Scale Efficiency 0.822 0.372 1.000 0.218 Panel C: 2002 Technical Efficiency 0.750 0.216 1.000 0.327 Pure Technical Efficiency 0.851 0.266 1.000 0.266 Scale Efficiency 0.861 0.438 1.000 0.222 Panel D: 2003 Technical Efficiency 0.506 0.188 1.000 0.320 Pure Technical Efficiency 0.894 0.429 1.000 0.201 Scale Efficiency 0.562 0.188 1.000 0.298 Panel E: 2004 Technical Efficiency 0.582 0.331 1.000 0.209 Pure Technical Efficiency 0.924 0.685 1.000 0.133 Scale Efficiency 0.636 0.386 1.000 0.226 Panel F: Merchant Banks All Years Technical Efficiency 0.696 0.188 1.000 0.295 Pure Technical Efficiency 0.897 0.266 1.000 0.190 Scale Efficiency 0.770 0.188 1.000 0.258 Table 4 presents mean efficiency scores of the finance companies for the years 2000 (Panel A), 2001 (Panel B), 2002 (Panel C), 2003 (Panel D), 2004 (Panel E), and All Years (Panel F). Similar to their merchant bank counterparts, the results from DEA Model A seem to suggest that the financial companies’ mean technical efficiency has been on a declining trend during the earlier part of the studies, before increasing during the latter years. The decomposition of technical efficiency into its pure technical and scale efficiency components suggest that scale inefficiency outweighs the pure technical inefficiency of the financial companies during all years. The results seem to suggest that the finance companies have exhibited a mean technical efficiency of 44.7 percent, which is lower compared to their merchant bank counterparts. The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 157 Likewise, the results suggest that the financial companies’ inefficiency was mainly due to scale, rather than pure technical albeit at a higher degree of 44.8 percent (merchant banks – 23.0 percent). The financial companies also seem to have exhibited a lower pure technical efficiency of 82.0 percent (merchant banks – 89.7 percent). Overall, the results suggest that compared to their merchant bank counterparts, the financial companies were relatively managerially inefficient in controlling their operating costs and have been operating at a relatively less optimal scale of operations. Table 4: Summary Statistics of Efficiency Measures – Finance Companies (DEA Model A) The table presents mean, minimum, maximum, and standard deviation of Malaysian NBFIs’ technical efficiency (TE), its mutually exhaustive, pure technical efficiency (PTE), and scale efficiency (SE) components derived from DEA Model A (excluding the risk factor). Panel A, B, C, D, and E shows the mean, minimum, maximum, and standard deviation of TE, PTE, and SE of the finance companies for the years 2000, 2001, 2002, 2003, and 2004, respectively. Panel F presents the finance companies’ mean, minimum, maximum, and standard deviation of TE, PTE, and SE scores, respectively. The TE, PTE, and SE scores are bounded between a minimum of 0 and a maximum of 1. Efficiency Measures Mean Minimum Maximum Std. Dev. Panel A: 2000 Technical Efficiency 0.538 0.350 1.000 0.216 Pure Technical Efficiency 0.811 0.466 1.000 0.197 Scale Efficiency 0.679 0.399 1.000 0.228 Panel B: 2001 Technical Efficiency 0.389 0.266 0.693 0.142 Pure Technical Efficiency 0.807 0.491 1.000 0.219 Scale Efficiency 0.489 0.342 0.693 0.124 Panel C: 2002 Technical Efficiency 0.248 0.058 0.589 0.149 Pure Technical Efficiency 0.828 0.530 1.000 0.186 Scale Efficiency 0.300 0.092 0.589 0.155 Panel D: 2003 Technical Efficiency 0.490 0.243 0.769 0.140 Pure Technical Efficiency 0.822 0.440 1.000 0.199 Scale Efficiency 0.599 0.446 0.769 0.104 Panel E: 2004 Technical Efficiency 0.625 0.296 0.974 0.188 Pure Technical Efficiency 0.835 0.428 1.000 0.209 Scale Efficiency 0.758 0.540 0.974 0.169 Panel F: Finance Companies All Years Technical Efficiency 0.447 0.058 1.000 0.205 Pure Technical Efficiency 0.820 0.428 1.000 0.193 Scale Efficiency 0.552 0.092 1.000 0.220 158 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 The findings are interesting in that although the merchant banks were small relative to their financial counterparts with having relatively limited operations, they seem to have exhibited higher efficiency levels. The findings support the divisibility theory, which holds that there will be no such operational advantage accruing to large NBFIs if the technology is divisible. That is, small scale NBFIs can produce financial services at costs per unit output comparable to those of large NBFIs, suggesting no or possibly negative association between size and performance. This was made possible as advances in technology reduced the size and cost of automated equipment; thus, significantly enhancing small NBFIs ability to purchase expensive technology, implying more divisibility in the banking industry’s technology (Kolari & Zardkoohi, 1987). Since the dominant source of the total technical X- (in) efficiency in the Malaysian NBFI sector seems to be scale related, it is worth investigating the composition of the efficiency frontier. Table 5 shows NBFIs that lie on the efficiency frontier under DEA Model A. Table 5: Composition of Production Frontiers (DEA Model A) Bank Type 2000 2001 2002 2003 2004 Count Affin Merchant Bank MB IRS DRS DRS DRS DRS 0 Affin-ACF Finance FC DRS DRS DRS DRS DRS 0 Alliance Finance FC DRS DRS DRS DRS 0 Alliance Merchant Bank MB DRS DRS DRS DRS 0 Arab-Malaysian Finance FC CRS DRS DRS DRS DRS 1 Arab-Malaysian Merchant Bank MB CRS DRS DRS DRS DRS 1 Aseambankers MB CRS DRS CRS DRS DRS 1 Bumiputra-Commerce Finance FC DRS DRS DRS DRS DRS 0 Commerce International Merchant Bankers MB CRS CRS DRS DRS DRS 2 EON Finance FC DRS DRS DRS DRS 0 Hong Leong Finance FC DRS DRS DRS DRS DRS 0 Malaysian International Merchant Bankers MB IRS CRS CRS 2 Mayban Finance FC DRS DRS DRS DRS DRS 0 Public Finance FC DRS DRS DRS DRS 0 Public Merchant Bank MB CRS CRS DRS DRS 2 RHB Delta Finance FC DRS DRS DRS DRS 0 RHB Sakura Merchant Bankers MB CRS CRS CRS DRS DRS 3 Southern Finance FC DRS DRS DRS DRS DRS 0 Southern Investment Bank MB CRS IRS DRS CRS IRS 2 Utama Merchant Bank MB IRS DRS CRS CRS CRS 3 Number of NBFI n 6 4 5 2 1 Note: CRS – (Constant Returns to Scale); DRS – (Decreasing Returns to Scale); IRS – (Increasing Returns to Scale); The NBFIs corresponds to the shaded regions which have not been efficient in any year in the sample period (2001-2005) compared to the other NBFIs in the sample; MB – Merchant Bank; FC – Finance Company International Journal of Banking and Finance, Vol. 5, Iss. 2 [2008], Art. 8 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 159 The composition of the efficiency frontier for DEA Model A suggests that the number of 100 percent efficient NBFIs [operating at constant returns to scale (CRS)], varies between one to six NBFIs. During the period of study, the merchant banks seem to have dominated the efficiency frontier for DEA Model A. It is also clear from the results that two merchant banks, namely RHB Sakura Merchant Bankers and Utama Merchant Bank, have appeared the most times on the efficiency frontier. A total of eight merchant banks have appeared at least once on the efficiency frontier, while only two merchant banks have failed to make it to the frontier. On the other hand, the results seem to suggest that only one financial company has managed to make it to the frontier, while nine others have never made it to the efficiency frontier throughout the period of study. 2.2 Non-performing Loans and the Gap Between the Two DEA Models Having established the basic DEA model, we now analyze the potential impact of risk and problem loans concerning the efficiency of Malaysian NBFIs. As indicated previously, these results are obtained by modifying the initial DEA model to incorporate an additional, non-discretionary input variable, in the form of provisions of loans losses. In general, the findings seem to suggest that controlling for problem loans resulted in a higher mean technical efficiency of Malaysian NBFIs during all years5. In line with the findings by Drake & Hall (2003) and Altunbas et al. (2000), the results seem to suggest that potential economies of scale may well be overestimated when risk factors are excluded. Likewise, it is clear that the inclusion of loan loss provisions has resulted in a higher mean pure technical efficiency of Malaysian NBFIs6. The results support earlier findings by Altunbas et al. (2000), who had suggested that the mean scale efficiency estimate is much more sensitive than the mean pure technical efficiency estimate to the exclusion of risk factors. We now turn to discuss the impact of the inclusion of loan loss provisions on the evolution of the merchant banks’ technical efficiency. The results from Table 6 suggest that the inclusion of risk factors has resulted in a higher technical efficiency for merchant banks. It is also apparent that the inclusion of loan loss provisions has had a greater positive impact on the merchant banks’ scale efficiency. Table 7 highlights the results for the financial companies. Similar to their merchant bank counterparts, the results from Table 7 suggest that the inclusion of risk factors has resulted in a higher technical efficiency for financial companies. Likewise, it is also apparent that the inclusion of loan loss provisions has had a greater positive impact on the financial companies’ scale efficiency. With a closer look at the results, it seems that the magnitude of the increase in the financial companies’ pure technical and scale efficiency is higher compared to their merchant bank peers. A plausible reason is that during the period of study, the financial companies had a higher amount of defaulted loans compared to their peers. 5 Except for the merchant banks during the year 2000. 6 Except for the merchant banks during the year 2000. 160 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 The empirical findings clearly demonstrate the importance of risk in explaining financial institutions’ efficiency, in particular scale efficiency. If anything could be deduced from the results, the omission of risk factors may significantly overestimate financial institutions’ potential economies of scale, which could lead to bias conclusions and consequently, policy recommendations. The findings are particularly important for the Malaysian policy makers in its quest to consolidate the banking system further to achieve greater economies of scale and efficiency. The Malaysian government has always believed that such a move would result in larger institutions, which could withstand greater competition from foreign players, as well as any shocks to the financial system. As the actual potential economies of scale may significantly be lower than initially expected, policy makers should be Table 6: Summary Statistics of Efficiency Measures – Merchant Banks (DEA Model B) The table presents mean, minimum, maximum, and standard deviation of Malaysian NBFIs’ technical efficiency (TE), its mutually exhaustive, pure technical efficiency (PTE), and scale efficiency (SE) components derived from DEA Model B (inclusive of the risk factor). Panel A, B, C, D, and E shows the mean, minimum, maximum, and standard deviation of TE, PTE, and SE of the merchant banks for the years 2000, 2001, 2002, 2003, and 2004, respectively. Panel F presents the merchant banks mean, minimum, maximum, and standard deviation of TE, PTE, and SE scores, respectively. The TE, PTE, and SE scores are bounded between a minimum of 0 and a maximum of 1. Efficiency Measures Mean Minimum Maximum Std. Dev. Panel A: 2000 Technical Efficiency 0.908 0.443 1.000 0.193 Pure Technical Efficiency 0.926 0.532 1.000 0.165 Scale Efficiency 0.973 0.834 1.000 0.058 Panel B: 2001 Technical Efficiency 0.818 0.437 1.000 0.219 Pure Technical Efficiency 0.897 0.551 1.000 0.179 Scale Efficiency 0.903 0.683 1.000 0.119 Panel C: 2002 Technical Efficiency 0.837 0.275 1.000 0.251 Pure Technical Efficiency 0.914 0.492 1.000 0.159 Scale Efficiency 0.905 0.559 1.000 0.189 Panel D: 2003 Technical Efficiency 0.885 0.504 1.000 0.204 Pure Technical Efficiency 0.912 0.513 1.000 0.181 Scale Efficiency 0.963 0.793 1.000 0.085 Panel E: 2004 Technical Efficiency 0.896 0.700 1.000 0.127 Pure Technical Efficiency 0.950 0.857 1.000 0.048 Scale Efficiency 0.946 0.700 1.000 0.110 Panel F: Merchant Banks All Years Technical Efficiency 0.869 0.275 1.000 0.199 Pure Technical Efficiency 0.929 0.492 1.000 0.151 Scale Efficiency 0.927 0.559 1.000 0.123 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 161 Table 7: Summary Statistics of Efficiency Measures – Finance Companies (DEA Model B) The table presents mean, minimum, maximum, and standard deviation of Malaysian NBFIs’ technical efficiency (TE), its mutually exhaustive, pure technical efficiency (PTE), and scale efficiency (SE) components derived from DEA Model B (inclusive of the risk factor). Panel A, B, C, D, and E shows the mean, minimum, maximum, and standard deviation of TE, PTE, and SE of the merchant banks for the years 2000, 2001, 2002, 2003, and 2004, respectively. Panel F presents the merchant banks mean, minimum, maximum, and standard deviation of TE, PTE, and SE scores, respectively. The TE, PTE, and SE scores are bounded between a minimum of 0 and a maximum of 1. Efficiency Measures Mean Minimum Maximum Std. Dev. Panel A: 2000 Technical Efficiency 0.823 0.560 1.000 0.174 Pure Technical Efficiency 0.902 0.561 1.000 0.162 Scale Efficiency 0.918 0.644 1.000 0.122 Panel B: 2001 Technical Efficiency 0.799 0.511 1.000 0.173 Pure Technical Efficiency 0.878 0.517 1.000 0.196 Scale Efficiency 0.918 0.747 1.000 0.094 Panel C: 2002 Technical Efficiency 0.643 0.324 1.000 0.212 Pure Technical Efficiency 0.860 0.533 1.000 0.181 Scale Efficiency 0.726 0.512 1.000 0.139 Panel D: 2003 Technical Efficiency 0.801 0.554 1.000 0.157 Pure Technical Efficiency 0.859 0.562 1.000 0.153 Scale Efficiency 0.940 0.752 1.000 0.085 Panel E: 2004 Technical Efficiency 0.963 0.764 1.000 0.097 Pure Technical Efficiency 0.982 0.769 1.000 0.098 Scale Efficiency 0.979 0.949 1.000 0.018 Panel F: Finance Companies All Years Technical Efficiency 0.795 0.324 1.000 0.189 Pure Technical Efficiency 0.882 0.517 1.000 0.161 Scale Efficiency 0.900 0.512 1.000 0.130 more cautious in promoting mergers as a means in achieving greater efficiency by attaining better economies of scale. Furthermore, most of the research conducted surrounding the explanation of bank or thrift industry failures had found that failing institutions carried a large proportion of non-performing loans in their books prior to failure [Dermiguc-Kunt (1989); Whalen (1991); Barr & Siems (1994); Berger & Humphrey (1992); Barr & Siems (1994); and Wheelock & Wilson (1995)]. Banks approaching failure tend to have low cost efficiency while experiencing high ratios of problem loans, as failing banks tend to be located far from the best practice frontiers. 162 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 Table 8: Composition of Production Frontiers (DEA Model B) Bank Type 2000 2001 2002 2003 2004 Count Affin Merchant Bank MB IRS DRS DRS CRS CRS 2 Affin-ACF Finance FC DRS DRS DRS DRS IRS 0 Alliance Finance FC DRS DRS DRS CRS 1 Alliance Merchant Bank MB DRS DRS IRS IRS 0 Arab-Malaysian Finance FC CRS DRS DRS DRS DRS 1 Arab-Malaysian Merchant Bank MB CRS DRS DRS DRS CRS 2 Aseambankers MB CRS DRS CRS CRS CRS 4 Bumiputra-Commerce Finance FC CRS CRS DRS DRS CRS 3 Commerce International Merchant Bankers MB CRS CRS CRS CRS DRS 4 EON Finance Berhad FC CRS DRS DRS DRS 1 Hong Leong Finance FC DRS DRS DRS IRS DRS 0 Malaysian International Merchant Bankers MB IRS CRS CRS 2 Mayban Finance FC DRS DRS DRS CRS CRS 2 Public Finance FC DRS CRS CRS CRS 3 Public Merchant Bank MB CRS CRS CRS CRS 4 RHB Delta Finance FC IRS DRS CRS CRS 2 RHB Sakura Merchant Bankers MB CRS CRS CRS DRS CRS 4 Southern Finance FC DRS DRS DRS DRS IRS 0 Southern Investment Bank MB CRS IRS IRS CRS IRS 2 Utama Merchant Bank MB IRS DRS CRS CRS CRS 3 Number of NBFI n 8 6 7 9 10 Note: CRS – (Constant Returns to Scale); DRS – (Decreasing Returns to Scale); IRS – (Increasing Returns to Scale); The NBFIs corresponds to the shaded regions which have not been efficient in any year in the sample period (2001-2005) compared to the other NBFIs in the sample; MB – Merchant Bank; FC – Finance Company Next, the composition of the efficiency frontier and the nature of the returns to scale for DEA Model B are discussed. Table 8 presents the results of the nature of returns to scale in the Malaysian NBFI sector, derived from DEA Model B. Unlike the results from DEA Model A, the composition of the efficiency frontier for DEA Model B suggests that the number of 100 percent efficient NBFIs had increased substantially to between six and ten NBFIs. The results from DEA Model B are very much similar to those from DEA Model A, where the merchant banks seem to have dominated the efficiency frontier. It is apparent from Table 8 that the global leaders under DEA Model B have increased to four merchant banks, while there was only one merchant bank that failed to appear on the efficiency frontier throughout the period of study. Unlike DEA Model A, the results from DEA Model B suggest that seven finance companies have managed to appear on the efficiency frontier, while there were only three finance companies that have never made it to the efficiency frontier throughout the period of study. T h e In tern a tio n a l Jo u rn a l o f B a n kin g a n d F in a n ce, 2007/08 V ol. 5. N um ber 2: 2008: 149-167 163 Table 9: Summary of the Null Hypothesis Tests of Identical Technologies between Merchant Banks and Finance Companies The table present results from the parametric (ANOVA and t-test) and nonparametric (Kolmogorov-Smirnov, Mann-Whitney and Kruskall-Wallis) tests. The tests are performed to test the null hypothesis that domestic and foreign banks are drawn from the same population (environment). Test methodology follows among others, Aly et al. (1990), Elyasiani and Mehdian (1992), and Isik and Hassan (2002). *** indicate significant at the 5% level. Test Groups Parametric Test Non-Parametric Test Individual Tests Analysis of Variance (ANOVA) test t-test Kolmogorov-Smirnov [K-S] test Mann-Whitney [Wilcoxon Rank-Sum] test Kruskall-Wallis Equality of Populations test Hypotheses Mean mb = Mean fc Distribution mb = Distribution fc Median mb = Median fc Test Statistics F (Prb > F) t (Prb > t) K-S (Prb > K-S) z (Prb > z) χ2 (Prb > χ2) Panel A: 2000 TE Model A TE Model B PTE Model A PTE Model B SE Model A SE Model B 15.606*** 0.851 1.107 0.083 14.699*** 1.318 -3.950*** -0.922 -1.052 -0.289 -3.834*** -1.148 1.572*** 1.000 0.857 0.500 1.601*** 0.750 8.500*** 21.500 22.000 26.000 7.500*** 22.500 7.316*** 1.387 2.128 0.524 7.868*** 1.136 Panel B: 2001 TE Model A TE Model B PTE Model A PTE Model B SE Model A SE Model B 0.634*** 0.045 0.040 0.051 17.639*** 0.095 -3.678*** -0.213 -0.996 -0.225 -4.200*** -0.308 1.342 0.671 0.671 0.447 1.565*** 0.447 11.500*** 44.000 35.000 44.000 12.000*** 47.000 8.541*** 0.211 1.541 0.283 8.314*** 0.053 16 4 T h e In te rn a ti o n a l Jo u rn a l o f B a n ki n g a n d F in a n ce , 2 00 7/ 08 V ol . 5 . N um be r 2: 2 00 8: 1 49 -1 67 Panel C: 2002 TE Model A TE Model B PTE Model A PTE Model B SE Model A SE Model B 19.470*** 3.911 0.053 1.530 42.959*** 3.494 -4.412*** -1.978 -0.230 -1.237 -6.554*** -1.869 1.565*** 1.342 0.671 0.894 2.012*** 1.342 11.000*** 24.000 40.000 34.000 2.000*** 28.000 8.824*** 4.033*** 0.685 1.753 13.367*** 2.887 Panel D: 2003 TE Model A TE Model B PTE Model A PTE Model B SE Model A SE Model B 0.021 0.866 0.614 0.489 0.140 0.034 -0.147 -0.753 -0.783 -0.699 0.374 -0.185 0.991 0.798 0.822 0.822 1.016 0.547 37.000 33.000 31.000 33.000 32.000 33.000 0.427 1.073 1.460 1.190 1.128 1.074 Panel E: 2004 TE Model A TE Model B PTE Model A PTE Model B SE Model A SE Model B 0.198 0.168 1.149 0.816 1.564 2.158 0.445 0.409 -1.072 -0.903 1.250 1.469 0.657 0.457 0.572 0.514 0.915 0.686 32.500 35.000 28.000 33.000 24.000 36.000 0.114 0.012 0.743 0.150 1.333 0.000 (continued) The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 165 4.3 Univariate Results After examining the DEA results, the issue of interest now is whether the two samples are drawn from the same population (i.e., whether the merchant banks and financial companies possess the same technology). The null hypothesis tested is that the merchant banks and financial companies are drawn from the same population or environment, having identical technologies. We have tested the null hypothesis by using a series of parametric (ANOVA and t-test) and non-parametric [Kolmogorov- Smirnov, Mann-Whitney, (Wilcoxon Rank-Sum), and Kruskall-Wallis] univariate tests. The results are presented in Table 9. Based on most of the results for DEA Model A, we failed to reject the null hypothesis at the 5 percent levels of significance that the merchant banks and the financial companies are drawn from the same population having identical technologies, while the results for DEA Model B failed to reject the null hypothesis during all years. This implies that there is no significant difference between the merchant banks and the financial companies’ technologies (frontiers); thus, it is appropriate to construct a combined frontier. Furthermore, the results from the Levene’s test for equality of variances do not reject the null hypothesis that the variances among the merchant banks and the financial companies are equal, implying that we can assume the variances between both groups to be equal. 5. Conclusion The preferred, non-parametric Data Envelopment Analysis (DEA) methodology allowed us to distinguish between three different types of efficiency: technical, pure technical, and scale efficiencies. During the period of study, the results suggested that the Malaysian merchant banks exhibited a mean technical efficiency of 69.6 percent, while the financial companies have exhibited a lower mean technical efficiency of 44.7 percent. Overall, the results suggest that scale inefficiency dominates pure technical inefficiency effects in determining Malaysian NBFIs’ total technical inefficiency. The findings also seem to suggest that scale efficiency tends to be much more sensitive to the exclusion of risk factors, implying that potential economies of scale may be overestimated when risk factors are excluded. The empirical findings clearly demonstrate the importance of risk in explaining financial institutions’ efficiency, particularly scale efficiency. If anything could be deduced from the results, the exclusion of risk factors may significantly overestimate the financial institutions potential economies of scale, which could result in bias conclusions and policy recommendations. The findings are important for policy makers in its quest to consolidate the banking system further to achieve greater economies of scale and efficiency. As the actual potential economies of scale may significantly be lower than initially expected, policy makers should be more cautious in promoting mergers as a mean to achieve greater efficiency by attaining better economies of scale. Author statement: Fadzlan Sufian is affiliated with the research department of a local bank, the CIMB Bank Berhad and is a staff member of the The University of Malaysia. E--mail: fadzlan14@gmail.com. 166 The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 References Altunbas, Y., Liu, M-H., Molyneux, P., and Seth, R. (2000). Efficiency and Risk in Japanese Banking, Journal of Banking and Finance 24 (10): 1605-1628. Aly, H.Y., Grabowski, R., Pasurka, C. and Rangan, N., (1990). Technical, Scale and Allocative Efficiencies in U.S. Banking: An Empirical Investigation, Review of Economics and Statistics 72 (2): 211-218. Avkiran, N.K., (2002). Productivity Analysis in the Service Sector with Data Envelopment Analysis. Camira: N.K. Avkiran. Barr, R. and Siems, T., (1994). Predicting Bank Failure Using DEA to Quantify Management Quality, Working Paper, Federal Reserve Bank of Dallas. Benston, G.J., (1965). Branch Banking and Economies of Scale, The Journal of Finance 20 (2): 312-331. Berger, A.N and Humphrey, D.B., (1992). Measurement and Efficiency Issues in Commercial Banking, in Z.Griliches, (eds.), Measurement Issues in the Service Sectors. National Bureau of Economic Research: University of Chicago Press, 245-279. Berger, A.N. and Humphrey, D.B., (1997). Efficiency of Financial Institutions: International Survey and Directions for Future Research, European Journal of Operational Research 98 (2): 175-212. Berger, A.N. and Mester, L.J., (1997). Inside the Black Box: What Explains Differences in the Efficiencies of Financial Institutions, Journal of Banking and Finance 21 (7): 895-947. Charnes, A., Cooper, W.W., Huang, Z.M., and Sun, D.B., (1990). Polyhedral Cone – Ratio DEA Models with an Illustrative Application to Large Commercial Banks, Journal of Econometrics 46 (1-2): 73-91. Coelli, T., (1996). A Guide to DEAP Version 2.1, CEPA Working Paper 8/96, University of New England, Armidale, Australia. Coelli, T., Rao, D.S.P. and Batesse, G.E., (1998). An Introduction to Efficiency and Productivity Analysis. Boston, MA: Kluwer Academic Publishers. Cooper, W.W., Seiford, L.M., and Tone, K., (2000). Data Envelopment Analysis. Boston: Kluwer Academic Publishers. Das, A., and Ghosh, S., (2006). Financial Deregulation and Efficiency: An Empirical Analysis of Indian Banks During the Post Reform Period, Review of Financial Economics 15 (3): 193-221. Dermiguc-Kunt, A., (1989). Deposit Institutions Failure: A Review of the Empirical Literature, Federal Reserve Bank of Cleveland Economic Review 25 (4): 2- 18. Drake, L., and Hall, M.J.B., (2003). Efficiency in Japanese Banking: An Empirical Analysis, Journal of Banking and Finance 27 (3): 891-917. Elyasiani, E., and Mehdian, S., (1992), Productive Efficiency Performance of Minority and Non-Minority Owned Banks: A Non-Parametric Approach, Journal of Banking and Finance 16 (5): 933-948. Isik, I., and Hassan, M.K., (2002). Technical, Scale and Allocative Efficiencies of Turkish Banking Industry, Journal of Banking and Finance 26 (4): 719-766. The International Journal of Banking and Finance, 2007/08 Vol. 5. Number 2: 2008: 149-167 167 Katib, M. N., and Mathews, K., (2000). A Non-Parametric Approach to Efficiency Measurement in the Malaysian Banking Sector, The Singapore Economic Review 44 (2): 89-114. Kolari, J., and Zardkoohi, A., (1987). Bank Costs, Structure and Performance. Lexington Books: USA. Levine, R., (2004). Finance and Growth: Theory, Evidence & Mechanism,in Aghion, P. and Durlauf, S., (eds), Handbook of Economic Growth, Amsterdam: North- Holland, pp. 81, in Reforming Corporate Governance in Southeast Asia, by Khai Leong Ho (2005), published by Institute of Southeast Asian Studies. McKinnon, P.I., (1973). Money and Capital in Economic Development, Washington D.C., The Banking Institution. Okuda, H., and Hashimoto, H., (2004). Estimating Cost Functions of Malaysian Commercial Banks: The Differential Effects of Size, Location and Ownership, Asian Economic Journal 18 (3): 233-259. Rajan R.G., and Zingales, L., (1998). Financial Dependence and Growth, American Economic Review 88 (2): 559-586. Sealey, C., and Lindley, J.T., (1977). Inputs, Outputs and a Theory of Production and Cost at Depository Financial Institutions, Journal of Finance 32 (4): 1251-1266. Shaw, E.S., (1973). Financial Deepening in Economic Development, New York, Oxford University Press. Singh, C., (2005). Financial Sector Reforms and State of Indian Economy, Indian Journal of Economics & Business 4 (1): 88-133. Sufian F., (2007). Trends in the Efficiency of Singapore’s Commercial Banking Groups: A Non- Stochastic Frontier DEA Window Analysis Approach, International Journal of Productivity and Performance Management 56 (2): 99 – 136. Thanassoulis, E., (2001). Introduction to the Theory and Application of Data Envelopment Analysis. Kluwer Academic Publishers: Boston. Weill, L., (2007) Is there a Gap in Bank Efficiency between CEE and Western European Countries? Comparative Economic Studies 49 (1): 101–127. Whalen, G., (1991). A Proportional Hazards Model of Bank Failure: An Examination of its Usefulness as an Early Warning Tool, Federal Reserve Bank of Cleveland Economic Review 27 (1): 21–31. Wheelock, D.C., and Wilson, P.W., (1995). Explaining Bank Failures: Deposit Insurance, Regulation and Efficiency, Review of Economics and Statistics 77 (4): 689-700. International Journal of Banking and Finance 3-1-2008 The efficiency of non-bank financial intermediaries: Empirical evidence from Malaysia Fadzlan Sufian Recommended Citation