American Finance & Banking Review; Vol. 2, No. 2; 2018 ISSN 2576-1226 E-ISSN 2576-1234 Published by Centre for Research on Islamic Banking & Finance and Business, USA 1 A Test of Miller and Modigliani Dividend Policy Irrelevance Theory in Nigerian Stock Market Udobi, Philomina I. 1 & Iyiegbuniwe, Wilfred I. 1 1 Department of Finance, Faculty of Management Sciences, University of Lagos, Akoka, Lagos, Nigeria Correspondence: Udobi, Philomina I. ,Department of Finance, Faculty of Management Sciences, University of Lagos, Akoka, Lagos, Nigeria,Email:udobiphil@yahoo.com Received: May 19, 2018 Accepted: May 25, 2018 Online Published: June 13, 2018 Abstract This study empirically tests for the validity of Miller and Modigliani’s dividend irrelevance proposition in the Nigerian Stock Exchange (NSE). Secondary data were obtained from the Nigerian Stock Exchange fact book and firms’ annual audited financial statements for fifteen years (2001-2015). Mediation Analyses, was used to measure the direct and indirect effects of dividend on stock price. Correction of the anomalous use of current dividend and current earnings by the use of naive expectation of dividend and earnings revealed that the direct effect of expected dividend on share price is significant but the indirect effect of expected dividend on share price through earnings is not significant. The implication of these results is that expected dividend has its unique (direct) effect on share price beyond the effect on share price which it shares with expected earnings (indirect effect). This conclusion suggests that dividend policy is relevant in valuation of shares in NSE. It was therefore recommended that company management should treat dividend as an active corporate finance decision-making variable and should employ dividend in information signalling to capital market investors. Keywords: Expected dividend, Expected earnings, Share price, Mediation, Relevance, Irrelevance. 1. Introduction Dividend policy is a major tool in decision making by corporate managers. It has received keen interest from scholars and researchers worldwide and this has led to the formulation of many theoretical models and testing of various variables. Huda and Farah (2011) affirmed that the development of theoretical models and variables has helped in determining the factors that assists managers in dividend policy decision making. Adesina, Uwuigbe, Uwuigbe, Asiriuwa and Oriabe (2017) suggested that corporate managers should use dividend as a vital tool in their firm’s decision making. This suggestion is contrary to the finding of their study that dividend does not have influence on firm’s value. The dividend relevance theory posited by Lintner (1956) and dividend irrelevance theory posited by Miller and Modigliani (1961) are the major contending theories on dividend policy and its impact on share price. The arguments and assertions of these theories did not rule out the influence of earnings on firms’ share prices. The dividend relevance theory agrees that despite the direct relationship between dividend and share price, there is still a possibility of an indirect effect of dividend on share price through earnings while the dividend irrelevance theory argues that the relationship that dividend tends to have on share price is as a result of its relationship with firms’ future earnings. Thus, the impact of dividend on share price is not direct but only indirect. In the light of the above arguments, various studies’ all over the world’s stock market has tried to test for the validity of dividend policy irrelevance with no consensus. Some of the challenges in the previous empirical tests of the effects of dividend on share price include the inaccurate measurement and definition of the earnings variable which is the unobservable but key variable in MM (1961) dividend irrelevance proposition. Amadasun, (2011); Toby, (2014) stated that the use of current dividend and current earnings in some previous tests of dividend irrelevance gave erroneous results that led to the conclusion that dividend is irrelevant in the valuation of share price. Udobi (2016) asserted that the use of current dividend and current earnings is flawed because share price is determined by www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 2 ex-ante (expected/future) information on determinant variables including dividend and earnings and not ex-post (current) information such as current dividend and current earnings. Another limitation in previous studies is the methodology used and its inability to decompose the total effect of dividend on share price into direct and indirect effects. In particular, is the inability of a methodology to isolate the direct effect from the indirect effect which it shares with earnings? This study revisits the Miller and Modigliani’s (1961) dividend irrelevance proposition. It objective is to overcome some of the limitations in previous studies and to determine if an M&M dividend irrelevance proposition is applicable in the Nigerian Stock Market. 2. Review of Existing Literature Dividend Relevance Miller and Modigliani (1961) arguments emphasized that the impact that dividend has on share price is due to its information contents about future earnings which is the real determinant of share price. Black and Scholes (1974) investigated the impact of dividend policy on firm’s value of companies listed in New York Stock Exchange. Capital Asset Pricing Model (CAPM) was used to analyze five years secondary data (Share prices and dividend) of twenty five quoted firms. They concluded that dividend policy of firms has no impact on their stock price. On the contrary, Aharony and Swary (1980) applied the Naïve Expectation model of quarterly dividend and earnings and found that share prices react to increase in dividend payment, coincident with earnings announcement. Their finding corroborates dividend relevance proposition. DE Angelo and DE Angelo (2006) also found that the information content of dividend is highly relevant. Adefila, Oladipo and Adeoti (2004) examined the effect of dividend policy on the market price of shares. The methodology used was Pearson’s product moment correlation of dividend with share price of fifteen companies. The study found that the correlation coefficients are statistically insignificant for most of the fifteen companies. It also found that the correlation between net profits (earnings) and share prices are statistically insignificant for all the companies analysed. The study concluded that there is no significant relationship between dividends and share prices, there is no significant relationship between net profits and share prices. This study was poorly designed and the use of correlation analyses to test dividend relevance is inadequate. Hence, the conclusion reached by this study is suspicious. Abor (2008) found out that there is a relationship between corporate earnings and dividend payout and concluded that both past and current earnings have impact on a firms’ dividend policy. Musa (2009) investigated the impact of dividend policy on the share prices of 53 quoted firms in Nigeria, applying parsimonious multiple regression model which employed five variables: current earnings, previous dividend, cash flow, investment and net current asset and three non-metric variables, growth, firm size and industry classification. The study found that the five metric variables have impact on dividend policy of firms in Nigeria. Adesola and Okwong (2009) tested the relevance of dividend theories of share prices in Nigeria with cross sectional data on twenty-seven companies for the period 1996 to 2006. They commented that their finding of positive and significant effect of dividend on share prices for the sample of Nigerian companies indirectly cast some doubt on the empirical validity of dividend irrelevance. Khalid, Chijioke and Aruoriwo (2010) investigated the impacts of dividend yield and dividend payout ratio on changes in share price of companies listed on the United Kingdom Stock Exchange. A regression model was used to analyze the data which revealed positive relationship between dividend yield and stock prices and revealed that dividend payout is statistically insignificant. Amadasun (2011) attempted to test the hypothesis that dividend does not increase stock price in Nigeria, using First Bank (Nig) plc. as a case study. The study used a regression model that included Dividend Per Share, Earnings Per Share, Return on Capital Employed, Retained Earnings and Price Earnings Ratio as explanatory variables of price per share. The results of the study had statistically insignificant regression coefficients for both dividend per share and earnings per share. Though this study concluded that “dividend does not lead to increase in stock value”, and thus purportedly lends support to Miller-Modigliani thesis of dividend irrelevance, the conclusion is consistent with the statistical insignificance of the dividend variables in the regression results. Furthermore, the study has the following deficiencies: - (1) it used a case study of a bank rather than cross sectional or panel data; and the regression model included four earnings variables; earning per share, return on capital employed, retained earnings and price earnings ratio. It therefore would have the problem of multi-collinearity. These observations indicate that the results and conclusion of the study are not reliable. Khan (2012) sampled of twenty-nine companies to explicate the effect of dividend on stock prices for the period 2001 to 2010. The study used Fixed and Random Effect Model on panel data and found that dividend policy has positive effect on share prices after controlling for the effects of earnings per share, profit after tax, and return on equity and concluded that dividend irrelevance theory is not applicable in case of Pakistan Stock Market. www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 3 Abubakhar (2012) examined the influence of dividend pay-out ratio on the share prices of listed non-service firms in Nigeria with a probabilistic sample of twenty-six firms using multiple regression models. The study found a statistically significant relationship between dividend pay-out and share prices. It also found that the size of the listed non-service firms significantly explains share prices in Nigeria. Both dividend pay-out and size of the firms had positive impact on share prices. The findings of the study do not however constitute valid test of the dividend irrelevance proposition became it did not consider the effect of earnings and dividend in the model. Rabindra (2012) examined the impact of dividend on stock price in Financial and non-Financial institutions of Nepal Stock Exchange, analyzing the secondary data with a regression model. Stock price being the dependent variable while Dividend Per Share (DPS), Retained Earnings Per Share, Lagged Price Earnings Ratio and Lagged Market Price Per Share are the explanatory variables. The result showed that dividend has impact on share prices more than retained earnings. Kanwal (2012) studied the impact of dividend on stock prices of chemical and Pharmaceutical companies in Pakistan Stock Exchange for the period 2001-2010. Secondary data of five variables: Stock dividend, Earning Per Share (EPS), Profit After Tax (PAT), Retention Ratio and Return on Equity (ROE) were analyzed with panel regression model. The study showed that Stock dividend, (EPS), (PAT) are statistically significant. In other words, these variables have positive impact on stock prices while Retention Ratio and Return on Equity have negative impact on share prices. The study asserted that changes in dividend policy provide statistically significant information content which can be used to make predictions about future stock prices”, and that the findings support the informational content of dividend hypothesis. These findings’ assertions indicate that changes in dividend payment merely create occasions for changes in stock prices and that there was no sufficient evidence to suggest that stock price changes are caused by dividend payments. The study did not include earnings in its analyses. Hence, its findings cannot be quite conclusive as to the impact of dividend on share price given the effect of earnings. Ozuomba, Okaro and Okoye (2013) carried out research to test the effect of dividend policy on shareholder’s wealth of public firms in Nigeria for a period of twelve years (2000-2011). Secondary data of ten randomly selected firms out of two hundred and sixteen public limited firms were analysed with multiple regression model using dividend per share as the dependent variable while earning per share (EPS) and market price per share (MPS) are the independent variables. The results showed that the EPS and MPS of eight firms are both statistically significant and have impact on shareholder’s wealth of the quoted firms while that of two firms are not statistically significant at 10% confidence interval. The model of this study is wrong as it used dividend as the dependent variable instead of share price to represent shareholders wealth as stated in its objective. The data on EPS, DPS and MPS are not synchronized. Data used is also scanty because instead of using panel data, it did the analyses as time series of individual companies, and therefore its findings are very misleading. Ordu, Enekwe and Anyanwaokoro (2014) conducted a study to find out the effect of dividend payment on the market share prices in Nigeria. Seventeen quoted firms were considered for a period of twelve years (2000-2011). Using the Ordinary Least Squares technique, positive effect was found between Market Share Price (MPS) and Dividend per Share (DPS). This result supports the dividend relevance theory, confirming that dividend increase results to an increase in market share price. Dada, Malomo and Ojediran (2015) support this argument and concluded that investors prefer dividend payment to future growth. Iqbal, Ahmed and Shafi (2014) looked at the effect of dividend bubble on share prices of thirty quoted firms in Karachi Stock Exchange for a time frame of eleven years. Time series data of the thirty listed firms were analysed with linear regression model. The result showed that Earning Per Share, Return on Equity, retention ratio are positively correlated with share price while Dividend yield and Price earnings ratio have a negative impact on share price. The study however concluded that dividend has a strong positive impact on share prices of KSE and thus, supports the dividend relevance theory. This study is faulted because of the use of time series of thirty firms for a period of eleven years. A panel data would have been more appropriate in order to get valid findings and conclusion. Oyinlola and Ajeigbe (2014) examined the impact of dividend policy on stock prices of quoted firms in Nigeria, using 22 companies listed on the Nigerian Stock Exchange over the period 2009 – 2013. It used panel regression model to determine the impact of dividend per share and retained earnings per share on share price. The results indicate that both dividend and retained earnings significantly impact on share price. In addition, Granger causality tests indicate that dividend per share granger cause share price. Toby (2014) studied the relevance of dividend policy in share price determination in the Nigerian Stock Market with a sample of twenty stocks within the period 2005-2012 with regression analyses of dividend and retained earnings time series data on individual companies. The study found that there is no significant relationship between change in dividend policy and change in share price. This surprising result differs from the extant literature on the impact of dividend on share price. The result carried out the analysis on company basis (separate regression analyses for each www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 4 stock in the sample) rather than use cross-sectional or panel data which will capture inter-company variations. Furthermore, the study did not include appropriately defined earnings variable in the analyses. It instead used retained earnings. In the results neither dividend nor retained earnings was a statistically significant determinant of share price. The conclusion of the study that “the results agree with the earlier research works which argue that dividend policy is irrelevant in determining enterprise value” is therefore very suspect and invalid. Edward’s (2014) investigated the effect of dividend on share price of some selected quoted firms in Ghana Stock Exchange for a period 2005-2009 using descriptive analyses on the primary data, found that dividend is highly correlated with share prices in Ghana. Ojeme, Mamidu and Ojo (2015) studied the impact of dividend policy on shareholders wealth in Nigerian quoted banks before and after the global financial meltdown within the period of four years (2007 – 2010). Secondary data of all 21 quoted banks during this period were gathered from NSE and firm’s published annual reports. The study concluded that positive correlation between average market value of share and dividend paid by banks is an indication that payment of dividend is relevant and the amount paid affect the market value of banks’ shares. This is not a valid test of dividend irrelevance theory. It used only correlation analysis and four years data. It did not consider the influence of earnings on share price as posited by the dividend irrelevance theory. Oduwole (2015) evaluated and compared the predictive power of earnings and dividends in Nigeria for a period of fourteen years (2001-2014). The quarterly data of the variable used (EPS, interim and final dividends) were collected from the Nigerian stock market. The study used portfolio evaluation measures (Sharpe ratio and Jensen alpha) to assess the investment performance of portfolio based on dividend yield and earnings yield, respectively. The results indicate that a portfolio formed using a market capitalization weighted approach for the highest quartiles of dividend yield outperformed buy the market and hold policy while similar portfolio based on earnings yield did not outperform the market. This is not a test of dividend irrelevance theory. Chirima (2015) investigated the impact of dividend pay-out on stock prices of quoted service firms of Zimbabwean Stock Exchange for a period of five years (2008-2012). The data were analysed with chi-square and regression model. The result showed a statistically significant relationship between dividend announcements and share prices. Egbeonu, Edori and Edori (2016) examined the weighted average of five year financial summary data of twelve listed firms from the various sectors of the Nigerian Stock Exchange in order to measure the impact of dividend policy on firms’ value. The study reported that internal rate of return is inversely insignificant, Dividend per share is inversely significant while earning per share is positively significant to share price. Adesina, Uwuigbe, Uwuigbe, Asiriuwa and Oriabe (2017) examined the impact of dividend policy on share price valuation in Nigeria. Data of four out of twenty two banks were analysed during ten years’ timeframe (2006-2016). They observed in their study that earning per share has a strong impact on share price while there is significant impact of dividend yield and retention ratio on share price. It was however concluded that there is need for Nigerian firms to consider dividend policy in other to increase the firm’s earnings and future performance. Iftikhar, Raja and Sehran (2017) established that dividend has a positive statistical influence on stock price after they investigated the impact of dividend policy on five state banks of the Karachi stock Exchange (KSE) for a period of ten years. They concluded that dividend is relevant in KSE. Budagaga’s (2017) study supported the dividend relevance theory after observing the effect of dividend payment on forty-four firms’ value of Istanbul Stock Exchange for duration of nine years. Udobi, Iyiegbuniwe & Ezike (2018) examined the impact of current dividend on market shares prices of the Nigerian Stock Exchange. The study analysed fifteen years (15) secondary data of NSE quoted firms with mediation analysis. Stock prices is the dependent variable while current dividend, current earnings, Asset-growth, sales- growth, insider-shareholding and Leverage are the independent variables. The findings indicate that current dividend has a direct (Unique) effect on share price, and at the same time has indirect effect on share price through current earnings. It concluded that current earnings partially mediate the effect of current dividend on quoted Nigerian firms. 3. Research Methods The purpose of this study is to test the validity of Miller and Modigliani dividend irrelevance theory in Nigeria by defining dividend and earnings as expected dividend and expected earnings. This study is based on all the shares listed on the Nigerian Stock Exchange (NSE). It covers a period of Fifteen years, (2001-2015). This time period is chosen because of the availability of data, to accommodate the pre and post consolidation of Nigerian’s financial institutions. It was further reduced because of firms with incomplete data points were deleted. Stock price data were obtained from the daily official price list of the Nigerian Stock Exchange. Data on earnings per share, dividend per share, insider-shareholding, assets, and sales turnover were collected from the Nigerian Stock Exchange Fact Book and published annual reports of the listed companies that constitute the sample of this study. Assets-growth (Proxy www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 5 for investment opportunity), sales-growth (proxy for investment growth), and leverage (proxy for risk) variables were computed from the earnings, sales turnover, and total assets data. The study uses a panel study to measure the impact of expected dividend and expected earnings on share price in Nigeria. 3.1. Model Specification Mediation analysis, a new methodology in finance adapted from the field of social psychology is used to determine the irrelevance of expected dividend on share price. it is imperative to estimate the direct effect of expected dividend on share price which is beyond the effect of expected dividend on share price that is due to the relationship between dividend and a mediator (such as earning). Mediation refers to a situation when the relationship between a predictor variable (expected dividend) and an outcome variable (share price), can be explained by their relationship to a third variable called the mediator (for example, expected earnings). Mediation is said to have occurred, if the strength of the relationship between the predictor variable (expected dividend) and the outcome variable (share price) is reduced by including the mediator variable (expected earnings). In the case of perfect or complete mediation, the effect of the predictor is completely wiped out by including the mediator. Baron and Kenny (1986), Judd and Kenny (1981), James and Brett (1984) discussed three models in establishing Mediation, These three models are needed in investigating the mediation of the effect of a predictor variable(dividend) on an outcome variable(share price) by a mediating variable(earnings):  Model 1 is to determine that the predictor (expected dividend) is correlated with the mediating variable (expected earnings). This step requires the regression of earnings on dividend to confirm that dividend is a significant predictor of earnings. The regression model is: M = β1 + β2 X1 + ei (1) Where: β1 = intercept β2 = Coefficient of the relationship between dividend and the mediating variable (earnings) X = Dividend M = Earnings ei = Error term  Model 2 is to determine that the predictor variable (expected dividend) is correlated with the outcome variable (share price). In other words, regress outcome variable (share price) on the predictor variable to confirm that dividend is a significant predictor of share price. The regression model is; Y = λ1 + λ2 Xi + ei (2) Where: Y = outcome variable λ1 = intercept; λ2 = total effect of dividend on share price Xi = predictor variable; ei = error term The value of the coefficient “λ2” is the total effect of expected dividend on share price. This step establishes that there is a dividend effect on share price that may be mediated.  Model 3 is to determine that the mediating variable (Expected earnings) affects Share Price. The regression model is Y = ɤ1 + ɤ2 X + ɤ3M + ei (3) Where: ɤ1 = intercept; ɤ2 = The direct effect of expected dividend on share price ɤ3 = The direct effect of expected earnings on share price Y = Share price; X = dividend; M = Earnings; www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 6 ei = error term, Equation (3) produces the direct effect of expected dividend on share price, ɤ2, and the direct effect of expected earnings on share price, “ɤ3”. The model controls for the influence of earnings on the effect of dividend on share price by establishing the effect of earnings on share price As suggested by Baron and Kenny (1986), Judd and Kenny (1981), Mediation analysis can be done through a series of regression analyses which reflect the above conditions necessary to demonstrate mediation. Fields (2012),however affirmed that by far, the best way to tackle mediation analyses is to use the PROCESS custom dialog box written by Hayes (2012) to wrap the Preacher and Hayes(2004, 2008a) mediation and moderation tools in a convenient menu and dialog box interface in IBM Statistical Package for Social Sciences statistical software (SPSS). This software was employed to do mediation analysis in this study. 3.2. Estimation Technique Field (2012) states that the three models discussed earlier (equations 1, 2 and 3) above test the four conditions of mediation:  Expected dividend must significantly predict Expected earnings in model 1 (equation 1).  Expected dividend must significantly predict share price in model 2 (equation 2).  Expected earnings must significantly predict share price in model 3 (equation 3).  Expected dividend must predict share price less strongly in model 3 than in model 2. To establish that earnings completely mediates the dividend-share price relationship, in model 3 (equation 3), the effect of dividend on share price after controlling for earnings (coefficient ɤ2) should not be significantly different from zero. Such result will uphold dividend irrelevance proposition. But if in model 3 (equation 3) the coefficient c' is significantly different from zero and the coefficient ɤ3 is not significantly different from zero, dividend irrelevant proposition is refuted. But if in model 3 (equation 3), the coefficient, ɤ2, is significantly different from zero and coefficient “β1” in model 1(equation 1) and “ɤ3” in model 3(equation 3), are significantly different from zero, then earnings has partial mediation effect on dividend, suggesting that dividend has both indirect (mediation) effect (through earnings) and direct effect on share price. Field (2012) further observed that although Baron and Kenny (1986) advocated looking at the size of the regression parameters, in practice, people tend to look for a change in significance. So mediation would occur, if the relationship between the predictor (dividend) and outcome (share price) was significant (P< 0.05) when looked at in isolation (model 1) but not significant (P > 0.05), when the mediator (for example, earnings) is included (Model 3). Where the coefficient of the predictor (dividend) is significant in both models 1 and 2, mediation can exist, if there is a reduction in the size of the coefficient of relationship between the predictor(dividend) and outcome (share price) in model 3, as compared to model 1. In other words, the predictor variable (dividend) predicts the outcome variable (share price) less strongly in model 3 than in model 1. Field (2012) remarked that the problem with Baron and Kenny (1986) test of mediation is the question of how much of a reduction is necessary to intermediation. 4. Data Analysis and Interpretation Descriptive Statistics and Correlation Table 1. Descriptive Statistics N Minimum Maximum Mean Std.Deviation Earnings 609 -20.000 28.000 2.065 3.724 Dividend 602 0.00 24.000 1.293 2.658 Share Price 608 0.50 1056.65 38.14 78.827 Insider -holdings (%) 612 5 88 54.44 20.317 Asset- Turnover (Nm) 592 57.04 3186128 34443 148532 Sales -Turnover (Nm) 596 93.17 673181 50580 84363 Change in Earnings 559 -20.00 25.00 0.133 2.287 Change in Dividend 548 -7.00 8.30 0.073 1.255 Change in Share Price 558 -214.02 483.43 4.115 32.254 Sales –Growth (%) 541 -97.44 3254.45 28.877 155.903 Leverage 423 -1460.01 803.67 -1.954 85.695 Asset Growth (%) 535 -3785.49 6091.16 40.546 332.678 Source: Author’s Computation, 2016. www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 7 4.1. Correlation of Expected Dividend, Expected Earnings, and Share Price A necessary condition for use of mediation analysis is that expected dividend is correlated with the mediator variables respectively (expected earnings, insider-shareholding, assets-growth, sales-growth, and leverage); and that the mediator variables are correlated with share price, respectively. The correlation matrix of expected dividend and the mediation variables (expected earnings, insider- shareholding, assets-growth, sales-growth, and leverage) and share price is presented in table 2. Expected dividend is correlated with share price with a coefficient of 0.82 which is statistically significant at p value of 0.00. Expected dividend and expected earnings have a correlation coefficient of 0.89 which is statistically significant at p-value of 0.000. The correlation of expected earnings with share price is 0.78, which is statistically significant at p-value of 0.000.These results indicate that expected earnings has the potential of mediating the impact of dividend on share price. These results confirm that mediation analysis is an appropriate technique to evaluate the influence of earnings on the impact of dividend on share price. The correlation coefficient of dividend and the other mediating variables (insider-shareholding, assets- growth, sales- growth, and leverage) are relatively low and suggestive that these variables may not be significant dividend signalling variables. Table 2. Correlation Matrix of Variables Expected earnings Expe cted divid end Share Price Insider share Holding Asset Sales Earnings Change Div. Change Share Price Change Sales Growth % Leve rage Asset Growth % Expected earnings 1 Expected dividend p-value 0.89 .000 1 Share Price p-value 0.78 .000 0.82 .000 1 Insider share – Holding p-value 0.08 0.39 0.10 0.16 0.09 0.23 1 Asset p-value 0.04 .338 0.01 .778 0.01 .844 -0.26 .000 1 Sales p-value 0.31 .000 0.31 .000 0.27 .000 -0.16 .000 0.45 .000 1 Expected earnings Change p-value 0.37 .000 0.17 .000 0.07 .000 0.02 .048 0.02 .575 -0.02 .000 1 Expected dividend Change p-value 0.15 .000 0.31 .000 0.01 .000 0.03 .028 0.00 .830 0.00 3 .936 0.28 .000 1 Share Price Change p-value 0.37 .000 0.41 .000 0.58 .000 0.04 .023 -0.02 .789 0.02 .600 0.07 .506 0.08 .496 1 Sales Growth (%) p-value -.02 .818 -.02 .956 -0.02 .496 0.03 .503 0.00 .998 0.00 .932 0.02 .504 0.03 .396 0.00 .786 1 Leverage 0.02 0.05 0.03 0.05 0.00 0.03 0.00 0.02 0.02 0.01 1 www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 8 p-value .666 .334 .520 .319 .973 .597 .949 .751 .660 .458 Asset Growth (%) p-value 0.02 .539 0.02 .883 0.00 .932 -0.06 .000 0.15 .000 0.09 .000 0.08 .862 0.04 .520 -0.03 .319 0.12 .597 0.01 .395 1 Source: Author’s computation, 2016 4.2. Relationship between Expected Dividend and Expected Earnings Table 3 presents the regression of expected dividend on expected earnings. The regression coefficient is 0.4649 and it is statistically significant at p-value of 0.0000. This confirms that expected dividend is a significant predictor of expected earnings and that expected earnings is a possible mediator in the relationship between dividend and share price. The positive sign of the regression coefficient confirms that increase in dividend pay-out is a signal of increase in future earnings as posited by the dividend information signalling theories (Bhattacharya, 1980). The estimated regression model is: Mi = Ɵ1+ Ɵ2Xi + ei (4) = 0.0686 + 0.4649Xi + ei Where: Ɵ1 = intercept Ɵ2 = the influence of expected dividend on expected earnings Xi = Expected dividend Mi = Expected earnings ei = Error term The null hypothesis is: Ho: Ɵ2 = 0. Table 3. Relationship between Expected Dividend and Expected Earnings Model Summary R R 2 MSE F df1 df2 p 0.2957 0.0875 4.2941 38.33 1 400 0.0000 Model coefficient S.E t p Constant 0.0686 0.1036 0.6627 0.5079 Expected Dividend 0.4649 0.0751 6.1913 0.0000 Source: Author’s computation, 2016. 4.3. Total Effect of Expected Dividend on Change in Share Price The total effect (without control of the effect of the moderators) of dividend on share price is shown in table 4. The regression coefficient of expected dividend is positive and statistically significant at p-value of 0.0018. This confirms that expected dividend is a significant predictor of changes in share price. The coefficient of determination (R 2 ), 3.13 per cent, is small and suggests that there are other variables that determine changes in share price other than expected dividend. The positive sign of the regression coefficient of dividend indicates that share price will increase with increase in expected dividend. The total effect of dividend on share price measures its influence without the mediating variables in the regression model. The total effect of expected dividend on change in share price is obtained by the regression of expected dividend on change in share price as in the model below. Yi = + Xi + ei = 5.2943 + 3.8465 Xi + ei (5) Where; α = intercept = Coefficient of expected dividend Y = Change in share price www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 9 Table 4. Total effect of expected dividend on change in share price Model Summary R R 2 MSE F df1 df2 p 0.1770 0.0313 1153.37 2.129 6 395 0.0492 Model: Outcome variable is Change in Share Price Coefficient S.E t p Constant 5.2943 1.6930 3.1273 0.0019 Expected Dividend 3.8645 1.2273 3.1489 0.0018 Source: Authors’ computation, 2016. As seen earlier, it is necessary to decompose the total effect of dividend into the direct and the indirect (mediated) effects of dividend on share price via the following mediating variables: expected earnings, insider- shareholding, assets-growth, sales-growth, and leverage. 4.4. Direct Effect of Expected Dividend and the mediators on Change in Share Price The direct effect of expected dividend on change in share price is the influence of dividend on share price in the presence of the mediating variables in the regression model: The regression model is as follows: Yi = α1 + α2 Xi + α3 M1i + α4 M2i + α5 M3i + α6 M4i + α7 M5i + ei (6) Yi = 0.59 + 3.5545Xi + 0.68M1i+ 0.09M2i – 0.005M3i – 0.005M4i + 0.006M5i + ei Where: α1 = Intercept α2 = direct effect of expected dividend on share price α3 = direct effect of expected earnings α4 = direct effect of insider-shareholding α5 = direct effect of assets-growth α6 = direct effect of sales-growth α7 = direct effect of leverage Y = Change in share price X = expected dividend M1 = expected earnings M2 = insider-shareholding M3 = assets-growth M4 = sales-growth; M5 = leverage; and ei = the error term. The null hypothesis is that: H0 : α1 = α2 = α3 = α4 = α5 = α6 = 0 Table 5. Direct Effect of Expected Dividend on Change in Share Price Model Summary: outcome variable is change in share price R R 2 MSE F df1 df2 p 0.1770 0.0313 1153.37 2.129 6 395 0.0492 Model Coefficient S.E t p Constant 0.5859 5.1726 0.1133 0.9099 Expected Earnings 0.6831 0.8255 0.8275 0.4084 Insider-Shareholding 0.0921 0.0881 1.0457 0.2963 Assets-Growth -0.0047 0.0053 -0.8838 0.3773 Sales-Growth -0.0054 0.0216 -0.2494 0.8032 Leverage 0.0064 0.0193 0.3286 0.7426 Expected Dividend 3.5545 1.2904 2.7545 0.0062 Source: Author’s Computation, 2016. www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 10 The results in table 5 show that only the regression coefficient of expected dividend is statistically significant with a p-value of 0.0062. The regression coefficients of all the mediators, particularly expected earnings, are not statistically significant. These results indicate that dividend has unique effect on share price because the direct effect of expected dividend was estimated with control on the influence of the moderators (particularly expected earnings) on share price. 4.5. Indirect Effect of Expected Dividends on Change in Share Price The verification of relevance or irrelevance of the influence of dividend on share price dictates that the respective indirect (mediated) effect of dividend on share price via the mediators (expected earnings, insider-shareholding, assets-growth, sales-growth, and leverage) are estimated and tested for statistical significance. The indirect effects of dividend on share price through earnings, insider-shareholding, assets-growth, sales-growth and leverage are shown in Table 6. Table 6. Indirect Effect of Expected Dividend on Change in Share Price Mediating Variable Effect Boot S.E Boot LLCI Boot ULCI Total Indirect Effect 0.3100 0.5015 -0.4850 1.6731 Expected Earnings 0.3175 0.4823 -0.4697 1.5961 Insider Shareholding 0.0454 0.0388 -0.0061 0.1599 Asset Growth -0.0382 0.0547 -0.1507 0.0312 Sales Growth -0.0212 0.0878 -0.2666 0.0925 Leverage 0.0064 0.0375 -0.0749 0.0566 Source: Authors’ Computation, 2016. As shown in Table 6, none of the indirect effect of dividend on share price via the mediating variables is statistically significant, since their respective bootstrapped confidence interval contain zero value. Hence, the null hypothesis that the indirect effect is zero cannot be rejected for all the mediating variables. Of particular note is that the indirect effect of dividend on share price via expected earnings is not statistically significant. This indicates that while the direct effect of dividend on share price is significant, the indirect effect of dividend on share price via expected earnings is not significant. Further insight into the statistical significance of the indirect effects is done with the normal theory (Sobel) test of indirect effect, shown in Table 7. This ultimate test of indirect effect shows that none of the indirect effects of the mediating variables is statistically significant. Table 7. Sobel Test of Indirect Effects of Expected Dividend on Change in Share Price Variable Effect S.E Z p Expected Earnings 0.3175 0.3921 0.8099 0.4180 Insider Shareholding 0.0454 0.0999 0.4543 0.6496 Asset Growth -0.0382 0.0939 -0.4067 0.6843 Sales Growth -0.0212 0.1060 -0.1997 0.8417 Leverage 0.0064 0.0677 0.0948 0.9244 Source: Authors’ Computation, 2016. Though none of the indirect effect of dividends on share price is statistically significant the relative importance of the mediating variable can be established by computing the completely standardized effect size of the mediating variables. The completely standardized effect coefficients are not dependent on the units of the mediating variables. They are measured in standard deviation units and so are directly comparable (Field, 2012). Standardized coefficients represent the number of standard deviations that the outcome variable will change as a result of one standard deviation change in a predictor variable. The completely standardized indirect effect of dividend on share price is exhibited in table 8. Table 8. Effect Size (Completely Standardized) Indirect Effect Mediating Variable Effect Boot SE Boot LLCI Boot ULCI Total 0.0125 0.0240 -0.0187 0.0815 Earnings 0.0128 0.0236 -0.0179 0.0787 www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 11 Insider-shareholdings 0.0018 0.0016 -0.0004 0.0070 Investment-growth -0.0015 0.0023 -0.0064 0.0013 Sales-growth -0.0009 0.0037 -0.0099 0.0044 Leverage 0.0003 0.0016 -0.0032 0.0027 Source: Author’s Computation, 2016. As shown in Table 8, the most important mediating variable is expected earnings with a standardized effect size of 0.0128. Insider-shareholding variable comes a distant second with standardized effect size of 0.0018. Next are asset- growth, sales- growth and leverage in a descending order of importance. The above findings are inconsistent with the dividend irrelevance proposition that the influence of dividend on share price is due to its relationship with expected earnings. Rather, the significance of the direct effect of dividend on share price suggests that dividend has unique influence on share price. The results of this study are therefore consistent with dividend relevance proposition. 5. Conclusion The argument of Miller and Modigliani (1961) dividend irrelevance proposition implies that the effect of dividend on share price is fully mediated by earnings. Full mediation of the effect of dividend on share price by earnings means that the direct effect of dividend on share price will not be significantly different from zero and indirect effect of dividend on share price via earnings will be significantly positive. On the other hand, if the mediation effect is partial, then both the indirect and direct effect of dividend on share price will be significantly positive. In this case, there will be a reduction from the total effects of dividend on share price, which is accounted for by the indirect effect of expected dividend on share price. If the effect of expected dividend on share price is not mediated by expected earnings, the indirect effect of expected dividend on share price via expected earnings will not be significantly different from zero, while the direct effect will be significantly positive. In this study,mediation analyses was applied to isolate and measure the direct, indirect and total impact of expected dividend on firm’s share prices with the mediation of expected earnings,insider shareholdings. asstes-growth, sales- growth and leverage. The results shows that the direct effect of dividend on share price is positive and statistically significant while the indirect effect of expected dividend on share price, through expected earnings is insignificant. Furthermore, mediation of the effect of expected dividend on share price by insider-shareholding, assets-growth, sales-growth, and leverage (risk) were found to be statistically insignificant, respectively. While the total effect of dividend on share price is positive and statistically significant. These results contradict dividend irrelevance proposition but are consistent with dividend relevance hypothesis. The conclusion following from these results is that earnings do not mediate the effect of dividend on share price. Dividend has significant direct effect on share price that is not due to the relationship between dividend and earnings and thus, the Miller and Modigliani’s dividend irrelevance proposition do not hold in the Nigerian Stock Market. Dividend is relevant for the valuation of shares. The study has contributed to the resolution of the puzzle about the relevance or irrelevance of dividend for pricing of shares in the Nigerian Stock Market. References Abdul, A., & Muhibudeen, L. (2015). Relationship between dividend payout and firms performance: Evaluation of dividend policy of OANDO plc. International Journal of Contemporary Applied Sciences, 2 (6), 56-71. Abdullahi, A. (2014).Dividend policy and its impact on stock price: A study on commercial banks listed in Dhaka Stock Exchange. Global disclosure of Economics and Business, 3 (1), 9-17. Abor, J. (2008). Determinants of capital structure of Ghanaian firms. African Economic Research Consortium Biannual Workshop (AERC) Research paper 176, Nairobi, Kenya. Abubakhar, S. (2012). Dividend payout ratio and share price movement of listed Non-service firms in Nigeria: Empirical study. www.abu.edu.ng/publications/2012-04-11-163138_1156.docx Adediran, S.A., & Alade, S.D. (2013). Dividend policy and corporate performance in Nigeria. American Journal of Social and Management Sciences, 4(2), 71-77. Adefila, J.J., Oladipo, J.A., & Adeoti, J.O. (2004). The Effect of dividend policy on the market price of shares in Nigeria: Case study of fifteen quoted companies. International Journal of Accounting. University of Ado- Ekiti, 2 (1), 1-10 Adesina, K., Uwuigbe, U., Uwuigbe, O.R., Asiriuwa, O. & Oriabe, S. (2017). Dividend policy and share price valuation in Nigerian banks. EuroEconomica, 36(1), 1-10 Adesola, W.A., & Okwong, A.E. (2009). An empirical study of dividend policy of quoted companies in Nigeria. Global Journal of Social Sciences, 8 (1), 85-101. www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 12 Adeyemi, S.L., & Adewale A. A. (2010). Evaluation of the dividend practices among selected Nigerian quoted firms. Maiduguri Journal of Accounting and Finance, 8 (1). Aharony, J., & Swary, I. (1980). Quarterly dividend and earnings announcements and stockholder’s returns: An empirical analysis. Journal of Finance, 35(1), 1-12. Amadasu, D.E. (2011).Dividend is relevant: A restatement. African Research Review Journal, 5(4), 60-72 Anil, K., & Kapoor, S. (2008). Determinants of dividend payout ratios- A study of Indian information technology sector, The International Research Journal of Finance and Economics, 15(2), 63-71 Arslan, I., Farooq, A., & Ali Raja, S. (2014).The effect of dividend bubble on share price: Evidence from KSE-30 index. Research Journal of Finance and Accounting, 5(13), 83-87. Baker, H.K., & Weigand, R. (2014). Corporate dividend policy revisited. Journal of Managerial Finance, 4(2), 126-144. doi.10.1108/MF-03-2014-0077. Baron, R.M. & Kenny, D.A. (1986).The moderator- mediator variable distinction in social psychological research: Concept, strategic and statistical considerations. Journal of personality and social psychology, 51(6), 1173-1182. Bernstein, P.L. (1996). “Dividend: The puzzle”, Journal of Applied Corporate Finance, 9, 16-22 Black, F. (1976). The dividend puzzle. Journal of portfolio management. 2(2), 5-8. Bougatef, K. (2014). How do dividend payments affect stock prices: The case of Tunisian firms. The Journal of Commerce. 3(2), 21-25 Bringham, E.F and Houston, J.F (1999). Fundamentals of financial management. Second Edition. The Dryden Press, Florida, U.S.A Budagaga, A. (2017).Dividend payment and its impact on the value of firms listed on Istanbul Stock Exchange: A residual income approach. International Journal of Economics and Financial Issues, 7(2), 370-376. http.www.econjournals.com Chirima, D.T. (2015). The Impact of dividend payments on stock prices in a developing economy: The case study of services firms listed on the Zimbabwe Stock Exchange. Case Studies Journal, 4(5), 126-133 De-Angelo, H., & De-Angelo, L. (2006). The irrelevance of the MM dividend irrelevance theorem. Journal of Financial Economics, 79(2), 293-315. Edward, A.B. (2014). The Impact of Dividend payment on share prices of some selected listed companies on the Ghana Stock Exchange. International Journal of Humanities and Social Sciences, 4(1), 179-190 Egbeonu, O.C., Edori, I.S., & Edori, D.S. (2016). Effect of dividend policy on the value of firms: Empirical study of quoted firms in Nigeria Stock Exchange. Research Journal of Finance and Accounting, 7(3), 17-24 Fields, A. (2012). Discovering statistics using IBM SPSS Statistics. London, Sage publications Limited. ………….. (2013). Discovering statistics using IBM SPSS Statistics (4th Edition) London, U.K: Sage Publication Limited. Franc-Dabrowska, J. (2009). Does dividend policy follow the capital structure theory? Managing Global Transitions, 7(4), 367-382. Hakeem, S.A., & Bambale, A.J. (2016).Mediating effect of liquidity on firm performance and dividend payout of listed manufacturing companies in Nigeria. Journal of Economics Development, Management, IT, Finance and Marketing, 8(1), 15-35. Huda, F., & Farah, T. (2011). Determinants of dividend decision: A focus on banking sector in Bangladesh. International Research Journal of Finance and Economics, 77, 33-46 Husan-Aldin, N.A., Rafforty M, and Pillqar, R. (2010). Dividend policy: A review of theories and empirical evidence, International Bulletin of Business Administration, 9, 171-200. Iftikhar, A.B., Raja, N.J., & Sehran, K.N. (2017). Impact of dividend policy on stock prices of firms. International Scientific Journal Theoretical and Applied Science, 3(47), 32-37. Iqbal, A., Ahmed, F., & Shafi, A.R. (2014). The effect of dividend bubble on share price: Evidence from KSE-30 index. Research Journal of Finance and Accounting, 5(13), 83-87 Inyiama, O.I., & Ugah, H. (2015). Interactions between earnings and dividend payout rate in the oil and gas sector: Evidence from Nigeria (2004-2014). International Journal of Finance and Accounting, 4(5), 253-261 James, L.R., & Brett, J.M. (1984). Mediators, moderators and tests for mediation. Journal of Applied Psychology, 69, 307- 321. Judd, C.M., & Kenny, D.A. (1981).Process analysis: Estimating mediation in treatment evaluations. Evaluation Review, 5(5), 602-619. Kanwal, I.K. (2012). Effects of dividends on stock prices: A case of Chemical and Pharmaceutical Industry of Pakistan. Journal of Management, 2(5), 141-148. www.cribfb.com/journal/index.php/amfbr American Finance & Banking Review Vol. 2, No. 2; 2018 13 Khalid, H., Chijioke, O.M., & Aruoriwo, M.C. (2010).Dividend policy and share price volatility: United Kingdom evidence. Journal of Risk Finance, 12(1), 57-68 Khan, K.I. (2012). Effect of dividends on stock prices– A case of chemical and pharmaceutical industry of Pakistan. Journal of Management, 2(5), 141-148. Available at doi: 10.5923/j.mm.20120205.02. Lintner, J. (1956). Distribution of income among dividends retained earnings and taxes. American Economic Review, 46(2) 97-113. Majanga, B.B. (2015). The dividend effect on stock price: An empirical analysis of Malawi listed companies. Accounting and Finance Research Journal, 4(3), 99-105. www.sciedupress.com/afr Miller, M., & Modigliani, F. (1961). Dividend policy, growth and the valuation of shares. Journal of Business, 34(4), 411-433. Musa, I.F. (2009).The Dividend policy of firms quoted on the Nigerian Stock Exchange: An empirical analysis. African Journal of Business Management, 3(10), 555-566. Nnadi, M.A., & Akpomi, M.E. (2005).The effect of taxes on dividend policy of banks in Nigeria. International Research Journal of Finance and Economics, 19(1), 48-55. http://dx.doi.org/10.2139/ssrn 873828 Oduwole, O. (2015).Earnings or dividends, which had more predictive power? Research Journal of Finance and Accounting.6 (8), 54 – 60. Ojeme, S., Mamidu, A.I., & Ojo, J.A., (2015).Dividend policy and shareholders’ wealth in Nigerian quoted banks. Canadian Social Science Journal, 11(1), 24-29. Ordu, M.M., Enekwe, C.I., & Anyanwaokoro, M. (2014).Effect of dividend payment on the market price of shares: A study of quoted firms in Nigeria. IOSR Journal of Economics and Finance, 5(4), 49-62. Oyinlola, O.M., & Ajeigbe, K.B. (2014).The impact of dividend policy on stock prices of quoted firms in Nigerian. International Journal of Economics, Commerce and Management, 2(9), 1-17. Ozuomba, C.N, Okaro, S.C., & Okoye P.V.C. (2013). Shareholders value and firm’s dividend policy: Evidence from public companies in Nigeria. Journal of Management Sciences, 2(12), 26-28. Preacher, K.J., & Hayes, A.F. (2004). SPSS & SAS procedures for estimating indirect effects in simple mediation models. Behaviour Research Methods, Instruments and Computers, 36(4), 717-731 Preacher, K.J., & Kelly, K. (2011). Effect size measures for mediation models: Quantitative strategies for communicating indirect effects. Psychological Methods.16 (1), 93-115. Rabindra, J. (2012). Effects of dividends on stock prices in Nepal. NRB Economic Review, 24 (2), 61-75 Rubinstein, M. (1976). The irrelevancy of dividend policy in an arrow- Debreu economy. Journal of Finance, 31, 1229-1230. Shrikant, P., & Yuserrie, Z. (2015). Dividend policy decisions: Theoretical views and relevant issues. Reports on Economics and Finance, 1(1), 43-58. http://dx.doi.org/10.12988/ref.2015.596 Sobel, M.E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models.in Sociological Methodology, Leinhardt (Ed.) San Franscisco CA: Jossey- Bass, 290-312. Suwabe, T. (2006). Dividend policy that boosts shareholders value. Security Analysts Journal, 44(7), 1-18 Toby, A.J. (2014). Empirical test of the dividend policy irrelevance hypothesis in the Nigerian context. Research Journal of Finance & Accounting, 5(6), 167-178. Udobi, P.I. (2016). An empirical analysis of Miller & Modigliani dividend policy irrelevance theory: Evidence from the Nigerian Stock Market. Unpublished Ph.D Thesis, University of Lagos. …………..,Iyiegbuniwe,W.I., & Ezike, J.E.(2018). Empirical analysis of the impact of current dividend policy on share price: Evidence from Nigeria. Unilag Journal of Business, 4(2),133-160 Walter, J.E. (1962). Dividend Policy: It’s influence on the value of the enterprise. Journal of Finance, 18(2), 280- 291. Copyrights Copyright for this article is retained by the author(s), with first publication rights granted to the journal. This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/).