^{1}

^{2}

^{2}

^{*}

^{2}

^{1}

Despite the growing debate on the stock price valuation, it has become a complex puzzle. Various theories, models and explanations have been provided to solve this confusing riddle. This study contributes to this debate by determining the relative importance of stock dividends and retained earnings regarding stock price valuation in Karachi Stock Exchange. Data for the analysis of this study were collected from the 66 nonfinancial companies that were included in KSE-100 index for a period from 2007 to 2010. This study found the evidence that dividends are more important variable than the retained earning regarding the explanatory power of stock prices in Karachi Stock Exchange. Practical implications are also provided in the study.

The active debate on the stock prices was started after the publication of the work of John Burr Willams in 1938, who for the very first time proposed about stock prices and the function of the dividends instead of the profits of the firm as it was believed at that time (Williams [

After the work of the Williams [

Retained earnings on the other hand are the profits of the organization that are reinvested in the business. According to Pardhan [

This literature review summarizes some important research studies relating to the valuation of stocks, considering dividends and retained earnings. Literature review provides a bird’s eye view regarding past developments and findings relating to the topic.

Havkevy [

Walter [

The most cited work in this regard is of Lintner [

A significant development on the determination of share price was done by David Durand [

Gordon [

Miller and Modigliani [

In response to the criticism of the MM, Gordon [

Lintner [

Friend & Puckett [

Another study conducted by Pardhan [

The research of Azhagaiah and Priya [

Khan [

This part explains the methodology of the research by specifying the model of the research and by describing the sample used in this study.

This section specifies the model relating to the determination of the relative importance of dividends and retained earnings in explanation of the stock prices. As the proposition entails that present value of the future cash flows should be the base of determination of the stock price so cash flows are dividends and the liquidation cash flow at some time in the future. This liquidation cash flow is composed of the value of the assets of the company with is financed by the debts and equity along with retained earnings. Retained earnings on the other hand could be a measure of the growth of the firm as greater retained earnings entail that firma has greater need of financing and thus growing. Relative importance of the retained earnings and dividends has been a matter of interest since the beginning. We follow the methodology of Friend & Pukett [

where

The above regression model however is said to be prone to bias of two types (Khan [

where

The other bias in the Equation (I) is that the reported income of the firms is a faction of several accounting and short run economic factors. The accounting procedure requires the corporations to estimate the earning of the corporations conservatively; the principle of conservatism. This gives rise to bias in the reported earnings. The dividend amount on the other hand is measured precisely and the impact of this measurement error is born by the retained earnings. This effect in this way could produce a downward bias towards the retained earnings (Friend and Pukett [

where

This is consistent with the price behavior as in Equation (III) price of the company is determined by the previous expectations on the current expectations of the price, dividends and retained earnings (Friend and Pukett [

For the empirical investigation a sample of 100 companies, which constitute KSE-100 index at Karachi Stock Exchange was selected. KSE-100 index is most widely used measure of stock market activity in Pakistan. The companies contained in the index are selected by their market capitalization and to assure the adequate representation of every industry sector, companies from each sector having highest market capitalization within the sector are also included in this index. By all means KSE-100 is considered an appropriate measure to represent all the variations in the overall prices of stock market in Pakistan. While all the industrial sectors are reflected in this index the companies included in the index could broadly be categorized in two type i.e. financial companies and non-financial companies. KSE-100 index is composed of 24 financial companies and 76 non-financial companies. Financial companies were excluded for the analysis because of their different type of nature. Data for the four other nonfinancial companies were also not available so they were also excluded from the analysis. A further six companies were also not included in the final sample due to the fact that they were not dividend paying firms and as the research is about the relative importance of dividends and retained earnings in determining the stock prices so companies with zero dividend for all time spam of the study could mislead the results. So, final sample constitutes 66 nonfinancial firms in total. The sample constitute all the significant firms repre- senting non-financial sector which pay dividends making the results of the study generalizable to the whole non- financial sector of Karachi Stock Exchange.

Data were collected from the analysis reports provided by the Karachi Stock Exchange on their website and from state bank publication namely “Balance Sheet Analysis of the Joint Stock Companies Listed in Karachi Stock Exchange” Issue (2003-2008) and (2004-2009). This study covers a time spam of four years from 2007 to 2010. Data was pooled for the analysis and OLS regression technique is used to draw inferences from the study.

This part of the study provides descriptive, correlation and regression analysis for this study.

(RE), lag value of price earnings ratio

cified only considers dividend per share and retained earnings per share as independent variables, while Model II considers lag of price earnings ratio along with dividend per share and retained earnings per share and Model III considers lag of market price along with the variables included in Model I. The mean value of the dependent variable i.e. MP is 319.26 along with a median of 138.46 and standard deviation of 568.91. The minimum market price in the whole of the data set was 4.89 and maximum market price was 4448. The overall, data set comprises companies with varying market price and characteristics making the analysis to be more generalizable. The first independent variable of the study is dividend per share (DPS) which had mean and median of 15.85 and 6 respectively along with a standard deviation of 29. The minimum and maximum of the variable were 0.5 and 299 respectively. The second independent variable of the study is retained earning which had mean and median of 15.85 and 5.88 respectively. The standard deviation for RE variable was 27.83 and minimum and maximum values are −56.54 and 27.83 respectively. The next variable inline is lag of price earning of the firm which had mean value of 7.13 and median of 10.55. The standard deviation for the variable is 51.72 along with minimum of −609.34 and maximum of 111.73. The last variable is lag of market price which had mean of 321.44 along with the median of 174.25. The standard deviation of the variable was 543.32 and minimum along with the maximum values were 3.55 and 4448 respectively. The overall data set indicates much variability which implies that the results of the study would be more generalizable.

No value of Pearson correlation coefficient increases the cut off value of 0.8 for all variables used in this study. So we may say that there is no high correlation among the independent variables used in the study. The highest correlation coefficient among independent variables is around 0.7 indicating a moderate correlation between lag of market price and dividend per share the reason for this correlation might be the fact that the prices of a security reflect the future dividends as well. This is a realistic situation so this correlation could be ignored for the purpose of the analysis.

Moreover, according to

MP | DPS | RE | |||
---|---|---|---|---|---|

Mean | 319.2625 | 15.84937 | 13.3681 | 7.13127 | 321.4357 |

Median | 138.46 | 5.99969 | 5.874912 | 10.54747 | 174.25 |

Maximum | 4448 | 229.0095 | 210.8868 | 111.7307 | 4448 |

Minimum | 4.89 | 0.499966 | −56.53852 | −609.34 | 3.55 |

Std. Dev. | 568.9087 | 28.99948 | 27.83083 | 51.72203 | 543.3233 |

Skewness | 3.92404 | 3.958715 | 3.010484 | −10.57026 | 4.263973 |

Kurtosis | 22.34909 | 22.89993 | 19.55701 | 126.4271 | 26.08894 |

Jarque-Bera | 2942.862 | 3096.178 | 2095.109 | 105847.9 | 4089.318 |

Probability | 0 | 0 | 0 | 0 | 0 |

Sum | 51720.53 | 2567.598 | 2165.632 | 1155.266 | 52072.58 |

Sum Sq. Dev. | 52108802 | 135396.2 | 124703.3 | 430702.1 | 47527226 |

Observations | 162 | 162 | 162 | 162 | 162 |

DPS | RE | ||||
---|---|---|---|---|---|

MP | Pearson Correlation | 0.6867^{**} | 0.3303^{**} | 0.74121^{**} | 0.0631 |

Sig. (2-Tailed) | 0.0000 | 0.0000 | 0.0000 | 0.4248 | |

DPS | Pearson Correlation | 0.2419^{**} | 0.7009^{**} | 0.0808 | |

Sig. (2-Tailed) | 0.0019 | 0.0000 | 0.3067 | ||

RE | Pearson Correlation | 0.2387^{**} | 0.0115 | ||

Sig. (2-Tailed) | 0.0022 | 0.8841 | |||

Pearson Correlation | 0.1513 | ||||

Sig. (2-Tailed) | 0.0545 |

^{**} Correlation is significant at the 0.01 level (2-tailed).

Model I established the primary relationship between the primary independent variables of dividend per share and retained earnings. This model is good fit as indicated by the F-statistic of 79.5689. This model predicts almost 50% variation in the stock price as the value of adjusted R square is 49.393. Durbin-Watson test value, on the other hand was found to be 2.034496 for the Model I, which indicates that no serial correlation exists in Model I as the test values are near the standard of 2. Model II on the other hand considers lag of price earnings ratio along with the main variables under consideration i.e. dividend per share and retained earnings per share to control for firm characteristics which are directly immeasurable otherwise. The usefulness of this inclusion would be judged by the relative increase in the goodness of fit of the model as indicated by the F-statistic and increase in the value of adjusted R square in comparison to the value in Model I. The inclusion of this variable has led to a decrease in the value of F-statistics to 52.7927 a value of 79.5689 in Model I. There is also no indication of improvement in the values of adjusted R square and a slight decrease in its value is also noted. The values of adjusted R square in the Model I was 0.49393 while its values in Model II is found 0.490811. The Durbin-Watson stat does not find trace of multicollinearity as calculated value i.e. 2.034608 is quite near to the standard of 2. While, Model III considers variable of lagged market price as a measure of the previous expectations to current expectations of the company prospects along with the main variables i.e. dividend per share and retained earnings per share. The significance of this variable inclusion is measured by increase in the F-statistic which is a measure of overall goodness of fit of model and increase in the value of adjusted R square which is a measure of predictability power of the model, from the value provided by Model I. The inclusion of this variable has significantly increased the F-statistics from 52.7927 to 86.7209. The adjusted R square has also shown im- provements and has grown from to 0.614982. A slight trace of positive autocorrelation in the Model III is found

Regression Analysis (Model I: | ||||
---|---|---|---|---|

Variable | Coefficient | Std. Error | t-Statistic | Prob. |

C | 71.20066 | 38.02481 | 1.87248 | 0.063 |

DPS | 12.64378 | 1.133567 | 11.154 | 0 |

RE | 3.565653 | 1.181167 | 3.01875 | 0.003 |

R-Squared | 0.500217 | Mean Dependent Var | 319.263 | |

Adjusted R-Squared | 0.49393 | S.D. Dependent Var | 568.909 | |

S.E. of Regression | 404.7137 | Akaike info Criterion | 14.8626 | |

Sum Squared Resid | 26043112 | Schwarz Criterion | 14.9198 | |

Log Likelihood | −1200.869 | F-Statistic | 79.5689 | |

Durbin-Watson Stat | 2.034496 | Prob(F-Statistic) | 0 | |

Regression Analysis (Model II: | ||||

Variable | Coefficient | Std. Error | t-Statistic | Prob. |

C | 70.69989 | 38.26786 | 1.8475 | 0.0665 |

DPS | 12.62898 | 1.140748 | 11.0708 | 0 |

RE | 3.567237 | 1.184842 | 3.01073 | 0.003 |

0.100145 | 0.620627 | 0.16136 | 0.872 | |

R-Squared | 0.500299 | Mean Dependent Var | 319.263 | |

Adjusted R-Squared | 0.490811 | S.D. Dependent Var | 568.909 | |

S.E. of Regression | 405.9589 | Akaike info Criterion | 14.8748 | |

Sum Squared Resid | 26038821 | Schwarz Criterion | 14.951 | |

Log Likelihood | −1200.856 | F-Statistic | 52.7297 | |

Durbin-Watson Stat | 2.034608 | Prob(F-Statistic) | 0 | |

(Model III: | ||||

Variable | Coefficient | Std. Error | t-Statistic | Prob. |

C | 19.98011 | 33.93348 | 0.5888 | 0.5568 |

DPS | 6.048764 | 1.352993 | 4.47065 | 0 |

RE | 2.827333 | 1.035434 | 2.73058 | 0.007 |

0.515242 | 0.072155 | 7.14078 | 0 | |

R-Squared | 0.622157 | Mean Dependent Var | 319.263 | |

Adjusted R-Squared | 0.614982 | S.D. Dependent Var | 568.909 | |

S.E. of Regression | 353.0067 | Akaike info Criterion | 14.5952 | |

Sum Squared Resid | 19688971 | Schwarz Criterion | 14.6715 | |

Log Likelihood | −1178.214 | F-Statistic | 86.7209 | |

Durbin-Watson Stat | 1.968415 | Prob(F-Statistic) | 0 |

but this autocorrelation is easily negligible as provided by Durbin-Watson stat of 1.968415, which are considerably near the standard of 2.

Turing towards the key point which is the determination of the relative importance of the dividends and retained earnings in prediction of the stock price, which would be determined by the relative higher coefficient value and t-statistic, we found that dividend has more influence over the stock price than retained earnings in all three Models. For the Model I, variables of dividend per share and retained earnings per share have positive coefficients with values of 12.64378 and 3.565653 respectively along with t-statistics of 11.154 and 3.01675 respectively, clearly indicating that dividends have more explanatory power with regard to the determination of the share price for the entire sample used in this study. The results with regard to magnitude and direction of the relationship between dividends, retained earnings and stock price are almost the same for Model II as were in Model I. Dividends again conform to be more important determinant of the stock price than retained earnings as provided by greater regression coefficient of 12.62898 for dividends as compared to coefficient value 3.567237 for retained earnings. The t-statistics were also found higher for the dividend per share 11.0708 as compared to its value for the variable of retained earnings per share i.e. 3.01073 respectively. The relationship of retained dividend per share and retained earnings per share with stock price was found positive for all three data sets. The lagged price earnings ratio variable on the other hand, was found insignificant in regression run for Model II with t-statics 0.16136. The direction of the relationship however, was found positive. This variable on the whole does not promise to have any significant change with regard to its impact on the estimation of relative importance of dividends and retained earnings. The question of relative importance of dividends and retained earnings still favors the dividends even after the inclusion of the lagged market price as control variable in Model III. The largest coefficient value of 6.048764 among all three variables and the second largest t-statistic value of 4.47065 strengthen the view that dividends probably are the most important factor that determines the stock prices in the stock markets. The variable of retained earnings per share provides a coefficient of 2.827333 along with a t-statistic of 2.37058 favoring the strength of dividends over retained earnings. The most significant variable in regression of Model III found to be lagged market price variable with the largest t-statistic of 7.14078 among all three variables used in this regression regression but a relatively smaller coefficient value i.e. 0.515242. All the variables in Model III indicated a direct relationship with stock price.

This section probed into the question of relative importance of dividend and retained earnings by applying three models. A total of 3 regressions were run and the results overall supported the hypothesis that dividends have a positive impact on the stock prices so, hypothesis H_{a1} is accepted, the association of the retained earnings on the other hand also provided a positive impact on the stock prices so, H_{a2 }is also accepted. Moreover, the results overall support the notion that dividends are more important than the retained earnings in determination of the stock prices, this leads towards the acceptance of hypothesis H_{a3} as Well. Apart from that, inclusion of the measure of the firm characteristics i.e. lagged price earnings ratio in the model as independent variable do not improve the estimation while inclusion of the lagged market price improves the estimation by increasing explained variance and overall goodness of fit of model.

The aim of this study was to seek out the question of relative importance of dividends and retained earnings in determination of stock price. A positive relationship of dividends and retained earnings was found for stock prices which are consistent with the literature on the topic. Dividend seems to be the most important variables with regard to the determination of stock prices in the stock market. Dividend is said to play a signaling mechanism for the investor to evaluate future prospects of the corporation thus reducing the information asymmetry problem. Theoretically, as proposed by Miller & Modigliani [

Moreover, the inclusion of lag of price-earnings ratio variable in the basic regression model specifying market price as a function of dividends per share and retained earnings per share did not improve the estimation for our sample. The inclusion of lag of market price variables although brought some improvements in the model with regard to predictability power of model. On the whole this variable could be used to improve the estimation with regard to the relative importance of dividends and stock prices. This is also consistent with the previous finings (Friend & Puckett [

This research study not only bridges the gap in the existing literature with regard to stock price evaluation in one of the best performing market of the developing counties in the world but also provides valuable practical insights with regard to stock price movements in the Karachi Stock Exchange. Findings of the study would not only be beneficial for the policy makers in sense as to leverage economic growth by supporting the financial activities such as devising monetary and fiscal policy in a way to boost the trading activity in the stock market which in turn could contribute significantly towards the economic growth and also attract foreign direct investment into the country but also for the individual investors who can use the findings of this study to device his investment strategy. Moreover, the findings of this study are also beneficial for the managers of the corporate sector who strive hard to build the value of the firm with a view to providing maximum benefits to the shareholders.

This study also has certain limitations which should be considered necessary. First of all the scope of the study is only limited to the establishment of the relative importance of the dividends and retained earnings and it does not consider other implications of the dividend policy like information content of the dividends, tax differentials between dividends & capital gains and agency cost implications of the dividends. These considerations clearly establish superiority of dividends over capital gains resulting from retained earnings. The empirical testing of these implications of the dividends is out of scope of this study and further research in the Pakistan would be more meaningful in this regard. Further, final sector is also excluded from the study due to distinct characteristics of financial sector and a subsequent study could also be conducted to test the relative importance of dividends and retained earnings for the financial sector and the findings of non-financial and financial sector could be compared for a better understanding of the dividends implications in both of the sectors.