Corporate dividend policy

Abstract

The primary objective of this study was to test Gordon's model on stock valuation in forecasting common stock prices under more recent world economic climate and together find an optimum dividend rate for a selected number of Canadian Companies. -- Computer program BMDO2R was employed to perform time series multiple regression analysis on the Gordon's model of stock price valuation. Share prices were predicted only for those companies which yielded significant results. The problems of auto-correlation and multicollinearity were recognized in this study. Auto-correlation was tested by means of Von-Neuman ratio and for multicollinearity, simple inter correlations between the three independent variables (i.e. growth in dividends, dividends and size variables) were looked at. -- Two of the seven companies gave insignificant results i.e. investors in these companies did not place importance on any of the three variables considered. Two other companies yielded statistically significant results but these were not acceptable as the signs of the related coefficients were negative and beyond any reasonable explanation. Share prices were, therefore, not predicted for these four companies and an optimum dividend rate could not be determined. One other company revealed that size was the predominant factor in explaining share price. This gave only one share price as size variable was independent of the earnings retention fraction and therefore an optimum dividend rate could not be determined. (Gordon's model assumes that maximization of share price is the sole criterion in formulating dividend policy. Earnings retention fraction was the variable used to generate different share prices). Two other oil companies yielded some encouraging results. In one of them only growth in dividend variable was preferred and in the other, dividend variable was preferred share price predictions compared favourably with the actual share prices in these two companies. -- Multicollinearity was almost absent in this study. Despite the presence of auto correlation in most of the companies, the results were accepted wherever these agreed with the theory. -- Two main conclusions were drawn from this study: (1) A general approach to the formulation of (a) the share price model i.e. Eq. (16) in section 4.3 and (b) rate of return on net worth i.e. Eq. (18) in section 4.3, cannot be taken as shown by the results obtained in this study. Therefore, for each company, variables affecting the share price need to be identified through trial and error. (2) Companies in which investors prefer growth in dividends, should retain the maximum possible amount of earnings and in companies where dividends are preferred, maximum possible amount of earnings should be distributed through dividends. Maximization of share price, however, has to be the sole criterion in formulation of dividend policy, for the above arguments to hold good.