Determining the probability of a company paying dividends: an exploratory study at the Nairobi Stock Exchange

Abstract

Previous studies on dividend at the Nairobi stock Exchange [Abdul( 1993), Njoroge (2001)] have used ordinary least square regression techniques which assumes among other things linearity and normal distribution. Yet Aduda (1993) demonstrated that the distributions of the financial ratios for the Nairobi Stock Exchange firms are not normally distributed. Hence models using the normality assumptions may not be appropriate for analysis.

Thisis thus an exploratory study that attempts to use alternative models that do not rely on the normality assumptions to analyse the determinants of dividends payment at the Nairobi stock exchange. This study uses the technique of binary logistic regression which makes no assumptions of normality and linearity to analyse the effect of different financial ratios on the listed firms' dividend policies. The aim is to establish the factors that may discriminate between the firms that are likely to pay dividends from the ones that are not.

Dependent variables consisted of the binary state of either yes or no for companies that paid and did not pay dividends respectively. Being an investigative study, various ratios were considered as the independent variables, then using backward (Wald) stepwise logistic regression analysis with SPSSsoftware, the least significant and correlated variables were eliminated in steps until the "best" resultant model was obtained.

Findingsfrom the study confirms that the distribution of ratios for the listed firms between 2000 and 2004 are not normally distributed; Dividend payout ratio, Dividend yield, Price Earning Ratio and Price to Book values were the most significant factors in discriminating the dividends paying firms from non-payers at the Nairobi Stock Exchange.

Policy implication of this study is that the existing and potential shareholders could use these findings to establish how likely their firms or the firms they intend to invest in would pay dividends and hence construct their investment portfolios appropriately. Furthermore, in an era of increasing call for transparency and accountability the firms' managers and directors could use the model to def~nd their dividend policy decisions. Hence the findings of this study could assist in solving part of the agency conflicts between the directors and shareholders of publicly quoted firms.