Predictability of ordinary stock returns at the Nairobi Stock Exchange in Kenya

Abstract

The study sought to determine the predictability of ordinary stock return, for selected securities listed on the NSE using recursive least squares regression. Monthly closing prices (mode price) for the selected securities were used to obtain dividend yield and earnings price ratio, which were exogenous variables in the model. The other independent variables in the model are Monthly Treasury bill rate, Month -on -month inflation rate, monthly percentage change in broad money supply and monthly percentage change in export earnings from coffee and tea as a measure of agricultural production. The period of study was January 1995 to December 2002 Although there were no significant differences between actual and forecast values generated by the model, we conclude that the predictability evidence for ordinary shares in the NSE is weak and not conclusive. This is due to the fact that only three samples had statistically significant sensitivity measures (coefficients) of the variables used in the model. Besides, the proportion of explained variations (R2) in ordinary stock return was low ranging from 3.8% to 20.9%. The implication of this finding to financial analysts is that at the end of the day, not all the variations in ordinary stock returns can be explained by changes in various macroeconomic indicators. This in turn confirms the assertion of Shiller (1989) that Short-term changes in stock market indexes may well be influenced by what he termed 'investor psychology'; and what other scholars such as Williamson (1993) have called 'herdlike' behaviour of investors in investment decision-making.