Predictive ability of selected asset pricing models on the Nairobi stock exchange.

Abstract

This study is on the predictive ability of three pricing models on the Nairobi Stock Exchange objective was to identify which of the selected highest predictive ability. selected asset (NSE). The main models had the The 57 companies quoted on the NSE as of February 1989 were surveyed. The focus was on companies that have ordinary shares in their capital structure. These companies were stratified into two groups:' the actively traded and the non-actively traded companies. Though there were sixteen companies in the former group, only twelve were studied. The major primary sources of data were the NSE daily price lists, and the annual reports and accounts of the companies. The data obtained from these sources were monthly prices; and the earnings per share (EPS), dividends per share (DPS) and capital employed respectively. Forecasting of future sh~re prices, DPS and EPS was done via the Box-Jenkins (1976) method of time-series analysis. The forecasted values for DPS and EPS were discounted and capitalized respectively for five separate periods to give rise to predicted share prices for the first five months after each company's financial year end. Each of the prices obtained was compared to the actual price for the same period. The differences between the • two prices were subjected to t-tests. From the results of these tests, it was generally concluded that under the prevailing conditions in the NSE, no model qualified as a good predictor of share prices.