Stock Market Over-reaction: The South African Evidence

Abstract

It has been suggested that stock markets over-react and that investors pay too much attention to recent “dramatic” news. If over-reaction does occur and prices overshoot then there should be a subsequent revision in the opposite direction. This paper outlines empirical research into the over-reaction hypothesis on the Johannesburg Stock Exchange using data over the period July 1974 to June 1989 for two hundred and four relatively well traded securities.The results are consistent with the over-reaction hypothesis and indicate substantial weak form inefficiencies in the South African stock market in the long-term. The performance of portfolios of shares formed on the basis of prior return data can be predicted and, on average, portfolios of prior ‘losers’ outperformed prior ‘winners’ by about twenty percent over the three years after portfolio formation. Finally, comparison between the empirical results and a similar study for the New York Stock Exchange calls into some question the hypothesis that exceptionally large returns in January in the USA are due to investor tax loss selling. There is evidence of both a January effect and an asymmetric excess returns effect for the South African market but it is less pronounced than for the American market.