Share market responses to large daily price declines: Evidence from the Johannesburg Stock Exchange

This study, utilising aspects of event methodology, focuses on the phenomenon of over-reaction which is defined as the over-response to new information. The over-reaction hypothesis suggests that the greater the magnitude of initial price change, the more extreme the offsetting reaction. Cox and Peterson's methodology, adapted to local conditions on the Johannesburg Stock Exchange (JSE), is used. Events are defined as single day price declines in excess of ten percent, fifteen percent and twenty percent for companies trading on the JSE between 1973 to 1998. Abnormal returns are computed for the twenty trading day period following the price decline. Abnormal returns are computed using both the trade-to-trade and the standard market model approaches. The regression parameters are estimated using a 150 day pre-event period and a 150 day post-event period. The period spanning six days before to twenty days after the event is not used for parameter estimation. Abnormal returns are then calculated for the event window, and average abnormal returns (AARs) and cumulative average abnormal returns (CAARs) computed and tested for significance over varying ‘time windows’ from the day after the event to twenty days thereafter. A series of cross-sectional regressions are also employed to control for liquidity (size), book-to-market and price earnings ratio effects. In contrast to Cox and Peterson who concluded that over-reaction per se does not independently and significantly account for abnormal returns, this study finds evidence of significant over-reaction with a positive price reversal over the three trading days following the significant price decline. The study therefore supports prior studies on the JSE suggesting the presence of market over-reaction.