Seasonal Patterns On the Johannesburg Stock Exchange: Some New Evidence

The purpose of this study is to revisit earlier findings on seasonalities in South African share markets to determine whether these anomalies still exist. The study stretches over two time periods, i.e. from 1978 to 1989 (sub-period 1) and from 1990 to 1998 (sub-period 2) and examines price changes in the All Share Index, Industrial Index, All Gold Index and Financial Index.In sub-period 1, significant day-of-the-week (or Monday) effects are found. Turn-of-the-year effects are also observed. In sub-period 2, significant week-of-the-month effects and strong positive turn-of-the-month effects are observed. The seasonalities observed in sub-period 1 are in line with earlier research but as none of these anomalies persist into sub-period 2 it appears that thes~ anomalies no longer exist in South African share markets. It therefore appears (with the exception of the Financial Index) that the popular day-of-the-week effect (or Monday effect) is no longer present in South African share markets.In sub-period 1, Mondays of all weeks contribute towards explaining the variation of returns. In sub-period 2, there is limited evidence of the Monday effect, but only in week 4 and the observed seasonal is not strong enough to cause a significant day-of-the-week effect. The evidence in sub-period 2 and, to a limited extent that observed for the All Share Index and Industrial Index in sub-period 1, indicates that there may be some support for the findings of Wang, et al. (1997) that the Monday effect is caused primarily by Monday returns in the latter part of the month.The anomalies observed in sub-period 1 are still evident when examining the variables concurrently. In sub-period 2, the turn-of-themonth effect remains strongly significant when combining the seasonalities. It is found that the week-of-the-month effect disappears when controlling for the turn-of-the-month effect. This suggests that the week-of-the-month effect is simply a manifestation of the turn-of-the-month effect.