Are African Stock Returns Significantly Different Over Time? Rolling ANOVA Tests

• Published in: The Journal of developing areas 2021-03, Vol.55 (2), p.131-145
• Main Authors: Obalade, Adefemi A, Muzindutsi, Paul-Francois
• Format: Article
• Language: English
• Subjects: Adaptive behavior; Behavior; Calendars; Efficiency; Electronic trading systems; Environmental conditions; Equality; Hypotheses; Kruskal-Wallis test; Rates of return; Researchers; Robustness; Seasonal variations; Securities markets; Statistics; Stochastic models; Stock exchanges; Stock markets; Stocks; Trading; Variance analysis
• ISSN: 0022-037X; 1548-2278; 1548-2278
• DOI: 10.1353/jda.2021.0036

Adaptive market hypothesis (AMH) holds that stock return behaviour is time varying due to changes in the environmental conditions surrounding the markets. With the advent of the AMH, researchers' attention has gravitated to the examination of time-varying elements of seasonality in stock returns. However, researchers have not been applying rolling ANOVA tests despite the fact that ANOVA tests constitute the earliest methods used to establish whether returns are statistically different across weekdays or months of the year. This study examined time-varying seasonality in the selected African stock markets, by testing significant differences in mean returns across days of the week, months of the years and parts of the months over time. The sample for this study is a random combination of largest and smaller markets (South Africa, Nigeria, Morocco, Mauritius and Tunisia) in the African continent. The above objective was achieved with the aid of rolling ANOVA tests, implemented within a five-year fixed-length rolling window, rolled one year forward over time. The three ANOVA tests implemented in this study, are the Kruskal-Wallis (KW) and F-test for the equality of mean and the Levene test for the equality of variance. The three tests are conducted for robustness purposes; alternatively, the KW is regarded as adequate to achieve the purpose. The study findings indicated that there are windows where weekdays' returns, months of the year returns, and parts of the month returns, are significantly different and there are windows when they are not. Results indicated that there could be time-varying seasonality. In this context, the ANOVA tests performed in this study provided information as to whether cycles of significant differences (inefficiency/anomaly) alternated with those of equality in mean returns (efficiency). This article showed that ANOVA tests could also provide a degree of insight into the possibility of adaptive behavior of stock returns in African stock markets. Investors should be aware of the non-static nature of stock return seasonality to prevent fundamental loss. Future research is recommended on the adaptive behaviour of trading strategy of calendar anomalies in African stock markets.