Stock market price dynamics in Africa: evidence from 14 countries

Purpose This paper deals with the stock market prices in Africa. In particular, we focus on data from 14 African countries that have stock exchanges with a market capitalisation of more than US$1bn also capturing stock market dynamics during the COVID-19 pandemic. Design/methodology/approach The methodology is based on the concept of fractional integration that indicates that the number of differences to be taken in a series to render it stationary I(0) may be a fractional value. Findings Assuming a white noise process, only the South African stock market displays transitory shocks. Allowing for autocorrelation, the time trend is statistically significant in four countries: Namibia, South Africa, Tunisia and Zimbabwe, although mean reversion is found in two countries: Morocco and South Africa. These two countries are the ones where the random walk can be rejected and thus are informationally inefficient. There is one country (Rwanda) with an estimated value of d above 1, while the unit root null hypothesis cannot be rejected in the remaining countries, implying that in all of them, other than Morocco and South Africa, the random walk hypothesis cannot be rejected. Research limitations/implications A limitation of this work is the number of countries examined, 14, due to the lack of available data. Dealing with the methodology, a limitation is also the linear nature of the trend structure examined, which may be extended to non-linear approaches. Practical implications The practical implication is the mean-reverting nature of some of the series examined, implying that no strong measures should be adopted in these cases if exogenous negative shocks occur. Social implications The academia and also practitioners can be interested in the present work, in particular in relation to the mean-reverting nature of the series and the efficient market hypothesis. Originality/value The originality in this work comes from the use of fractional integration, a methodology that is not very usual in the analysis of time series data.