Modelling Long Memory in Volatility in sub-Saharan African Equity Markets

Abstract

This study examines the long memory properties in the second moments of the return series in the equity markets of Ghana, Kenya, Nigeria and South Africa. Using 5219 daily observations of thin-trading adjusted total equity market return data, we find the presence of long-range dependency in the second moments of the return innovations in all the four countries’ equity markets in the full sample. To isolate spurious long memory, we perform structural breaks analysis to guide us in splitting the data for further examination. We find that all the four countries exhibited two structural breaks each during the sample period, and long memory identified in these two markets were not influenced by the structural breaks. This finding may have an influence on portfolio diversification and risk management. The long memory characteristics in the conditional volatility may also provide useful information to market participants in pricing long-term derivative contracts.