A multifactor asset pricing model of the natural resource sector in Africa

Student thesis: Doctoral ThesisDoctor of Philosophy


The importance of the African market has increased in the past few years thanks to the continent’s fast-growing economies. However, research on asset pricing within the market remains very low. This study comprehensively investigates asset pricing within the resource sector of the African equity market.

To achieve a robust analysis, data problems within the African market had to be addressed. To do this, I formed indices of African markets; this was done by the creation of two major indices – the emerging African market index and the frontier African market index. A further two indices were created – the South African market index and the emerging African market excluding South Africa index. I also employ this classification because I expect differences in the results within these markets.

One major problem identified in the literature review regarding previous research in the African market is the lack of adjustments for survivorship bias. In analysing survivorship bias, I used the Jensen alpha approach and the mean difference approach, which identified survivorship bias of 297.47 basis points per week and 359.00 basis points per week for the emerging African market, using each approach respectively.

In analysing the performance of asset-pricing models within the African continent, I find significant differences across all four market indices, due to the varying levels of integration with world markets. I find that beta is consistently positive and significant, however, while size and liquidity are both significant but their direction depends on the characteristics of the surveyed market. I also find that value and momentum factors have a positive relationship with returns, but their importance depends on the level of integration with world markets.

The coskewness measure was found to be important only in the frontier African market, while the cokurtosis measure is important in an emerging African market context (including when South Africa is excluded). For the contagion factor, there seems to be an offsetting effect between the dummy and higher-order moments in the emerging African market; otherwise, contagion is negative and significant.

This contagion factor accounts for the financial crisis and the Arab Spring. This offered a unique opportunity to test the impact of contagion on unconditional, as well as conditional, asset-pricing models.

In analysing the conditional model, I employed GARCH-type models and found beta in the frontier African market to be unstable, while the high alpha parameter values in the South African market, the emerging African market and the emerging African market excluding South Africa showed no significance.

In testing for contagion using the conditional beta, I employed the dummy variable test and the comparison-of-means test; both showed evidence of contagion within the emerging African index, the emerging African index excluding South Africa and the frontier African market index. There was no evidence of contagion within the South African market and I attribute this to the interdependence between the South African market and Western markets.
Date of Award2016
Original languageEnglish
Awarding Institution
  • Coventry University
SupervisorTimothy Rodgers (Supervisor)

Cite this