Abstract:
The Capital Asset Pricing Model (CAPM) has its fair share of weaknesses and problems such as its well documented series of unrealistic assumptions. As a response Fama and French (1992) introduced the Fama and French three-factor model (FF3FM), but it remains unpopular among investors, practitioners and academics compared to the CAPM because it is deemed not cost effective and thought of as not being better than the CAPM. In 2015, Fama and French introduced the Fama and French five-factor model (FF5FM) that augmented profitability and investment into their FF3FM. Cakici (2015), Jiao and Lilti (2017), Foye (2018) and others have tested the five-factor model using data from their respective stock markets. The findings of these studies may not necessarily apply to South Africa because of institutional differences between countries. However, South African studies used different testing methods compared to this study. Therefore, given this background, this study sought to test the effectiveness of the FF5FM against the CAPM and the FF3FM in estimating stock returns on the Johannesburg Securities Exchange Limited (JSE Ltd).
This study tested the performance of the FF5FM against the CAPM and the FF3FM using data from the JSE-listed firms. The study sought to find out if the FF5FM performs better than the CAPM and the FF3FM when estimating stock returns of JSE-listed firms. Specifically, this study tested the CAPM, FF3FM and FF5FM using all the JSE-listed firms to determine which model explains better the common variation and the cross section of expected future stock returns. In addition, the study investigated whether the value factor became redundant when the additional factors, profitability and investment were added to the FF3FM as per Fama and French (2015).
Using the bespoke Generalized Method of Moments (GMM) of Hansen (1982) to carry out the regressions with data from the JSE for the period 2003 – 2019, the results show that profitability is a more reliable factor than investment in explaining share returns. The results also show that the FF5FM performs better than the other two models in estimating returns based on the assumption that most holding periods are significantly shorter than 16 years. Furthermore, the test results rejected the hypothesis that the value factor becomes redundant in explaining stock returns when more factors are added to the FF5FM.
