Crash testing the Fama-French factor model in emerging stock markets

Within just a few years, the Fama-French five-factor model¹ has become a standard model in the academic literature on asset pricing. Yet out-of-sample tests on the model in the emerging markets investment universe remain scarce. In a recent paper², Robeco’s Matthias Hanauer and fellow researcher Jochim Lauterbach, from the Technische Universität München, addressed this issue. Using monthly stock returns for a total of 28 emerging market countries over the period from July 1995 to June 2016, they examined the predictive power of an extensive set of factors, including value, size, profitability, investment and momentum variables.

In so doing, they were able to confirm the key results of Fama and French’s research on their five-factor model in developed markets. In other words, they found that the value, profitability, and investment factors (of which the combination of the latter two is often referred to as ‘quality’) can also be documented in emerging markets. Moreover, in contrast to conclusions deriving from financial theory, but in line with previous findings³, they did not find a positive relationship between risk and return.

However, the two researchers also found that the exact factor definitions used by Fama and French are less robust than alternative factor definitions and that a momentum factor should be added to the model4. As a result, they argue that the factor definitions of the Fama-French five-factor model are not ideal for emerging markets and propose a different set of variables they dubbed the “strongest emerging market variables”.

More specifically, they find that the anomalous returns associated with cash flow-to-price, gross profitability, composite equity issuance, and momentum are pervasive, as they are found in the calculation of both equal- and value-weighted returns, as well as cross-sectional regressions. Based on these findings, they proposed a new emerging markets factor model and derived out-of-sample return forecasts with it, based on current firm characteristics.

The forecasts based on the alternative factor definitions are superior to those derived with the five-factor model for both types of returns and cross-sectional regressions. Furthermore, these conclusions hold true not just for theoretical long-short portfolios, but also when taking common practical investment hurdles, such as short-selling constraints, transaction costs, and investments limited to large stocks into account. Under these conditions, the revised factor model proves to deliver a higher Sharpe ratio than value- and equal-weighted strategies as well as the minimum volatility and an optimized portfolio based on return forecasts derived from the five-factor model.

Ultimately, this study raises the question of how factors ought to be defined, not just in emerging markets but also in developed markets. The results presented by Hanauer and Lauterbach provide strong evidence to support the idea that the factor definitions of the five-factor model may not be optimal not only for emerging markets, but also for developed markets, including the US market.

Although the value, profitability and investment factors are now commonly accepted in the research community, both from an academic and a practical point of view, the debate over which variables serve as the best proxies for measuring these factors remains. In fact, in a paper published in 2017 in the Journal of Financial Economics, Fama and French acknowledged that future research might further refine the definition of factors in asset pricing models, including their own five-factor model.

Read the related research paper ‘The cross-section of emerging market stock returns’.