Empirical Tests of Multifactor Capital Asset Pricing Models and Business Cycles. U.S. Stock Market Evidence Before, During and After the Great Recession

Abstract

I estimate and perform empirical tests on the three most commonly used multifactor capital asset pricing models - Fama and French three-factor, Carhart four-factor and Fama and French four-factor models - in the U.S. stock market before, during and after the Great Recession. I prove that the critique directed at each of these models is fair, and none of the models is able to deliver persistent results. I demonstrate that the Fama and French threefactor model has a better performance compared to the four- and five-factor models. RMW and CMA factors are shown to be statistically insignificant in the Fama and French five-factor model, what reduces it to the traditional Fama and French three-factor model. I conjecture that the reason for that so many researchers have failed to recreate Fama and French’s results might lie in that Fama and French omit some steps when describing how they construct the models’ factors. Both the Fama and French three-factor and Carhart four-factor models demonstrate worse results when applied to the U.S. stock portfolio returns during the Great Recession than otherwise, which is first of all observed in the increased redundant variable problem. Surprisingly, it is the relative distress variable, HML, that tends to be redundant in the recession state of economy. Based on the achieved results, I point out the need for a more reliable asset pricing model, a model that would demonstrate robust results at explaining stock portfolio returns both in normal times and in crisis times in the economy.