Asset Pricing in Western Europe: An Inquiry into the Merits of the Fama-French Five-factor Model

Abstract

Factor models are at the core of asset pricing seeking to explain asset returns. The Fama-French five-factor model is an extension of perhaps the most appraised multi-factor model, namely the Fama-French three-factor model (Fama & French, 2015, 1993). While the asset pricing literature on the equity market in the United States is exhaustive, the Western European sphere is largely unexplored. This dissertation pursues the objective of contributing to filling that space in the literature by scrutinizing the external validity of the five-factor model by testing its merits in describing average excess stock returns in Western Europe in the period July 1999 to December 2021. The factors and test assets are constructed from scratch, and the R code is shared, motivated by the debate on the factor replication crisis (Jensen, Kelly & Pedersen, 2021) and to ease future research. The suitability of the model is scrutinized by estimating time-series factor loadings and crosssectional risk premiums for six sorts of test assets that comprehend 177 portfolios that are formed on independent double and triple sorts on different firm characteristics. The estimation of the regressions follows the procedure by Fama and Macbeth (1973) and applying Shanken’s correction to the standard errors (Shanken, 1992). The absolute and relative model performance is evaluated by testing if pricing errors are jointly equal to zero by means of the Gibbons, Ross, and Shanken (1989) test statistic. The results provide evidence that the Fama-French five-factor is suitable for describing average excess stock returns in Western Europe in the period July 1999 to December 2021. Although the five-factor model does not fully capture variation in average excess returns, it fares better vis-á-vis the Fama-French three-factor model across evaluation measures. The results allude to the portfolios exhibiting patterns in returns with respect to the factors, and the robust-minusweak (RWM) factor along with the high-minus-low (HML) factor are priced and generate significant risk premiums. However, the results outline that the conservative-minus-aggressive (CMA) factor is redundant in explaining variation in average returns, and the quest for a parsimonious factor model leads to a four-factor model that drops CMA. A principal component analysis asserts that broadening the factor universe beyond that of Fama and French can add to the description of returns. Finally, the suitability of the five-factor model is not sensitive to the choice of the sample period nor how the factors are defined.