Are Fama–French and momentum factors really priced?

Abstract

This paper investigates whether the existence of pricing anomalies represents compensation for bearing extra-market risks by directly testing a version of Merton's (1973) Intertemporal Capital Asset Pricing Model (ICAPM), allowing for both time-varying first and second moments of asset returns. The conditional ICAPM is estimated using multivariate GARCH in mean modeling strategy, and then it is used to examine three popular anomalies— size, book-to-market, and momentum. The results indicate that all the four risk factors (market portfolio (MKT), size portfolio (SMB), book-to-market portfolio (HML), and momentum portfolio (UMD)) are not only significantly priced, but also time-varying. The empirical results documented in this study provide a strong support of risk-based explanations for the anomalies. That is the apparent high average returns represent compensation for bearing extra-market risks, which are not captured by the CAPM. The results also shed a new light on how the momentum effect should be interpreted since previous studies fail to detect a priced momentum factor in addition to book-to-market factor. Furthermore, because the ICAPM estimated in this study is fully parameterized, I am able to decompose the total time-varying risk premia into four components: market, size, book-to-market, and momentum. Among those four component risk premia, the market risk premium is the dominant one in describing the return dynamics of portfolios sorted based on the size, book-to-market, and industry.