Microfounding the Fama-MacBeth Regression

Abstract

In a new methodological approach to empirically test which factors determine equilibrium asset prices, Berk and Van Binsbergen (2016) find that the Capital Asset Pricing Model would better explain how capital flows into and out of mutual funds. This result is puzzling given that empirical asset pricing tests based on cross-sectional returns favor multifactor models `a la Fama and French (1993). In this paper, we focus on mutual funds managers’ investment decisions instead of mutual funds investors’ decisions. Using a structural model, we recover the factor risk premiums from observed fund managers’ investment decisions, for a set of well published asset pricing models. We find that Fama and French (1993) expanded by the momentum factor of Carhart (1997), and the liquidity factor of Pastor and Stambaugh (2003), can explain the largest fraction of funds’ investment decisions. Most notably, we also find that the macro-finance model a la Chen et al. (1986) is the one that best explains the observed mean-variance efficiency of the market portfolio. We reconcile this apparent paradox, by showing that factors that influence mutual fund managers’ asset allocation decisions (eg. value-size investing) may not survive at the aggregate level.