New Evidence on Conditional Factor Models

Abstract

We estimate conditional multifactor models over a large cross-section of stock returns associated with 25 alternative CAPM anomalies. Using conditioning information significantly improves the performance of the multifactor models of Hou, Xue, and Zhang (2015, HXZ) and Fama and French (2015, 2016). Both models have a similar global fit, yet HXZ clearly dominates other models in terms of pricing the extreme portfolio deciles and the cross-sectional dispersion in equity risk premia. Thus, the asset pricing implications of alternative investment and profitability factors differ significantly in the large cross-section of stocks. The HML factor is largely redundant when using conditioning information.