Estimation and test of a simple model of robust capital asset pricing: An info-metrics approach

Abstract

We introduce the info-metrics approach to empirical asset pricing under ambiguity. We apply relative entropy as a pseudo-metric of model discrepancy, and generalized maximum entropy as a principle of statistical inference, to cross-sectional asset pricing tests. We show that a single-factor market representation of the CAPM under ambiguity can explain the cross-section of U.S. stock returns without the aid of additional risk factors. The additional factors can be interpreted as compensations for idiosyncratic ambiguity. The approach can also recover the market price of ambiguity that sets a lower (entropy-based) bound on stock prices, which can be understood as investors' “margin of safety” against extreme market events.