Robust inference for consumption-based asset pricing with power

Abstract

Kleibergen and Zhan (2020) propose a new approach to test consumption-based asset pricing models that is robust to the “useless” factor problem, i.e., concluding too often that a factor is priced when the factor is actually uncorrelated with the test assets and is not priced. I show that even when factor correlation is economically large and significant (think of 0.40 and larger), their testing approach lacks power in small samples to detect sufficient factor correlation or to find that a factor is priced. I propose simple remedies that help to achieve robust and powerful asset pricing tests.