News‎ > ‎

New version of Elephants paper online

posted Nov 22, 2018, 8:12 AM by Julian Thimme   [ updated Nov 22, 2018, 8:14 AM ]
In the new version of my paper "Elephants and the Cross-Section of Expected Returns" with Nora Laurinaityte, Christoph Meinerding, and Christian Schlag, we show that there is a flaw in the design of GMM cross-sectional tests. As a result, factors with little explanatory power can seem highly significant and the explanatory power can be spuriously high. We show that this effect is present for weak and for strong factors and that irrelevant factors (betas to these are not related to expected returns in the cross-section) have the potential to drive out relevant ones. We suggest that authors should always report and discuss their estimates of the factor means which happen to be far away from the sample averages in case of a spuriously high pricing performance.
The new version of the paper can be downloaded here.