GMM Weighting Matrices in Cross-Sectional Asset Pricing Tests
30 Pages Posted: 21 Nov 2017 Last revised: 20 Mar 2024
Date Written: June 9, 2021
Abstract
The estimation of misspecified linear factor models for the cross-section of expected returns with GMM can result in a spuriously high explanatory power when the estimated factor means are allowed to deviate substantially from the sample averages. In fact, by shifting the weights on the moment conditions, any level of cross-sectional fit can be attained. The mathematically correct global minimum of the GMM objective function can be obtained at a parameter vector that is far away from the true parameters of the data-generating process. This is not a property of small samples, but holds in population. It is a feature of the GMM estimation design and applies to strong as well as to weak factors, and to all types of test assets.
Keywords: Asset pricing, cross-section of expected returns
JEL Classification: G00, G12, C21, C13
Suggested Citation: Suggested Citation