My new paper, coauthored by Nora Laurinaityte, Christoph Meinerding, and Christian Schlag, shows that the population growth of captive Asian elephants explains the cross-section of expected returns with a cross-sectional R^2 of 93% and a t-statistic of 4.0 for the market price of risk... NOT! We rather show that standard GMM cross-sectional asset pricing tests can generate spurious explanatory power for factor models when the weight on certain moment conditions is set inappropriately. In fact, by shifting the weights in the GMM, any desired level of cross-sectional fit can be attained at the price of not matching the factor means. We run placebo tests with factors that by construction do not explain the cross-section of expected returns and obtain spuriously high cross-sectional R^2's. Finally, we document examples of factor models proposed in the literature that suffer from this bias. The paper can be downloaded here.