We demonstrate analytically that cross-sectional variation in the effects of events, i.e., in true abnormal returns, necessarily produces event-induced variance increases, biasing popular tests for mean abnormal returns in short-horizon event studies. We show that unexplained cross-sectional...