A robust and powerful test of abnormal stock returns in long-horizon event studies

This paper proposes a novel standardized test for abnormal returns in long-horizon event studies that takes into account cross-sectional correlation, autocorrelation, and heteroskedasticity of stock returns. Extensive simulation analyses demonstrate improved size and power of testing relative to existing long-run test methodologies. Application to initial public offerings and seasoned equity offerings further demonstrates robustness to extreme return outliers inherent in these long-run studies.