Bias-Coppected Estimation for singles-event studies

Single-firm event studies used in securities fraud litigation are prone to statistical bias. Brav and Heaton (2015) describe three potential issues: low statistical power to detect price impacts, confounding effects that cannot be averaged out, and an effect bias in the upward direction. The effect bias is the difference between the measured effect size and the true effect size, with the measured effect size being biased upward relative to the true effect size. However, Brav and Heaton have underestimated the effect bias. In this paper, I quantify the bias and offer a nonparametric bias corrected estimator for single firm event studies.