Consequences of Outlier Returns for Event Studies: A Methodological Investigation and Treatment
本文提出一种抗异常值的最大似然估计方法，将股票回报分解为常规和异常成分，发现传统OLS方法在事件研究中会漏判约37%的负面和43%的正面事件，质疑了以往事件研究结论的有效性。
Stock returns are decomposed into their regular and outlier components using a maximum likelihood outlier-resistant estimation method. Analytical results depicting the impact of outliers on the ordinary least square (OLS) estimated models and cumulative abnormal return (CAR) statistics are derived and validated using Monte Carlo simulations. The implications of outliers for past event studies are investigated using samples drawn randomly from the universe of stocks in the CRSP database. The OLS-CAR statistics fail to forecast about 37% of the negative-impact and 43% of the positive-impact events. These results raise serious concerns about the validity of conclusions of past event studies, especially those that rejected the hypothesis of significant-impact events.