MEASURING LONG-HORIZON SECURITY PRICE PERFORMANCE

Our simulation results show that rests for long-horizon (i.e., multi-y ear) abnormal security returns around firm-specific events are severel y misspecified. The rejection frequencies using parametric tests somet imes exceed 30% when the significance level of the test is 5%. Our res ults are robust to many different abnormal-return models, Conclusions from long-horizon studies require extreme caution. Nonparametric and b ootstrap tests are likely to reduce misspecification.