The Impact of Non-Normality and Misspecification on Merger Event Studies

Financial event studies using daily stock returns are frequently used to evaluate antitrust policy and to 'predict' the consequences of mergers. Although there is ample evidence that daily stock returns are not normally distributed, traditional asymptotic results are often used for hypothesis testing. We suggest some general results concerning the conditions under which using ±1.96 as critical values for hypothesis testing under or over state the true significance levels. Further, we investigate the cause of non-normal event study residuals and find that a possible alternative explanation to one in which non-normality of residuals is the consequence of omitted events.