Assymetrical Response to Earnings and Dividend Announcenments
Although there has been a proliferation of papers reporting results of event studies in recent years, few papers provide an analysis of the underlying distributions of the data employed and utilise parametric tests regardless of normality violations although the parametric tests tend to overstate rejection rates when data are non-normal. This paper inspects abnormal returns around earnings and dividend announcements for normality and provides evidence that daily excess return data are non-normal. Two parametric tests and a non-parametric test are then applied to determine whether the abnormal returns are significantly different to zero and is found that the non-parametric test is better specified when the underlying distribution violates normality assumptions.