Mitchell and Sta¤ord (1997) o¤er concrete perspective on how signiÞcancelevels can be overstated because of the failure to adjust for the correlation acrossÞrms of post-event abnormal returns. Using the three-factor model (1), theycalculate the standard deviations of abnormal returns for portfolios of Þrmswith an event during the most recent 36 months. The proportions vary somewhat through time and across their three event classes (mergers, share repurchases, and SEOs), but on average the covariances of event-Þrm abnormalreturns account for about half the standard deviation of the event portfolioÕsabnormal return. Thus, if the covariances are ignored, the standard error of theabnormal portfolio return is too small by about 50%! This estimate need notapply intact to the exchange listings of Dharan and Ikenberry (1995), but itsuggests that a full adjustment for the cross-correlation of post-listing abnormalreturns could cause the statistical reliability (t"!2.78) of their!7.02%post-event three-year CAR to disappear
đang được dịch, vui lòng đợi..
