Adjusted beta responses to dividend announcements

Abstract

In this study, the impact of regression tendency and order bias on the calculated average residuals and around the event period is examined. The findings indicate that the abnormal return calculated with unadjusted parameters may lead to a misestimation of the true average residual. However, the bias due to regression tendency and order bias may not change the essential conclusions about the significance of the residuals if they are highly significant. However, for residuals that are borderline statistically significant, compensation for order bias and regression tendency may be important. In addition, if the magnitude of the abnormal return is of interest, residuals calculated with biased betas may over- or understate the true abnormal return. For small non-randomly selected samples, the differences in the residuals using unadjusted and order-bias-adjusted betas are more prominent. Therefore, the researcher should be aware of the impact that regression tendency and order bias can have on the interpretation of results.