Information transfer effects associated with dividend initiation announcements

In this thesis, new evidence is provided on the information effects associated with dividend initiation announcements. The evidence shows that dividend initiation announcements are not purely firm specific events as implied in prior studies and that the announcements have effects on other industry counterparts. The information transfer phenomenon is investigated using four estimation methods. The results based on seemingly unrelated regression (SUR) and a simultaneous equation system show that dividend initiation announcements are risk-altering events and that both announcing and non-announcing firms in the same industry experience changes in risk during the announcement period. This finding is consistent with the hypothesis that dividend initiation announcements convey information about other firms in the industry. The analysis also shows that more firms exhibit information transfer effects when a simultaneous equations system is used than when SUR is employed. This suggests that estimating information transfer effects using a method that takes account of interdependence among firms in the same industry not only improves the CAPM parameters, but also it increases the likelihood of detecting information transfer effects. The causality analysis shows that for more than half of the sample firms that exhibit evidence of information transfer effects, the observed effect is caused primarily by the announcements. For these firms, the information effects are either contemporaneous or unidirectional, with the effects flowing from the announcing firms to nonannouncing firms. Interestingly, over half of the firms that experienced a significant information effect (in the form of risk change) had significant betas only during the event period but not in the comparison (non-dividend initiation) period. This is strong evidence of an information effect. That the announcement caused a significant beta change only during the information period is evidence of how informative the market perceived the dividend initiation announcements to be. This result is powerful because it suggests that for those firms, security risk premium comes from information and not from a market risk premium. The finding is consistent with the proposition that although generally, market risk premium does not explain returns of these firms, the dividend information effect was incorporated into the firms' beta during the information period. Further analysis based on the event study methodology also indicates that dividend initiation announcements elicit abnormal price performance for both announcing and nonannouncing firms. While on average, the announcing firms' shareholders earn significantly positive standardised abnormal returns that were 4.20 times bigger than the residual standard deviation on the announcement day, the non-announcing firms' shareholders also realised positive and significant abnormal returns 3.40 times greater than the residual standard deviation. The results of the study have implications for prior empirical studies on dividend announcements. They imply that the information effects associated with dividend initiation announcements are greater than have been reported in the literature. By focusing on only the announcing firms, prior dividend information-based studies have underestimated the information effects associated with dividend initiation announcements.