Investors reaction to dividend announcements: parametric versus nonparametric approach

Since the seminal works of Ball and Brown (1968) and Fama et al. (1969), event studies have served an important purpose in capital market research as away of testing stock price reaction towards a particular event. The ordinary least squares estimation method (OLS) has been employed in most extant research. However, if the assumed normality assumption is violated, then the OLS procedure will produce bias estimate to abnormal returns. Therefore, some authors (e.g. Dombrow et al., 2000) suggest a nonparametric approach to estimate the market model such as Theil (1950). This paper aims to examine investors' behavior prior to dividend announcements. The results indicate that the daily abnormal returns depart considerably from normality and that Theil's estimation procedure produces higher standardized abnormal returns than that of the OLS estimation procedure. Furthermore, the results confirm that investors achieve positive abnormal returns in the pre-announcement period. This paper concludes that some investors have access to information prior to become public.