Validity of Market Model in the Event-Study Methodology: Evidence from Dividend Announcements in Colombo Stock Exchange (CSE)

Abstract

Market model is widely used to predict stock returns under the event-study methodology, especially when testing the news impact of stock markets. The underline assumptions of the model have the potentiality in producing robust empirical findings in case of efficient stock markets. However, those assumptions tend to hinder its applicability for stock markets, which are yet to reach strong form efficiency. Therefore, this study assesses the validity of Market model considering its good fit and robustness in case of final dividend announcements in CSE. On this regard, 211 final dividend announcements are considered during 2009 to 2013. 120 days prior to the announcement are used to predict expected returns on the event day. The results of the study indicate that the explanatory power of the Market model in almost all these cases are below 5%. This seriously questions the accuracy prediction, which is crucial in the event-study methodology. Further, when the model is robusted with time series forecasting, two models provide abnormal gains which are significant but varied in terms of their signs over 25% of the dividend announcements. Thus, findings of the study evidences that Market model, in the event-study methodology, fails in terms of both accuracy and robustness.