Hypotheses Development and Research Design
5.4 Method of analysis
5.4.2 Event study
The event study methodology is the most powerful tool for researchers to assess the finan-cial impact of a particular event or news announcement. The event study methodology is made popular and well-known by Fama et al. (1969) who examine the effect of the announcement of a stock split on stock prices. Since then, event study methodology is developed by financial economists to assess the performance of securities markets and its application can be seen in the studies such as to estimate the impacts of regulation and other exogenous events on firms or industries (Rucker et al., 2005). The event study methodology is now widely used because of its general capability of being applied in many areas such as accounting and finance and laws.
Event study has become a standard method of measuring security price reactions to some announcement or event. In the study of price reaction to news, event study has been used for two major reasons, firstly, to test the null hypothesis that the market efficiently incorporates information into prices and secondly, to examine the impact of some event on the wealth of the firm’s security holders, under the condition that the hypothesis of market efficiency is maintained (Binder, 1998). Due to the wide application of event study methodology, there seems to be relatively little controversy about statistical properties of event study methodology. The conditions under which event studies provide information and permit reliable inferences are well-understood.
In general, in order to assess the impact of an event, we require a measurement whether
there is any abnormal movement of stock return surrounding the date of the event. It means that the stock return will be compared with the ‘normal’ stock return during the event window. A significant abnormal return reflects the impact of the event to stock prices, hence the event is considered an informative event. The normal return is defined as the return would be if the event did not take place. Returns during the event window are compared with normal returns. The differences between event window returns and normal returns are the abnormal returns (MacKinlay, 1997). This section is devoted to describing the model and statistical test that have been used in the literature.
Event definition and sample selection
The initiation of financial analysts coverage organised by a stock exchange is a relatively new practice. As mentioned earlier in Section 2.6, the analysts coverage incentive scheme is a new scheme introduced by Bursa Malaysia. The importance of analyst coverage, and hence the analysts’ initial reports is discussed in Chapter 4.
Identification of the correct event date is crucial in event study. In the study clearly identified. The ‘event’ is the release date of the financial analysts’ initial reports of companies that are involved in the incentive scheme in the Malaysian stock market. We define the ‘financial analysts’ initial reports’ as the first or the initiation reports produced by the analysts involved in the incentive scheme, and not the continuation reports. As the analysts recommendations are usually disclosed in the same analysts reports, the release of the analysts recommendations is also an important event accompanying the release of the analysts’ initial reports. This means the event study also examines the impact of analysts recommendations that go together with the analysts reports.
Companies that are selected in the event study are companies participating in the incentive scheme and their reports are already available. As mentioned earlier in Section 5.3.4, a matching control group is formed so that the results would be comparable. The event date is obviously identified as the date of submission of the analysts’ initial reports.
Even though some sophisticated investors may have an idea in which months the analysts reports of a particular company would be released, the exact event date is unanticipated nor the content of the reports.
In the study, we define ‘normal’ time as 220 days before the event to 121 days before the event (day −220 to −121), or 100 trading days of the estimation period. This is the period when the parameters are estimated and the normal return is calculated. It means that this is the return we are expecting if there is no event during this period of time.
The event window is the period the study is interested in. A highly deviated return in the event window from the return in normal period suggests that the event has a significant impact.
The normal days or estimation windows usually range from 100 to 300 days for daily studies and from 24 to 60 months for monthly studies. Lengthening the estimation windows involves a trade-off between greater precision of coefficients estimation and the coefficients becoming less representative. In some studies, the event window or the test period is a subset of the estimation windows (Armitage, 1995).
McWilliams and Siegel (1997) suggest that the events should be clearly unanticipated, while the event date should be clearly identified. Some investors may receive the infor-mation on the event day prior to the announcement to the public, and therefore trading may take place before the event. Due to the inability to exactly identify the date when the news were released, McWilliams and Siegel (1997) propose a short event window of two days. They argue that the event window should be very short because the underlying assumption in the event study is that the market is efficient that the market could be expected to react very quickly if the event is judged to be informative.
Moreover, McWilliams and Siegel (1997) suggest that a long event window is only justified for events that may have been leaked to or predicted by some traders. In this case, the event window should include a period before the event. Event window that goes beyond the event date is justified for events in which the impact of the event is uncertain.
Therefore, in the study various event windows were examined from a long event window of 120 days before and 120 days after the event, i.e. CAR [−120, 120] to a short event window, i.e. CAR [−1, 2]. The purpose of using a long event window is to capture any leaking of information or insider trading, while a short event is to capture the immediate impact of the event. Using a long and short event windows provides an insight of the properties of stocks involved in the incentive scheme.
As described in Section 2.6, there are 100 companies initially participating in the incentive scheme. As of the third quarter of 2005, there are 300 companies participating in the scheme out of a total of more than 1,000 companies listed on the stock exchange.
After performing the screening tests as described earlier in Section 5.3.2, we obtained the final sample of 55 participating companies.