predict based on broad market movements in that period, given the stocks sensitivity to the market. For example, suppose that the analyst has estimated that a .5% and b .8. On a day that the market goes up by 1%, you would predict from equation 12.1 that the stock should rise by an expected value of .5% .8 1% 1.3%. If the stock actually rises by 2%, the analyst would infer that firm-specific news that day caused an additional stock return of 2% 1.3% .7%. We sometimes refer to the term et in equation 12.2 as the abnormal return-the return beyond what would be predicted from market movements alone. The general strategy in event studies is to estimate the abnormal return around the date that new information about a stock is released to the market and attribute the abnormal stock performance to the new information. The first step in the study is to estimate parameters a and b for each security in the study. These typically are calculated using in- dex model regressions as described in Chapter 10 in a period before that in which the event occurs. The prior period is used for estimation so that the impact of the event will not affect the estimates of the parameters. Next, the information release dates for each firm are recorded. For example, in a study of the impact of merger attempts on the stock prices of target firms, the announcement date is the date on which the public is informed that a merger is to be attempted. Finally, the abnormal returns of each firm surrounding the announcement date are computed, and the statistical significance and magnitude of the typical abnormal return is assessed to determine the impact of the newly released information. One concern that complicates event studies arises from leakage of information. Leakage occurs when information regarding a relevant event is released to a small group of in- vestors before official public release. In this case the stock price might start to increase (in the case of a "good news" announcement) days or weeks before the official announcement date. Any abnormal return on the announcement date is then a poor indicator of the total impact of the information release. Abetter indicator would be the cumulative abnormal return, which is simply the sum of all abnormal returns over the time period of interest. The cumulative abnormal return thus captures the total firm-specific stock movement for an entire period when the market might be responding to new information. Figure 12.5 presents the results from a fairly typical event study. The authors of this study were interested in leakage of information before merger announcements and con- structed a sample of 194 firms that were targets of takeover attempts. In most takeovers, stockholders of the acquired firms sell their shares to the acquirer at substantial premiums over market value. Announcement of a takeover attempt is good news for shareholders of the target firm and therefore should cause stock prices to jump. Figure 12.5 confirms the good-news nature of the announcements. On the announce- ment day, called day 0, the average cumulative abnormal return (CAR) for the sample of takeover candidates increases substantially, indicating a large and positive abnormal return on the announcement date. Notice that immediately after the announcement date the CAR no longer increases or decreases significantly. This is in accord with the efficient market hypothesis. Once the new information became public, the stock prices jumped almost im- mediately in response to the good news. With prices once again fairly set, reflecting the ef- fect of the new information, further abnormal returns on any particular day are equally likely to be positive or negative. In fact, for a sample of many firms, the average abnormal return will be extremely close