Detecting Contemporaneous Security Market Reactions to a Sequence of Related Events
研究两种新统计量H和G2，用于检测证券市场对一系列相关事件的同期反应，扩展了传统事件研究方法，适合金融计量和实证研究者参考。
This paper describes results of a study of the empirical properties of two statistics, which extend security-return-based test statistics used in previous studies of contemporaneous market reactions to events.1 The first, called the H statistic, is a generalization of the Cumulative Average Residual (CAR) statistic. The procedure employed to test the significance of the H statistic extends the generalized least squares approach of Collins, Rozeff, and Salatka [1982] and Collins and Dent [1984]. The second, called the G2 statistic, is a generalization of the idea first introduced by Beaver [1968], who squared residuals in order to detect market reactions even if the average reaction was zero. The procedure we employed to test the significance of the G2 statistic has its closest antecedents in Patell [1976] and in Gonedes [1975].2 We might also note here that Schipper and Thompson [1983; 1985] propose an alternative framework within which tests similar to ours can be conducted. In our study, empirical distributions of H and G2 statistics were generated under the null hypothesis that there was no market reaction.