Measuring and Testing Abnormal Operating Performance in the Accounting-based Event Study
Hyung-Chan Jung
This paper investigates the empirical power and misspecification of event study methods detecting accounting-based measures of abnormal operating performance following major corporate events, such as stock repurchases, security offerings, mergers and acquisitions, or earnings announcement, in the Korean Stock Market. Based on the simulation analysis that is similar to that of Brown and Warner(1985), this paper examines the choice of a model of operating performance and a statistical test of abnormal operating performance. We find that the level model of expected operating performance that matches sample firms to reference portfolios of one-digit KSIC code and similar pre-event performance yields well specified, powerful test statistics in the random samples. We also find that nonparametric Wilcoxon signed-rank test statistics are uniformly more powerful than conventional t-statistics. We document, however, that there is no method that yields test statistics that are well specified in every sampling situation due to the impact of size or performance based sampling biases on statistical tests.