A Bayesian Approach to Event Studies for Securities Litigation

摘要

As described in Gelbach and Hawkins (2020), when considering claims of securities fraud involving publicly traded securities, U.S. courts generally assume, based on the theory of efficient markets, that securities markets respond rapidly - within a single trading day - to release of information that the firm fraudulently represented its financial position. Under this rule, plaintiffs must establish by the legal standard of the "preponderance of the evidence" that the public release resulted in a one-day decline in the stock's price in excess of "normal" variation in the stock's price. By current practice, to establish such a decline, the plaintiff generally must show that the excess decline was statistically significant, presumably by the most widely applied standard of α = 0.05. Gelbach and Hawkins correctly point out that placing the burden of demonstrating statistical significance on the plaintiff is inconsistent with judging the case based on the preponderance-of-the-evidence standard. They instead lay out an elegant Bayesian alternative that they argue is consistent with the preponderance-of-the-evidence standard.