The Logic and Limits of Event Studies in Securities Fraud Litigation

Event studies are ubiquitous in securities fraud litigation.  Litigants use economist-prepared event studies to address several key elements, including loss causation, reliance, materiality and damages.  As one trial court recently explained: “To show that a corrective disclosure had a negative impact on a company's share price, courts generally require a party's expert to testify based on an event study that meets the 95% confidence standard.” Erica P. John Fund, Inc. v. Halliburton Co., 309 F.R.D. 251, 262 (N.D. Tex. 2015) (emphasis added).

The United States Supreme Court’s 2014 decision in Halliburton Co. v. Erica P. John Fund, Inc., 134 S. Ct. 2398 (2014) (Halliburton II) increased the importance of event studies both by reaffirming their role in enabling a plaintiff to obtain a presumption of reliance at the class certification stage and by opening the door for defendants to use them to rebut this presumption.  Litigants use event studies in the reliance inquiry to prove or disprove price impact, that is, whether allegedly fraudulent statements significantly affected the market price of the stock.  Event studies address price impact by determining whether a highly unusual price movement occurred relative to the return that would have been expected for the stock in the absence of fraud. 

Key aspects of event study methodology are misunderstood.  Courts and commentators repeatedly fail to recognize critical differences between the litigation context and the academic setting in which event studies were developed, differences that require adjustments to the event study methodology.  As the lower courts struggle to implement and address the legal issues raised in Halliburton II, a better understanding of event studies is critically important. 

In our working paper, “After Halliburton: Event Studies and Their Role in Federal Securities Fraud Litigation,” which is available here, we explore a variety of considerations related to the role of event studies in securities fraud litigation. The paper starts by explaining the building blocks of the event study methodology.  We demonstrate how economists determine the functional components of an event study –the expected return, the excess return on the event date, and whether the excess return is statistically significant. 

We then explain four common problems with the way event studies used in securities fraud litigation are conducted, and we show how to fix them.  First, the testing approach used in standard event studies must be modified to reflect the fact that, in securities fraud litigation, a stock price movement must not only be large, but also occur in the correct direction to show the relevant kind of price impact; thus, one-sided hypothesis testing, rather than the conventionally used two-sided testing, is the appropriate statistical method.  Second, litigation event studies incorrectly assume that excess returns are normally distributed. This mistake can be corrected using the SQ test proposed by two of us in other work. This test works whether excess returns are normally distributed or not.[1]  Third, event studies need to reflect the fact that litigation frequently involves testing for price impact on multiple dates.  Fourth, litigation event studies need to incorporate adjustments the problem of dynamic changes in the excess return’s volatility.

We use six event dates from the Halliburton litigation to illustrate these points and to demonstrate their potential effect on the litigants’ ability to demonstrate price impact.  In Halliburton itself, we show that the district court erroneously found a statistically significant price impact on only one of the six dates.  A correct application of the event study methodology, as described in our paper, results in a showing of price impact on a second date as well. 

Finally, the paper highlights certain broader limits of event studies.  One such limit is that event studies simply demonstrate the statistical significance of price movements, but do not address the rationality of those price movements.  Another is that standard event studies cannot reliably detect the effect of false confirmatory disclosures or the event of multiple disclosures that are made simultaneously or bundled together (though it may be possible to do so using modified approaches). Perhaps most important, in an effort to reduce the incidence of false positives, experts testifying about event studies used in securities fraud litigation usually adopt the 95% standard for statistical significance, reflecting typical practice in academic event studies.  But in the litigation context, this standard has the effect of producing a test with very low power – that is, a very high risk of false negatives.  Consequently, even experts who use the modifications we propose in this paper will fail to find a statistically significant price impact in a substantial number of cases where actionable fraud really did occur, if they continue to use the 95% standard for significance.  The policy implications of imposing this standard of proof warrant further scrutiny by courts and policymakers.

Jill Fisch is the Perry Golkin Professor of Law, Jonah Gelbach is Professor of Law, and Jonathan Klick is Professor of Law, all at the University of Pennsylvania Law School.