Guest Column: On the Misuse of Event Studies to Examine Market Efficiency

Dr. David Tabak is a Senior Vice President in the New York office of NERA Economic Consulting.

Dr. David Tabak

Proof of market efficiency is often important for securities class actions at both the class certification and merits stages.  Under the Supreme Court’s ruling in Basic v. Levinson, if a market is efficient, there is a rebuttable presumption that investors relied on market prices, meaning that there is no need for each potential plaintiff to demonstrate individual reliance on any misrepresentation or omission, as those would be reflected in the market price.  Courts have recommended various tests, or factors to consider, to assess whether a financial market is efficient.  Unfortunately, the primary test that is designed to be consistent with the academic literature on market efficiency, an examination of whether prices respond to news, is often performed in ways that render it meaningless.  This tide could turn, due to recent decisions recognizing some of the flaws in certain analyses as well as decisions that require courts to engage in more careful examinations of market efficiency.  On the latter point, the Ninth Circuit held last week that “Plaintiffs pleading fraud-on-the-market, on the other hand, may have to establish an efficient market to even raise common questions or show predominance.” Dukes v. Wal-Mart Stores, Inc., Nos. 04-16688 & -16720 (9th Cir. Apr. 26, 2010).

This article reviews a flawed manner in which this test is often implemented, while a fuller version of this article (available here) discusses how the test can be performed in a manner that is statistically valid.

I. Market Efficiency in the Courts

In Basic v. Levinson, the United States Supreme Court provided plaintiffs in securities class actions with a rebuttable assumption of reliance if the security that is the subject of the suit traded in an efficient market.  Basic v. Levinson, 485 U.S. 224, 108 S.Ct. 978 L.Ed.2d. 194 (1988).  Analysis of market efficiency has been performed at both the class certification and merits stages of securities fraud cases.

The seminal case on testing for market efficiency, Cammer v. Bloom, identified five factors for which the court felt analysis would be helpful in alleging market efficiency.  Cammer v. Bloom, 711 F. Supp 1264 (D.N.J. 1989).  The Cammer Court noted that one factor, an empirical response of prices to new information was “the essence of an efficient market and the foundation for the fraud on the market theory.”  Other courts have recognized the supremacy of this factor as well.  See, for example, PolyMedica, in which the Court stated that the cause-and-effect relationship “is ‘in many ways, the most important,’ [citing In re Xcelera.com Securities Litigation]…”  In re PolyMedica Corp. Sec. Lit., 453 F. Supp. 2d 260, 266 (D. Mass. 2006).  To a financial economist, this is indeed correct, as an efficient market is defined as one that quickly and fully incorporates all information of a certain type (pricing information for the security in weak-form efficiency; all public information for semi-strong-form efficiency; and all information for strong-form efficiency).

II. How Has the Price Response to News Been Examined?

There are various ways in which one can attempt to examine whether stock prices respond to news.  Perhaps the easiest, and the least reliable, is a form of “proof by example,” in which one finds a number of days in which there is news followed by a relatively large stock price movement.  The court in PolyMedica, properly noted that the “mere listing of five days on which news was released and which exhibited large price fluctuations proves nothing.”  Though this proof-by-example may be fancied up with various statistics (e.g., making sure that the stock price movement is statistically significant under some form of measurement) or even by looking at all of the days with price movements at or above some level, this form of proof by example still proves nothing, or perhaps more accurately proves merely that there were a few days when there was news and a large price movement without proving any reliable correlation between those two factors.  Also proving nothing are other non-comparative analyses such as determining the percent of all days with a significant price change that were associated with news or the percent of all days with news that were associated with a significant price change.

To see why these analyses are not meaningful, consider an example in which there are sixteen news days and four non-news days.  Of the sixteen news days, assume that four, or 25%, are associated with a significant price change; of the four non-news days, assume that one, or 25%, is associated with a significant price change.  Because the same percent of the days with and without news is associated with a significant price movement, one cannot say that news is correlated with a significant price movement.  Yet, this would not prevent someone from noting the four examples of days with news and presenting them as examples, or worse, proof, that the security’s price responds to news.

While real-world situations will typically be more complicated, often, once one peels away the calculations of market-adjusted returns and statistical significance, the same flawed “proof-by-example” methodology underlies some claimed results.   See, for example, the recent Countrywide case in which the Court recognized these flaws, and consequently found an event study to be “too subjective to be very helpful to the Court.  First, the event study used a very small sample of only ten days within the class period.  Second, those days may not be a representative sample because they were some of the biggest return days.”  In re Countrywide Financial Corporation Securities Litigation, Lead Case No. CV-07-05295-MRP (MANx) (December 9, 2009).  (For purposes of disclosure, the author was a consultant to defendants in this litigation.)

III. Conclusion

There are various ways to test if the price of a security responds to news.  For any such test to yield valid evidence in favor of market efficiency, however, it cannot be some form of proof by example or an examination of only days with large price movements.  Proper forms of event studies to assess market efficiency are indeed possible, and are used in the academic literature and in litigation, and can be useful to a court in examining whether a security’s price responds to news.