Judicial Competence in Statistics Will Govern the Future of Securities Class Actions

In today’s post we dig into the weeds of one aspect of what it takes to pursue a class action for alleged fraudulently misleading disclosures by public companies in violation of the federal securities laws.  The recent Supreme Court decision in Erica P. John Fund Inc. v. Halliburton Co., 134 S. Ct. 2398 (2014) (often referred to as Halliburton II because it was the second Supreme Court decision addressing the class action issues in that case), reemphasizes the importance of expert evidence that must be presented by a plaintiff to prove that a class action is appropriate.  That places great weight on the evaluation of such evidence.  There is a small body of case law addressing how to go about doing that, and on August 4, 2014, another case was decided that adds a little to that body of law: Brown v. China Integrated Energy Inc., No. CV 11-02559 (C.D. Cal.).  (Click here to see a copy of Brown v. China Integrated Energy).

First, some background.  A securities fraud action for alleged violations of section 10(b) of the Securities Exchange Act of 1934 requires, among other elements, proof that the plaintiff suffered losses by acting in reliance on the defendants’ alleged fraudulent communications.  In order to proceed as a class action – that is, an action in which a representative plaintiff pursues a case on behalf of many other similarly-situated persons – the plaintiff must show that the key evidence used to prove a case will apply equally to the cases of the “absent class members,” the other plaintiffs who would win or lose based on the representative plaintiff’s case.  Because “reliance” is a subjective concept, to proceed as a class action, a putative class representative must somehow show that its theory of “reliance” would apply equally to show reliance by the absent class members.

As we discussed in our previous post on the Halliburton II decision, in Basic Inc. v. Levinson, 485 U.S. 224 (1988), the Supreme Court decided that private federal securities fraud damage actions for violations of section 10(b) could proceed as class actions if the evidence showed that the security in question traded on an “efficient market” for those securities.  The Court accepted that for such markets, the price of the security necessarily would reflect all material public information about the security, and as a result, both the representative plaintiff and those who engaged in market transactions in the same time-frame would, by completing the transaction, be effectively “relying” on all material information in the marketplace at that time, which would have been incorporated into the price of the security.

As we described in the earlier post:

Basic became the Creator of the Securities Class Action because it formulated a theory by which reliance could be proved en mass without the usual showing that each plaintiff actually relied on the alleged misrepresentation.  It did this by accepting the viability of the “efficient capital markets hypothesis” (ECMH), an economic construct that open and developed securities markets are reasonably efficient marketplaces, which incorporate all publicly available information into stock price reasonably quickly.  485 U.S. at 243-45.  On top of this, it concluded that investors assumed that the prices of stocks on these exchanges fairly reflected accurate publicly-available information about the issuers of those stocks, i.e., that they traded shares “in reliance on the integrity of the price set by the market.”  Id. at 245.  Using these assumptions, the Court created a “presumption” – founded on “considerations of fairness, public policy, and probability, as well as judicial economy” (id.) that allowed securities plaintiffs to prove reliance in private section 10(b) actions en mass: “An investor who buys or sells stock at the price set by the market does so in reliance on the integrity of that price.  Because most publicly available information is reflected in market price, an investor’s reliance on any public material misrepresentations, therefore, may be presumed for purposes of a Rule 10b-5 action.”  Id. at 247.  However, that presumption can be rebutted: “Any showing that severs the link between the alleged misrepresentation and either the price received (or paid) by the plaintiff, or his decision to trade at a fair market price, will be sufficient to rebut the presumption of reliance.”  Id. at 248.

Halliburton II reaffirmed the approach adopted in Basic, leaving in place the basic elements of a class action for alleged violations of section 10(b).  In particular, it left in place the concept that there could be classwide proof of reliance by showing that a security traded on an “efficient market.”  The Halliburton II opinion added to this that a defendant could still avoid class certification in such cases by proving there was, in fact, no price impact from the misleading disclosures alleged in the complaint (and therefore the stock price did not implicitly convey those disclosures).  Both the proof that the security traded on an efficient market and any proof that the security price nevertheless was unaffected by the alleged misrepresentations require expert evidence.  So Halliburton II doubled down on the importance of this type of expert evidence as predicate for securities class action cases.

Experts are notoriously flexible in their ability to perform a supposedly rigorous analysis and yet reach contrary conclusions.  Plaintiffs can be expected to present “experts” whose studies will show the market for the securities at issue was efficient.  Defendants will likewise be expected to present experts who will either use studies to show the opposite, or will poke holes in plaintiff’s expert analysis to show it is unreliable, and therefore the plaintiff failed to satisfy its burden of proving an efficient market.  This places a lot of weight on how courts evaluate and rule on this battle of experts.

The experts in this area rely on statistical or “econometric” analysis to support an opinion.  The statistical analysis involved is called an “event study.”  It involves looking at how the price of the security at issue typically moves in relation to a set of factors in a period during which no fraud is alleged or known.  Usually the experts use a statistical tool called a “regression analysis,” which is a method of estimating the impact on a “dependent variable” (here, changes in the security price) of a range of “independent variables.”  The expert would then look to see how much, if any, the security price departed from the price “predicted” by this formula on days when company-specific information was released.  If the departures from the predicted price are large enough, they would be found “statistically significant” and the expert would opine that the market had characteristics showing efficiency and/or that released information had a specified impact on the price of the security.  

The key word here is “estimate,” because this type of analysis proceeds with a high degree of uncertainty.  The regression analysis looks at a lot of data points and tries to map out a formula (or “model”) of how the dependent variable is affected by the independent variables that comes close to matching the data points.  (Often that model tries to map a straight line through the data points which minimizes how far the data points are from the line, but the same analysis can be used with other possible models such as logarithmic or parabolic relationships between the independent and dependent variables).  One interesting measure, called the “coefficient of determination” or “r-squared,” shows how well the data actually fit the model.  It provides a measure of how well observed outcomes are replicated by the model, i.e., what proportion of total variation of outcomes is, or is not, “explained” by the formula.  Strikingly, many of these analyses in securities cases result in a r-squared in the range of 0.2 to 0.3, which means that the model relied on by the expert for his or her conclusions only explains 20% to 30% of stock price changes, leaving 70% to 80% of the stock price movement unexplained by the analysis.

Not only is there a high degree of uncertainty because of unexplained price movement, but there are many judgmental decisions about how to go about building this “model.”  These would include what time period to use as the “control” period; what independent variables to use; what to do about dates during the control period when there were company-specific disclosures that could impact security price; what period of time after a disclosure to use to determine whether or not there was a price impact (i.e., does the price impact have to occur within one minute of a disclosure, one month of the disclosure, or somewhere in between?); how to attempt to differentiate between the impact of the tested disclosure and other simultaneous events (e.g., disclosures not alleged to be inaccurate) that could be the cause of price movement, etc.  Different choices will often yield different expert opinions.  The courts are just beginning to develop standards by which to judge the efficacy of these kinds of expert opinions.  They will have to become much more proficient at this in the post-Halliburton II era.

It is in this context that Judge Beverly Reid O’Connell waded into the competing expert reports addressing plaintiffs’  motion to certify a class in Brown v. China Integrated Energy.  Plaintiffs presented experts to support their contention that the company’s stock traded on an efficient market.  The court rejected one of the two plaintiff experts for lack of expertise – it turned out that he had nearly zero prior experience addressing the market efficiency issue.  The court permitted expert evidence from plaintiff’s other expert but found his methodology so flawed that his expert opinion on the issue was essentially worthless.  The flaw in his methodology was in how he conducted his event study.  The defense (through its own expert) convinced the judge that key decisions in constructing the event study model were totally subjective, unsupported by any governing standards, and thus “unreliable.”  He used an “entirely subjective approach to selecting events to include” in the event study.  Even though he testified that the conclusions would not have been altered  by changes in the events used, the court found that he “did not follow commonly accepted methods in performing his event study” and therefore “his declaration and testimony are unreliable and should be excluded.”  Without that evidence, plaintiffs could not sustain their burden under Basic and Halliburton II of showing the stock traded on an efficient market, thereby triggering the “fraud on the market” presumption of reliance that would make the claims certifiable as a class action.

Securities lawyers out there had better brush off their college statistics books and learn more about regression analysis.  Especially following the decision in Halliburton II, these will be the front lines in securities class action litigation for some time to come, and federal judges will need a lot of help to determine when putative experts give reliable evidence in this arcane area.

Straight Arrow

August 5, 2014

Contact Straight Arrow privately here, or leave a public comment below: