Saturday, August 1, 2015
The Halliburton district court finally issued its decision (.pdf) on the plaintiffs’ renewed motion for class certification – and I’m afraid it’s exactly as incoherent as the most pessimistic predictions might have anticipated.
The district court recognized that, after Halliburton I, it was prohibited from making loss causation determinations as part of the class certification inquiry. However the district court did hold that if there is no price movement in response to an alleged disclosure (and there is also no price movement when the misstatement is first made), that fact establishes there was no artificial inflation originally.
In other words, the court believed that the absence of affirmative evidence of price inflation is evidence of absence, and sufficient to carry the defendants’ burden to prove that any misstatements had no effect on prices. (To be fair, from the opinion, it appears the plaintiffs themselves urged this position).
The court went on to find that the Halliburton defendants had shown there was no price movement on almost all of the alleged corrective disclosure dates – and so class certification would be denied as to those dates. However, because there was price movement for one disclosure date – December 7, 2001, the last day of the proposed class period – class certification would be granted as to that date.
The problem is, the plaintiffs were seeking class certification for all persons who purchased Halliburton stock from the start of the fraud until the end of the fraud. The court’s ultimate determination – that one disclosure date was sufficient and others were not – tells the reader nothing about what class can be certified, as though the court had forgotten the entire purpose of the exercise. I suppose we’ll find out the class definition if an order follows, but for now, the opinion sheds almost no light on that subject.
[More under the jump]
(Obligatory disclaimer: I worked on amicus briefing in support of the plaintiffs for both of Halliburton’s trips to the Supreme Court.)
To understand the situation, it’s important to keep in mind that the Halliburton case consists of two sets of allegations of different frauds, both occurring at roughly the same time. (A third set of allegations was dropped from the motion for class certification.)
First, the plaintiffs allege that Halliburton misaccounted for certain customer receivables starting in mid-1999. As a result, its financial statements were false until – allegedly – the truth was revealed in December 2000, and Halliburton acknowledged losses.
Second, the plaintiffs allege that Halliburton understated its likely liability for certain asbestos-related claims at a subsidiary. The first misstatements were made in mid-1999, and Halliburton allegedly continued to falsely minimize its liabilities through May 2001. The truth was alleged revealed over time, in a series of announcements of new exposures and adverse jury verdicts stretching between June 2001 and December 2001.
Because the earliest alleged misstatements occurred in 1999, and the latest disclosure occurred in December 2001, the plaintiffs sought certification of a class of persons who purchased Halliburton stock between July 1999 and December 2001.
I will not summarize the entire travel of the Halliburton case and the various Supreme Court rulings on the subject – you can read my earlier posts here and here – but after Halliburton II, plaintiffs were entitled to a presumption that all of the alleged misstatements inflated Halliburton’s stock price (and thus, a that anyone who bought at the market price had “relied” on the fraud). To defeat class certification, the defendants would have to rebut the presumption that the misstatements had inflated Halliburton’s stock price.
So how do you prove a negative? How do you show that a misstatement did not inflate prices?
The court held – and the parties apparently agreed – that if there was no market reaction to the revelation of the truth, that would be enough to carry the defendant’s burden. I.e., the logic is, if the lie inflated the market price, we’d expect the truth to deflate it, and if that didn’t happen, obviously, the price was never inflated to begin with.
That’s actually faulty logic on several levels.
First, as has been extensively discussed elsewhere, see Alon Brav & J.B. Heaton, Event Studies in Securities Litigation: Low Power, Confounding Effects, and Bias, the tools we have for identifying price impact – event studies - are extremely imprecise. Among other things, they are particularly likely to yield false negatives when used in litigation.
Additionally, event studies cannot disprove that an announcement impacted a security’s price – they can only fail to rule out the possibility that the announcement did not have an effect on price. (That’s an issue that I believe Jill Fisch at Penn is currently writing about for the forthcoming Halliburton symposium issue of the Duke Journal of Constitutional Law and Public Policy, so I’ll just tell you to stay tuned).
Finally, when it comes to studying price reaction, a perennial issue is “how fast is fast enough?” The Supreme Court in Halliburton II was very clear that markets need not be perfectly efficient to justify a presumption of price impact; indeed, the Court recognized that markets may absorb information more or less quickly. Yet in its class cert decision, the Halliburton district court continued to insist on an extreme level of efficiency, by refusing to consider evidence that the corrective disclosures took up to 2 days – rather than a few minutes – to affect prices.
So that’s the first set of problems – how do you measure any kind of price impact.
But the second and larger issue is that the court’s entire analysis was conceptually flawed. Remember, the only reason even to look at the corrective disclosures is to draw the inference that the earlier misstatement impacted prices; for class certification purposes, that’s the only relevance that corrective disclosures have. And yet somehow, the Halliburton district court lost sight of this critical fact.
The court’s confusion is evident in its treatment of the asbestos claims. The original misstatements were alleged to have been made from 1999 through May 2001. According to the plaintiffs, the truth was revealed incrementally, on June 28 2001; August 9, 2001; October 30, 2001; December 4, 2001; and December 7, 2001. For various reasons, the court concluded that there had been no price reaction on any of these dates except for the final one, on December 7, 2001. And so it issued the following ruling: “the Court GRANTS in part Plaintiffs’ Motion for Class Certification, only with respect to the alleged corrective disclosure of December 7, 2001.”
Which only begs the question: what exactly is the court’s finding as to the impact of the original misstatements?
The court doesn’t say. It simply holds that it will deny certification as to all of the plaintiffs’ alleged corrective disclosures, except for the corrective disclosure on December 7.
So ... is there price impact from the original misstatements or not? The misstatements were made way back at the beginning of the class period and lasted for years; the court agreed that there was a disclosure on the last day of the class period that resulted in a price drop. So doesn’t that mean there must have been a price impact at all times prior to that date? If so, what is the relevance – for class certification purposes – of the earlier disclosures? Does the court think that there was no price impact of the original misstatements until December 5, 2001? Does the court think that only plaintiffs who held through December 7, 2001 can stay in the class because the fraud was not revealed until then? If so, isn’t that a loss causation ruling rather than a price impact ruling – which the court explicitly acknowledged it was forbidden to make?
I have no idea who the court believes should remain in the class, or why, because having wandered down the garden path of corrective disclosures, the court entirely forgot the reason for its journey.
Which only highlights the fallacy of relying on corrective disclosures exclusively as a mechanism for identifying whether misstatements made much earlier had an impact on price. As the Supreme Court acknowledged in Halliburton I, there are lots of reasons why a misstatement might inflate prices at Time 1, but revelation of the truth would not result in deflation at Time 2. Most obviously, an intervening problem might drive the price down so far that new revelations become irrelevant. (Like, in extreme cases, a bankruptcy). Or maybe the market began to suspect the truth and the price slowly fell over time, such that the truth was already known at the time of the big reveal, and the corrective disclosure was not, in fact, a disclosure at all. Or maybe the plaintiffs are just wrong about whether a particular disclosure was revelatory. Maybe there was never a revelation of the truth. None of this tells you whether the price was initially inflated; at best, it creates an issue of fact as to how long the harm lasted – something that, whatever the answer, is true classwide.
To be sure, there might be situations where, upon a full consideration of the facts, the lack of price movement upon revelation of the truth does suggest there was no impact earlier. But that’s going to be a very contextual inquiry, requiring an analysis of the nature of the statement, any potential intervening events, the amount of time to elapse between the original misstatement and the disclosure, analyst/public comments about both the statement and the disclosure, and so forth. And most importantly, it requires that the court always be attentive to its ultimate inquiry: whether the original misstatement, not the subsequent disclosure, affected prices. None of which the court did here.
As for the plaintiffs’ second set of claims, concerning the customer receivables, the plaintiffs only alleged a single disclosure date. The court concluded that there was no price drop on that date, so it issued its ruling: “Defendants have rebutted the Basic presumption as to the allegedly corrective disclosure made on that date.”
I can only assume that what the court meant was that the associated misstatements regarding customer receivables had no impact on price, and therefore any claims pertaining to the associated misstatements could not be part of the class case. But the court didn’t say so explicitly, because it was only focused on corrective disclosures. And the varying evidence with the asbestos claims – one disclosure that caused a price drop, others that didn’t – should have given the court pause regarding the probative value of disclosure evidence in the first place. Sadly, it did not.
Anyway, when I look at this decision in the broader context of private securities actions, I cannot help but be struck by how all of our measures of impact and harm and loss are, at this point, so far removed from reality as to border on complete legal fiction. Materiality is a construct from case law, with numerous additional doctrines piled on to it by courts without any heed for actual evidence of how markets behave. Price impact is presumed for all lies, big or small, obvious or obscure. Courts cannot and do not take into account the very real possibility that the effects of a lie told on Day One will eventually dissipate if not repeated, simply because the information will become stale - and because loss causation has become so stylized (and god knows no one can measure damages), those concepts cannot make up the difference. The doctrine has not even begun to deal with problems like, say, disentangling the effects of multiple lies told over the course of a class period when only some are associated with scienter (and are thus actionable). In other words, what we call "harm" and "damage" for the purpose of private securities fraud lawsuits have become so artificial that it no longer seems as though we're even trying to measure the actual real-world effects of fraud. I believe private lawsuits are an essential supplement to SEC action but a system of fines or statutory damages would make so much more sense.