NVIDIA

In 1995, Congress passed the Private Securities Litigation Reform Act (PSLRA), which dramatically heightened the pleading burden for plaintiffs bringing securities fraud cases.  At the same time, the PSLRA also instituted a mandatory stay on discovery until resolution of any motions to dismiss, which means plaintiffs have to use their own investigation – relying on public information, confidential sources, and the like – to draft a complaint that is sufficiently particular to satisfy PSLRA standards.

In 2002, Steve Bainbridge and Mitu Gulati published How Do Judges Maximize? (The Same Way Everybody Else Does – Boundedly): Rules of Thumb in Securities Fraud Opinions.  The paper explained that, given the high pleading standards of the PSLRA, judges deciding motions to dismiss lighten the workload by coming up with various rules of thumb for determining whether a complaint pleads materiality or scienter.  For example, they identified the puffery doctrine (presuming that investors treat vague statements of optimism as immaterial), the bespeaks caution doctrine (predictions of the future are immaterial if they are caveated by warnings of future uncertainty), and fraud by hindsight (refusing to draw inferences about what the company knew at an earlier time due to negative disclosures at a later time), as new bright line rules that judges employ to dismiss complaints, disconnected from the general fact pattern.

I can add a bunch more.  For example, a refusal to infer knowledge from one’s position in the company (except in generally rare instances where the “core operations” doctrine kicks in), a refusal to treat bonuses and similar compensation as a motive for fraud, a refusal to treat corporate departures as indicative of fraud unless there are “particularized allegations connecting the departures to the alleged fraud,” In re Hertz Global Holdings, 905 F.3d 106 (3d Cir. 2018), and a refusal to treat insider trades by non defendants as indicative of scienter (an issue I blogged about here).  These are more “rules” for complaints that, I’d argue, contradict our common sense understanding of how to go about inferring other people’s states of mind on a day to day basis just, you know, as a consequence of living as human beings on this planet.

Additionally, a couple of years ago, I published Fact or Fiction: Flawed Approaches to Evaluating Market Behavior in Securities Litigation, which identified still more rules of thumb that judges have developed over time.  For example, when addressing loss causation, courts often require revelation of the fact of falsity (rather than a revelation of problems traceable to the fraud), and will further hold that notice of an investigation, without more, categorically cannot cause a loss. 

You see the problem.  The question whether a complaint pleads falsity or a strong inference of scienter or loss causation or materiality is a holistic factual inquiry; but as these rules build up, pleading or litigating a securities case becomes about whether plaintiffs have complied with an increasingly arcane set of rules.  I like to quote then-Chief Judge William Young from 1999, describing the PSLRA as a “Byzantine pleading code …for securities actions.”  In re Number Nine Visual Tech. Corp., 51 F. Supp. 2d 1 (D. Mass. 1999).

Which is what I was thinking about when going over NVIDIA’s opening brief in NVIDIA Corporation v. E. Ohman J:or Fonder AB, currently pending before the Supreme Court. 

The plaintiffs pled falsity by relying, in part, on an expert report that drew inferences based on market information.  The assumptions and inferences on which the report was based were in the complaint.  Nonetheless, NVIDIA seeks to have the Supreme Court adopt a bright line rule that an expert opinion does not constitute an allegation of fact sufficient to satisfy PSLRA pleading standards.  NVIDIA also challenges the market evidence and factual assumptions that underlay the report as too generic.  Finally, NVIDIA admits that sometimes expert reports may be helpful, but here, the complaint rests too much on the expert report alone, so that the other facts surrounding it are not sufficient.

I see no way the Supreme Court can give NVIDIA what it wants without either acting as a district court and weighing this specific complaint and these specific allegations without regard for future cases, or - more likely - adding more artificial rules to securities pleadings.  Expert reports are still pretty rare in complaints, but if the rule is you can’t have them be the sole support of a complaint, then we’ll see endless litigation over what counts as “sole” (especially since, in this very case, the plaintiffs offer additional allegations).  And if plaintiffs can’t use expert reports that draw inferences based on detailed market information, can they use short seller reports that do the same?  (The Ninth Circuit, for example, has held that short seller reports analyzing public information may be a basis for alleging loss causation if they involve “extensive and tedious research involving the analysis of far-flung bits and pieces of data.” In re BofI Holding, Inc. Securities Litigation, 977 F.3d 781 (9th Cir. 2020)).

If plaintiffs can’t use short seller reports, what happens if the plaintiffs’ attorneys draw inferences based on public information – does that go too far?  Suddenly what started out as a simple rule about experts becomes whole new categories of prohibited inferences being added to the list.  And ruleification ends up obscuring the ultimate inquiry, namely, whether there is a sufficient basis to infer that the defendants’ statements were false.

There’s a similar issue with respect to the other question presented in this case, which is about pleading scienter.  In NVIDIA, the plaintiffs offer a lot of circumstantial evidence of scienter, plus make allegations regarding the existence of documents allegedly reviewed by the defendants, but the plaintiffs haven’t seen the documents so they can’t plead their exact contents.  NVIDIA wants the Court to adopt a bright line rule that the contents of the documents must be described, because “Plaintiffs built their entire scienter case around NVIDIA’s internal documents and data.  By choosing this route, Plaintiffs were required by the PSLRA’s pleading standards to allege with particularity what those documents and sources said and how they supported Plaintiffs’ preferred inference of scienter.”  Which means, were the Court to accept NVIDIA’s argument, the next step is litigation over what counts as the “entire” case, and how much description of a document is enough, and how much support a generally-described document can provide when there are other allegations of scienter.  The argument becomes about the contours of the rule, not whether the substantive standard for pleading scienter is met.

Securities complaints are now hundreds of pages long – and sometimes, ironically, dismissed for being too long and confusing, because Goldilocks-like, they must be just right.  See, e.g., Macovski v. Groupon, Inc., 2021 WL 1676275 (N.D. Ill., Apr. 28, 2021).  (Apparently, plaintiffs must also take a class in narrative exposition before they can survive a motion to dismiss.)  Selecting a lead plaintiff takes months, drafting an amended complaint typically is 60 days or so after that, plus another 60 for the motion to dismiss, 60 for the opposition, and 30-60 for a reply – weeks or months before an oral argument and god knows how long before a decision, and all of this before there’s been any discovery.  By the time you get to depositions, witnesses will be able to say “I don’t remember” and be completely credible.  The last thing this system needs is more up front rules about what plaintiffs can and cannot plead.