Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc.

77 F.4th 74

Opinion

22-484 Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc.

UNITED STATES COURT OF APPEALS FOR THE SECOND CIRCUIT

______________

August Term 2022

(Argued: September 21, 2022 | Decided: August 10, 2023)

Docket No. 22-484

ARKANSAS TEACHER RETIREMENT SYSTEM, WEST VIRGINIA INVESTMENT MANAGEMENT BOARD, PLUMBERS AND PIPEFITTERS PENSION GROUP,

Plaintiffs-Appellees,

PENSION FUNDS, ILENE RICHMAN, Individually and on behalf of all others similarly situated, PABLO ELIZONDO, THOMAS DRAFT, Individually and on behalf of all others similarly situated,

Plaintiffs,

HOWARD SORKIN, Individually and on behalf of all others similarly situated, TIKVA BOCHNER, Individually and on behalf of herself and all others similarly situated, DR. EHSAN AFSHANI, LOUIS GOLD, Individually and on behalf of all others similarly situated, Consolidated Plaintiffs,

v.

GOLDMAN SACHS GROUP, INC., LLOYD C. BLANKFEIN, DAVID A.

VINIAR, GARY D. COHN,

Defendants-Appellants

SARAH E. SMITH,

Consolidated Defendant. †

______________

Before:

WESLEY, CHIN, and SULLIVAN, Circuit Judges.

Shareholders of Defendant-Appellant Goldman Sachs Group, Inc. brought this class action lawsuit against Goldman and several of its former executives, claiming defendants committed securities fraud in violation of § 10(b) of the Securities Exchange Act of 1934 and Rule 10b–5 promulgated thereunder by misrepresenting Goldman’s ability to manage conflicts of interest in its business practices. After a number of appeals and subsequent remands, including an appeal to the Supreme Court, the district court once again certified a shareholder class under Federal Rule of Civil Procedure 23(b)(3).

For the reasons that follow, we reverse the district court’s class certification decision with instructions to decertify the class.

Judge Sullivan concurs in the result in a separate opinion.

_________________

ROBERT J. GIUFFRA, JR., Sullivan & Cromwell LLP, New York, NY (Richard H. Klapper, David M.J. Rein, Benjamin R. Walker, Julia A. Malkina, Jacob E. Cohen, Sullivan & Cromwell LLP, New York, NY; Morgan L. Ratner, Sullivan & Cromwell LLP, Washington, D.C., on the brief) for Defendants-Appellants.

KANNON K. SHANMUGAM, Paul, Weiss, Rifkind, Wharton & Garrison LLP, Washington D.C. (Audra J. Soloway, Paul, Weiss, Rifkind, Wharton & Garrison LLP, New York, NY, on the brief) for Defendants-Appellants.

THOMAS C. GOLDSTEIN, Goldstein & Russell, P.C., Bethesda, MD (Kevin K. Russell, Goldstein & Russell, P.C., Bethesda, MD;

† The Clerk of the Court is directed to amend the official caption as set forth above.

3

Spencer A. Burkholz, Joseph D. Daley, Robbins Geller Rudman & Dowd LLP, San Diego, CA; Thomas A. Dubbs, James W.

Johnson, Michael H. Rogers, Irina Vasilchenko, Labaton Sucharow LLP, New York, NY, on the brief) for Plaintiffs-Appellees.

Todd G. Cosenza, Willkie Farr & Gallagher LLP, New York, NY, for Amicus Curiae Former United States Securities and Exchange Commission Officials and Law Professors in Support of Defendants-

Appellants.

Carmine D. Boccuzzi, Jr. (Victor L. Hou, Jared Gerber, on the brief)

Cleary Gottlieb Steen & Hamilton LLP, New York, NY, for Amicus Curiae Economic Scholars in Support of Defendants-

Appellants.

Jonathan K. Youngwood, Simpson Thacher & Bartlett LLP, New York, NY (Craig S. Waldman, Joshua C. Polster, Daniel H.

Owsley, Simpson Thacher & Bartlett LLP, New York, NY; Ira D. Hammerman, Kevin Carroll, Securities Industry and Financial Markets Association, Washington, D.C.; Thomas Pinder, American Bankers Association, Washington, D.C.;

Gregg Rozansky, Bank Policy Institute, Washington, D.C.;

Tyler S. Badgley, U.S. Chamber Litigation Center, Washington, D.C.; Kenneth Stoller, American Property Casualty Insurance Association, Washington, D.C., on the brief) for Amicus Curiae Securities Industry and Financial Markets Association, Bank Policy Institute, American Bankers Association, Chamber of Commerce of the United States of America, American Property Casualty Insurance Association in Support of Defendants-Appellants.

Christopher E. Duffy, Vinson & Elkins LLP, New York, NY (Jeremy C. Marwell, James T. Dawson, Vinson & Elkins LLP, Washington, D.C.; Darla C. Stuckey, Randi V. Morrison, Society for Corporate Governance, New York, NY, on the brief)

4

for Amicus Curiae Society for Corporate Governance in Support of Defendants-Appellants.

Lyle Roberts, Shearman & Sterling LLP, Washington, D.C. (George Anhang, Shearman & Sterling LLP, Washington, D.C.; William Marsh, Shearman & Sterling LLP, Dallas, TX; Cory L. Andrews, John M. Masslon II, Washington Legal Foundation, Washington, D.C., on the brief) for Amicus Curiae Washington Legal Foundation in Support of Defendants-Appellants.

Ernest A. Young, Apex, NC (Jeremy Lieberman, Emma Gilmore, Pomerantz LLP, New York, NY; Andrew L. Zivitz, Kessler Topaz Meltzer & Check, LLP, Radnor, PA; Stacey M. Kaplan, Kessler Topaz Meltzer & Check, LLP, San Fransisco, CA, on the brief) for Amicus Curiae Financial Economists in Support of Plaintiffs-Appellees.

Deepak Gupta, Gupta Wessler PLLC, Washington, D.C. (Linnet R.

Davis-Stermitz, Gupta Wessler PLLC, Washington, D.C.;

Salvatore J. Graziano, Jai K. Chandrasekhar, Bernstein Litowitz Berger & Grossmann LLP, New York, NY; Joseph E. White, III, Saxena White P.A., Boca Raton, FL, on the brief) for Amicus Curiae Securities Law Scholars in Support of Plaintiffs-Appellees.

John Paul Schnapper-Casteras, Schnapper-Casteras PLLC, Washington, D.C., for Amicus Curiae Better Markets, Inc. in Support of Plaintiffs-Appellees.

Carolyn E. Shapiro, Schnapper-Casteras PLLC, Washington, D.C.

(John Paul Schnapper-Casteras, Schnapper-Casteras PLLC, Washington, D.C.; Daniel P. Chiplock, Lieff Cabraser Heimann & Bernstein, LLP, New York, NY, on the brief) for Amicus Curiae Former SEC Officials in Support of Plaintiffs-Appellees.

_________________

5

WESLEY, Circuit Judge:

This class certification dispute has been laboring in federal court for nearly a decade. It raises challenging questions about how defendants in securities fraud class actions, having lost a motion to dismiss, can rebut the legal presumption of reliance established in Basic Inc. v. Levinson, 485 U.S. 224 (1988), at the class certification stage. The case is before us again: for a third time, the district court certified, under Federal Rule of Civil Procedure 23(b)(3), a shareholder class, and, for a third time, we granted defendants leave to pursue an interlocutory appeal of that order under Rule 23(f).

Some context is required at the outset. The Basic presumption excuses classes of securities fraud plaintiffs from proving that each class member individually relied upon a defendant’s alleged misrepresentations. Courts can instead presume that stock trading in an efficient market incorporates into its price all public, material information—including material misrepresentations—and that investors rely on the integrity of the market price when they choose to buy or sell that stock. At the same time, defendants can rebut the presumption and defeat class certification by demonstrating, by a preponderance of the evidence, that the

6

misrepresentations did not actually affect, or impact, the market price of the stock. This legal terrain under Basic is familiar, and, in this appeal, uncontested.

From there, however, the journey becomes difficult. Analyzing whether a defendant has proved a lack of price impact is complicated by the fact that a misrepresentation can affect a stock’s price either by causing the stock to trade at an inflated price, or as is alleged here, by maintaining inflation that is already built into the stock price. See In re Vivendi, S.A. Sec. Litig., 838 F.3d 223, 258 (2d Cir. 2016). In the latter scenario, the misrepresentation prevents preexisting inflation in a stock price from dissipating, but does not cause a price uptick. Instead, the back-end price drop—what happens when the truth is finally disclosed—operates as an indirect proxy for the front-end inflation, or the amount that the misrepresentation fraudulently propped up the stock price. Simply put, the theory goes: back-end price drop equals front-end inflation.

Fair enough. But what happens when the match between the contents of the price-propping misrepresentation and the truth-revealing corrective disclosure is tenuous? Consider two examples. In the first, an automobile manufacturer’s earlier statement to the market that its best-selling vehicle passed all safety tests is followed by later news that, in fact, the car failed several crash

7

tests. A price drop follows. There, the earlier statement is precisely negated, or rendered false, by the later news—a clean match. In the second example, however, the same back-end news (and the same price drop) is instead preceded by the manufacturer’s statement to the market that it strives to ensure that all its vehicles are road-ready, that it has an elaborate testing protocol to that effect, but that the task is tall, the goal difficult to achieve. There, it is less apparent the market would understand the later news of failed crash tests revealed that, in fact, there was no protocol, or that, in fact, the manufacturer did not seek to make its automobiles safe to drive. The match between the more specific “corrective disclosure” and the earlier, more generic statement is on shakier ground. Can courts still infer that the back-end price drop equals the front-end inflation?

The Supreme Court answered that commonsense question. It explained that the “inference [] that the back-end price drop equals front-end inflation [] starts to break down” when the earlier misrepresentation is generic and the later corrective disclosure is specific, and that, “[u]nder those circumstances it is less likely that the specific disclosure actually corrected the generic misrepresentation . . . .” Goldman Sachs Grp., Inc. v. Ark. Tchr. Ret. Sys. (Goldman), 141 S. Ct. 1951, 1961 (2021).

8

Following Goldman, courts are now directed to compare, at the class certification stage, the relative genericness of a misrepresentation with its corrective disclosure, notwithstanding that such evidence is often also highly relevant to the closely related merits question of whether the misrepresentation would have been material to a shareholder’s investment calculus—which, under other Supreme Court guidance, a court may not resolve at class certification. See Amgen Inc. v. Conn. Ret. Plans & Tr. Funds, 568 U.S. 455 (2013). In short, Goldman’s mismatch framework requires careful trekking: district courts must analyze the price impact issue without drawing what might appear to be obvious conclusions for off-limits merits questions such as materiality. As Judge Hamilton, writing for the Seventh Circuit, put it, courts must analyze this issue “without . . . thinking about a pink elephant.” In re Allstate Corp. Sec. Litig, 966 F.3d 595, 602 (7th Cir. 2020).

The question in this case is whether, in applying the Supreme Court’s mismatch framework, the district court clearly erred in finding that Goldman failed to rebut the Basic presumption by a preponderance of the evidence, and, therefore, abused its discretion by certifying the shareholder class. It did.

9

Accordingly, we reverse the district court’s order and remand with instructions to the district court to decertify the class.

BACKGROUND

Factual Background The facts underlying this lawsuit have been discussed at length in our prior opinions, see, e.g., Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc. (ATRS I), 879 F.3d 474, 478 (2d Cir. 2018), but are nonetheless recounted here.

Plaintiffs-appellees are individuals and institutions who acquired shares in The Goldman Sachs Group, Inc. between February 5, 2007, and June 10, 2010 (the “Class Period”). Their claims are being pursued by three pension funds—the lead plaintiffs—each of which acquired Goldman common stock within the same period. Plaintiffs filed a consolidated class action complaint in July 2011 against Goldman and a handful of its former executives (collectively, “Goldman” or “defendants”), accusing Goldman of violating Section 10(b) of the Securities Exchange Act and Rule 10b–5 promulgated thereunder. See 15 U.S.C. § 78j(b); 17 C.F.R. § 240.10b–5. Plaintiffs allege defendants made material misrepresentations about Goldman’s business practices and its approach to conflicts-of-interest management.

10

The Challenged Statements The alleged misrepresentations generally fall into two categories. First, plaintiffs point to statements relating to Goldman’s business principles, which were included in the company’s annual report to shareholders and made by Goldman executives at various conferences:

• We are dedicated to complying fully with the letter and spirit of the laws, rules and ethical principles that govern us. Our continued success depends upon unswerving adherence to this standard.

• Most importantly, and the basic reason for our success, is our extraordinary focus on our clients.

• Our clients’ interests always come first. Our experience shows that if we serve our clients well, our own success will follow.

• Integrity and honesty are at the heart of our business.

Joint Appendix (“J.A.”) at 97.

Second, plaintiffs challenge statements contained in the “Risk Factors”

portion of Goldman’s Form 10-K, filed every year during the Class Period with the Securities Exchange Commission (“SEC”), concerning the management of conflicts of interest. With respect to this conflicts disclosure, plaintiffs focus on the emphasized language below:

11

Conflicts of interest are increasing and a failure to appropriately identify and deal with conflicts of interest could adversely affect our businesses.

Our reputation is one of our most important assets. As we have expanded the scope of our businesses and our client base, we increasingly have to address potential conflicts of interest, including situations where our services to a particular client or our own proprietary investments or other interests conflict, or are perceived to conflict, with the interests of another client, as well as situations where one or more of our businesses have access to material nonpublic information that may not be shared with other businesses within the firm.

The SEC, the NYSE, FINRA, other federal and state regulators and regulators outside the United States, including in the United Kingdom and Japan, have announced their intention to increase their scrutiny of potential conflicts of interest, including through detailed examinations of specific transactions. There have been complaints filed against financial institutions, including Goldman Sachs, alleging the violation of antitrust laws arising from their joint participation in certain leveraged buyouts, referred to as “club deals,” as discussed under “Legal Proceedings—Private Equity-Sponsored Acquisitions Litigation” in Part I, Item 3 of the Annual Report on Form 10-K. In addition, a number of class action complaints have also been filed in connection with certain specific “club deal” transactions which name the relevant “club deal” participants among the defendants, including Goldman Sachs affiliates in several cases, and generally allege that the transactions constitute a breach of fiduciary duty by the target company and that the “club” participants aided and abetted such breach. We cannot predict the outcome of the litigation to which we are a party, and we may become subject to further litigation or regulatory scrutiny in the future in this regard.

12

We have extensive procedures and controls that are designed to identify and address conflicts of interest, including those designed to prevent the improper sharing of information among our businesses. However, appropriately identifying and dealing with conflicts of interest is complex and difficult, and our reputation could be damaged and the willingness of clients to enter into transactions in which such a conflict might arise may be affected if we fail, or appear to fail, to identify and deal appropriately with conflicts of interest. In addition, potential or perceived conflicts could give rise to litigation or enforcement actions.

J.A. 3278 (emphasis added). 1 It is undisputed that the challenged statements did not cause statistically significant increases in Goldman’s stock price. Instead, plaintiffs say, the statements maintained an already-inflated stock price. According to plaintiffs, that balloon popped when news of undisclosed conflicts of interest revealed the falsity of the challenged statements and caused the stock to drop.

The Corrective Disclosures

1 Plaintiffs also include a footnote in their brief to remind us they challenged in their complaint a December 2009 Goldman press release—issued in response to a December 24, 2009, New York Times article detailing Goldman’s questionable business practices—in which Goldman asserted, inter alia, that “its CDOs were fully disclosed and well known to [CDO] investors.” Appellees Br. at 45 n.7 (quoting Compl. ¶ 124). Plaintiffs make little attempt to flesh out their theory of liability based on this statement, perhaps because the district court previously rejected plaintiffs’ claim for relief based on the press release. See Richman v. Goldman Sachs Grp., Inc., 868 F. Supp. 2d 261, 274 (S.D.N.Y. 2012). In any event, we generally regard an argument as waived when it appears only in a footnote. See United States v. Botti, 711 F.3d 299, 313 (2d Cir. 2013).

13

Specifically, plaintiffs target three dates in 2010 when they claim the false nature of the business principles and conflicts disclosure statements was revealed to the market. Broadly, they focus on what they characterize as the disclosure of concealed conflicts of interest infecting several collateralized debt obligation (“CDO”) transactions involving subprime mortgages. In essence, they allege that, publicly, Goldman touted various CDOs as long-term investment opportunities to investors when, in fact, Goldman was betting on them to fail.

First, and featured most heavily throughout this litigation, on April 16, 2010, the SEC initiated an enforcement action against Goldman and one of its employees regarding a CDO transaction known as Abacus 2007 AC-1 (the “Abacus Complaint”). See generally Press Release, SEC, Goldman Sachs to Pay Record $550 Million to Settle SEC Charges Related to Subprime Mortgage CDO (July 15, 2010), https://www.sec.gov/news/press/2010/2010-123.htm. The SEC accused Goldman and its employee of committing securities fraud. It targeted Goldman’s failure to disclose in its marketing materials to various institutional customers that the hedge fund Paulson & Co. played an active role in the CDO’s asset selection process, and for telling those investors that Paulson held a long interest in the

14

Abacus CDO when, in fact, Paulson was short. The next day, Goldman’s stock price declined 12.79% from $184.27 to $160.70 per share.

Second, on April 30, 2010, Goldman’s stock price dropped another 9.39% following a report from The Wall Street Journal that Goldman was under investigation by the Department of Justice (“DOJ”) for its purported role in unspecified CDOs. Finally, on June 10, 2010, various media outlets reported that the SEC was investigating Goldman’s conduct in another transaction, Hudson Mezzanine Funding 2006; a further 4.52% decline in the price of Goldman stock followed.

Neither the DOJ nor the SEC took further action related to the second two corrective disclosures. As to the first corrective disclosure, the Abacus Complaint culminated in a consent judgment under which Goldman agreed to pay $550 million, and, without “admitting or denying the allegations in the complaint . . . acknowledge[d]” that the Abacus marketing materials were “incomplete” and that it was a “mistake” for Goldman to state that the reference portfolio was “selected by” ACA Management LLC “without disclosing the role of Paulson.” J.A. 665.

15

In plaintiffs’ view, these corrective disclosures revealed to the market that Goldman’s statements about its conflicts management practices and business principles were false. Goldman, they say, lied about having extensive practices and procedures in place to manage its conflicts of interest, or otherwise knowingly failed to disclose mishaps in their conflicts protocol. 2 As a result of Goldman’s fraud, plaintiffs claim that they lost over $13 billion.

2Plaintiffs’ theory of falsity has evolved throughout this lawsuit. For example, although plaintiffs alleged in their complaint that “Goldman’s warnings to shareholders regarding potential conflicts of interest omitted the fact that it was indeed aware of the existence of such conflicts at the time,” J.A. 53, plaintiffs’ counsel appeared to abandon that theory at oral argument, acknowledging that “everybody knew that [Goldman] had conflicts,” Oral Arg. Audio at 1:11:30, Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc (No 22-484). On appeal, plaintiffs instead claim that the filing of the Abacus Complaint revealed to the market that “Goldman doesn’t have effective practices and procedures in place” to manage conflicts. Id. at 48:30. They press the same argument in their brief. See Appellees Br. at 44, 57. It strains credulity to say that the corrective disclosures revealed to the market that Goldman lied about having extensive procedures and controls designed to address conflicts of interest. The district court did not make such a finding. Nor does the post-disclosure market commentary offered by plaintiffs come anywhere close to supporting that inference; no report cited by them questions the extensiveness of Goldman conflicts procedures. Dr. Finnerty, plaintiffs’ class certification expert, did not espouse that view. The record, in short, provides no support for that theory. Accordingly, we consider as plaintiffs’ theory that the challenged statements were misleading because Goldman failed to disclose, in choosing to speak on its business practices and, in particular, its approach to conflicts management, that it was actively mismanaging conflicts—a theory plaintiffs have offered throughout this litigation, and which the district court considered. See, e.g., Special Appendix (“S.A.”) at 17; J.A. 4707.

16

Litigation History 1. Goldman’s Motion to Dismiss Much of the early action in this case proceeded in line with a typical securities litigation. Following the filing of plaintiffs’ complaint, Goldman moved to dismiss under Federal Rules of Civil Procedure 9(b) and 12(b)(6). In pertinent part, it pressed a materiality argument: the alleged misrepresentations, Goldman argued, were too vague and general for a reasonable shareholder to have relied on them in determining the value of Goldman’s stock. Thus, it continued, those statements did not influence plaintiffs’ investment decision-making, and any loss they suffered was unrelated to them.

The district court saw it differently. Although it agreed that some of Goldman’s statements were immaterial as a matter of law—and dismissed the complaint to the extent it relied upon those statements—it held that the business principles and conflicts statements were not “so obviously unimportant to a reasonable investor” as to be immaterial as a matter of law. Richman v. Goldman Sachs Grp., Inc., 868 F. Supp. 2d 261, 271, 280 (S.D.N.Y. 2012). With respect to those statements, the district court denied Goldman’s motion to dismiss, and thereafter denied Goldman’s motions for reconsideration of, and an interlocutory appeal

17

from, that order. See In re Goldman Sachs Grp., Inc. Sec. Litig., No. 10 Civ. 3461, 2014 WL 2815571, at *6 (S.D.N.Y. June 23, 2014) (reconsideration); In re Goldman Sachs Grp., Inc. Sec. Litig., No. 10 Civ. 3461, 2014 WL 5002090, at *3 (S.D.N.Y. Oct. 7, 2014) (interlocutory appeal).

2. Class Certification Having survived defendants’ threshold attack, plaintiffs moved to certify a class of shareholder plaintiffs. Class certification under Federal Rule of Civil Procedure 23 is dictated by certain requirements, many of which are not at issue here. To the point, Goldman did not dispute that (1) the named plaintiffs’ class is so numerous that joinder is impracticable, (2) at least one question of law or fact is common to the class, (3) the class representatives’ claims are typical of the class wide claims, and (4) the class representatives, here, the pension funds, will be able to fairly and adequately protect the interests of the class. See Fed. R. Civ. P. 23(a). Goldman did, however, maintain that plaintiffs failed to satisfy Rule 23’s additional hurdle for classes primarily seeking money damages. That requirement, set forth in Rule 23(b)(3), demands that common questions of law or fact predominate over individual questions that pertain only to certain class members.

18

In this lawsuit, Rule 23(b)(3)’s predominance requirement has been, and in this appeal remains, center stage. Under the Rule, analysis of whether questions of law or fact common to class members predominate “begins, of course, with the . . . underlying cause of action.” Erica P. John Fund, Inc. v. Halliburton Co. (Halliburton I), 563 U.S. 804, 809 (2011). Like many securities scrap-ups, the parties here join issue on the element of reliance—that is, whether plaintiffs relied upon the alleged misrepresentations. 3 As previewed above, to satisfy their class certification obligation of demonstrating class-wide reliance, plaintiffs invoked the Basic presumption, asking the district court to presume that all class members relied upon defendants’ misstatements, as reflected in its price, in choosing to buy Goldman stock. 4

3The six elements of securities fraud are “(1) a material misrepresentation or omission by the defendant; (2) scienter; (3) a connection between the misrepresentation or omission and the purchase or sale of a security; (4) reliance upon the misrepresentation or omission; (5) economic loss; and (6) loss causation.” Halliburton I, 563 U.S. at 810 (internal citations omitted). 4For purposes of this appeal, the parties do not dispute that the other prerequisites to the Basic presumption are satisfied, that is, that defendants’ purported misstatements were publicly known, its shares traded in an efficient market, and plaintiffs purchased the shares at the market price after the misstatements were made but before the truth was revealed. Additionally, although materiality is an additional prerequisite under Basic, class members need not prove it prior to class certification. See Halliburton Co. v. Erica P. John Fund, Inc. (Halliburton II), 573 U.S. 258, 276 (2014).

19

Again, Basic rests on what is referred to as the “fraud-on-the-market”

theory—that a stock trading on theoretically efficient markets like the New York Stock Exchange or Nasdaq, incorporates all public, material information, including material misrepresentations, into its share price. Basic, 485 U.S. at 246. More simply, the misrepresentation—the fraud—is “on the market.” See id. Without the Basic presumption, classes pursuing claims of securities fraud would face the onerous task of demonstrating each class member was aware of, and bought the company’s stock based on, an alleged misrepresentation. That burden would splinter classes along class member-specific lines, undermining the purpose of the class action device, and all but dooming securities claims from proceeding under Rule 23. Basic is therefore a saving grace for classes: they need not directly prove that the defendant’s statements impacted its share price. Instead, satisfaction of Basic’s prerequisites serves as an “indirect proxy” for a showing of price impact. See Halliburton II, 573 U.S. at 278–81.

Importantly, however, the presumption is rebuttable. “[A]n indirect proxy should not preclude . . . a defendant’s direct, more salient evidence showing that the alleged misrepresentation did not actually affect the stock’s market price and, consequently, that the Basic presumption does not apply.” Id. at 281–82 (emphasis

20

added). Throughout what the district court aptly characterized as a “prolonged interlocutory appeals saga,” In re Goldman Sachs Grp., Inc. Sec. Litig., 579 F. Supp. 3d 520, 522 (S.D.N.Y. 2021), Goldman has steadfastly attempted to do just that.

a. Round One: Goldman’s First Appeal In its initial response to plaintiffs’ Rule 23 motion, Goldman laid the groundwork for the evidence that, in the present appeal, it continues to rely on to show an absence of price impact.

Goldman introduced, first, an event study 5 conducted by its chief price impact expert, Dr. Paul Gompers, demonstrating that the business principles statements and conflicts disclosure did not cause a significant uptick in Goldman’s stock price. Second, Goldman identified 36 dates—all prior to the corrective

5 As we previously explained, an event study “isolates the stock price movement attributable to a company (as opposed to market-wide or industry-wide movements) and then examines whether the price movement on a given date is outside the range of typical random stock price fluctuations observed for that stock.” Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc. (ATRS II), 955 F.3d 254, 261 n.4 (2d Cir. 2020), vacated and remanded, 141 S. Ct. 1951 (2021) (citing Mark L. Mitchell & Jeffry M. Netter, The Role of Financial Economics in Securities Fraud Cases: Applications at the Securities and Exchange Commission, 49 BUS. LAW. 545, 556–69 (1994)); In re Vivendi, 838 F.3d at 253–56. If the isolated stock price movement falls outside the range of typical random stock price fluctuations, it is statistically significant. ATRS II, 955 F.3d at 261 n.4. If the stock price movement is indistinguishable from random price fluctuations, it cannot be attributed to companyspecific information announced on the event date. See id.

21

disclosure dates—on which media outlets discussed, in varying degrees of detail, transactions which, according to the reports, raised questions about Goldman’s ability to manage conflicts of interest.

Goldman’s view on the significance of these pre-disclosure reports was fleshed out by Dr. Gompers. He explained that the pre-disclosure reports implicated the same topics covered by the challenged statements and, just like plaintiffs’ alleged corrective disclosures, called the reliability of the challenged statements into question. Building from there, Dr. Gompers opined that because these pre-disclosure reports—viewed by him as alternative corrective disclosures—caused no statistically significant price decrease, the price drop that did occur following plaintiffs’ offered corrective disclosures must have been caused by something other than any corrective effect that they had upon the challenged statements.

Goldman relied on another expert, Dr. Stephen Choi, to press an alternative explanation. Dr. Choi conducted an event study focusing on the first corrective disclosure, the April 2010 filing of the SEC’s Abacus Complaint. He pointed to qualities of that enforcement action—so-called “severity factors”—which, in his view, accounted for the entirety of the price decline that followed. To buttress that

22

opinion, he identified four out of a group of 117 enforcement events bearing similar qualities, whose announcements to the market resulted in significant drops in those companies’ stock prices. Goldman relied on Dr. Choi’s submissions to argue that the price drop in April 2010 was caused entirely by the news of the enforcement action itself, rather than the revelation of Goldman’s client conflicts.

Of course, plaintiffs countered defendants at every turn. They did so through their sole expert, Dr. John D. Finnerty, who, as discussed in more detail below, disputed the methods and conclusions of Goldman’s experts.

The district court disagreed with Goldman and certified the class. See In re Goldman Sachs Group, Inc. Securities Litig., No. 10 Civ. 3461, 2015 WL 5613150 (S.D.N.Y. Sept. 24, 2015). In relevant part, the district court discredited Dr. Gompers’ event study, observing that under plaintiffs’ inflation-maintenance theory, the challenged statements could have maintained, rather than caused, an already inflated stock price. Id. at *6. It also declined to consider Goldman’s evidence regarding the pre-disclosure reports, concluding that such evidence was either “an inappropriate truth on the market defense” or an argument for materiality that the court “w[ould] not consider” at the class certification stage. Id. (internal quotation marks omitted). Finally, it found Dr. Choi’s submissions

23

unconvincing, explaining that alternative explanations regarding the cause of the price declines did not rule out that the corrective effect of each disclosure on the challenged statements may have been a contributing cause. Id.

Ultimately, the district court held that while a defendant can rebut the Basic presumption by a preponderance of the evidence, Goldman had failed to do so because it did not provide “conclusive evidence that no link exists between the price decline [of Goldman’s stock] and the misrepresentations.” Id. at *4 n.3, *7.

ATRS I. The first time this case arrived at our doorstep, we vacated and remanded. Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc. (ATRS I), 879 F.3d 474 (2d Cir. 2018). We held that defendants seeking to rebut the Basic presumption must do so by a preponderance of the evidence, and that it was unclear whether the district applied that standard. Id. at 485.

Second, we held it was error for the district court to conclude that it could not consider the pre-disclosure reports. Id. We also encouraged the district court to hold an evidentiary hearing, which, in advance of its initial class certification decision, it had deemed unnecessary. Id. at 486.

24

b. Round Two: We Affirm On remand, the district court received supplemental briefing, held a class certification evidentiary hearing, and, ultimately, certified the class a second time. In re Goldman Sachs Grp., Inc. Sec. Litig., No. 10 Civ. 3461, 2018 WL 3854757, at *2 (S.D.N.Y. Aug. 14, 2018). Defendants called Drs. Gompers and Choi, who offered testimony in line with their expert submissions. Plaintiffs, meanwhile, called Dr. Finnerty, who, consistent with his submissions, offered rebuttals to defendants’ experts.

In the end, the district court again credited Dr. Finnerty’s opinion that the alleged misrepresentations maintained an already-inflated stock price, finding that he had established a causal link between the alleged misrepresentations and the price decline following the three alleged corrective disclosures. Id. at *4.

Defendants’ experts, it continued, did not sufficiently sever that link. In pertinent part, the district court distinguished the pre-disclosure reports from plaintiffs’ corrective disclosures; it found that although the former may have reported and suggested “Goldman’s conflicts in the ABACUS deal, the ABACUS Complaint was the first to detail it.” Id. at *4. Those details—and the fact that the charges were brought by Goldman’s principal regulator—“obviously rendered the

25

[Abacus Complaint] more reliable and credible than any of the 36 media reports . . . .” Id.

As for Dr. Choi’s event study, the district court again largely discounted it.

It noted that the study concerned only the Abacus Complaint, but not the second and third corrective disclosures, and that, in any event, the severity factors were arbitrary and not well-established methods of measurement. It concluded that defendants had failed to rebut the Basic presumption. Id. at *5–6.

ATRS II. We granted Goldman leave to pursue another interlocutory appeal, and, ultimately, affirmed. See Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc. (ATRS II), 955 F.3d 254 (2d Cir. 2020), cert. granted, 141 S. Ct. 950 (2020), and vacated and remanded, 141 S. Ct. 1951 (2021).

That time, however, Goldman principally pressed a hardline rule: general statements, as a matter of law, are incapable of maintaining inflation in a stock price. Id. at 266. We disagreed; in our view, Goldman’s proposed rule too closely resembled a materiality analysis, which, as we then understood Supreme Court precedent, was off-limits at the class certification stage. Id. at 269.

We also concluded that the district court did not abuse its discretion in certifying the class. Id. at 274. Goldman primarily took issue with the district

26

court’s analysis of the 36 dates of pre-disclosure reporting, but we found no clear error in the district court’s findings.

Judge Sullivan dissented. He would have accorded more weight to those pre-disclosure reports, which he said demonstrated that when the market learned about Goldman’s conflicts, it did not negatively react. See id. at 278 (Sullivan, J., dissenting). In his view, “the generic quality of Goldman’s alleged misstatements, coupled with the undisputed fact that Goldman’s stock price did not move on any of the 36 dates on which the falsity of the alleged misstatements was revealed to the public, clearly compels the conclusion that the stock drop following the corrective disclosures was attributable to something other than the misstatements alleged in the complaint.” See id. at 278–79 (Sullivan, J., dissenting) (internal quotation marks and citation omitted).

c. The Supreme Court’s Decision in Goldman The Supreme Court granted Goldman’s petition for certiorari. Before the Court, however, defendants abandoned their rule-based argument, and, notably, plaintiffs conceded that, as a factual matter, the generic nature of a misrepresentation often is important evidence of price impact that courts should consider at class certification. Goldman, 141 S. Ct. at 1958. Plaintiffs further

27

conceded that courts can consider expert testimony and use “their common sense in assessing whether a generic misrepresentation had a price impact,” id. at 1960, and that such considerations are appropriate at class certification even though they might also be relevant to materiality, see id.

As previewed above, the Court agreed with the parties, and offered guidance as to how genericness concerns should fit into cases proceeding under the inflation-maintenance theory of price impact. It acknowledged that, under the theory, courts generally look to the back-end price drop as a proxy for front-end inflation.

However, the Court added:

[T]hat final inference—that the back-end price drop equals front-end inflation—starts to break down when there is a mismatch between the contents of the misrepresentation and the corrective disclosure. That may occur when the earlier misrepresentation is generic (e.g., “we have faith in our business model”) and the later corrective disclosure is specific (e.g., “our fourth quarter earnings did not meet expectations”). Under those circumstances, it is less likely that the specific disclosure actually corrected the generic misrepresentation, which means that there is less reason to infer front-end price inflation—that is, price impact—from the back-end price drop.

Goldman, 141 S. Ct. at 1961. As such, it explained, the “generic nature of a misrepresentation often will be important evidence of a lack of price impact, particularly in cases proceeding under the inflation-maintenance theory” id., and

28

that is true “regardless whether that evidence is also relevant to a merits question like materiality,” id. at 1960. Concluding that it was unclear whether we considered that evidence, the Supreme Court vacated our judgment and remanded for further proceedings consistent with its opinion. See Goldman, 141 S. Ct. at 1963.

ATRS III. Upon remand, we noted that in evaluating the parties’ competing price impact evidence, the district court did not discuss the generic nature of Goldman’s alleged misrepresentations, nor the submissions of a third Goldman expert, Dr. Laura Starks, relevant to that inquiry. Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc. (ATRS III), 11 F.4th 138, 143 (2d Cir. 2021) (internal citation omitted). We concluded that the fact intensive questions raised by Goldman were better evaluated by the district court in the first instance. We vacated the district court’s order and remanded, directing the district court to “consider all record evidence relevant to price impact and apply the legal standard as supplemented by the Supreme Court.” Id. at 143–44.

29

3. Round Three: The Decision Below That brings us to the present appeal. On remand, the district court stayed the course and—in the decision Goldman now appeals—certified plaintiffs’ class for a third time. In re Goldman, 579 F. Supp. 3d at 520.

Much of the evidence before the district court, as well as the district court’s analysis of it, should by now be familiar. Because it is discussed extensively below, it needs only brief mentioning here. On plaintiffs’ side of the ledger, the district court again found “persuasive[]” plaintiffs’ evidence establishing a link between (a) the revelatory nature of the corrective disclosures regarding Goldman’s conflicts of interest and (b) the subsequent stock price declines. Id. at 531. Specifically, it credited Dr. Finnerty’s focus on the “conduct underlying the reported enforcement actions, not merely the actions themselves.” Id. at 532 (alteration omitted).

Turning to defendants’ experts, the district court noted it had previously declined to credit Dr. Gompers’ opinion regarding the lack of abnormal price movement associated with the pre-disclosure reports, and reasoned that neither the Supreme Court’s nor our remand had any bearing on its previous findings. Thus, the district court “again decline[d] to credit Dr. Gompers’ conclusions.” Id.

30

On the same tack, reconsideration of Dr. Choi’s event study did not alter the district court’s view of it, which remained “unaffected by the updated direction from above.” Id. It reiterated that “Dr. Choi’s methodology was novel, unreliable, and thoroughly outpaced by the conclusions he derived therefrom.” Id.

The court then turned to “the heart of the parties’ post-appeal dispute: the extent of the alleged misstatements’ generic nature.” Id. at 533. Noting that defendants had abandoned their “‘genericness’-as-a-matter-of-law” test, it began by considering the genericness, “as a matter of fact,” of the challenged statements. Id. On this issue, the district court considered, for the first time, the opinions offered by Goldman’s expert, Dr. Laura Starks, as well as Dr. Finnerty’s rebuttals to them. In the end the district court sided with Dr. Finnerty, finding that the statements’ generic nature did not render them incapable of inducing investor reliance. See id. at 534.

Finally, the district court applied the Supreme Court’s mismatch sliding scale and found that the alleged misstatements “are not so exceedingly more generic than the corrective disclosures that they vanquish the otherwise strong inference of price impact embedded in the evidentiary record.” Id. at 537. The “comfortable, though certainly not boundless, gap in genericness,” it explained,

31

“fails to satisfy Defendants’ burden to demonstrate a complete lack of price impact attributable to the alleged misstatements.” Id. at 538. It certified the class.

For a third time, we granted defendants leave to pursue an interlocutory appeal of that order.

DISCUSSION

“We review a district court’s grant of class certification for abuse of discretion,” Levitt v. J.P. Morgan Sec., Inc., 710 F.3d 454, 464 (2d Cir. 2013), reviewing de novo “the conclusions of law underlying that decision” and “‘for clear error the factual findings underlying’” its ruling, such as the court’s price impact determination, id. (quoting Teamsters Loc. 445 Freight Div. Pension Fund v. Bombardier Inc., 546 F.3d 196, 201 (2d Cir. 2008)). “Under the clear error standard, we may not reverse [a finding] even though convinced that had [we] been sitting as the trier of fact, [we] would have weighed the evidence differently.” Atl. Specialty Ins. Co. v. Coastal Envtl. Grp. Inc., 945 F.3d 53, 63 (2d Cir. 2019) (alterations in original) (internal quotation marks and citations omitted). Rather, a finding is clearly erroneous only if although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a

32

mistake has been committed.” Id. (internal quotation marks and citations omitted); see also ATRS III, 11 F.4th at 142.

Goldman presses three principal arguments on appeal. First, it contends the district court understated the genericness of the alleged misrepresentations and, in setting them against the more detailed corrective disclosures, failed to meaningfully apply the Supreme Court’s mismatch framework. Second, Goldman challenges the district court’s application of the inflation-maintenance theory; it claims that by using the price drop following the detailed, specific corrective disclosures as a proxy for the inflation-maintaining capacity of the broad, generic misrepresentations, the district court improperly extended the theory. These arguments have merit.

Finally, though less forcefully this time around, Goldman maintains the district court again misweighed Dr. Gompers’ and Dr. Finnerty’s expert submissions, and in doing so made untenable credibility findings. We begin there, because although that argument does not carry the day, the district court’s analysis on this front gives important context to why we agree with Goldman’s first two arguments. In the end, the district court’s class certification decision cannot stand.

33

I. The district court did not clearly err in rejecting defendants’

characterization of the 36 dates of pre-disclosure reporting as alternative corrective disclosure dates.

Careful application of the Supreme Court’s guidance in Goldman requires a clear understanding of plaintiffs’ theory regarding the tie between the corrective disclosures and the alleged misrepresentations—why, according to them, there are grounds to infer that the back-end news actually corrected the front-end misstatements. The dueling submissions of Drs. Finnerty and Gompers regarding the significance of the 36 dates of pre-disclosure reporting bear directly on that issue.

Although the two experts maintained differing views on the significance of the pre-disclosure reports, their respective analyses shared common ground: the price declines on the alleged corrective disclosure dates, they agreed, were attributable to “Goldman-specific” information. J.A. 3908, 3912, 3915. However, in order to determine what Goldman-specific information caused the stock price decline on the corrective disclosure dates, Dr. Gompers focused on 36 dates on which various articles, all published before the filing of the Abacus Complaint, reported broadly on Goldman and concerns of conflicts of interest.

34

Dr. Gompers viewed the 36 pre-disclosure dates as “alternative corrective disclosure dates,” J.A. 3806, because the information contained in the articles “was similar to the information released on the alleged corrective disclosure dates in that it indicated to market participants that Goldman allegedly favored itself over its clients, or favored one client over another,” J.A. 1532. Building from there, Dr. Gompers explained that because Goldman’s stock did not decline in response to similar information revealed by the pre-disclosure articles, the price decline on the three disclosure dates must have been due to news of the enforcement action in and of itself.

In that sense, Dr. Gompers opined that the pre-disclosure reports “disentangle[d] how much [of the price decline] was due to [the] conflict news.” J.A. 4602. Unlike the three corrective disclosure dates, which contained both “conflicts news” and “news of an enforcement action,” id., the pre-disclosure reports discussed only news implicating Goldman’s conflicts management. The pre-disclosure reports, for Dr. Gompers, are simply a better match.

Through Dr. Finnerty, plaintiffs offered various rebuttals. For example, Dr. Finnerty argued that any potential price impact was “thwarted by Goldman’s repeated denials” as set forth in many of the articles. J.A. 2043. Dr. Finnerty also

35

opined that the Abacus Complaint revealed “significant new information concerning the severity of Goldman’s misconduct in issuing the Abacus CDO,” J.A. 2044, which, for him, uncovered for the first time “the truth about Goldman’s fraudulent conduct regarding its conflicts of interest,” id., and the fact that Goldman had “failed to manage its conflicts of interest,” J.A. 4707.

On the whole, Dr. Finnerty pegged as futile Dr. Gompers’ efforts to disentangle the price impact caused by the news of the enforcement action itself from the conduct underlying it. Dr. Finnerty explained that “[t]he enforcement actions or investigations are inextricably tied to the content [and] . . . the fact that [the SEC] embodied [the conduct] in an enforcement action document raises . . . in the minds of investors, the severity level.” J.A. 657.

The district court ultimately credited Dr. Finnerty’s opinion. It noted that it had previously declined to credit Dr. Gompers’ view of the pre-disclosure reports, and that because “the updated direction from the Supreme Court and Second Circuit has no bearing on these factual findings,” it would “reiterate[], and restate[], its grounds only in brief.” In re Goldman, 579 F. Supp. 3d at 532.

It found that (1) unlike the pre-disclosure reports, the Abacus Complaint was the first public account to detail and document those conflicts with hard

36

evidence, including incriminatory emails and memoranda authored by Goldman employees; (2) “the underlying source of the disclosure—the SEC—lent extra credibility and gravitas unequaled in the prior reports; and (3) the disclosure was unencumbered by any of the denials or mitigating commentary that had rendered prior reports less jarring.” Id.

Goldman begins its press by arguing that the district court erred in crediting Dr. Finnerty’s views of the pre-disclosure reports. It did not. The district court recognized correctly—or, at least, not clearly erroneously—a qualitative difference in the respective buckets of news. The Abacus Complaint contained details which substantiated its allegations of wrongdoing; the pre-disclosure reports, meanwhile, discussed the transactions, but only generally alleged that Goldman had acted unlawfully, or was otherwise guilty of wrongdoing. It was not clearly erroneous to recognize that, in pointing its finger at Goldman, the SEC had details to back it up.

Nor did the district court err in finding that the SEC’s name lends a certain amount of credibility or gravitas to the allegations underlying the Abacus Complaint, and, therefore that, as a matter of common sense, denying wrongdoing in the face of an SEC enforcement action is likely to have less of a thwarting effect

37

on potential stock price declines than denying more general claims made by media outlets. The district court recognized what, at minimum, is not clearly erroneous: the filing of an enforcement action is a different kind of event than the publishing of a news story. The two events ring of different tenors.

By the same token, the district court did not misstep in finding unpersuasive Goldman’s efforts to disentangle and separately quantify the price decline attributable to, on the one hand, the conduct underlying the enforcement action, and, on the other hand, the news of the enforcement action itself. The SEC’s decision to charge Goldman was precisely because of the nature of the conduct. As Dr. Finnerty opined, the enforcement action “signals the greater severity than if an enforcement action hadn’t been filed, but an enforcement action is never going to get filed unless the misbehavior or alleged misbehavior occurred in the first place . . . [t]hat’s why you can’t separate them.” J.A. 658. It was not clear error to credit that opinion.

Still, that gets us only so far. While the district court did not clearly err in rejecting Goldman’s invitation to view the pre-disclosure reports as alternative corrective disclosure dates, that focuses our analysis on the corrective disclosures as alleged by plaintiffs—but it does not resolve it. Likewise, even accepting as true

38

(or not clearly erroneous) the district court’s view that the conduct (conflicts management) described in the Abacus Complaint is intertwined with the charge— in other words, that the price drop occurred because of both—that establishes, at most, that the corrective disclosures, like the alleged misrepresentations, concern the subject of conflicts management.

The question remains whether, in light of the Supreme Court’s guidance in Goldman, it was clear error for the district court to rely on that subject-matter match to use the back-end price drop as a proxy for front-end inflation allegedly maintained by what the district court acknowledged were comparatively generic misstatements. Again, a back-end price drop is, at most, “backward-looking, indirect evidence,” In re Allstate Corp. Sec. Litig., 966 F.3d at 613, of the price impact “at the time of purchase,” id. at 611. “Data from later times may be relevant to this inquiry, but only insofar as they help the district court determine the information impounded into the price at the time of the initial transaction.” Id. at 612. In our view, the genericness and mismatch inquiries go to the value of the back-end price drop as indirect evidence of a front-end, inflation-maintaining price impact, an issue at which the parties direct most of their efforts.

39

II. The district court clearly erred in assessing the generic nature of business principles statements, but not the conflicts disclosure.

Goldman argues the district court failed to appreciate the generic nature of the challenged statements. It faults the district court for discrediting Dr. Laura Starks, who, Goldman says, correctly observed that the alleged misrepresentations “do not provide information that bears on a company’s future financial performance or value” and “are also too general to convey anything precise or meaningful” that can be used in investment decision-making. J.A. 2608. Goldman insists that in finding as a matter of fact that “[t]he alleged misstatements were not so generic as to diminish their power to maintain pre-existing price inflation,” In re Goldman, 579 F. Supp. 3d at 534, the district court glossed over and minimized the genericness analysis.

The district court’s findings on this issue go to a baseline question: how generic are the alleged misrepresentations? Beginning there makes sense; it is a practical, threshold factual inquiry to the ensuing Goldman-driven analysis of whether there is a gap in specificity between a set of misstatements and corrective disclosures.

The district court conducted that initial inquiry by separating the statements in two buckets, one consisting of the business principles statements—such as

40

“integrity and honesty are at the heart of our business,” which it acknowledged “present as platitudes when read in isolation”—and the other containing the “more specific” conflicts disclosure. 6 In re Goldman, 579 F. Supp. 3d at 534. With respect to the former, it found that “even the more generic statements, when read in conjunction with one another (and particularly in conjunction with statements specifically concerning conflicts), may reinforce misconceptions about Goldman’s business practices, and thereby serve to sustain an already-inflated stock price.” Id. As to the more specific conflicts disclosure, the court found that “the statements concerning Goldman’s conflicts . . . are quite a bit more specific in form and focus than, say assurances that ‘[i]ntegrity and honesty are at the heart of our business.’” Id. The district court’s answer to the preliminary inquiry: not so generic.

Business Principles Statements With respect to the business principles statements, the district court’s genericness analysis is untenable.

6 Again, plaintiffs focus on Goldman’s representation in its conflicts disclosure that it “ha[s] extensive procedures and controls that are designed to identify and address conflicts of interest, including those designed to prevent the improper sharing of information among our businesses.” J.A. 3278.

41

The district court found that the business principles category of statements—statements such as “integrity and honesty are at the heart of our business”—“present as platitudes when read in isolation.” Id. There is no need to second-guess that factual finding; nor was it clearly erroneous to find, as the district court did, that when read as a whole the business principles statements are somewhat more specific. See id. From there however, the district court overstated their specificity by finding that these “more generic statements, when read . . . particularly in conjunction with statements specifically concerning conflicts[], may reinforce misconceptions about Goldman’s business practices.” Id.

That finding was clearly erroneous. The business principles and conflicts statements were separately disseminated to shareholders in separate reports at separate times, 7 and plaintiffs offered no evidence, either through Dr. Finnerty or otherwise, to support a finding that, notwithstanding that space in medium and time, investors would still conjunctively consume those statements. True, a statement can be materially misleading when “the defendants’ representations,

7 For instance, plaintiffs allege in their complaint that Goldman released its 2007 Form 10-K, containing the conflicts disclosure, on January 29, 2008, J.A. 139, and released its 2007 Annual Report, containing its business principles, on March 8, 2008, J.A. 141.

42

taken together and in context, would have mislead a reasonable investor.” Altimeo Asset Mgmt. v. Qihoo 360 Tech. Co., 19 F.4th 145, 151 (2d Cir. 2021) (quoting Rombach v. Chang, 355 F.3d 164, 172 n.7 (2d Cir. 2004)). But the relevant “context” is not a separately disseminated misstatement—at least where, as here, the statements do not obviously compliment or implicate the same topics—but the reality of the company’s affairs or condition at a time when a misstatement was made.

So, for example, a company’s statement that its distribution market is “highly competitive,” might be actionable when considered within the context that the company did not actually operate in a competitive market and instead colluded with its competitors to fix prices. See In re Henry Schein, Inc. Sec. Litig., No. 18 Civ. 01428, 2019 WL 8638851, at *12 (E.D.N.Y. Sept. 27, 2019). Or, a company’s statement that it has “demonstrated successful acquisition and integration capabilities” might be actionable when, at the time the statement was made, the company had already fired key integration staff and was dealing with a poor integration of a newly acquired company. City of Omaha Police & Fire Ret. Sys. v. Evoqua Water Techs. Corp., 450 F. Supp. 3d 379, 412 (S.D.N.Y. 2020). Case law does not suggest, however, that investors read one statement in conjunction

43

with separately disseminated statements, at least where, as here, those statements do not obviously build off one and other.

Plaintiffs offer no meaningful rebuttal. Instead, they claim that “the business-principle[s] statements, while more generic, are not challenged standing alone but as reinforcing the conflict statements.” Appellees Br. at 39. That bald assertion is unsupported by any citation to the record, nor, upon our independent of review of it, is there any suggestion that the business principles statements were consumed by investors as piggybacking off the conflicts disclosure. To the contrary, plaintiffs’ complaint alleges that they comprise, on their own, the “third category of false and misleading statements.” J.A. 95. In any event, by that logic, an exceedingly generic statement could always withstand, for example, motions to dismiss or for summary judgment by seeking shelter under a more specific statement, so long as the more specific statement implicates broad topics such as integrity or honesty. Securities law provides no such cover.

“A finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” Dist. Lodge 26, Int’l Ass’n of Machinists & Aerospace Workers, AFL-CIO v. United Techs. Corp., 610 F.3d 44, 51 (2d

44

Cir. 2010) (internal citation omitted). The record evidence here provides no support for reading the business principles statements in conjunction with the conflicts disclosure. Accordingly, doing so was clear error.

Keeping in mind that a class certification genericness analysis pursuant to Goldman was, for the district court, and is, for us, new and uncharted territory, it is appropriate to pause to consider the implications of the error identified. In the normal course, that error would almost certainly require remand. Erroneously assessing a misrepresentation’s genericness would necessarily infect the ensuing mismatch inquiry—it would proceed from the wrong starting point.

However, the balance of the district court’s analysis, including its mismatch inquiry, centers on the conflicts disclosure. Apart from acknowledging that the business principles statements “equate roughly, in terms of genericness, to the Supreme Court’s prototype,” 8 In re Goldman, 579 F. Supp. 3d at 538, it did not meaningfully discuss them further. To be sure, the district court’s choice in focus is no fault of its own; throughout this litigation the conflicts disclosure has been center stage. Still, the district court acknowledged a gap in genericness even

8 That protype: “we have faith in our business model.” Goldman, 141 S. Ct. at 1961.

45

between the two sets of alleged misrepresentations—that is, that the conflicts disclosure is “quite a bit more specific in form and focus,” id. at 534, than the business principles statements. Plaintiffs likewise agree that their best shot at success is the conflicts disclosure; their counsel conceded at oral argument that, standing on its own, a claim based on the business principles statements would face a decidedly tough road to recovery. See Oral Arg. Audio at 1:07:16, Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc (No 22-484). 9 In short, if the district court’s mismatch analysis, centered as it is on the conflicts disclosure, cannot withstand scrutiny—and, as explained below, it cannot—then plaintiffs’ claim based on the business principles statements must also fail. Accordingly, there is no need to remand for the district court’s reconsideration of the genericness of the business principles statements.

Conflicts Disclosure As an initial matter, however, there is no merit to Goldman’s claim that, in labeling the conflicts disclosure as, essentially, less generic than the business principles statements, the district court similarly understated that statement’s

9 Plaintiffs’ counsel acknowledged that “if this was a case just about the business principles, we would have a very, very significant problem.” Id.

46

generic nature. Not so. The district court assessed the conflicts disclosure, and, again, found that it was “quite a bit more specific in form and focus” than the business principles statements. In re Goldman, 579 F. Supp. 3d at 534. Goldman insists that is not enough; it contends that the district court failed to meaningfully consider our materiality case law, which, it claims, would have spotlighted the generic nature of the conflicts disclosure.

It is true that Goldman gives courts a green light to assess a statement’s generic nature by referencing case law bearing on materiality. More specifically, Goldman permits courts to look to those cases for guidance as to whether, as a factual matter, courts have labeled comparable statements as generic. For example, our own materiality cases often feature claims based on a company’s risk disclosures, and our discussion in those cases often centers on whether the risk disclosures are sufficiently specific to evoke investors’ reliance. 10 Those cases have examined, on one end, a detailed description of a company’s environmental compliance efforts, recounting the company’s pollution abatement equipment,

10See, e.g., Plumber & Steamfitters Loc. 773 Pension Fund v. Danske Bank A/S, 11 F.4th 90, 103 (2d Cir. 2021); Singh v. Cigna Corp., 918 F.3d 57, 60 (2d Cir. 2019); Meyer v. Jinkosolar Holdings Co., 761 F.3d 245, 247–50 (2d Cir. 2014); ECA, Loc. 134 IBEW Joint Pension Tr. Of Chicago v. JP Morgan Chase Co., 553 F.3d 187, 205–06 (2d Cir. 2009).

47

water treatment efforts, and around-the-clock environmental monitoring teams, Jinkosolar Holdings, 761 F.3d at 247, and, on the other, more generic representations that a company has “established policies and procedures to comply with applicable requirements,” Singh, 918 F.3d at 61.

Of course, the overarching question in those cases—materiality—differs from the price-impact analysis at issue here, yet both inquiries task courts with considering an alleged misrepresentation’s generic nature. Courts can look to those cases to answer whether a set of challenged statements are, as a matter of fact, generic.

Goldman complains that the district court failed to do that here.

Again: not so. The district court made clear that it considered cases bearing on materiality to the extent they presented issues overlapping with the price impact analysis. See In re Goldman, 579 F. Supp. 3d at 535 n.17. We take the district court at its word. Goldman bemoans that the district court relegated that point to a footnote, but that provides no occasion to impose specific stylistic mandates on district courts as they navigate this tricky area of law.

48

III. The district court’s price impact analysis was based on an erroneous application of the inflation-maintenance theory.

Although its attack on the district court’s threshold inquiry regarding the generic nature of the conflicts disclosure is without merit, we agree with Goldman that, having conducted that factual assessment, the district court then erred in applying Vivendi’s inflation-maintenance theory to weigh the parties’ evidence regarding the extent to which that disclosure might, in practice, maintain Goldman’s stock price. Review of the district court’s factual findings is limited to clear error, but whether a legal standard—here, the inflation-maintenance theory—has been incorrectly applied to those findings is an issue of law to be reviewed de novo. See In re Initial Public Offerings Sec. Litig., 471 F.3d 24, 32 (2d Cir. 2006) (“We will apply the abuse-of-discretion standard both to [the district court’s] ultimate decision on class certification as well as her rulings as to Rule 23 requirements, bearing in mind that whether an incorrect legal standard has been used is an issue of law to be reviewed de novo.”).

In this portion of its analysis, the district court began by crediting Goldman’s expert, Dr. Starks, who opined that the alleged misrepresentations were “unlikely, in a vacuum, to consciously influence investor behavior . . . .” In re Goldman, 579 F. Supp. 3d at 534. Ultimately, however, the district court found Dr. Starks’

49

opinion to be of “limited use[].” Id. It explained that “the proper measure of inflation maintenance by a company that chooses to speak ‘is not what might have happened had a company remained silent, but what would have happened if it had spoken truthfully.’” Id. (quoting In re Vivendi, 838 F.3d at 258).

Proceeding from that principle, the court credited “Dr. Finnerty’s analysis that truthful, contrary substitutes for the alleged misstatements would have impacted investors’ subsequent decision-making,” and found that “as Dr. Finnerty concluded, ‘[t]his is precisely what happened here when investors learned in April and June 2010 the details and severity of Goldman’s misconduct, and Goldman’s stock was devalued accordingly.’” Id. (quoting J.A. 2816). To the same tune, it faulted Goldman for failing to present evidence “purporting to demonstrate, under the test set forth in Vivendi, that if Goldman had replaced the alleged misstatements with the alleged truth about its conflicts, its stock price would have held fast.” Id. at 535.

Goldman contends that the district court’s rendition of the inflation-maintenance theory is overly expansive. Correct. Specifically, our cases applying the theory establish its limits; the district court’s interpretation pushed the inflation-maintenance theory well beyond them.

50

a. The inflation-maintenance theory under Vivendi, Waggoner, and Goldman

Waggoner. Our recent application of the inflation-maintenance theory in Waggoner v. Barclays PLC highlights the tension at work with applying the inflation-maintenance theory to the facts of this case. There, in order to quell “concerns that high-frequency traders may have been front running” other traders on a specific Barclays trading platform, Barclays’ officers made numerous statements to assure investors the platform was “safe from” aggressive trading practices, and that it “was taking steps to protect” institutional investors on those platforms by monitoring and removing aggressive traders who violated the platforms’ special protections. 875 F.3d 79, 87 (2d Cir. 2017). In the end, upon the filing of a complaint by the New York Attorney General (the “NYAG Complaint”) alleging securities fraud under state law, investors learned that those representations were allegedly false because, according to the State, no special protections existed, and, in fact, Barclays favored rather than removed aggressive traders. Id. at 88.

Waggoner is particularly illuminating given the similarity between the corrective disclosure there and here; both took the form of an enforcement action. Unlike here, however, Waggoner presented a tight fit between corrective disclosure

51

and misrepresentation: the NYAG Complaint targeted the same trading platform discussed by Barclays in their misleading statements, and took aim at the same or similar statements underlying the claims subsequently pressed by plaintiffs in Waggoner, alleging that those statements were false or misleading. Compare id. at 87–88, with Summons and Complaint at 6, 8–11, People ex rel. Schneiderman v. Barclays Capital, Inc. et al., No. 451391/2014 (N.Y. Sup. Ct. June 25, 2014), Dkt. No. 1, 2014 WL 2880709. There was no question in Waggoner that the corrective disclosure directly implicated not just the same topic, but the alleged misstatements themselves—a notable distinction from the misrepresentation-corrective disclosure relationship here.

Waggoner is therefore an easy inflation-maintenance case. The company’s affirmative false statements were expressly identified as such by the corrective disclosure. By expressly and specifically negating the alleged false statement, the truthful substitute for the lie was identified by the corrective disclosure itself.

Vivendi. The link between misstatement and disclosure was equally strong in Vivendi. There, the company’s repeated statements regarding its comfortable liquidity situation were later contradicted by a body of information—including several downgrades to its debt rating, public reports regarding the company’s lack

52

of transparency about its large debt obligations, and, ultimately, the announcement that the company faced massive refinancing needs that would require a fire sale of assets—all of which revealed that its cash flow was anything but strong. See In re Vivendi, 838 F.3d at 235–37.

As in Waggoner, among the various disclosures identified by the plaintiffs in Vivendi were back-end reports or investigations expressly implicating the alleged misstatements. See Amended Complaint at ¶¶ 132, 146, In re Vivendi Universal, S.A. Sec. Litig., No. 02 Civ. 5571, (S.D.N.Y. July 8, 2009), ECF No. 904, 2009 WL 2611656. Simply, the company’s lie that it had abundant cash flow was made apparent by a cascade of news revealing its crippling debt obligations. In both Vivendi and Waggoner, the strong link between misrepresentation and corrective disclosure provided sturdy ground to use the back-end price drop as a proxy for front-end inflation. The back-end disclosures’ corrective effect upon the affirmative misrepresentations was obvious.

To be sure, not all the corrective disclosures in Vivendi expressly referenced the alleged misrepresentations. For example, the company’s back-end announcement revealing its massive refinancing needs did not expressly recant its earlier statements regarding its comfortable cash situation. See In re Vivendi, 838

53

F.3d at 237. Yet there can be little doubt that even those corrective disclosures directly rendered false the company’s affirmative misrepresentations. It is also true that Vivendi’s misrepresentation regarding its “strong free cash flow,” id. at 235, or its “free operational cash flow . . . far above . . . objectives,” id., might plausibly be labeled more generic than, for example, later announcements from the company describing the company’s specific refinancing needs. We do not suggest that the inflation-maintenance theory requires a precise match. It may frequently be the case that what is corrective about a “corrective disclosure” is situated among details which, in the aggregate, make for a somewhat more specific back-end disclosure.

And yet, Vivendi accounts for that possibility. Its application of the back-end—front-end inference rested on a finding that, had the company spoken truthfully regarding its debt problems at an equally generic level, the market would have reacted. We described, as a truthful substitute for the company’s “rosy picture of its liquidity state,” the “misgivings its executives were sharing behind the scenes,” which included statements—less specific than the corrective disclosure news—that the company was in “danger” of a downgrade, or that its liquidity situation was “tense.” Id. at 235, 258. Vivendi therefore directs that where

54

the corrective disclosures do not expressly identify the alleged misrepresentation as false (as in Waggoner), the “truthful substitute” should align in genericness with the alleged misrepresentation. 11 Goldman. The Supreme Court’s guidance in Goldman adds more to the mix.

Even under a proper application of Vivendi’s equally-generic-truthful-substitute formula, plaintiffs might still attempt to (a) identify a highly specific corrective disclosure, and (b) identify and extract a generic truth purportedly embodied therein, in order to (c) craft a link between a generic misrepresentation and specific corrective disclosure. For example, plaintiffs could point to news detailing a company’s commission of securities fraud, and then claim that nestled therein was the more generic revelation that the company’s earlier, general statement that it aims to act lawfully was a lie. From there, they might still contend that the back-end price drop is an appropriate proxy for front-end inflation.

11To be sure, with respect to the loss causation element of securities fraud—that is, the causal link between the alleged misconduct and the loss ultimately suffered by plaintiff— the “basic [] calculus” remains the same whether the truth is revealed in “a corrective disclosure describing the precise fraud” or through “events constructively disclosing the fraud.” Vivendi, 838 F.3d at 262. Yet the question here—whether there is a basis to infer that the back-end price equals front-end inflation—is a different question than loss causation, and, in light of Goldman, requires a closer fit (even if not precise) between the front- and back-end statements.

55

Goldman dispels that notion. The Court explained that a gap in genericness between misrepresentation and corrective disclosure reduces the likelihood that investors would understand the “specific disclosure [to have] actually corrected the generic misrepresentation,” Goldman, 141 S. Ct. at 1961, and, in such a scenario, the back-end–front-end inference starts to break down. In other words, although Vivendi somewhat solves for a back-end–front-end space in genericness by asking whether an equally generic, truthful substitute would have dissipated inflation, Goldman requires that any gap among the front- and back-end statements as written be limited. 12

b. The district court erroneously applied Vivendi’s “truthful substitute” inquiry

The Goldman-Vivendi-Waggoner trio spotlights the district court’s error below. First, unlike both Vivendi and Waggoner, not one of the corrective disclosures here expressly identifies either the business principles statements or conflicts disclosure. Second, the district court acknowledged a considerable gap

12Of course, Goldman does not call into question Vivendi itself. There, any mismatch in specificity between the misrepresentations and disclosures was minimal. Vivendi’s affirmative, repeated representations regarding its cushy cash flow were directly contradicted by news portraying its cash situation as anything but that. See In re Vivendi, 838 F.3d at 234–36.

56

in specificity between the corrective disclosures and alleged misrepresentations. Therefore, the district court should have asked “what would have happened if [the company] had spoken truthfully,” In re Vivendi, 838 F.3d at 258, at an equally generic level. However, in what amounts to the crux of the district court’s misstep, the district court allowed the “details and severity,” In re Goldman, 579 F. Supp. 3d at 534, of the corrective disclosure to do the work of proving front-end price impact, notwithstanding that the front-end statements are, according to the district court’s own findings, “comfortabl[y]” more generic than the back-end disclosures, id. at 538. The district court’s formulation of Vivendi outpaces Vivendi itself. It also fails to heed Goldman’s cautionary guidance that the back-end—front-end inference starts to “break down,” Goldman, 141 S. Ct. at 1961, when there is a mismatch in genericness at the front and back ends.

Utilizing a back-end price drop as a proxy for the front-end misrepresentation’s price impact works only if, at the front end, the misrepresentation is propping up the price—that is, in the district court’s words, if “Goldman’s alleged misstatements reinforced [the market’s] misconception.” In re Goldman, 579 F. Supp. 3d at 536. In other words, reinforcement requires some indication that investors relied upon the conflicts disclosure as written, and, here,

57

the district court credited Dr. Starks’ opinion that investors did not. Although the theory’s starting point is that a misrepresentation need not be “associated with an uptick in inflation,” Vivendi, 838 F.3d at 259, a misrepresentation must actually maintain inflation; it must, in other words, hold its weight in propping up the price.

Consider, for example, an investor who reads certain statements in a company’s Form 10-K, and then thinks “Things seem to be going well; I think I’ll hold onto my shares.” Although the statements did not cause that investor to buy more stock, they informed or influenced her decision. And if the company’s statements are later revealed as false, liability might follow not because the statement caused new or more inflation—that is, caused investors to purchase more stock (thereby increasing demand and, ultimately, raising the share price)— but instead because the statement maintained inflation, or influenced the investor’s decision to hold tight.

Viewed against that backdrop, the district court’s finding that the challenged “statements were [not] consciously relied upon, in the moment, by investors evaluating Goldman,” In re Goldman, 579 F. Supp. 3d at 535, begs the following question: how, then, can it be that the statement impacted Goldman’s stock price?

58

The district court’s answer: because they would be relied upon had Goldman disclosed “the details and severity of Goldman’s misconduct,” id. at 534, which no one doubts did impact the price. The details and severity of the misconduct, it says, should be substituted in place of the challenged, more generic statements. Again, that substitution stretches the “back-end price drop equals front-end inflation” inference beyond its breaking point—and certainly beyond how we’ve previously construed it. Vivendi requires that the “truthful substitute” align in genericness with the alleged misrepresentation. Here, however, the district court’s substitute looks nothing like the original.

Likewise, Goldman requires that courts pay special attention to mismatches in specificity between a misstatement and corrective disclosure. But by reimagining a more specific misstatement, the district court failed to sufficiently follow that guidance. Again, the district court found a “comfortable” gap in genericness between the alleged misstatements and subsequent corrective disclosures, but then found that it was not “boundless” given the fact that both implicated conflicts of interest at Goldman and “Goldman’s infrastructure for managing them.” Id. at 538. It recognized, in other words, that the corrective disclosures bore on the same subject—conflicts of interest management—but did

59

not expressly or otherwise clearly refer to Goldman’s cautionary language regarding conflicts in the “Risk Factors” portion of its 10-K.

Proceeding from that match in subject matter, the district court moved forward with its Vivendi analysis by finding that, if Goldman had disclosed in its Risk Factors the “details and severity of Goldman’s misconduct” as set forth in the Abacus Complaint, a price drop would have followed. However, again, requiring only a general front-end—back-end subject matter match to, effectively, concoct a highly specific truthful substitute does not meaningfully account for the Supreme Court’s guidance in Goldman.

c. Our materiality cases contextualize the district court’s misapplication of inflation-maintenance theory, and the heightened relevance of Goldman’s guidance in this case.

Nor, in any event, does securities law permit plaintiffs to target generic risk disclosures on the theory that, had the risk disclosures contained a detailed admission of severe wrongdoing, a price drop would follow. There is no “affirmative duty to disclose any and all material information.” Vivendi, 838 F.3d at 239. 13 Instead, “disclosure is required when necessary to make statements

13Vivendi explains: “Absent an actual statement, a complete failure to make a statement— in other words, a ‘pure omission’—is actionable under the securities laws only when the

60

made, in light of the circumstances under which they were made, not misleading.” Jinkosolar Holdings, 761 F.3d at 250 (internal citations and alterations omitted). But the duty to disclose more is triggered only when that which is disclosed is sufficiently specific to evoke a reasonable investor’s reliance. See, e.g., Caiola v. Citibank, N.A., 295 F.3d 312, 331 (2d Cir. 2002).

For example, in ECA shareholders alleged JP Morgan made numerous misrepresentations in its annual report to shareholders regarding its “highly disciplined” risk management process. 553 F.3d at 205. The plaintiffs claimed those statements were revealed as false in light of, in their view, the bank’s poor financial discipline and the bank’s liability arising from the infamous WorldCom and Enron scandals. See id. at 205–06. In rejecting plaintiffs’ claim, the court in ECA held that “the statements are too general to cause a reasonable investor to rely upon them . . . these statements did not, and could not, amount to a guarantee that its choices would prevent failures in its risk management.” Id. at 206. Indeed, since ECA we have reaffirmed that a company’s 10-K disclosures regarding, for example, its compliance efforts “can be materially misleading if ‘the descriptions

corporation is subject to a duty to disclose the omitted facts.” 838 F.3d at 239 (internal citations omitted). There is no claim in this case that Goldman was under such a duty.

61

of compliance efforts’ are ‘detailed’ and ‘specific.’” Plumber & Steamfitters Loc. 773 Pension Fund, 11 F.4th at 103 (quoting Singh, 918 F.3d at 63).

In such cases, the disclosure itself acts as a gatekeeper: courts ask whether investors would even rely on that disclosure—as written—such that they would be misled by an omission. Were it otherwise, securities plaintiffs could find a road to success in the rearview mirror: they would need only find negative news, such as the revelation that a company may have committed securities fraud, and then point to any previous disclosure from the company which touches upon a similar subject, such as that company’s commitment to complying with the law—no matter how generic that statement is. Asking whether the disclosure as written is specific enough to evoke investor reliance avoids turning securities claims into a game of litigation-by-hindsight.

These risk-disclosure cases highlight why the Supreme Court’s concerns in Goldman loom especially large here. As in ECA, plaintiffs contend that having generally discussed risks related to conflicts of interest, Goldman should have divulged the details surrounding its mismanagement of conflicts with respect to, for example, the Abacus transaction. In failing to do so, plaintiffs claim, Goldman’s risk disclosure was misleading by omission. See J.A. 52, 477. However,

62

as explained above, such claims require special attention to the generic nature of the disclosure. Again: the duty to disclose more is triggered only where that which is disclosed is sufficiently specific.

Of course, in ECA and its progeny, analysis of the level of detail in the disclosure answered whether, as written, the disclosure was material. Class certification litigation provides no forum to relitigate materiality. See Amgen, 568 U.S. at 468. But the Supreme Court’s guidance in Goldman directs courts to consider issues bearing on materiality to the extent those issues overlap with the price impact analysis, and with respect to genericness concerns, the overlap is substantial. In cases based on the theory plaintiffs press here, a plaintiff cannot (a) identify a specific back-end, price-dropping event, (b) find a front-end disclosure bearing on the same subject, and then (c) assert securities fraud, unless the front-end disclosure is sufficiently detailed in the first place. The central focus, in other words, is ensuring that the front-end disclosure and back-end event stand on equal footing; a mismatch in specificity between the two undercuts a plaintiff’s theory that investors would have expected more from the front-end disclosure.

Goldman’s guidance involves similar concerns. If a stock price decline follows a back-end, highly detailed corrective disclosure—containing, for

63

example, “hard . . . incriminatory” evidence regarding the company’s wrongdoing, In re Goldman, 579 F. Supp. 3d at 532—courts must be skeptical whether the more generic, front-end statement propped up the price to the same extent. As in ECA, Vivendi does not authorize plaintiffs to beef up a generic disclosure with a healthy dose of detail and thereby transform it into something that, as then consumed by investors, it was not.

d. Guidance moving forward Accordingly, a searching price impact analysis must be conducted where (1) there is a considerable gap in front-end–back-end genericness, as the district court found here, (2) the corrective disclosure does not directly refer, as it did in Waggoner, to the alleged misstatement, and (3) the plaintiff claims, as plaintiffs claim here, that a company’s generic risk-disclosure was misleading by omission.

In such cases, case law bearing on materiality can help guide courts in considering, as a factual matter, the generic nature of the alleged misrepresentation. Where a gap exists, courts should ask, under Vivendi, whether a truthful—but equally generic—substitute for the alleged misrepresentation would have impacted the stock price. Importantly, unlike the classic inflation-maintenance case—where the back-end price drop is itself the evidence

64

(albeit indirect) of the front-end price impact—the value of the back-end proxy, given the gap in specificity, will be diminished.

As such, courts should consider other indirect evidence of price impact, directed at either the inflation-maintaining nature of the generic misstatement, or the price-dropping capacity of an equally generic corrective disclosure. Ultimately, a court must determine not just whether the defendant spoke on topics generally important to investment decision-making, but instead whether the defendant’s generic statements on that topic were important in that regard. 14 For instance, pre- or post-disclosure discussion in the market regarding a generic front-end misstatement can be a useful indicator of its inflation-maintaining

14Indeed, litigants already appear to be offering this kind of evidence. For example, in Ferris v. Wynn Resorts Ltd., 2023 WL 2337364 (D. Nev. Mar. 1, 2023), the defendants attempted to exploit a front-end back-end mismatch. At the front end, defendants denied allegations of misconduct by the company’s former chief executive office. Back-end news detailing an alleged decades-long pattern of sexual harassment by the executive was followed by a price decline. Defendants argued that because neither their denials nor the allegations specifically referenced sexual misconduct, the back-end news bearing specifically on sexual harassment could not be used as a proxy for front-end inflation. In rejecting that argument, the district court pointed to evidence that “the media, market participants, and the defendants themselves immediately made the connection between the revelations” and the allegations. Id. at *10. The district court identified post- disclosure commentary specifically discussing both the back-end news, as well as the front-end misstatements. See id. We express no view on the district court’s analysis, but mention that case simply to note the kind of post-Goldman evidence parties are offering at the class certification stage.

65

capacity, as well as of the fact that a truthful, equally generic substitute would likewise not go unnoticed by the market as an inflation dissipator.

Much of this analysis fits comfortably within the prototypical class certification skirmish. In claims proceeding under the inflation-maintenance theory, plaintiffs relying on the Basic presumption will likely face opposition from defendants, who will attempt to rebut that presumption by demonstrating that the alleged misrepresentations did not impact share price. Parties will then join issue with respect to the generic nature of both the misstatements and corrective disclosures, whether they match in specificity, and, if not, whether truthful, equally generic substitutes for the challenged statements would have impacted the stock price. Evidentiary submissions are likely to follow. Ultimately, the court must still find whether the defendants have demonstrated, by a preponderance of the evidence, that the alleged misstatements did not in fact impact the price of the stock. 15

15It may be true that class certification litigation following Goldman is likely to involve evidence that, at the summary judgment stage, might also be relevant to materiality. But that’s by design; Goldman directs courts to consider “all record evidence relevant to price impact, regardless whether that evidence overlaps with materiality or any other merits issue.” Goldman, 141 S. Ct. at 1961.

66

Although, of course, the district court did not have the benefit of our analysis, the parties submitted a mountain of evidence that bears directly on whether an equally generic substitute for the conflicts disclosure would have dissipated the inflation allegedly maintained by that statement, making remand for further factfinding unnecessary.

For example, the district court credited Dr. Finnerty’s analysis of the news coverage and commentary surrounding the alleged corrective disclosures, which it found “convincingly links Goldman’s post-disclosure plight back to the alleged misstatements.” In re Goldman, 579 F. Supp. 3d at 536. But, while this commentary certainly touches on the subject of conflicts of interest and suggests that the management of them is important, it does not suggest that the market relied on the conflicts statements to assess Goldman’s conflicts management procedures.

To put a finer point on it, Dr. Finnerty cited a pre-disclosure investor report providing that “Goldman is very careful” about conflicts and “actually has a full-time partner monitoring” conflicts “according to [a Goldman executive],” J.A. 1228 (emphasis added)—but not according to the conflicts disclosure. Similarly, a pre-corrective disclosure 2007 Merrill Lynch report offered by plaintiffs says that the “consistency with which the firm has avoided crossing the

67

line and damaging its reputation is such that it must be doing something right” with respect to managing conflicts. J.A. 782. But there is no indication that the analyst drew that conclusion from the conflicts disclosure—nor could it, as the conflicts disclosure says nothing about how Goldman manages its conflicts other than through “extensive procedures.” That same article says that “the conflict management process is clearly taken extremely seriously at the firm, since it is viewed as not just a by-product but a key pillar of the firm’s franchise business.” J.A. 782. Yet there is no discussion in the conflicts disclosure of “key pillars.”

On the same tack, the pre-disclosure Merrill Lynch report says that “the scale and growth of its client trading and investment-banking franchise make it clear that [] conflicts have overall been well managed.” J.A. 1228, 4825 (emphasis added). The report itself suggests that the analyst relied on something other than the conflicts disclosure.

With respect to the post-disclosure Wall Street Journal article, the article notes that CDOs are “riddled with potential conflicts,” and that “[t]his territory is especially dangerous for Goldman because of the perception that it is an elite adviser and an elite trader that can do both simultaneously while managing the

68

conflicts to the satisfaction of its clients.” J.A. 3773. But again, nothing suggests that the market came to that realization by relying on the conflicts disclosure.

In short, although market commentary can provide insight into the kind of information investors would rely upon in making investment decisions—and therefore can serve as indirect evidence of price impact—commentary touching upon only the same subject matter, given the contours of this case as discussed above, cannot be enough. As we have previously put it, albeit in another context: “[p]laintiffs conflate the importance of a bank’s reputation for integrity with the materiality of a bank’s statements regarding its reputation. While a bank’s reputation is undeniably important, that does not render a particular statement by a bank regarding its integrity per se material.” ECA, 553 F.3d at 206.

On the other side of the ledger, defendants managed to sever the link between back-end price drop and front-end misrepresentation. Goldman introduced Dr. Starks’ analysis of 880 analyst reports published during the Class Period (both before and after the filing of the Abacus Complaint), none of which reference the conflicts disclosure. J.A. 2617–18. Likewise, the pre-disclosure reports identified by Dr. Gompers, even if falling short as alternative corrective disclosures, touch on the subject of conflicts of interest (and do so in more generic

69

terms), but do not expressly nor impliedly refer to the conflicts disclosure. As with the market commentary identified by Dr. Finnerty, the pre-disclosure reports might suggest that the market cared generally about how mismanaged conflicts could damage Goldman’s reputation, but they do not suggest that investors were misled by Goldman’s conflicts disclosure. Indeed, as provided above, the district court found that investors would not consciously rely on the conflicts disclosure in making investment decisions. 16 In summary, a searching review of the record leaves us with the firm conviction that there is an insufficient link between the corrective disclosures and the alleged misrepresentations. Defendants have demonstrated, by a preponderance of the evidence, that the misrepresentations did not impact Goldman’s stock price, and, by doing so, rebutted Basic’s presumption of reliance.

16At most, plaintiffs identify a single post-disclosure news article expressly mentioning the business principles statements. See J.A. 2317. Yet, as provided above, the district court found that those statements, which it characterized as platitudes, would not have been relied on by investors absent inclusion of the details and severity in the Abacus Complaint. The Vivendi error remains. Moreover, defendants offered 880 pre- and post-disclosure analyst reports which make no mention of the business principles statements. Finally, the considerable front-end–back-end genericness gap is, based on the district court’s findings, even more pronounced with respect to the business principles statements. Against that backdrop, the article fails to do the work plaintiffs ask of it. In other cases—without that backdrop—it might.

70

The district court clearly erred in concluding otherwise, and therefore abused its discretion in certifying the shareholder class.

e. The concurrence

Judge Sullivan criticizes the above analysis as “circuitous.” Concurring Op.

at 7–8. This case has a long and difficult history. The parties have substantially changed their arguments along the way. In Judge Sullivan’s view, the linchpin remains the 36 dates of pre-disclosure reporting and the accompanying expert testimony. We’ll agree to disagree on that score. In any event, this is a complex case and whatever analytical approaches might be warranted in future cases remains to be seen. However, we take no issue with Judge Sullivan’s observation regarding the difficult task of thinking about materiality but not ruling on it. Not easy stuff. Someday the Supreme Court will revisit the issue. In the meantime, we have work to do.

CONCLUSION

We REVERSE the district court’s class certification order, and REMAND with instructions to decertify the class.

RICHARD J. SULLIVAN, Circuit Judge, concurring in the judgment:

I join the majority in its bottom-line conclusion that the district court improperly granted class certification in this long-running case. Nevertheless, I disagree with Part I of the majority opinion, since my position all along has been that the district court committed clear error in assessing Goldman’s expert evidence. I likewise disagree with Part II of the majority opinion, given my view that the district court also clearly erred in evaluating the “generic nature” of both Goldman’s business-principles statements and its conflicts disclosures. Finally, while I concur in the majority’s ultimate conclusion that the district court erred in its class-certification ruling, I fear that the majority’s approach needlessly complicates what, to my mind, should be a straightforward balancing of the several factors that bear on the question of reliance. Let me explain why.

I.

When this case came before us in 2020, I took the position that we should reverse the lower court’s class-certification ruling. See Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc. (ATRS II), 955 F.3d 254, 275, 279 (2d Cir. 2020) (Sullivan, J., dissenting). This was because, in my view, “Defendants offered persuasive and uncontradicted evidence that Goldman’s share price was unaffected by earlier

2

disclosures of Defendants’ alleged conflicts of interest,” which thereby “sever[ed] the link between the alleged misrepresentation and the price paid by Plaintiffs for Goldman shares.” Id. at 275, 278–79 (internal quotation marks and alterations omitted). Back then, I felt compelled to credit Dr. Paul Gompers’s testimony demonstrating that “prior disclosures – as set forth in [thirty-six] separate news reports over as many months – had no impact on Goldman’s stock price,” id. at 278, and Dr. Stephen Choi’s testimony that “the stock drop following the corrective disclosures was attributable” entirely to the “news that the SEC and DOJ were pursuing enforcement actions against Goldman,” id. at 279. The weight of this expert evidence, “coupled with” “the generic quality of Goldman’s alleged misstatements,” persuaded me that Goldman had succeeded in rebutting the presumption of reliance outlined in Basic Inc. v. Levinson, 485 U.S. 224 (1988). Id. at 278–79.

After taking up this case, the Supreme Court did not ultimately determine whether Goldman had rebutted all evidence of price impact. See Goldman Sachs Grp., Inc. v. Ark. Tchr. Ret. Sys., 141 S. Ct. 1951, 1961, 1963 (2021). It nevertheless instructed that, in “assessing price impact at class certification, courts should be open to all probative evidence on that question – qualitative as well as quantitative

3

– aided by a good dose of common sense.” Id. at 1960–61 (internal quotation marks omitted).

In light of the Supreme Court’s guidance, we agreed that remand was appropriate so that the district court could reassess the price-impact evidence relating to the allegedly false statements. First, we acknowledged that the district court was required “to take into account all record evidence relevant to price impact, including the generic nature of Goldman’s statements.” Ark. Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc. (ATRS III), 11 F.4th 138, 143 (2d Cir. 2021) (internal quotation marks omitted). Second, we impressed upon the district court the need to consider the report of Goldman’s expert, Dr. Laura Starks, “which focused on the generic nature of Goldman’s statements.” Id. And third, we recognized that the district court should consider not only “expert testimony,” but also apply “common[-]sense” reasoning to “assess all the evidence of price impact.” Id. (internal quotation marks omitted). I agreed to remand the case, believing that each of these factors reinforced my prior conclusion that Goldman had rebutted all evidence of price impact.

II.

Now that the district court has certified the class yet again, the majority rightfully concludes that the latest class-certification order should be reversed because Goldman “rebutted Basic’s presumption of reliance” by “demonstrat[ing], by a preponderance of the evidence, that the [alleged] misrepresentations did not impact Goldman’s stock price.” Maj. Op. at 69. While I agree with this ultimate conclusion, I worry that the majority has charted a meandering course that, in addition to contradicting my earlier conclusions, obscures what should be an uncomplicated inquiry. That is, we are to weigh all types of evidence of price impact – including the “generic nature” of the disputed statements, “evidence . . . relevant to . . . materiality,” and “all [other] probative evidence” – whether presented as “expert testimony” or revealed as a simple matter of “common sense.” Goldman, 141 S. Ct. at 1960.

A.

As one might expect, my longstanding position in this case is fully at odds with Part I of the majority opinion, which concludes that the district court did not clearly err in rejecting the testimony of Dr. Choi and Dr. Gompers. On this point, not much more needs to be said beyond what has already been covered in my prior dissent. See ATRS II, 955 F.3d at 275–79 (Sullivan, J., dissenting). It simply

5

bears noting that Dr. Choi’s event study established that the drop in Goldman’s share price following a corrective disclosure was entirely attributable to the announcement of an SEC enforcement action against the company. Moreover, as Dr. Gompers recounted without contradiction, the releases of thirty-six news reports, beginning three years before the first corrective disclosure, revealed Goldman’s conflicts of interest (including ones concerning the Hudson and Abacus transactions specifically), and yet had no measurable impact on Goldman’s share price. Based on this expert testimony, my view was – and remains – that Goldman “rebut[ted]” the “presumption of reliance” by “demonstrating that news of the truth credibly entered the market” through these prior disclosures and “dissipated the effects of [any] prior misstatements.” Amgen Inc. v. Conn. Ret. Plans & Tr. Funds, 568 U.S. 455, 481–82 (2013) (internal quotation marks and alterations omitted); see also Halliburton Co. v. Erica P. John Fund, Inc. (Halliburton II), 573 U.S. 258, 279–84 (2014). 1

1 The majority correctly recognizes that, in my view, the “linchpin” of Goldman’s defense “remains the [thirty-six] dates of pre-disclosure reporting and the accompanying expert testimony.” Maj. Op. at 70. While this may be true, it is the overwhelming evidence offered by Goldman in its totality – that is, these prior disclosures, the exceedingly generic nature of the alleged misstatements, its mismatch with the genericness of the more specific corrective disclosures, and the evidence relevant to materiality – that makes this truly not a “close case[].” Goldman, 141 S. Ct. at 1970 (Gorsuch, J., concurring).

B.

I also disagree with Part II of the majority opinion, which finds “no merit to Goldman’s claim that . . . the district court . . . understated [the conflicts] statement[s’] generic nature.” Maj. Op. at 45–46. 2 For starters, common sense tells us that the alleged misstatements in Goldman’s Form 10-K filings were highly generic, as they merely stated that Goldman “ha[d] extensive procedures and controls that [were] designed to identify and address conflicts of interest,” while warning that “a failure to appropriately identify and deal with conflicts of interest could adversely affect [the company’s] business” and lead to “litigation,” “enforcement actions,” and “damage[]” to its “reputation.” J. App’x at 3278. According to the district court, this language attested to “Goldman’s specific approach to conflicts management” and its “sufficient conflicts procedures.” Sp. App’x at 23, 26–27 (emphasis added). And while the majority avoids reversal of this finding by insisting that it was not clearly erroneous, I am still convinced that such exceptionally “general” statements were not capable of “affect[ing]” the

2I concur with the majority that the district court clearly erred in evaluating the genericness of the business-principles statements. Accordingly, my concurrence is limited to a discussion of the conflicts-of-interest statements that Goldman made in its Form 10-K filings.

“price” of Goldman’s “securit[ies].” Goldman, 141 S. Ct. at 1960 (internal quotation marks omitted).

Indeed, when placed side-by-side, these Form 10-K statements and the corrective disclosures are a study in contrasts. Unlike the Form 10-K statements, the corrective disclosures – relating to the selection of the underlying assets for the Abacus CDO by Paulson & Co. and the purchases of the Hudson CDO by hedge-fund Dodona I LLC – were far more specific, given that they referred to particular transactions, financial products, and industry participants. Accordingly, “[a] good dose of common sense” leads to the obvious conclusion that the Form 10-K statements were highly “generic,” the corrective disclosures were appreciably more “specific,” and the “mismatch” between the “contents” of the two was striking. Id. at 1960–61 (internal quotation marks omitted).

C.

To be clear, my disagreement with Parts I and II of the majority opinion extends beyond these specific findings about Goldman’s price-impact experts and the “generic nature” of the statements at issue; it also goes to the heart of the majority opinion’s approach in assessing reliance under the Basic presumption at class certification.

For one thing, the majority’s stepwise consideration of Goldman’s expert testimony and the “generic nature” of the statements is difficult to square with existing precedents, which have never required courts to consider these price-impact factors in isolation. See, e.g., Halliburton II, 573 U.S. at 279–282, 284; Goldman, 141 S. Ct. at 1960–61. In the same vein, the circuitous path taken by the majority – i.e., first rejecting Goldman’s expert evidence, then affirming in part the district court’s genericness and materiality analyses (without addressing mismatch), before eventually doubling back to reach the opposite conclusion after considering mismatch and materiality in the context of our inflation-maintenance precedents – strikes me as unnecessary and overthinks what, in my view, was a relatively straightforward directive from the Supreme Court to assess “all probative evidence” of price impact. 3 Goldman, 141 S. Ct. at 1960. Finally, to the extent that the majority implies that a more “searching” price-impact analysis is required for inflation-maintenance cases only, Maj. Op. at 63, I strongly disagree. Based on my reading of the Supreme Court’s decision, courts must apply the

3 As but one example, the majority purports to stand behind the district court’s materiality analysis in Part I, where it “t[ook] the district court at its word” “that it considered cases bearing on materiality to the extent they presented issues overlapping with . . . price impact,” only to later reverse course in Part III and observe during its discussion of the inflation-maintenance theory that our materiality precedents do in fact indicate that a reasonable investor would not rely upon the alleged misstatements. Compare Maj. Op. at 45–47, with id. at 59–62.

genericness, mismatch, and materiality analyses to all questions of reliance, and not just to ones related to the inflation-maintenance theory.

D.

I offer a final observation, not as a criticism of the majority opinion, but simply as an acknowledgment of the predicament that the Supreme Court has created for lower courts tasked with assessing reliance at the class-certification stage of securities actions like this one. In the span of a decade, the Supreme Court has held that defendants may not challenge materiality at class certification, Amgen, 568 U.S. at 480–82, while also acknowledging that materiality evidence may be introduced to rebut price impact and reliance, Goldman, 141 S. Ct. at 1960–61. This instruction places district courts in a peculiar position.

As an initial matter, it’s hard to imagine how class-wide reliance based on the Basic presumption can be established under Federal Rule of Civil Procedure 23 without consideration of the statements’ materiality. To be sure, proof of materiality for class-certification purposes is not needed because plaintiffs’ materiality claims all rise and fall in unison, leaving no room for individualized concerns to predominate over class ones. See Amgen, 568 U.S. at 459–60. This is logical enough. That said, the inescapable reality is that plaintiffs must also satisfy class-certification requirements under Rule 23 for the element of reliance. See id.

at 466–67, 473. Crucially, the Basic presumption is appropriate only if the “fraud-on-the-market theory” holds true – that is, “investors” “rel[ied] on the market price’s integrity” and material statements were readily incorporated into share prices in an “efficient market.” Id. at 462–63, 466–67. Since, by definition, only material statements are reflected in market prices, a showing of materiality at the class-certification stage is needed – not to answer the merits question of materiality – but to satisfy a precondition to the very fraud-on-the-market theory that props up the Basic presumption of reliance. See id. at 490–91 (Thomas, J., dissenting).

It is difficult to conceive how a statement that is immaterial under an objective “reasonable[-]investor” standard could ever be relied upon by rational investors. Basic, 485 U.S. at 231–32; see also Louis Loss et al., Fundamentals of Securities Regulation 896 (7th ed. 2018). Further still, permitting the use of materiality evidence to resist class certification only when it is dressed in reliance’s clothing strikes me as needlessly exalting form over substance, since moving forward, it stands to reason that materiality evidence will virtually always be presented to courts tasked with resolving class-certification motions. Although “Goldman d[id] not ask [the Supreme Court] to revisit these precedents,” Goldman,

11

141 S. Ct. at 1962, the tension between Amgen and Goldman cries out for the Court to take another look at these decisions, since courts are now forced to navigate a materiality-reliance twilight zone that is shrouded in considerable confusion.

Fortunately, the fog is not so dense in the case before us. I agree with the majority that our materiality precedents support Goldman’s position that the corrective disclosures had no discernible price impact. See Singh v. Cigna Corp., 918 F.3d 57, 63–64 (2d Cir. 2019) (holding that no “reasonable investor would” “rely on” one defendant’s “Form 10-K statements as representations of satisfactory compliance,” since they were only “simple and generic assertions” about it “having” and “allocating significant resources” to compliance “policies and procedures” (internal quotation marks omitted)); see also ECA, Loc. 134 IBEW Joint Pension Tr. of Chi. v. JP Morgan Chase Co., 553 F.3d 187, 205–06 (2d Cir. 2009) (finding no materiality with regard to bank’s statements about its integrity and risk-management capabilities). Those precedents demonstrate that “no reasonable investor would have attached any significance to the generic statements on which Plaintiffs’ claims are based,” ATRS II, 955 F.3d at 278 (Sullivan, J., dissenting), and along with the expert testimony presented to the district court and a healthy dose

12

of common sense, lead us all to agree that the Basic presumption has been sufficiently rebutted.

III.

For the reasons discussed above and in my prior dissent, see ATRS II, 955 F.3d at 275–79 (Sullivan, J., dissenting), I remain persuaded that Goldman carried its burden of severing the link between the alleged misstatements and the price paid by Plaintiffs for their shares, thus rebutting “the presumption of reliance” that adheres where rational investors transact in an efficient market for securities, Basic, 485 U.S. at 248. I therefore join in the majority’s conclusion, though not its precise reasoning, and vote to reverse the district court’s class-certification ruling.