On July 23, 2020, Judge Paul A. Englemayer of the U.S. District Court for the Southern District of New York denied a motion to certify a proposed class of direct purchasers of aluminum in a decision that may signal a trend toward courts rejecting the use of statistical averages to demonstrate a classwide impact of alleged anticompetitive conduct. In re Aluminum Warehousing Antitrust Litig., No. 13-MD-2481 (PAE), 2020 WL 4218329 (S.D.N.Y. July 23, 2020). In so ruling, Judge Englemayer wrote that the use of averaging “flouts the requirement that an expert’s model reliably prove that each putative class member suffered individual injury.”[i]
Many of the most significant class actions in the Southern District of New York over the past eight years have been antitrust cases involving alleged manipulation of a financial benchmark or rate by large financial institutions. These lawsuits can generate significant defense costs – a familiar trend that shows no signs of abating in the current climate.[ii] But in their pursuit of strategies to more efficiently plead and litigate rate-manipulation cases, the plaintiffs’ bar has relied extensively on the use of statistical and econometric analyses of market behavior to underpin allegations of classwide impact.
This strategy follows a similar storyline: (1) plaintiffs file a class action complaint using statistical analysis to allege that purported pricing anomalies demonstrate the existence of anticompetitive manipulative conduct that caused antitrust injury to all participants in that market; (2) survive, “albeit barely,” a Rule 12 motion challenging these statistics-driven pleadings, which often lack allegations of misconduct against a specific defendant or particularized harm to the named plaintiff(s);[iii] and (3) secure settlements prior to class certification or summary judgement.
As a result of this pattern, statistics-driven rate-manipulation theories have generally gone untested beyond the pleadings stage, leaving key questions as to whether these theories can support certification of a class unresolved. But in In re: Aluminum Warehousing Antitrust Litigation,[iv] Judge Englemayer noted that “courts have disdained models that have found classwide price impact by means of averaging impact across a class period[,]” and held that plaintiffs’ averaging was not “common proof that conspiratorial conduct caused pricing injury to all purchasers during the more than six-year class period.”[v] This ruling highlights the vulnerability of reliance on statistical averages at the class certification stage.
DENIAL OF CLASS CERTIFICATION IN IN RE: ALUMINUM WAREHOUSING ANTITRUST LITIGATION
In re: Aluminum Warehousing Antitrust Litigation is an expansive multidistrict litigation in which a putative class of plaintiffs alleges they purchased aluminum at inflated prices as a result of a rate-manipulation conspiracy between several major financial institutions and aluminum warehouse operators.[vi] Aluminum in the relevant market is typically purchased by contracts in which one component of the final purchase price is a regional premium associated with the cost of delivery.[vii] Plaintiffs allege that, in order to increase the value of aluminum derivatives tied to the underlying asset prices of the metal, the defendants engaged in a variety of tactics to drive up the cost of regional premiums, thereby causing plaintiffs to pay inflated costs for aluminum they purchased by contract.
At the class certification stage, Judge Englemayer subjected plaintiffs’ theory to a rigorous Rule 23 analysis centered on whether a chain of statistical models was capable of reliably demonstrating classwide antitrust injury and damages, i.e., how the final purchase prices paid by proposed class members were impacted by the defendants’ alleged manipulation of regional premiums.[viii] Plaintiffs, who had relied on a statistics-driven theory of classwide impact throughout multiple stages of motion practice and discovery, argued that their expert models could produce classwide proof by estimating the “average effect” of increases in the regional premium at issue on the final purchase price paid by each proposed class member.[ix]
But Judge Englemayer found that the statistical models developed by plaintiffs’ expert to measure antitrust injury and damages suffered from a “range of significant methodological infirmities” owing to their improper use of statistical averages.[x] More specifically, the use of averages failed to account for the “lumpiness” of the alleged conspiratorial conduct and “raise[d] questions about whether all direct purchases were in fact harmed throughout the lengthy conspiracy.”[xi] In the court’s view, one would need to conduct an individualized analysis of each proposed class member’s transactions to determine antitrust impact and injury.[xii] Based on this determination, Judge Englemayer found that the proposed class failed to satisfy the predominance requirement of Rule 23, and therefore denied the plaintiffs’ motion for class certification.[xiii]
KEY TAKEAWAYS FOR ANTITRUST DEFENDANTS AND THEIR COUNSEL
- Class action plaintiffs are increasingly using statistical and econometric analyses to plead antitrust claims alleging rate-manipulation conspiracies.
- Class action pleadings based on statistical analyses enable plaintiffs to plead broad conspiracies across entire markets and lump together differently-situated groups of market participants under the same banner.
- Courts are critical of class action pleadings that rely heavily on statistical averages to support allegations of classwide impact, but generally allow such pleadings to survive Rule 12 motions.
- Based on a trend toward settlement after the Rule 12 stage, there is very little precedent addressing the use of statistical averages to support rate-manipulation conspiracy claims at the class certification stage.
- In re: Aluminum Warehousing Antitrust Litigation is an example of the potential shortcomings of statistical averages to satisfy the requirements of Rule 23 and may signal a general reluctance by courts to allow plaintiffs to rely on statistical analyses rather than particularized allegations of injury and damages.
[i] In re Aluminum Warehousing Antitrust Litig., No. 13-MD-2481 (PAE), 2020 WL 4218329, at *48 (S.D.N.Y. July 23, 2020).
[ii] Dan Clark, Legal Departments Continue to Spend More on Class Action Litigation, Law.com (July 29, 2020), available at https://www.law.com/corpcounsel/2020/07/07/legal-departments-continue-to-spend-more-on-class-action-litigation/.
[iii] Courts have criticized pleadings that rely heavily on generalized statistical analyses that fail to distinguish between individual defendants. See In re GSE Bonds Antitrust Litig., 396 F. Supp. 3d 354, 364 (S.D.N.Y. 2019) (statistical averages “can flatten or hide trends that might tell a different story”); Fire & Police Pension Ass’n of Colorado v. Bank of Montreal, 368 F. Supp. 3d 681, 698 (S.D.N.Y. 2019) (rejecting use of aggregate data purportedly showing defendants’ net trading positions as not specific to any single defendant). However, courts have generally permitted these statistics-based pleadings to survive past the Rule 12 stage. See, e.g., In re Commodity Exch. Inc., 213 F. Supp. 3d 631, 660–66 (S.D.N.Y. 2016) (plaintiffs’ use of statistical averages of gold pricing fluctuations “barely nudged” antitrust claims over the line of plausibility); In re GSE Bonds Antitrust Litig., No. 19-CV-1704 (JSR), 2019 WL 5791793, *4 (S.D.N.Y. Oct. 15, 2019) (aggregated statistical analysis of bond pricing data plausibly established duration of alleged conspiracy).
[iv] In re Aluminum Warehousing Antitrust Litig., 2020 WL 4218329, at *1-2.
[v] Id. at *48.
[vi] Id. at *1–2.
[viii] Id. at *16.
[ix] Id. at *53.
[x]Id. at *48, *53–54.
[xi] Id. at *48.
[xii] Id. at *54.
[xiii] Id. at *45–48.