Class Actions Brief Blog

A leading feature of the Supreme Court’s decision in Wal-Mart Stores Inc. v. Dukes is the “rigorous scrutiny” the trial court must apply to determine whether the evidence plaintiffs offer to support class certification meets the requirements of Rule 23. Following the Supreme Court’s decisions in Wal-Mart and Comcast, “[i]t is now indisputably the role of the district court to scrutinize the evidence before granting certification, even when doing so ‘requires inquiry into the merits of the claim.’” Rail Freight, 725 F.3d at 253 (D.C. Cir. 2013) (quoting Comcast, 569 U.S. at 28); see also Wal-Mart, 564 U.S. 338, 351–52 (2011).

In re: Pre-filled Propane Tank Antitrust Litigation, decided November 9, 2021 by Judge Gary A. Fenner in the Western District of Missouri, lays out a view of such scrutiny. A purported class of propane tank consumers brought this antitrust action to recover based on allegations they paid inflated prices due to the defendant gas producers’ agreement to sell under-filled tanks to retailers that were subsequently purchased by the plaintiffs. The monetary stakes were significant: the plaintiffs’ expert estimated potential class wide damages of $117 million, before trebling.

The design of the class action covered consumers in Arizona, Nevada, North Carolina, New Mexico, Maine, Minnesota, Michigan, South Dakota, Utah, California, Iowa, North Dakota, and West Virginia. To support their motion for certification, and to demonstrate common proof of class-wide injury and damages, plaintiffs depended on economic analyses and opinions by Daniel A. Ackerberg, a professor of economics at the University of Texas. Likewise, defendants depended on a rebuttal analysis by Kevin Murphy, a professor of economics at the University of Chicago.

Maintaining that plaintiffs had failed to satisfy their burden to establish predominance under Rule 23(b)(3), Ferrellgas relied on an independent analysis of the market, providing evidence of defendants’ varied pricing to individual retailers throughout the proposed class period followed by a number of critiques and economic analyses testing the reliability of the common proof of a class-wide overcharge to the putative class plaintiffs had designed.

Pursuant to this effort, Professor Murphy also disaggregated Professor Ackerberg’s model, which was based on average wholesale prices, by re-running Professor Ackerberg’s own overcharge regressions separately for Ferrellgas’s largest retailers. In doing so, Professor Murphy concluded that Professor Ackerberg’s use of a single average overcharge estimate wrongly attributed an overcharge to a significant percentages of sales. The opening exchange of expert reports led to a “Round 2,” with plaintiffs moving for a reply and defendants moving for a sur-reply, where Professor Ackerberg submitted extensive exhibits and analyses, and Professor Murphy addressed these supplemental materials in his sur-reply.

After laying out an exacting standard for certification based on Comcast, Wal-Mart, and other key precedents, Judge Fenner explained that the predominance requirement cannot be met “in circumstances where a plaintiff’s common proof of impact fails to show that all or nearly all class members were in fact injured.” Judge Fenner then aimed at economic evidence central to the class certification question: “the Court must scrutinize [economic evidence] rigorously to ensure that it is capable of demonstrating impact on a class-wide basis: ‘[i]t is now clear . . . that Rule 23 not only authorizes a hard look at the soundness of statistical models that purport to show predominance—the rule commands it.’”

Applied here, these principles led Judge Fenner to articulate a two-part burden to establish predominance. First, the plaintiffs must show a common impact—here, an overcharge that affected all or nearly all retailers purchasing gas from defendants. Second, they must demonstrate that the retailers passed on those overcharges to all or nearly all of the plaintiff end-customers who purchased the under-filled tanks.

In evaluating whether plaintiffs met this burden, Judge Fenner dissected both experts’ opinions, including the details of the regression models, with a focus on the sub-regressions that disaggregated plaintiffs’ expert Professor Ackerman’s model to test for any measured overcharge to individual retailers. Judge Fenner’s comparative observation on the aggregation issue illustrates the deep dive he made into the expert reports: “Murphy’s disaggregation demonstrate[s] that significant proportions (over 30%) of the at-issue class sales did not exhibit an overcharge, even though Professor Ackerberg’s application of that same model on ‘aggregated’ data otherwise indicates that all of those sales were subject to an overcharge.” This reasoning led him to conclude that plaintiffs had not satisfied their burden to show a common overcharge.

In a companion to his common overcharge analysis, Judge Fenner found that plaintiffs’ reliance on averaging and aggregated data “masks” variation and “skews” the evidence. Judge Fenner supported this analysis by highlighting the business relationship between Ferrellgas and a large retailer, Wal-Mart. Ferrellgas’s negotiations with Wal-Mart, resulting in multiple cost reductions, were unique to Wal-Mart but “pooled” with pricing and sales to all other retail customers in Professor Ackerman’s model. A similar example of individualized cost concessions affecting hundreds of independent retailers in California was also part of the record and emphasized in the decision.

Evaluating these examples, Judge Fenner spoke to the hazard in “overcharge models that rely on averaging” because they are bound to “conceal variation in pricing and other individualized issues that may predominate.” Drawing from Eighth and Third Circuit rulings, Judge Fenner reasoned that, when the market for a product is grounded in individual negotiations, the risk is significant that any injury is individualized. He concluded that, by failing to account for these individual differences, Professor Ackerberg’s model “suffered” from a failure to employ common or class-wide evidence that could establish that defendants overcharged any particular retailer.

To engage Professor Ackerberg’s regression model, Judge Fenner relied on an econometrics treatise by Dr. Thad Mirer and a publication by the ABA’s Antitrust Section. Professor Ackerberg maintained that regression can evaluate price per gallon as a function of explanatory variables that purportedly represent “costs, demand, and the nature of competition.” Judge Fenner observed that his model was “notably bare” with respect to variables that purport to control for all factors impacting price other than the alleged conduct—notably, the individualized prices negotiated by Ferrellgas’s and AmeriGas’s tens of thousands of customers. Such individualized prices arising from these negotiations run counter to the predominance requirement of Rule 23(b)(3).

The order concluded with an analysis of the “pass-through” theory advanced by Professor Ackerberg, which the Court found insufficient on predominance grounds. Due to “reliance on averages” and “miniscule sample size,” Judge Fenner viewed the theory as unable to account for individual variation, reinforcing his conclusion that the plaintiff purchasers could not satisfy Rule 23(b)(3)’s predominance requirement.

In thoroughly evaluating the economic evidence presented by both parties’ experts, Judge Fenner offered a robust example of the “rigorous analysis” required at the class certification stage following Wal-Mart and Comcast. In re: Pre-filled Propane Tank Antitrust Litigation serves as a cautionary tale for parties seeking certification of a class action when their experts’ models may not withstand such demanding scrutiny.