United States - Economist's Perspective: Class actions – litigation, policy and latest developments

This is an Insight article, written by a selected partner as part of GCR's co-published content. Read more on Insight


For a matter to proceed as a class action in the US, it must be first certified per the criteria listed in Rule 23 of the Federal Rules of Civil Procedure. Namely, a class must satisfy all criteria of Rule 23(a) and any one of the criteria listed in Rule 23(b). Although these criteria are questions of law ultimately answered by the trier of fact, expert economic analysis often is required to determine whether the proposed class indeed satisfies those criteria. Recent judicial decisions have helped to clarify the role of an expert economist in the class certification process and have helped to impose more analytic rigour in evaluating the Rule 23 criteria.

Requirements for class certification

The criteria listed in Rule 23(a) are summarised as: (i) numerosity, (ii) commonality, (iii) typicality and (iv) adequacy.[1] While economists might be asked to opine on one or more of those criteria, they most often focus on analysing the question of ‘predominance’ stemming from classes brought under the third prong of Rule 23(b).[2] In particular, to certify a class under Rule 23(b)(3) (cases where damages are involved), the court must find that ‘questions of law or fact common to class members predominate over any questions affecting only individual members, and that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy’.[3]

In evaluating predominance, economists are generally asked to analyse whether: (i) the challenged conduct, if true, would have affected all, or nearly all, proposed class members in a ‘common’ manner; and (ii) a formulaic method, common to all, or nearly all, proposed class members exists to measure damages stemming from the challenged conduct at a professionally reasonable degree of reliability. ‘Common’ here does not mean that all proposed class members must have been affected by the same magnitude. Instead, it means the challenged conduct, if true, would have had a measurable effect on all, or nearly all, proposed class members, and the impact on class members is directionally consistent. By way of example, in an alleged price-fixing case, even if each proposed class member may have paid different prices, common impact could still be demonstrated by showing that at least some purchases for each class member were purchased at increased prices as a result of the challenged conduct, regardless of any disparity in the increase.

The no-injury debate and expert evidence

An increasing area of debate in class actions litigation in the US involves claims for which some material proportion of class members has suffered no obvious harm or economic losses. This is occurring not only with antitrust cases, but also in matters alleging product defects, consumer fraud and statutory violations of state and federal laws with small or no concrete consequences for individual plaintiffs.

The US Supreme Court acknowledges the ‘importance’ of properly addressing ‘whether uninjured class members may recover’ a damages award.[4] In Tyson Foods, although the Supreme Court affirmed the Eighth Circuit’s ruling certifying the proposed class, Justice Kennedy, writing for the majority, noted that Tyson Foods could ‘raise a challenge to the proposed method of allocation [of damages] when the case returns to the District Court for the disbursal of the award’.[5] Kennedy’s statement was in response to the argument advanced by Tyson Foods that the plaintiffs had ‘not demonstrated any mechanism for ensuring that uninjured class members do not recover damages here’.[6] In a concurring opinion, Chief Justice Roberts wrote to ‘express [his] concern that the District Court may not be able to fashion a method for awarding damages to only those class members who suffered an actual injury’.[7] Roberts further opined that article III of the Constitution ‘does not give federal courts the power to order relief to any uninjured plaintiffs’ and that ‘if there is no way to ensure that the jury’s damages award goes only to injured class members, that award cannot stand.’[8]

Antitrust cases and expert evidence

Economic and statistical analyses are typically performed in support of certifying antitrust or anticompetitive class actions in the US. The First Circuit concluded in the 2015 Nexium decision that certification of a class ‘is permissible even if the class includes a de minimis number of uninjured parties’.[9] In certifying the class, the First Circuit laid out three principles necessary for class certification: (i) the theory of liability must conform to the injury caused to plaintiffs; (ii) the definition of the class must be ‘definite’; and (iii) the amount of the damages award must be limited to injured parties.[10]

The First Circuit returned to the issue of no-injury plaintiffs in its more recent decision in Asacol.[11] In that matter, the plaintiffs alleged that the defendant, Warner Chilcott, impermissibly delayed the entry of a generic competitor to Asacol, an anti-inflammatory drug used to treat colitis, by withdrawing Asacol from the market in lieu of a similar drug for which patent protection was longer than for Asacol.[12] Neither the plaintiffs nor Warner Chilcott disputed that around 10 per cent of the proposed class would not have switched to the generic version of Asacol had it been able to launch, despite the fact that the generic version would have been a lower-priced alternative.[13] The district court ruled that these uninjured proposed class members could be removed from the case through the assistance of a claims administrator and pointed to the prior Nexium ruling as its justification.[14]

The First Circuit rejected the district court’s view that the claims administration process would provide either a feasible or equitable method for identifying uninjured plaintiffs. In reaching this conclusion, the First Circuit observed that ‘this is not a case in which a very small absolute number of class members might be picked off in a manageable, individualized process at or before trial.’[15] The First Circuit further observed that ‘[p]laintiffs’ proposed claims process provides defendants no meaningful opportunity to contest whether an individual would have, in fact, purchased a generic drug had one been available.’[16]

More recently, the Ninth Circuit in Olean Wholesale Grocery v Bumble Bee Foods addressed the issue of uninjured plaintiffs.[17] The case followed the US Department of Justice’s 2015 investigation of price fixing by large tuna suppliers.[18] In its first review, the Ninth Circuit vacated the Southern District of California’s decision certifying three proposed classes of tuna buyers in the multidistrict litigation because the district court did not determine the number of uninjured parties in the proposed class.[19] Subsequently, an en banc opinion of the Ninth Circuit affirmed the district court’s certification of three subclasses of tuna purchasers that had alleged price fixing by tuna suppliers.[20] In particular, the en banc court ruled that the direct purchaser (DPP) class expert’s

[. . .] findings about the tuna market and tuna suppliers’ collusive behavior, pricing correlation test, regression model, and robustness checks confirmed his theory that the price-fixing conspiracy resulted in substantial price impacts, and that the impact was common to the DPPs during the collusion period. [… the] pooled regression model, along with other evidence, was capable of answering the question whether there was antitrust impact due to the collusion on a class-wide basis, thus satisfying this prerequisite of Rule 23(b)(3).[21]

Judge Lee and Judge Kleinfeld dissented based on three issues: (i) the defendants’ econometrician expert, if correct, showed that almost a third of class members may not have been injured, and thus the plaintiffs did not show the predominance of common issues; (ii) because class action cases typically settle following certification, a district court must resolve Rule 23 issues such as whether too many class members were uninjured; and (iii) the rejection of allowing only a de minimis number of uninjured class members created a split between the circuits.[22]

Focusing on the first criterion of predominance, the direct purchasers’ expert analysed the economic structure of the tuna market and performed a pricing correlation test purportedly showing that tuna suppliers’ prices tended to move together, indicating collusion could plausibly exist and would have a common, supracompetitive pricing impact.[23] The expert also provided a statistical regression model that used explanatory variables to isolate what he claims to have been the degree to which prices were higher because of the alleged collusion.[24] Based on his model, the expert found that 94.5 per cent of DPPs had a least one purchase above the predicted but-for price, further supporting common impact from the alleged conspiracy.[25]

The defendants’ expert argued that ‘it was not statistically appropriate to use a pooled regression model for transactions in the tuna market, given the multiple individualized differences among class members, such as disparities in negotiating tactics and bargaining power.’[26] In his opinion, ‘tuna transactions should not be pooled due to individual differences in each purchaser’s transactions,’[27] and he claims to have found that 28 per cent of DPPs did not have a positive, statistically significant overcharge.[28]

The en banc decision found the pooled regression models could control for individualised differences, stating that ‘[i]n antitrust cases, regression models have been widely accepted as a generally reliable econometric technique to control for the effects of the differences among class members and isolate the impact of the alleged antitrust violations on the prices paid by class members,’[29] and that a uniform overcharge was not implausible.[30]

Moreover, the en banc court ruled that the defendants’ expert did not show that 28 per cent of the class was potentially uninjured, because some purchasers simply had too few transactions to generate statistically significant results from his model.[31] The en banc court stated that ‘even class members with limited transactions during the class period can rely on the pooled regression model as evidence of impact on similarly situated class members.’[32] Whether or not the plaintiffs’ analysis would have been persuasive to a jury, the en banc court ruled that it was capable of showing class-wide impact.[33]

Consumer protection cases

Many class actions brought under state consumer protection laws allege economic losses are due when companies make false representations or omit information about their products to gain a market advantage over their competitors. In cases where the evidence indicates that buyers received substantial benefits from the products and would likely purchase the same product absent the challenged conduct, a pricing premium is the proper measure of economic loss. Defendants often challenge pricing premium measures because whether the market value of products would change in the absence of the challenged conduct depends on many factors, including exposure to the alleged omissions or misrepresentations, preferences of buyers, value in use and the costs of substitutes.

Damage models can be based on market information to estimate the pricing premium due to the alleged misconduct. Regression models often are used to isolate the effects of certain market factors and product features on product prices. Reliable results isolating the marginal effect of a particular marketing representation can be confounded, however, by the available market information, level of aggregation of the pricing information or randomness of the purchasing environment. Courts have found that plaintiffs failed to satisfy Comcast using these models.[34]

An alternative approach to isolating the alleged pricing premium relies on survey methods and market simulations. Conjoint methods are a particular type of survey to elicit consumers’ valuations of product attributes (which can include representations) and can measure incremental willingness to pay (WTP).[35] Conjoint methods can be combined with a market simulation to incorporate supply-side considerations in damages estimates.

Whether combined with a market simulation or not, these methods also have reliability limitations that have concerned the courts, especially in the context of complex product supply chains. For example, during In re Fluidmaster, Inc consumers filed a consolidated class action complaint in multidistrict litigation against the upstream manufacturer of allegedly defective water supply lines connecting to plumbing fixtures.[36] The court rejected the proposed conjoint survey to support the damages claim because ‘[a]sking an unrepresentative group of purchasers to artificially assign values among an arrangement of potentially unimportant attributes that fails to approximate real-world purchasing decisions does not seem designed to produce a reliable WTP estimate that can be used to calculate class-wide damages.’[37]

Benefitting from alleged conduct and expert evidence

Some putative class members not only are not injured by the conduct at issue but may have actually benefitted from that conduct. Recently, several market manipulation cases have been brought alleging that some form of anticompetitive conduct affected both the price of a particular commodity and the associated derivatives price for that commodity – one example is the In re Aluminum Warehousing Antitrust Litigation.

In Aluminum Warehousing, the defendants (Goldman Sachs, JP Morgan and Glencore, along with their associated warehouses) were accused of anticompetitive conduct that had the supposed effect of increasing prices for physical aluminium. According to the plaintiffs’ theory of the case, first-level purchasers of physical aluminium – ie, entities that purchased aluminium directly from a smelter – were harmed by incurring higher prices. But as the plaintiffs themselves readily conceded in their motion practice, large holders of physical aluminium would have benefitted from increased aluminium prices to the extent that such elevated prices would have boosted the value of their holdings. Such holders would have included many of the putative class members.

The defendants’ expert further identified other possible benefits experienced by putative class members. As explained by Judge Engelmayer in his opinion denying class certification, the expert found that some putative class members, such as entities trading in aluminium, may have generated trading profits from speculating on future prices of aluminium. Some of these same entities may also have received from the defendants certain incentive payments, which were themselves alleged to have been anticompetitive. These trading profits and incentive payments would, in the view of the expert, represent benefits that would have to be offset against any claimed overcharges when assessing the issue of common impact. Such an analysis would, fundamentally, require fact-intensive investigation of an individualised nature.

The facts and circumstances of the Aluminum Warehousing matter also raised the possibility of economic ‘conflicts’ among putative class members. One source of these potential conflicts resulted from some putative class members’ association with the very conduct that the named plaintiffs were challenging as anticompetitive. As mentioned above, some putative class members received incentive payments from the defendants to store additional aluminium in the defendants’ warehouses. Payment of these incentives was alleged to have contributed to elevated prices for physical aluminium. Thus, had the putative class been certified, at the merits phase of the litigation different putative class members may have wanted to pursue different theories of antitrust liability – for example, whether to claim that incentive payments were illegal or not – and those theories may have been in economic tension with one another. This would be particularly true with respect to developing a potential damages model accounting for differing theories of antitrust liability.

More broadly than the Aluminum Warehousing matter, any number of alleged market manipulation schemes exist in which the challenged conduct supposedly causes artificial deviations in the price to physically acquire some commodity. A putative class of plaintiffs would then likely bring suit alleging injury on the basis of those supposed artificial prices. However, many of the same entities included in the putative class may also have engaged in speculative trading on that commodity. By way of example, in an alleged scheme resulting in higher prices for a commodity, an entity may have been harmed by physically purchasing that commodity at too high of a price, while simultaneously benefitting from buying futures or other financial instruments that would increase in value as the price of the underlying commodity increased. From an economic perspective, then, it would be inconsistent for that entity to claim harm without also acknowledging a benefit stemming from that same challenged conduct. In the extreme, some entities so situated may have, on the net, benefitted from the conduct at issue.

In a market manipulation (or other) matter involving potential offsetting benefits, deducing who benefited and by how much would most likely require fact-intensive discovery and analysis on a class member-by-class member basis. Such individualised inquiry might be a sufficient basis for denying class certification.


Court decisions in US class actions continue to emphasise the importance of expert economic evidence, and more rigorous standards for that evidence, in the class certification process. In particular, for a class to be certified, rigorous economic analysis based on common evidence is required to demonstrate that all, or nearly all, proposed class members were impacted by the challenged conduct and that a formulaic method exists by which to measure damages stemming from that conduct. Due to the increasing rigour required of economic experts, more attention is given to demonstrating the presence or absence of uninjured plaintiffs and plaintiffs who might actually benefit from the alleged conduct.


[1] Fed R Civ P 23(a).

[2] Economists may also be asked to opine on issues related to ascertainability, or whether proposed class members can be identified ‘with reference to objective criteria’. See, for example, In Re Nexium Antitrust Litigation, 777 F3d 9, 19 (First Cir 2015).

[3] Fed R Civ P 23(b)(3).

[4]Tyson Foods, Inc v Bouaphakeo et al, 136 S Ct 1036, 194 L Ed 2d 124 (2016) at 1050.

[5] id at 1050.

[6] id 1049.

[7] id at 1050 (Roberts, CJ, concurring).

[8] id at 1053.

[9]In re Nexium Antitrust Litigation, 777 F3d 9 (1st Cir 2015).

[10] id at 18–19.

[11]In re Asacol Antitrust Litigation, 907 F3d 42 (1st Cir 2018).

[12] id at 44.

[13] id at 47.

[14] id.

[15] id at 53.

[16] id.

[17]Olean Wholesale Grocery v Bumble Bee Foods, F4th, 2022 WL 1053459 (9th Cir 8 April 2022) (en banc). This decision has been appealed to the Supreme Court. See petition for a writ of certiorari, StarKist Co et al v Olean Wholesale Grocery Cooperative et al (US 8 August 2022).

[18]Olean Wholesale Grocery v Bumble Bee Foods, F4th, 2022 WL 1053459 (9th Cir 8 April 2022) (en banc), at 15–16.

[19]Olean Wholesale Grocery v Bumble Bee Foods, 993 F3d 774 (9th Cir 2021), at 34–35.

[20]Olean Wholesale Grocery v Bumble Bee Foods, — F4th —, 2022 WL 1053459 (9th Cir 8 April 2022) (en banc), at 8.

[21] id at 10–11.

[22] id at 12.

[23] id at 34.

[24] id at 35.

[25] id at 37.

[26] id at 38–39.

[27] id at 39.

[28] id at 39–40.

[29] id at 47.

[30] id at 48.

[31] id at 40, 54.

[32] id at 54.

[33] id at 54.

[34] See, for example, Werdebaugh v Blue Diamond Growers, No. 12-CV-02724-LHK, 2014 WL 7148923 (ND Cal 15 December 2014). Comcast determined that Rule 23(b)(3) required courts to examine whether the damages model modelled ‘only those damages attributable to’ the theory of the unlawful conduct alleged. Comcast Corp v Behrend, 569 US 27 (2013).

[35] See, for example, Daniel McFadden, ‘The Choice Theory Approach to Market Research,’ Marketing Science 5(4) (Fall 1986).

[36]In re Fluidmaster, Inc, Water Connector Components Product Liability Litigation, No. 14-CV-5696, 2017 WL 1196990 (ND Ill 31 March 2017).

[37] id at *31.

Unlock unlimited access to all Global Competition Review content