The United States Requires a Rigorous Analysis of Common Impact at the Class Certification Stage
In the United States, courts place a heavy burden on plaintiffs to prove that class certification is appropriate. Plaintiffs must, in the first instance, show that: “the class is so numerous that joinder of all members is impractical” (the “numerosity” requirement); “there are questions of law or fact common to the class” (the “commonality” requirement); “the claims or defenses of the representative parties are typical of the claims or defenses of the class” (the “typicality” requirement); and “the representative parties will fairly and adequately protect the interests of the class” (the “adequacy” requirement).
In addition to these requirements, a U.S. plaintiff must also meet one of the requirements of Rule 23(b), which is where the fight over class certification usually occurs. In cases where damages are sought, as is common in antitrust class actions, Rule 23(b)(3) governs class certification. Rule 23(b)(3) requires a showing that “questions of law or fact common to class members predominate over any questions affecting only individual members” (the “predominance” requirement) and that “a class action is superior to other available methods for fairly and efficiently adjudicating the controversy” (the “superiority” requirement).
Under U.S. procedural rules, plaintiffs must prove with evidence that they meet each requirement, and that proof is subject to “rigorous analysis” at the certification stage, particularly as to predominance. To meet this burden of proof, plaintiffs must demonstrate that common evidence will establish that all (or nearly all) class members suffered damage from the alleged conspiracy. This stringent standard often requires U.S. courts to decide a “battle of the experts” before they certify the class and may “entail some overlap with the merits of the plaintiff’s underlying claim.”
Countries Outside the U.S. Typically Apply a Less Stringent Standard
Historically, class actions have been much less prevalent outside of the United States. As class action is a relatively new legal vehicle in other jurisdictions, new decisions and legislation continue to modify the bounds of such actions. Even those countries with relatively well-defined class certification regimes—such as Canada and the United Kingdom—have only a fraction of the number of class actions as the United States has. And both Canada and the United Kingdom employ a standard for evaluating evidence at the class certification stage that is less stringent than the U.S.-style rigorous analysis. Other countries approach collective actions more generally without a formal certification stage, particularly in continental Europe.
Canada Requires a Showing that is “Sufficiently Credible or Plausible to Establish Some Basis in Fact.”
Canada has one of the longest-standing class action regimes outside of the United States. Canadian courts require significantly less evidence than in the United States and do not weigh the merits of the expert opinions at the class certification stage. To demonstrate common impact among class members, a Canadian plaintiff must present an expert methodology that is “sufficiently credible or plausible to establish some basis in fact for the commonality requirement.” In a leading case on class certification, Pioneer Corp. v. Godfrey, the Supreme Court of Canada held that it was not necessary at the class certification stage to have a methodology that could show that each class member suffered a loss. The Godfrey decision underscored Canada’s view that factual disputes over whether common proof predominates over individualized proof to establish injury need not be determined at the class certification stage.
The United Kingdom Considers Whether a Proceeding is “Suitable” for Collective Action
The United Kingdom first allowed antitrust collective actions in 2015, with the passage of the Consumer Rights Act 2015. Under this new class action regime, a class representative can commence proceedings for breaches of competition law on behalf of a defined class of claimants. A class action claim may only proceed, however, if the Competition Appeal Tribunal (CAT) issues a collective proceedings order which requires, among other things, that the Tribunal is satisfied that the individual claims of the class members raise “the same, similar, or related issues of law.” But the U.K. has no analogue to the U.S. predominance requirement. The U.K. requires only the presence of common issues for which collective proceedings provide “an appropriate means for the fair and efficient resolution.”
The main case establishing the standards in the United Kingdom is the U.K. Supreme Court’s decision in Merricks v. Mastercard. The Merricks case involved a proposed collective proceeding on behalf of a class of some 46.2 million people claiming losses over a 16-year period. At the class certification stage, the CAT relied on Canadian jurisprudence for guidance. It accepted that the proposed expert methodology was sound in theory, but refused to issue a collective proceeding order because it was not persuaded that there was sufficient data available to adequately apply that methodology to determine the level of pass-on, and thus the amount of aggregate damages.
The English Court of Appeal set aside the order refusing certification and held, consistent with Canadian jurisprudence, that a proposed class representative need only show “a real prospect of success” for each of the certification requirements. It criticized the CAT for effectively conducting a mini-trial that required Mr. Merricks to establish more than a “real prospect of success” and stated that the court is not required to resolve at the certification stage conflicting issues of fact and evidence that can only be properly resolved at trial when pleadings, and fact and expert discovery are complete.
The Supreme Court broadly upheld the decision of the Court of Appeal and remanded the case back to the CAT to be evaluated applying a different and, in certain respects, lower threshold test. In particular, the Supreme Court held that the CAT erred when it held that difficulties in quantifying damages (because of the likely non-availability of data) were sufficient to require class certification to be denied. The Supreme Court held that, at the class certification stage, the CAT should ask whether the claim is relatively more suited for treatment as a class action or as an individual action. Hence, difficulties quantifying loss that would equally arise in individual proceedings do not, by themselves, provide a basis for refusing certification.
Economic Analysis: A Framework for Considering Optimal Similarity Standards for Class Certification
The different treatment of class actions around the world raises difficult choices for policymakers seeking to craft their own legal regimes. Economic analysis may help policymakers more systematically consider the tradeoffs inherent in designing such a regime, such as setting too low or too high a bar to show that prerequisites for class certification have been met, or imposing that bar at relatively earlier or later stages of the case.
While antitrust policies across jurisdictions differ in their goals and objectives, the discussion that follows is focused on maximizing economic efficiency. It assumes that the goal of an optimal antitrust class action regime will be to minimize the sum of: (1) deadweight loss resulting from anti-competitive behavior; and (2) the costs of dealing with class actions. However, the framework is general enough (and the mathematical notation is simple enough) to accommodate differing objectives. In some parts of the discussion that follow, we introduce some simple formal mathematical notation to give the framework some flexibility. However, by using formulas we do not intend to suggest that there exists a single, one-size-fits-all answer to this Issue.
For purposes of our analysis, we assume that the hypothetical policymakers are trying to identify the proper level of proof to assess whether members of a potential class have sufficiently similar characteristics, claims and evidence to proceed as one group. The policymaker is trying to decide where to set this Similarity standard.
While there are potentially multiple dimensions to Similarity standards, for the sake of the mathematical framework, we simplify it to a single dimension S. We call the lowest imaginable standard, perhaps equivalent to not having a Similarity hurdle to clear at all, s0 . We also posit a “maximum” Similarity standard, which can be thought of as a hurdle that no class action could conceivably clear, smax > s0. Between these two extremes lie all potential values of s, with lower values corresponding to less strict standards, and the strictness of the standards increasing as s approaches smax. We also assume that there is an optimal value, s*, which corresponds to the optimal standard, in the sense that economic efficiency is maximized, after taking into account considerations related to litigation costs and deterrence.
Policymakers will likely also give some weight to equity considerations, as the relative increase or decrease in litigation awards and settlements for certain values of s could benefit certain parties up and down the supply chain at the expense of others. Economists generally consider these to be transfers, without direct implications for economic efficiency. Still, as a non-economic point, policymakers will value fairness and equity, so any complete theory should take account of these factors as well. Economic transfers may also be relevant to the extent that they alter incentives, and thus behavior, and potentially efficiency.
Our discussion of the costs associated with differing standards will begin with the most direct effects and then consider how the behavior of the relevant entities would differ at different levels of s. We thus need to begin by specifying different outcomes, in terms of what happens to various claims when they are or are not certified.
Consider a putative class action (i.e., a lawsuit that has been brought on behalf of a class of plaintiffs but has not yet been certified by the court) and assume that a sufficiently large subset of the claims in the class are different enough that the putative class would not be certified under a higher standard sH, but that the putative class would be certified under a lower standard s L. In the case where the class is certified, we assume that discovery proceeds and that the parties eventually settle.
We turn next to the case where the standard is higher and class certification is denied. Class actions typically arise when economies of scale make it efficient to aggregate common claims. Accordingly, the ultimate outcome when a putative class action is not certified depends in large part on the number of putative plaintiffs, the Similarity of their claims, and the magnitude of individual damages. One potential outcome of non-certification is that some narrower alternative to the original putative class forms, where the narrower class has a stronger Similarity profile than the original class, still satisfies numerosity standards, and is still collectively large enough to justify litigation.
Plaintiffs that were part of the original class but are not part of a narrower class will generally fall into one of two groups: (1) those whose claims are too small to justify individual actions; or (2) those who, either in isolation or in conjunction with other similarly situated plaintiffs, are large enough that they can be brought as individual or joint actions involving a smaller number of plaintiffs. The denial of class certification on Similarity grounds need not end the claims but may result in a reorganization of the claims. For the former group, it may result in a narrower action, whereas for the latter, it may result in multiple or individual actions.
On the other hand, denial of class certification will often cause the putative class action to simply end, without any reorganization. While it is possible that individual purchasers might continue after a denial of class certification—as was seen in the Rail Freight case when a substantial number of large purchasers brought individual actions —the denial of class certification often dooms the case. This occurs when the costs of litigating the smaller, individual claims become prohibitive in light of the expected recovery, so that the plaintiffs and their attorneys are forced to abandon the case altogether. If the underlying claims are meritorious, then causing the entire litigation to end because of defects at the certification level is another social cost that would undermine the compensatory and deterrence goals of private antitrust litigation.
Wherever the standard is set, the relevant parties—potential plaintiffs, potential defendants, and the attorneys who represent them—will adjust their behavior. These changes in behavior have implications for the efficiency of various levels of the Similarity standard, which we turn to now.
Finding the Optimal Standard. It is possible to set too high or too low a bar for class certification, thereby creating a Similarity burden that is not socially optimal. In general, as the Similarity standards bar increases, the probability that a given case will successfully clear the class certification hurdle should decrease. If the bar is too high, then potentially meritorious class actions will not be brought, or they will be brought in a fashion that does not define the class to capture injured parties, or they will be brought and fail at class certification. If the bar is too low, this invites additional, potentially less meritorious class actions to be brought, or class actions where the class is defined improperly to include uninjured parties. To see why, consider the case of a potential class action for which, under the optimal standard s*, the costs associated with bringing a class action just exceed the benefits (in terms of the expected settlement or reward). A tightening of s means that for a given level of cost and effort, the potential for either a settlement or a reward has decreased, creating barriers to bringing meritorious claims. A relaxation of s means that for a given level of cost and effort, the potential for either a settlement or a reward has increased, inducing “entry.”
Costs Associated with Setting Strict Similarity Standards. There are potential costs associated with setting an overly strict Similarity standard. Using our previous notation, these are situations where s > s*.
Many Harmed Parties May Lack Recourse
A potential direct consequence of an increased incidence (or even risk) of “false negatives” is that fewer putative class actions will be filed in the first place. Since a denial of class certification often ends a case altogether, increasing the likelihood of denial will discourage many plaintiffs from bringing class actions, even when the underlying merits of a case are sound. If even a fraction of the marginal putative class actions—those that would have been brought and cleared the class certification stage under a less strict standard, but that would either not be brought, or would fail under the stricter standard—are meritorious, but involve claims that are not sufficiently large to bring individually, or cannot otherwise be efficiently litigated, the ramifications are potentially significant. Specifically, harmed parties would lack recourse and some legitimate claims that do not on their own warrant the expense of litigation could lose their only viable avenue for compensation. The net result is a potential transfer from those who were harmed to those who committed the antitrust violation.
Loss of Deterrence Due to False Negatives
The discussion above illustrates how a higher standard for Similarity could reduce the likelihood that firms committing antitrust violations face some monetary punishment. As firms incorporate this information into their decision-making, it is likely that some decrease in deterrence would follow. Undetected and unpunished violations of antitrust law can yield significant economic gain to firms. On the other hand, the risks associated with violations of the antitrust laws can act as a significant deterrent. Ultimately, the deterrent effect depends on whether the magnitude of expected punishment (in the form of fines and penalties, damages awards, or settlement payments) is large enough to offset the increased profits from collusion.
The higher incidence of false negatives would cause firms to lower their expectations of the monetary punishment associated with a collusive act. This may be particularly true for firms with sufficiently distinct customer bases, where the Similarity hurdle is likely to have the largest impact. In some situations, this could tip the balance so that the decreased expectation of a monetary penalty/punishment would mean that collusion is worth the risk, which could result in more price fixing at lower cost to the colluding firms, a social negative.
Possible Reduction in Class Action-Related Economies of Scale with Respect to Litigation Costs
All things being equal, as the Similarity standard increases, class actions that successfully clear that hurdle will become more homogeneous and the opportunity for inclusion of potential plaintiffs who were legitimately harmed (but in ways that are marginally different from other class members) decreases. Thus, in the course of applying a more rigorous analysis of common impact, there is an increased possibility of “false negatives.” In other words, a more stringent standard increases the risk that plaintiffs who were in fact harmed and for whom the relevant factual and legal questions predominate over individualized questions are, at the class certification stage, found to be sufficiently different to warrant their exclusion or to warrant a denial of class certification.
In such circumstances, where the putative class action fails to clear the Similarity hurdle, one of two things can happen. First, if plaintiffs’ individual or collective claims are large enough to justify stand-alone litigation, potential class members who were found to differ from the rest of the class may file parallel litigation. Similarly, as plaintiffs’ attorneys react to a higher standard, potential plaintiffs who, under the more stringent standard might jeopardize the chances of success at the class certification stage, may be excluded from the class through a narrowing of the class definition, again potentially resulting in an increased number of parallel actions. This potential increase in parallel actions would have the effect of undoing some of the inherent efficiency in class action regimes. Cases with common issues of law and fact would be duplicated, increasing discovery and other litigation costs relative to those that would prevail under the optimal Similarity standard.
Costs Associated with Setting Similarity Standards Too Low
On the other hand, there are costs associated with setting a Similarity hurdle too low.
Settlements and Damages Awards Get Paid Inefficiently
Lower standards can lead to parties who were not actually harmed getting rewarded, or to parties obtaining less compensation than they should. This can happen through at least two channels.
As previously discussed, the vast majority of certified class actions are resolved through settlement, which is the presumptive outcome of class actions that clear class certification. If some of the class actions achieving class certification are overly broad (in that some members of the class were not harmed) yet eventually settle, plaintiffs who were not actually harmed may receive a settlement award, and defendants who did not actually harm those plaintiffs may pay those settlement awards. While these harms could be ameliorated through settlement agreement provisions or claims administration procedures that screen out uninjured members, as the class gets larger and more complex, these procedures may become difficult to administer.
The corollary to the above point is that payments to plaintiffs who were included in an overly broad (but ultimately meritorious) class, but who were not actually harmed, may result in under-payment to those who actually were harmed. For instance, if a class action contains one million individuals and a settlement agreement provides $100 million to be equally divided among the class, then each individual would receive $100. But if 20 percent of the class was not actually harmed, then the harmed class members receive less than they otherwise would. Excluding the unharmed members but keeping the payment the same would lead to the same $100 million being provided to 800,000 individuals, or $125 each. To the extent that a lower standard leads to a broadening of putative classes, the prevalence of this unintended outcome is likely to increase.
Further, a payment to settle a less meritorious class action that under a strict Similarity standard would not have been certified raises its own equity and fairness concerns for later-settling plaintiffs. To the extent that defendants are forced to deplete their resources settling claims in class cases that would not have been certified under an optimal Similarity standard, they will have fewer total resources to settle different, possibly more meritorious claims in the future.
Higher Costs Associated with Discovery and Litigation of Cases That Lack Merit
A higher number of cases clearing class certification would, all else being equal, also increase the costs spent by firms on discovery and litigation in class action matters. The expenditure of some of these sums may be worthwhile if an otherwise meritorious case would only receive class certification under the looser standard. If, however, as discussed above, more classes lacking merit get certified, then litigants will also expend resources on discovery and litigation of frivolous cases that do not right wrongs or deter future wrongs, and may result in inequitable transfers of wealth.
An Increase in “False Positives” Could Undermine the Deterrence Effects Associated with Antitrust Laws
A simple model can help illustrate the concept. Assume that there are two possible actions for the firm to take—it can either collude or not collude. Assume that the operating profits (ignoring any risks of detection or punishment) from collusion πc are greater than those from not colluding πn. Assume further that there is some positive probability of the collusive acts being detected and leading to some large costs in the future. Let E(pc ) denote the present value of its expected punishment from collusion, taking into account the uncertainty of detection and punishment. In this simplified world, the firm will choose not to collude as long as πn ≥ πc − E(pc) (eq. 1) i.e., the profits from not colluding exceed the profits from collusion, once the expected punishment is taken into account.
Now, suppose that relaxing the Similarity standards makes it more likely that frivolous cases will be brought and will clear the class certification hurdle. The firm, wanting to avoid the risk associated with a trial, will settle. Under these assumptions, it now faces some risk, and therefore an expectation of some positive monetary cost or “punishment,” which we will denote E(pn), even when it chooses not to collude. Under this new scenario, the firm’s indifference equation becomes πn − E(pn) ≥ πc − E(pc) (eq. 2) i.e., the firm’s profits from not colluding are also offset by an expected “punishment.”
While E(pn) will likely be significantly smaller than E(pc), it nevertheless may be sufficient to tip the balance in some cases. That is, some firms for whom equation (1) holds may find that equation (2) does not hold, meaning that they would not collude under the optimal standard but would under the lower standard. Accordingly, it is plausible that an increase in frivolous lawsuits actually weakens the deterrence effect.
An Increased Risk of Lower Merit Claims May Chill Procompetitive Activities
The basic model presented above is an over-simplification: in reality, firms’ management of antitrust related risks cannot be distilled down to binary decisions. The example above does illustrate, however, that firms will need to be concerned, under some circumstances, not only with whether their activity is actually illegal, but also with the perception that their activity runs afoul of antitrust rules. In an environment where the probability of claims with little merit achieving class certification could be higher, this may be particularly true, as the risks relating to perceived violations are likely higher as well. Claims in which questions related to damages are “sufficiently credible or plausible” to clear class certification, but where those claims are found to be meritless can nevertheless impose costs on firms.
Accordingly, when the bar for class certification is lowered, firms may be less willing to engage in procompetitive or efficiency-enhancing activities that are harmless, if that harmlessness is not sufficiently evident for a court to throw out the case on motions to dismiss or motions for summary judgment.
For instance, suppose that oil and gas companies are accused of colluding to raise prices, after several firms imposed price increases, all of which went into effect simultaneously after an industry-wide trade association meeting. While there is no direct evidence of collusion, the facts are sufficiently circumstantial to allow the plaintiffs to prevail in early motion practice (for example, defeating a motion to dismiss). Let us also assume that the class was certified under a lower certification standard, but has clear Similarity problems that would have precluded certification under a higher standard (for example, it includes both entities whose purchase prices were locked into long-term contracts, and so were not affected by the increase, and those who buy on the spot market, who would be affected).
If we assume that the lack of evidence of collusion suggests parallel pricing rather than illegal price-fixing, the defendants would have a good chance of prevailing on the merits if they went to trial. Nonetheless, a sufficiently risk-averse firm would likely prefer to settle the case to avoid the uncertain outcome of a trial and potentially massive liability to the class. Thus, the expected litigation-related costs associated with benign activities, like attending a trade association event or independently matching competitors’ prices have increased, in part due to the lower hurdle for class certification.
A Summary of the Tradeoffs
The process of litigating class actions acts as a tax that imposes costs on society. The justification for this tax is that the threat of litigation and possibly damages awards incentivizes firms to comply with the antitrust laws. Generally speaking, these tax-like costs are closely related to the level of rigor required to meet the Similarity standard. This defines the central trade-off in setting Similarity standards: a higher standard for obtaining class certification generally results in fewer class actions (or successful class actions) and therefore lowers the “class action tax” but likely weakens the deterrent effect, imposing another set of costs on society.
The table below reorganizes and summarizes the key effects discussed above. Determining the relative importance of these costs is the policymakers’ task. However, the framework presented here should help clarify tradeoffs as antitrust authorities approach this complex, multi-layered problem.
