The Undemocratic Class Action


Class actions can have profound effects. But theorists, policymakers, and judges have long worried that attorneys can use them for their own advantage, reaping generous rewards for themselves while class members receive next to nothing. Unlike citizens or shareholders, members of a class cannot exercise democratic control over the attorney that nominally works on their behalf. I label this the democratic critique of class actions, and it has been the dominant framework for understanding class actions, shaping both case law and reform proposals.

The democratic critique is based on a false premise, though, because it does not take into account the downsides of democratic control. Drawing insight from the political economy literature, I show that there is an inevitable trade-off between control and expertise: if class members had control, they would lose the specialized knowledge that the attorney brings to the table. In other words, voting over litigation decisions would actually make class members worse off. Strategic features of the relationship make it hard for them to accept even well-meaning advice from their attorney: the mere possibility that the attorney might mislead them prevents information from being conveyed. Perversely, honest advice actually leads class members to make especially bad decisions. Given the choice, class members would rather delegate matters to the attorney instead of exercising class action democracy.

By blunting the dominant critique of class actions, this Article goes a long way to reestablishing their legitimacy. It also suggests better ways to protect class members’ interests. Rather than looking to institute democratic procedures in class actions or find some substitute for them, judges should engage in careful scrutiny of attorney fee arrangements and the structure of settlements. Class action law should ensure that the attorney does well if and only if their client, the class, does.


Class actions are at once among the most consequential and the most controversial elements of civil procedure. A single class action can “hold the fate of an industry in the palm of its hand”[2] or reshape civil rights on a national scale.[3] Yet at the same time, class actions provoke a range of concerns, both constitutional[4] and practical.[5] The locus of this controversy is the complicated relationship between the class and their attorney.[6] Class actions proceed on behalf of members of the class who typically have little to no direct involvement in the litigation. Their interests are represented by “named plaintiffs,” but they often serve as no more than figureheads because they lack the means and inclination to supervise the attorney.[7] There is the persistent worry that class members cannot control their lawyers, who, in turn, are free to turn the litigation to their own personal advantage. The concern is that “monsters are loose in the land”[8] as class actions end up lining the lawyers’ pockets while the actual plaintiffs receive little to no benefit. Furthermore, class members look to be sacrificing their rights and autonomy in exchange for efficiency and economies of scale.[9]

It is hard to overstate the impact of this concern. Many class action doctrines and statutory fixes[10] developed out of a desire to hold class action attorneys accountable.[11] In light of this worry it is hardly surprising that courts have sharply limited class actions,[12] treating the procedure as something of a last resort and seemingly preferring individual litigation to class actions wherever feasible.[13] It has also been the primary focus of theorists, leading to an ongoing crisis of legitimacy for class actions[14] and prompting a range of proposals to increase the active participation of class members, such as: enhancing class members’ exit rights,[15] leveraging mass communication technologies,[16] creating an aggregate litigation market,[17] and substituting non-governmental organizations in place of entrepreneurial attorneys.[18] This idea is not tied to an ideology—defenders and opponents of class actions alike worry about whether the class can really direct the litigation that is supposed to vindicate its rights.[19]

At bottom, this argument identifies a principal-agent problem: the principal (members of the class) cannot adequately control their agent (the attorney). Principal-agent problems are everywhere, though. They dominate both corporate law[20] and politics.[21] But shareholders and citizens both get to hold their agents accountable through elections—they can “throw the bums out of office” or things to that effect. These options are conspicuously absent in class actions or are, for one reason or another, rendered ineffective.[22] Most notably, even when class members have the power to exit the class and potentially “vote with their feet,” they rarely have the incentive to actually use it.[23]

Class actions therefore look unusually bad compared to other kinds of collective action because the principal lacks the means to control their agent.[24] I label this argument the democratic critique of class actions, and it has long been the dominant framework for understanding class actions.[25] I refer to this critique as democratic not only because class actions are collective institutions that are often compared to ones that have explicit democratic procedures. The problem as I see it is also that the class members themselves cannot steer their own litigation.[26] The kind of accountability enforced by elites like judges is different, though as we shall see, judges have a role to play.[27] The critique is pervasive, applying to nearly all class actions and definitely to those cases where the class action would be most important—those where each potential plaintiff has such a small stake that it would not be worth litigating on its own. The democratic critique thus renders all class actions suspect.

The democratic critique, however, relies on a mistaken premise: democratic accountability is not always the best way to address principal-agent problems.[28] Drawing on the political economy literature, this Article shows that the democratic critique suffers from two main flaws. First, it focuses too much on ex post controls like voting.[29] Second, and more fundamentally, the democratic critique neglects the complexities of the principal-agent relationship. Numerous studies show that control comes at a cost: the principal will not be able to take full advantage of the agent’s expertise, usually the reason they enlisted the agent in the first place.[30] If the principal retains final say over the decision, perhaps through some kind of vote, they end up losing access to a substantial amount of that expertise.[31] All of this occurs even if everyone involved is both scrupulously honest and perfectly rational.[32]

Accordingly, we should ask: if class members could vote on litigation matters, i.e., exercise the democratic accountability that appears to be lacking, would that option actually make them better off? If not, then this more nuanced understanding of principal-agent dynamics robs the democratic critique of much of its force. In class actions, when faced with this unavoidable trade-off between control and expertise, class members are generally better off surrendering control. That is, if class members were given the choice they would plausibly opt for the less democratic or representative procedure. Furthermore, democratic procedures bring complications and pathologies of their own—like pandering.[33] Rather than concentrate on implementing democratic procedures or some substitute for them in class actions, a more fruitful avenue for reform is to ensure that courts are vigilant about the structure of both lawyer fee awards and settlements.[34] These improvements are achievable and would further ensure the attorney is a faithful agent of the class. This approach, rather than class action democracy, better ensures class counsel will serve the best interests of the class. The best answer to the democratic critique, or any principal-agent problem, is not necessarily more democracy. My arguments also suggest the possibility for different rules in the rare cases where class members are exceptionally well-informed,[35] such as when local governments are members of the class.[36]

This Article makes several contributions. Most centrally, by exposing the flaws in the dominant critique, it offers a rejoinder to the widespread wariness, if not outright hostility, to class actions that pervades the law.[37] It also explains why reform proposals have yielded such mixed results:[38] the problems they aim to solve, inspired by the democratic critique, have to some degree been misidentified. Finally, it helps reorient attention to structures that would ensure fair, efficient, and legitimate means of aggregate litigation. Class action law is far from perfect, but it is imperative to correctly diagnose the problems if we mean to improve it. On a practical level, this means, among other things, equipping judges with standards for evaluating class action settlements they can both actually use and that protect the integrity of the litigation.[39]

This Article proceeds in three Parts. Part I details the democratic critique and the endemic principal-agent problem underlying it. Part II further situates the democratic critique in the case law and the literature and explains the relationship between my arguments and deterrence. With the distinct problem of class action democracy clearly identified, Part III reexamines it using the tools of political economy and modern social science, illustrating the trade-off between control and expertise and its consequences in the specific context of class actions.

Before going any further, though, a caveat is in order. Nothing here should be construed as a general statement against democracy. It all depends on the context, and the focus throughout this Article is class action litigation. Litigation is a rarefied environment where expertise is extremely important and, if the class action’s procedures are functioning well,[40] the members of the class share roughly the same goal (win the case and receive a large award). This certainly does not describe ordinary politics, where people may disagree about fundamental conceptions of the good, priorities, and so forth.[41] In those situations, democratic procedures are often essential.

I. Class Action Accountability

A. The Endemic Principal-Agent Problem in Class Actions

The democratic critique of class actions consists of two main parts: (1) the presence of a principal-agent problem, and (2) that class members cannot resolve that problem because they lack the opportunity or incentive to substantially participate in the litigation. Principal-agent problems are fundamentally problems of accountability and control; the problem is how to hold the agent accountable for their actions and control them so that they act in the best interests of the principal.[42] Such problems are common in a variety of contexts, including ordinary run-of-the-mill litigation. By their very structure as an aggregation vehicle, though, class actions introduce additional complications, especially when we consider the archetypal class action where numerous small claims are aggregated.

A necessary ingredient in any principal-agent problem is divergent interests: if the principal and the agent have identical interests, then the problem evaporates; the agent can be trusted to look after the principal’s interests since those are really her interests, too. In fact, one of the time-tested mechanisms for mitigating principal-agent problems is to align the interests of the parties involved.[43] However, class actions lead the interests of the attorney (the agent) and the class members (the principal) to diverge because the attorney unavoidably has so much more at stake than any single class member.[44] Their fees, especially under the near universal contingency fee arrangement where they are paid a portion of the plaintiffs’ recovery, are based on the class as a whole and, by definition, classes involve numerous class members.[45] To put things more concretely, imagine a case where each class member has $30 at stake. With that relatively small claim, class members would have no problem going to a trial where they only have a 50% chance of winning, and would happily opt for trial rather than accepting a $10 settlement. The expected value of the trial—50% of $30, or $15—is greater than the $10 settlement offer, and the values involved are small enough that considerations like risk aversion should not play any significant role. All else being equal, class members are content with higher risk, higher reward litigation strategies and choices.[46] Or more to the point, they are more content with such choices than their attorney will be.

The attorney has a different perspective. Continuing the example, suppose there are a million class members each with $30 at stake.[47] Under something like a 30% contingency fee, then the litigation is worth $10 million to the attorney, a sum that dwarfs the $30 that each class member had at stake and is likely to lead the attorney to be risk averse. Thus, the attorney views a safer option like settlement far more favorably than a high risk one like a trial; risk aversion affects her assessment. Similarly, the attorney might prefer lower risk theories of the case and litigation strategies—settlement is simply a clear illustration of the phenomenon. Moreover, we do not need to posit that it is psychological tendencies that steer the attorney’s interests. With a contingency fee—which, again, almost all class actions that aggregate small individual claims utilize—the attorney is effectively advancing the litigation expenses for the class action. They are expending substantial resources in working on the case and undertaking various opportunity costs; each hour they spend working on the class action is an hour that cannot be used on another case, investments they will not recoup unless they settle or win the case. According to one dataset of federal securities class action cases that settled between 2007 and 2012, the average case lasted four years, amounting to a massive amount of lost working time; the expenses associated with litigating the case were on average more than $700,000.[48] So, receiving nothing from the case usually carries additional costs for the attorney—they may have to close their firm or lay off employees—all of which leads them to behave in a risk averse manner.[49] Unlike the principal, then, the agent in a class action is not perfectly content adopting high risk, high reward strategies; she views things differently.

Modern class action law exacerbates all this. Class actions are very front-loaded: to make it past the motion for class certification—the first step in a class action proper—the plaintiffs must satisfy demanding procedural requirements.[50] As the Supreme Court has explained: “Rule 23 does not set forth a mere pleading standard. A party seeking class certification must affirmatively demonstrate his compliance with the Rule—that is, he must be prepared to prove that there are in fact sufficiently numerous parties, common questions of law or fact, etc.”[51] This will involve discovery, the main source of litigation costs,[52] extensive briefing, and even the plaintiffs establishing parts of their case-in-chief at an early stage in the litigation,[53] all increasing the expense that the attorney is effectively advancing or investing in the class action.

This differential valuation of the case leads to a fundamental, structural difference in the interests of the attorney and the class members. It is an endemic problem in class actions. As John Coffee puts it: “Across a broad range of cases, plaintiffs’ attorneys will be more risk averse than class members in considering settlement offers and will wish to accept many offers that the class will rationally wish to reject. . . . [T]his conflict over risk is basic and recurring.”[54] Therefore, class actions have one necessary ingredient of a principal-agent problem—the interests of the principal and the agent differ.

In and of itself, the attorney urging settlement or otherwise “playing it safe” is not much cause for alarm. Settlement is ubiquitous and often encouraged.[55] The class action attorney, however, is willing to pay a premium for it, leading the class to recover less than their case is worth. Continuing the numeric example of this Section, if each class member has a claim worth $15, taking into account the risks of losing the case, and the attorney settles for $10 each, they have lost a full third of the value of their case due to the attorney’s distinct attitude toward risk. If class members shared the attorney’s approach to risk, i.e., if their interests do not diverge, then this nominal “loss” is worth it to them—they would also be willing to accept less money in exchange for reducing risk and increasing certainty. But class members are not subject to the specific incentives described above.

Note that the differences between class members and counsel persist even if the class members have more individually at stake. Suppose that rather than $30 at stake class members have $300,000.[56] Since the class action still requires numerous class members, the attorney’s stake is still far more than theirs, even if the class is of a much more modest size than the one million class members I used in the previous example. Special circumstances may exist where the class members would be similarly risk averse;[57] they might be in dire straits and have special need of some recovery, making risky strategies less attractive. That would be a very particular set of circumstances, though (and, incidentally, one that law does not do a good job taking into account).[58] As a general matter, and in the vast majority of cases, the attorney is going to view risk differently, leading her interests to diverge from the class members’.

B. Controlling Class Counsel

Acknowledging the principal-agent problem in class actions is only part of the democratic critique since such problems are ubiquitous. What distinguishes class actions and makes them a particular cause for concern is that the principal lacks the means to really address this problem. Even if class members formally have some tools at their disposal—such as opting out of the litigation—the argument is that something prevents them from using them.[59] According to the democratic critique, the means to control class action attorneys either do not exist or are rendered moribund.

At the outset, I should note that simply legislating that class counsel should work in the principal’s interests is unlikely to resolve the problem. Lawyers are, of course, subject to an array of ethical duties. They must provide competent representation[60] and avoid conflicts of interest.[61] The prevailing view is that these duties are not up to the task of resolving principal-agent problems in class actions.[62] As Elliott Weiss and John Beckerman bluntly put it: “The rules governing attorneys’ professional conduct, which bar an attorney from allowing her own interests, financial or otherwise, to influence how she serves her client’s interests, do not effectively constrain plaintiffs’ attorneys in class actions.”[63] A basic problem is enforcement costs, including monitoring the lawyer’s behavior in order to detect wrongdoing. These costs would be substantial, possibly amounting to duplicating the litigation costs themselves.[64] Someone with a tiny stake in the litigation, like the archetypal class member, will simply not be willing to invest that much in holding the attorney to account—something that will become a recurrent theme.

I believe there is a more basic point, though: ethical duties do not genuinely address the endemic principal-agent problem in class actions. A comparison to corporate law usefully illustrates this point, especially since principal-agent problems have been the explicit focus of that field for half a century.[65] Such comparisons are also common in the class action literature.[66] Like attorneys, corporate directors, the relevant agents in that context, have extensive duties imposed on them.[67] The most demanding of these is the duty of loyalty,[68] which guards against self-dealing, i.e., conflicts of interest.[69] What light does corporate law shed on class actions? It shows that fiduciary duties, even the robust duty of loyalty, do not genuinely address the endemic principal-agent problem in class actions. That problem is a product of class counsel’s and class members’ different levels of investment in the litigation: the attorney’s greater investment led her to be risk averse when class members were not. Imposing something like the corporate duty of loyalty on class counsel would make little difference—that kind of behavior does not violate that duty. If it did then corporate executives, who almost always have a far greater stake in the firms they manage than run of the mill shareholders do, would be in constant danger of violating this relatively strict fiduciary duty.[70] Yet, there is no serious contention that directors are always violating their duty of loyalty. Indeed, that would be perverse. The entire point of things like stock options and other tools that increase the executives’ stake in the firm is to help ensure their loyalty; if following those incentives subjected directors to legal challenges under the demanding duty of loyalty standard, the entire regime of modern corporate governance would be self-defeating.[71] I consider other comparisons between class actions and corporate law below in Part III.

The primary focus of the democratic critique is the absence or ineffectiveness of means for class members to hold their attorneys accountable. The clearest example is that there is nothing like formal democracy in class actions. Shareholders and citizens both vote on their respective agents, a long-established method for holding agents accountable. If the agent displeases them, voters can “throw the bum out of office.”[72] This threat disciplines the agent: to the extent she wants to keep her job, she does what the voters want. The vote therefore functions like a collective performance review, with the voters expressing approbation or disapproval for the agent’s actions. As an accountability mechanism, voting has undeniable appeal. It embodies democratic ideals of collective decision-making and is designed to represent and respect the interests and preferences of the voters. Voting in this way, referred to as retrospective voting, is an iconic feature of many institutions.

At present, there are few opportunities for democratic participation by class members. Class counsel do not stand for election; they do not face the threat of recall if class members are displeased; class members do not hold a plebiscite to determine their theory of the case, and so on. Modern class action law provides for two avenues of participation: objecting and opting-out.

Class members can object to and possibly appeal a settlement in their case, an option that could potentially be expanded to involve other litigation decisions. There are drawbacks to relying on objectors to address the principal-agent problem in class actions stemming from collective action problems within the class[73] as well as the mixed incentives of the objectors themselves.[74] More fundamentally, though, objection does not give class members per se the ability to control the attorney because the ultimate decisionmaker is the court, not the class members themselves. Objecting to a settlement triggers searching review of it, but the court can still approve of the settlement despite substantial objections. Objection is therefore not a veto of a settlement, far from it, and courts have approved of settlements to which a clear majority of class representatives have objected.[75] The power to object or appeal, even if greatly expanded from its current form to apply to litigation decisions beyond settlement, is not a straightforward substitute for democratic participation.

The closest analogue to voting in class actions is the opportunity to opt-out, or “exit” one, available in most claims for damages.[76] These exit rights have been seen as essential to class action litigation[77] and figure prominently in the democratic critique of class actions.[78] In the limit, exit and a democratic procedure like electing class counsel have similar effects: a mass exodus from a class action shrinks its potential recovery, something the attorney would want to avoid and would alter her behavior in anticipation of. En masse, then, exit and voting converge. Therefore, much of what can be said about exit applies to the possibility of simply adding some explicit democratic procedure to class actions.

The worry is that the small claims of individual class members limit the efficacy of exit. For class members with relatively small claims, exit is something of an empty threat. Opting-out of the class action leaves them the alternative of pursuing their claim individually, but they have little incentive to take that route as the litigation costs would dwarf any possible recovery. Furthermore, just as small stakes make class members unwilling to take on the costs involved in enforcing ethical rules, it is frequently not worth it for class members to evaluate the litigation. If the claim is small enough, even filling out the paperwork to opt out might not be worth a class member’s while.

Class members can be rationally ignorant and rationally apathetic where “rationally” here indicates that the expected benefits from gathering information or bothering to opt out do not exceed the costs.[79] These costs can also lead to collective action problems. If monitoring the class counsel benefits all class members (making it a public good within the class), then class members will be tempted to free-ride on the work of others, enjoying the benefit without enduring the cost. Without some way to coordinate their efforts and distribute the costs, the monitoring needed to hold the agent accountable is unlikely to occur as everyone hopes that some other member of the class will do it.[80]

Class members rarely exercise their exit rights, leading to questions about whether they can actually play much of a role.[81] In and of itself, the low rate of opt-outs does not prove that it has little effect: the fact that an accountability mechanism is not utilized is equally consistent with it being ineffective or too burdensome to use and with the agent being effectively controlled by it. Suppose that the agent desperately wants to avoid being sanctioned so she always acts according to the principal’s wishes. Then, the accountability mechanism would be working quite well, even if it was never used. Indeed, the fact that it was never used ends up being evidence of its force. That being said, class members with small claims likely will not put much effort into monitoring the litigation, and there is empirical evidence associating larger claims with greater opt-out rates.[82] Modest stakes in the litigation limit the impact of exit.

The foregoing would apply to a hypothetical class action voting system, as well. Given a small claim it would not be worth it to bother figuring out how to vote (rational ignorance) or to simply bother going through the effort of voting (rational apathy). In addition, Samuel Issacharoff argues that voting is an awkward fit for class actions because, unlike other institutions that rely on democratic procedures, they are ephemeral. While this may seem odd to say given that some class actions drag on for years on end, once the case is resolved the class, as a legally meaningful entity, disappears. Other venues where we see voting do not typically have a clearly defined end.[83] Indeed, the rule of law is somewhat predicated on that fact.[84] And perpetual life is one of the defining characteristics of corporations. The ephemeral nature of class actions undermines a direct application of retrospective voting as an accountability mechanism; by the time the class members have the chance to “throw the bum out of office,” the office itself has disappeared.[85] Issacharoff also argues that the potential for exchange in office—that today’s losers can become tomorrow’s winners—is essential for a stable democratic system, something that the nature of class actions does not readily allow.[86]

I have reservations about these arguments. Exit, for instance, has potential as an accountability mechanism, especially as it can be more nuanced than a typical winner-takes-all election. In such an election, winning a narrow majority, even when 49% of the voters utterly despise the result, means capturing the whole prize. A class action attorney does not want class members to opt-out because that reduces the overall recovery and, usually, their fee along with it. So, the attorney has incentive to win the support of as many of them as she can, rather than a slim majority, likely inducing behavior that benefits the entire class over and above what would be needed to win a majority.

The lynchpin of the foregoing critiques of class actions Is rational ignorance or rational apathy. But this is an argument that proves too much because it implies that democratic participation is by and large useless. One of the most famous results in modern political science is that it is irrational to vote in a population of any significant size because the probability of that vote making a tangible difference is so small.[87] While voting in a state or national election affects serious matters—the economy, justice, foreign relations—the probability of being pivotal in any given election is so infinitesimal that, in order for voting to be rational (in the sense used here), the benefits the voter would have to receive from her preferred candidate winning would “stagger the imagination.”[88] Further, the rational apathy and ignorance observations are framed as the class members having too small a stake to make voting worth the effort, yet this actually makes little difference. In the case of national elections, voters have a lot at stake, but voting is still “irrational.” If we take the rational apathy and ignorance points seriously, then they should, for all practical purposes, bite equally when class members have $10 at stake and $10,000.[89] (Incidentally, this exposes one of the differences between opting out and class action democracy—with a sufficiently large claim, exit is perfectly rational if the attorney is not acting as one wishes.) Furthermore, these arguments would indicate that class actions were not especially worse in these regards than any other collective institution. Class actions might conspicuously lack a voting procedure, but rational ignorance and rational apathy suggest that such procedures are of dubious utility anyway. If this is a major problem with class actions, it is one that they have in common with many institutions.

Intermediaries, party identification, and media coverage all lower the costs of becoming informed in the political arena, although not enough to render voting rational. Some proposals for class action reform seek to create analogous features.[90] What ought to be avoided, though, is holding class actions to a wildly unrealistic standard, one far higher than other institutions that facilitate collective action and encounter their own principal-agent problems. That being said, the democratic critique has persuasive force and has been a longstanding feature of the class action debate for a reason. It captures something significant, namely the anxiety that the attorney is free to take advantage of the class. Moreover, it sets up a ready comparison—proponents of the critique assert that class actions perform substantially worse than similar institutions with respect to accountability. Part of the critique is also that democratic procedures cannot be simply imported into class actions in order to resolve the principal-agent problem at the heart of class actions. This forms the second part of the longstanding criticism of class actions: there exists the endemic principal-agent problems and the class members are distinctly unable to counteract it.

II. The Democratic Critique in Context

A. The Law of Class Interest

The democratic critique of class actions is ultimately based on accountability—class members cannot hold their attorneys accountable, i.e., they cannot subject their agent to (effective) control.[91] Accordingly, the critique explains, class actions are suspect and should either be cabined, such as being limited to cases where no other form of litigation would be possible, or reformed so as to grant class members greater control. These ideas are not confined to theoretical or policy debates—they reach outside law reviews and find their way into judicial decisions.[92] Since there is no explicit rule mandating voting or some other procedure in class actions, the closest things being opt-outs and objections, the democratic critique calls out for some grounding in law. A variety of grounds have been cited. These include constitutional due process[93] as well as various parts of Rule 23’s multipart, though often vague, requirements[94] such as adequacy of representation, that common questions of law and fact “predominate,”[95] and the court’s authority to approve class action settlements.[96] Courts are actually quite cavalier about what part of Rule 23 is relevant,[97] which can lead to confusion as not all of Rule 23’s requirements apply to all class actions.[98] Arguably, some class actions have been inappropriately stymied by misreading these complex requirements.[99]

The Supreme Court’s Ortiz v. Fibreboard Corporation[100] decision is especially explicit on these points. The Court invoked the endemic principal-agent problem in class actions, distinguishing a class action settlement from one in ordinary litigation where it seems fair to assume the settlement represents the fair value of the litigation because the parties vigorously bargained over it.

But no such assumption may be indulged in this case, or probably in any class action settlement with the potential for gigantic fees. . . .

In a strictly rational world, plaintiffs’ counsel would always press for the limit of what the defense would pay. But with an already enormous fee within counsel’s grasp, zeal for the client may relax sooner than it would in a case brought on behalf of one claimant.[101]

Further, the lawyers for the class had also arranged for a side settlement of 45,000 separate pending claims contingent on the resolution of Ortiz.[102] The lawyers thus had an endemic conflict of interest and a far more overt one, and the Court did not allow the class action and settlement to go forward. Similarly, the settlement in Amchem Products, Inc. v. Windsor[103] was rejected because there was “no structural assurance of fair and adequate representation for the diverse groups and individuals affected.”[104]

In Ortiz and Amchem, the Supreme Court drew attention to the issue underlying the democratic critique. The Court halted these class actions because it was not convinced that the basic principal-agent problems were sufficiently resolved so that litigation decisions (like a settlement agreement) represent the interests of the class and not just the attorney’s. Had the class as a whole ratified the decision, or even done so implicitly by not exercising a robust exit option, that would provide the missing structural assurance. As Issacharoff sums up: “The key issue is the guarantee that the agent be the faithful guardian of the interests of the class.”[105]

The exact language varies, but the normative framing of the democratic critique is consistently oriented around the interest of the class. Coffee, for instance, couches his version of the critique in terms of client autonomy, but he defines it as “the idea that the attorney should abide by the client’s preferences and litigation goals.”[106] This amounts to the same thing as class interest. It may seem odd to see the argument for democracy amount to the interests of the class, but democratic procedures are widely-regarded as an instrumental good—a means to an end. Instrumental justifications for democracy have a venerable pedigree, dating back to at least Aristotle and are a key part of modern debates in political and legal theory.[107]

This attention to class interest has been connected to a broader due process tradition.[108] I want to suggest that the democratic critique can also be tied to a wide-ranging trend of deferring to democratic institutions, which sometimes sounds in due process but can also be connected to other areas of law. The well-known Carolene Products footnote four calls for more exacting due process review when the democratic process is blocked or in some way breaks down—then (and only then) should courts intervene in ordinary legislation.[109] In administrative law, the Chevron doctrine instructs courts to defer to agency interpretations of their empowering statutes.[110] This doctrine rests, in part, on a democratic argument: administrative agencies are subject to control by the executive, which is more democratically-accountable than the federal courts, so courts should defer to the agencies by virtue of their superior democratic pedigree.[111] In other work, I have argued for an alternative, though not mutually-exclusive, democratic argument for Chevron deference based on Congress rather than the President.[112] Recent administrative law decisions by the Supreme Court emphasize democratic procedures,[113] although the Court does seem to neglect the role of Congress. This pattern also extends into private law. Courts, especially the Supreme Court, have put overwhelming faith in corporate democracy. The landmark decisions of Citizens United[114] and Hobby Lobby[115] both depend on corporate principals (shareholders) being able to control their agents.[116] These are complicated cases, but they share a common thread in that the Supreme Court relied on accountability to justify deferring to corporate leadership, even when their choices implicate fundamental constitutional rights.[117]

The handful of cases briefly canvased above mark out a subtle but consistent trend in jurisprudence the way a few points describe a line—all else being equal, if an institution exhibits more democratic procedures, if the Court does not identify a glaring principal-agent problem, it will be treated as more legitimate and be subject to fewer legal restrictions.[118] Class actions, with their prominent lack of democratic procedures and principal-agent problems, are instead causes for concern and courts keep a tight rein on them; the democratic critique provides a pervasive general rationale for this treatment.

B. Normative Frameworks: Deterrence & Class Interest

The normative framework the democratic critique adopts is the interests or well-being of the members of the class.[119] For the purposes of this Article, I do the same for reasons that are, for lack of a better term, rhetorical. It is the position used by the democratic critique and I seek to answer that critique on its own terms. Focusing on class interest is unremarkable in and of itself. We would hope that procedural devices serve the interests of the plaintiffs, and it is a common approach throughout the literature.[120] But, it is certainly not the only normative framework available. At least as important to class well-being is deterring wrongdoing.[121] Deterrence takes on special importance in class actions because the awards each plaintiff individually receives are frequently small. They often make little practical difference to the members of the class, though it bears reiterating that not all class actions are for small claims.[122]

When considering the core principal-agent problem in class actions and the democratic critique based on it, deterrence and class member well-being actually converge.[123] Assuming the underlying substantive law is reasonably well-calibrated to achieve deterrence, it would make no material difference if my arguments were framed in terms of optimal deterrence instead of class members’ interest.[124] This is not the case with every analysis of class actions. Brian Fitzpatrick, for example, argues that compensation can impede deterrence and recommends that, in some cases, where the class members have very small stakes, lawyers should receive the entire judgment, pulling apart the goals of class member well-being (however trivial the increase to their well-being their share of the judgment would be) and deterrence.[125]

At its simplest, deterrence means forcing the defendant to bear the full social costs of their wrongdoing; they must internalize those costs. This implies, for instance, that “[o]ptimal tort deterrence threatens firms with liability for the total costs of their tortious conduct.”[126] Overdeterrence can stifle innovation and lead to wasteful precautions.[127] So, ideally, damages would be set more or less equal to the harm incurred. There are numerous complications that can be added to this very simple account, such as the probability of enforcement, but this is the basic idea. To illustrate how class well-being and deterrence converge, suppose that the defendant has committed fraud (an activity we wish to deter) that netted it $5 million in profits. Deterrence instructs that the damages should be at least $5 million, otherwise the defendant will not be adequately deterred. Now suppose that the case is a class action and, owing to the endemic principal-agent problem the attorney wishes to take the safer route and settle for only $4 million. This takes away from the class’s well-being (as they have lost 20% of their recovery) and, simultaneously, from the deterrent effect of the lawsuit. Faced with this kind of litigation, the defendant would not be deterred, as they have a net profit of $1 million and, all else being equal, would continue with their fraud.[128] Again, all this presumes that damages reasonably estimate the social harm of the defendant’s conduct. But if they do not, that is not an issue with class actions but rather with the underlying substantive law.

Put succinctly, the endemic principal-agent problem in class actions leads to underdeterrence in just the same way it threatens to undermine the class’s interest. Left to their own devices, the risk-averse lawyer will take safer options that will reduce defendant’s expected costs and, therefore, fail to deter wrongdoing. From the perspective of class well-being, this is also a problem simply because it reduces the amount each class member receives—both normative frameworks point in the same direction.[129] Someone concerned solely with deterrence thus ought to be just as worried about the democratic critique as someone concerned solely with class well-being.

Class actions do complicate deterrence because they can be especially expensive to litigate.[130] So, even if the damages are well-calibrated for deterrence, the addition of significant litigation costs could lead to overdeterrence. Continuing the example above, if the defendant will have to pay $5 million in damages and another $1 million in litigation costs, then the defendant’s activity will be over-deterred, with $6 million in total cost against $5 million in harm. That might not matter much in an example of willful fraud, but we could imagine more borderline cases where the defendant ends up taking excessive cumbersome precautions (passing those costs and inconveniences onto consumers or the public at large) to guard against potential legal repercussions.[131] This kind of overdeterrence does not seem to be much of a genuine problem, though. Despite the high litigation costs, a class action represents an enormous economy of scale for the defendant. A class action is assuredly far more efficient than dozens of separate trials, not to mention hundreds or thousands of them. Furthermore, a class action offers the potential of “global peace,” resolving all of the claims against it once and for all. Realistically, particularly when we consider the archetypal small claims class action, the alternatives are not a class action and myriad individual cases. Rather it is between the class action and no litigation because bringing individual cases is not economical for the plaintiff—any recovery would be dwarfed by their litigation costs. Now, compared to no litigation at all, the class action is obviously far more expensive for the defendant. But this argument is singularly unpersuasive from the perspective of deterrence since it means that the defendant’s wrongful conduct would go entirely undeterred—a major failing in the law that class actions are designed to correct.[132]

III. The Political Economy of Class Actions

To restate the democratic critique, it asserts that current class action law does not provide many opportunities for class members to hold their attorneys accountable. They lack some mechanism like voting that would help them control their agent. At best, they have attenuated options, like objecting to a settlement or exiting the litigation, and even those, the argument goes, do not work well. Consequently, class action attorneys are free to pursue their own interests rather than those of the members of the class. And because the attorney has so much more at stake in the litigation than any of the class members, their interests are bound to diverge. In short, the attorney has the means and the motivation to pursue their own interests, even at the expense of the interests of the members of the class itself. An important aspect of this critique is that it makes no assumptions about the attorney’s motives. The problem is not that plaintiffs’ attorneys are, as a group, somehow greedy or venal or irresponsible. These potential issues are inherent in the structure of class actions; they are “baked in.” This explains why the democratic critique is pervasive: the very nature of the class action gives rise to the endemic principal-agent problem, and the critique holds that the class action lacks sufficient means to counteract that problem.

A. The Fundamental Trade-Off: Expertise vs. Control

This democratic deficit creates the problem that courts, theorists, and lawmakers have all tried to solve in various ways. It motivates class action legislation,[133] proposals for procedural reform,[134] and widespread skepticism or hostility towards the class action device, especially in courts.[135] Yet here I want to question the critique itself. Rather than propose another solution to the problem the democratic critique poses, I argue that the problem is overstated. In this Part, I will explain that this democratic deficit is not as bad for class members as it appears. In fact, if class members had access to robust accountability procedures, if, for example, they could vote on litigation decisions or to retain their attorneys, that option would generally not make them better off. Since the normative framework adopted by the democratic critique is the well-being of the class, a position that the courts have also adopted, we should rethink the democratic critique and its implications.[136] Moreover, my conclusions go a long way towards explaining the mixed results of many attempts at class action reform—they are solving the wrong problem.

This counterintuitive result about the consequences of voting is based on a branch of the principal-agent literature addressing the political economy of communication and eliciting the information an agent possesses. The canonical study in this area is by Vincent Crawford and Joel Sobel, which has been subsequently tested and expanded.[137] A crucial feature of the litigation environment is that the attorney (the agent) has valuable expertise. They have access to private information, about both the law in general and the specific case, that the class members (the principal) want to take advantage of. The principal might not know the best theory of the case, available defenses, and other key elements of the litigation without having legal training or conducting their own costly investigation. Furthermore, the agent’s information cannot be easily, conveniently, and verifiably transferred to the principal. Thus, there is no simple way for the principal to “check the agent’s work.” To take a non-legal example, scientists may conclude that there is a lake on Mars, but few people will be able to determine whether their research was reliable. An attorney’s private information is the same. While the client may be well-informed about the facts—as they experienced them firsthand—how those facts affect the litigation will be largely mysterious to them. Making that assessment involves, inter alia, looking to precedent, judging some facts relevant and others irrelevant, and identifying areas where the law might reasonably be modified, all things that lack an objective, easily accessed rubric.

A consistent finding in the principal-agent literature is that there is a trade-off between control over the agent and access to its expertise.[138] The principal might make the decisions, but it also might not know what the best course of action for it is. Sometimes, then, surrendering control over the agent can be good for the principal. Control in and of itself is not always in the principal’s best interests. It is a means to an end: sometimes the principal “sacrifices control because doing so gives the agent greater incentive to use its information” which ultimately makes the principal better off.[139]

The qualifier above is important, though. The principal-agent framework is extremely flexible, and such relationships are very common, but the specific characteristics of each relationship can lead to different results. Indeed, one of the strengths of the political economy methodology is identifying which characteristics matter and why. Which is to say that not everything stated here applies to all principal-agent relationships. Or, put another way, in some circumstances, the principal-agent problem has been mitigated or resolved to some degree. A doctor, for example, is different from class counsel.[140] Both have relevant expertise, but the incentives differ. The doctor’s incentives are not systematically misaligned with the patients’, there does not seem to be a clear analogue to the endemic principal-agent problem in class actions. When there are clear conflicts of interests—which would be more acute and more blatant than different tastes in risk—that amounts to a scandal and is subject to enforcement mechanisms generally deemed effective.[141] The information available differs as well. While both law and medicine are technical fields, patients do know their personal medical history and have a basic sense of how they feel—information relevant to assessing a doctor’s work. Class actions can turn on things as arcane as specific interpretations of defamation law, on which even the Supreme Court cannot agree.[142] A second opinion is also a time-honored tradition in medicine, and can be acquired at relatively low cost, which makes the agent’s information verifiable in a way that it is typically not in litigation. There is far more reasonable disagreement about the law, and the costs of running multiple trials is far greater than a second medical exam or similar investigation.[143] Perhaps most importantly, the principal’s stakes are really different—even costly investigations, monitoring, and enforcement are worthwhile (and thus credible threats) in the medical context. The phenomena I describe are tied to features of the specific principal-agent relationship—albeit ones that are actually quite common.[144] All else being equal, though, there is a persistent trade-off between control and expertise that must be addressed. I explain the intuitions behind this finding in the next Section.

B. Delegation vs. Democracy in Litigation

In this section, I describe the intuitions behind the principal-agent literature. To illustrate the effects of democratic accountability on class actions and the trade-off between control and expertise, let us assume that we could institute some sort of idealized procedure whereby class members vote on litigation decisions. There exist an array of practical reservations with this proposal, such as the timing of these votes or how we would go about conducting them efficiently, not to mention the challenges of instituting democratic procedures in class actions outlined previously. But for the purposes of this discussion, we will bracket all of that. The voting procedure under consideration is thus an unrealistic best-case scenario for class action democracy. This stacks the deck against my conclusion—if even that sort of idealized arrangement does little to improve the lot of class members, then instituting real-world versions of it cannot do better. Also, I choose this hypothetical where voters directly vote on litigation decisions for ease of exposition. We could alternatively suppose that class members decide whether to retain the attorney or pick attorneys amongst a slate of candidates; it does not materially affect the analysis. With this idealized voting in place, class counsel presents her arguments in favor of her proposed strategy, acting essentially as an advisor to the class who then makes the ultimate decision. Equivalently, we can think of the attorney’s choices of litigation strategy as something like a political platform—it is what she would do if elected. We can then compare this to circumstances without class action democracy, so the litigation decisions are delegated to the attorney. Delegation captures, in a simplified and stark form, the democratic critique’s account of the current state of class action law as there is no democratic check on counsel. For instance, the power of class members to opt-out and thereby exit the litigation is not included, taking on board the arguments about the weakness of this as a means to hold class counsel accountable.[145]

So, we have two scenarios: one where litigation decisions are delegated to the attorney and one where class members get to vote on them after receiving some advice from the attorney. Suppose that there is a class action and that the litigation decision at hand is setting a settlement threshold—any settlement offer from the defendant greater than that threshold will be accepted. Let us consider what might happen using some very simple abstract models.[146] The class action attorney’s ideal settlement threshold is depicted in Figure 1 as A. She arrives at that value by making full use of the information available to her, her analysis of the case, likely defenses and procedural hurdles, and so forth. In short, she uses her expertise. Point C on the figure depicts the class members’ ideal settlement threshold if they also had access to the attorneys’ expertise, i.e., if they knew everything that she did. These values are not identical because of the different stakes that the attorney and the class members have in the litigation. There will be some settlements that the attorney would accept that the class members, if they were fully informed, would reject, preferring the riskier option of a trial to the proffered settlement. Hence, in this example, there exists a genuine principal-agent problem between the class and their attorney; the endemic principal-agent problem of class actions is in full effect. It produces the space between A and C. What happens in the delegation scenario is straightforward. The attorney makes the litigation decisions, so she adopts the settlement threshold A.

Figure 1: Delegation




The voting scenario is more complicated. The attorney knows that the class members are a bit more aggressive, or comfortable with risk, than she is. There are settlement offers she would happily accept that they would reject. In this scenario, class members are the ones ultimately making the decision. All the attorney can do is make suggestions. Consistent with her role as class counsel, she can advise the class members about the strength of their case and explain what settlement threshold they should choose to adopt. Because of the endemic principal-agent problem in class actions, though, she has incentive to be strategic in her advice. She wants the class members to adopt settlement threshold A, (or as close to it as she can manage) so in advising the class, she could downplay the strength of the case, emphasizing its possible flaws in an effort to encourage class members to adopt a threshold more like the one she prefers. She has the ability to do this because, as noted above, class members cannot easily verify her advice. If the attorney tells the class that they are unlikely to prevail because of a contrary Supreme Court precedent or because the judge is especially hostile to their interest, they cannot readily determine if she is telling them the truth or lying.

As depicted in Figure 2, the attorney might suggest that class members choose settlement threshold A’, less than even the attorney’s ideal settlement threshold of A, in order to induce class members to vote the way she wants them to. She can craft her advice in order to account for the class members’ greater tolerance for risk. For their part, class members can anticipate this; they are aware that the attorney’s interests are not identical to theirs, so they take the attorney’s advice with a grain of salt. In the language of economics, the attorney’s advice becomes a “noisy signal” that does not convey as much information. So, when the attorney gives them advice to the effect of A’, all the class members can really infer from it is something in the “grain of salt” interval. The attorney’s advice communicates some information, but it ends up being imprecise. The attorney’s incentive to downplay the strength of the case gets in the way of effective communication—the nature of the interaction prevents her from sharing all the information at her disposal. On the basis of the attorney’s advice that they should vote for settlement threshold A’, the class members might end up picking settlement threshold C’, which is greater than C (their ideal settlement threshold if they were fully informed). They end up making a mistake, picking a settlement threshold that is too high, rejecting some settlement offers they would have wanted to accept.

Figure 2: Voting & Advice

“grain of salt” interval






The problem of strategic communication is fundamental and, to some degree, unavoidable. The scenario described in the previous paragraph and Figure 2 involved the attorney trying to manipulate the class. Rather than tell them exactly what she would do (A), she tried to use her advice strategically by suggesting a lower settlement threshold (A’). Consider a variation of this scenario where the attorney was scrupulously honest in her advice, telling the class members exactly what she would do if she were the one setting the settlement threshold.[147] In this “honest advice” scenario, her advice is naturally A, her ideal settlement threshold. Yet there is still no way for the attorney to prove her honesty to the class members.[148] The fact that the attorney might use her advice strategically means that the class members have to treat it with the same grain of salt, leading them to again adopt a settlement threshold that is too high. As shown in Figure 3, this honest advice leads class members to adopt C*, which is even greater than C’ (because A is greater than A’). So honest advice actually makes class members worse off than the strategic advice did—they end up selecting a far too stringent settlement threshold, rejecting perfectly reasonable settlement offers and opting for a risky and costly trial that they would prefer to avoid.[149] The honest advice variant shows that even well-meaning attorneys (and agents in general, for that matter), have to confront the difficult realities of strategic communication. We cannot rely on them simply being honest without some sort of (effective) enforcement mechanism behind it or some way to credibly convey that honesty to their principal.[150]

Figure 3: Honest Advice

“grain of salt” interval







These are fairly simple scenarios that involve only a single litigation decision, and I picked settlement thresholds for ease of exposition. Unlike the theory of the case, the extent of discovery, and other choices that have to be made during the lifecycle of a court case,[151] it is easy to see how preferences over risk influence settlement thresholds; there is a natural ordering of them. This is not to say that, if we were to somehow implement this sort of idealized voting procedure in class actions, these exact scenarios would play out. They are designed to explain the effects of different procedures or other institutional arrangements and are thus highly simplified and abstract.[152] My aim is to illustrate the well-established conclusion that there is a trade-off between expertise and control and the causes of that trade-off.[153]

Democratic procedures thus come at a cost. Even idealized democratic procedures—ones that sidestep rational apathy or practical implementation inconveniences—would entail a loss in terms of the expert information that the attorney possesses and would make full use of if she were making the decision on her own. Moreover, it bears keeping in mind that this is embedded in the relationship between the class and their attorney—the same divergent interests that formed the basis of the endemic principal-agent problem in class actions complicate the communication between class members and their counsel.

C. Reevaluating Accountability in Class Actions

This close examination of the interaction between members of a class and their attorney raises the natural question: which is better, delegation or voting? If presented with the choice between delegation or voting, which one would the class members themselves choose and which one would make them better off? Put another way, once we acknowledge the drawbacks associated with democratic control, the products of strategic interaction between the principal and the agent, the question then becomes whether the information lost outpaces the loss of control entailed in delegation. The complaint at the heart of the democratic critique is that class action attorneys are not serving the interests of the class, pointing to the need for democratic procedures or some adequate substitute for them. We are now in a better position to evaluate that claim.

As is often the case, it depends. Specifically, delegation is better for class members when:[154] (1) the agent’s specialized information is sufficiently valuable to the principal and (2) the divergence interests between the principal and the agent—or the agent’s bias—is not too great.[155] These conditions are related; they can substitute for each other, so the more one is satisfied the less the other needs to be for delegation to outperform voting. Thus, the more important the agent’s expertise (Condition (1)), the less their interests need to align with those of the principal (Condition (2)), and vice-versa. While the main conclusion that surrendering control can be useful to a principal seems odd at first,[156] there is something intuitive about both conditions. The more valuable the agent’s private information, their expertise, the more loathe the principal will be to throw it away. The “grain of salt” ends up being more costly to them. Likewise, the more the agent and principal share the same interests the less dangerous delegation becomes. Delegating to someone with identical interests is perfectly safe, as they would make the same decision as you would, anyway. In fact, it renders control mechanisms like retrospective voting superfluous. When the agent is relatively unbiased—so that her interests and the principal’s are closely aligned—any advice given will be more informative. But Wouter Dessein finds that, even in those cases, the principal is still better off delegating. If there is any divergence in interests, then there will still be some “noise” in the advice. The principal will still have to take the advice with a grain of salt, just a smaller one, so some information is lost. With delegation, there is still the agent’s bias, but this additional distortion, the noise that is a byproduct of communication, is not present, and the agent is able to use all of the information at her disposal.[157] If the agent’s interests are mostly, but not completely, the same as the principal’s, this loss of valuable information is not worth the minor increase in control.

Our assessment of the need for democratic participation in class actions, and, in turn, the force of the democratic critique, depends on the extent to which these conditions are typically satisfied in that context. Condition (1) is reasonably straightforward: members of the class, especially those with small stakes, will assuredly know less about the case than the attorney. To the extent that rational ignorance or apathy has any impact, they will preclude class members from bothering to become fully informed. Moreover, the attorney has specialized legal training, so they can far better understand the relevant nuances and precedents. This kind of training cannot be easily transferred to class members, arguably, they would have to attend law school themselves, so class members will find it difficult to evaluate the attorney’s assessments. In class actions, (and in litigation, generally) the role of counsel as an expert agent with greater information at her disposal is uncontroversial.[158]

Condition (2) is more complicated. If the attorney and the class had identical interests, this condition would be easily satisfied; but then, of course, we would have no principal-agent problem in the first place. The endemic principal-agent problem in class actions indicates that there is some bias. The question is how much. With one important caveat I will address in Section III.C, Condition (2) is likely to be sufficiently satisfied. That is, when it comes to class actions, the divergence in interests between members of the class and the class counsel should not be too great, especially taking into account how significantly the class action environment satisfies Condition (1). First, and this bears emphasizing, given the overall tenor of the discussion around class actions, both the class and their attorney want to win the case. They share this basic goal, so they are not truly at loggerheads. Furthermore, the customary contingency fee arrangement—which was part of the cause for the endemic principal-agent problem—does do a decent job at aligning the interests of the attorney with those of her client: she does better as the class does better.

It is hard to definitively say that a contingency fee does “enough” to connect the interests of the class to their attorney. But the situation closely resembles the also ubiquitous practice of compensating executives according to the firm’s performance, usually through stock. This is one of the main tools of corporate governance and commonly considered quite effective: “Although one can find outliers, there is no empirical basis for assuming any general divergence between the CEO’s incentives and shareholder value.”[159] This comparison is especially apt because the principal-agent problems involved with corporate executives and class counsel are strikingly similar. The principal-agent problem in class actions is caused by differential stakes in the litigation—the attorney has far more at stake in the litigation than any individual class member, leading them to have substantially different tastes in risk. Likewise, executives have far greater stakes in the firms they manage than the vast majority of shareholders (their principal). The typical shareholder’s investment in any single corporation is quite small; the corporation’s executives are nearly always far more sensitive to changes in the firm’s value or the risk of bankruptcy. In modern corporations, stock options are pervasive and account for most of an executive’s compensation. According to one study, the median chief executive officer’s stock options compensation was six times their salary.[160] Furthermore, the executive has less tangible considerations that tie her to the firm: if it goes bankrupt, that would represent a major disruption in her life, prompting the search for another job and possibly harming her reputation. This is not to imply that corporate governance is without its problems. There are persistent worries that corporate leadership ends up ignoring important stakeholders other than the shareholders[161] or overemphasizing shareholder value over all else.[162] Besides, even if the compensation structure aligns the agent’s interests with the principal’s, the respective agent may still be overpaid: they may be receiving too big a slice of the shared pie.[163] But, the executive compensation scheme does a good job at reducing the fundamental problem of bias.

Given the prevailing view that executive compensation successfully aligns the interests of the executive with those of the shareholders, it stands to reason that the similar arrangements in class actions do a respectable job of doing the same. The effects should be similar. There are additional structures that help limit the amount of bias of class counsel. Being able to tout successes and developing a reputation for winning large awards is valuable in enticing new clients. Along similar lines, class actions are a specialized practice and they not infrequently involve a small, tightly-knit legal community.[164] Not only does this foster reputational effects, but a repeat player has a separate incentive to forcefully litigate: the value of this particular case can affect future similar cases, so the class action attorney wants to avoid setting a bad precedent.[165] Rules of professional responsibility, possibly buttressed by Rule 23’s specific requirements[166] and due process,[167] while perhaps insufficient to resolve principal-agent problems in and of themselves, guard against egregious instances of attorney shirking or misconduct. All of this together keeps class counsel’s interests reasonably well-aligned with the class as a whole. And, as explained above, their interests do not need to be identical for delegation to be the optimal strategy—just not too divergent considering counsel’s (not insignificant) expertise. Moreover, we should not hold class actions to some overly idealized, impossible to satisfy standard; the principal-agent problems that exist in class actions appear no worse than the ones common in other legally recognized institutions.

At a high enough level of abstraction, all mechanisms for controlling agents seek to align interests with the principal’s. The point behind threatening to “throw the bum out of office” is that they want to hold onto their position and will behave accordingly. We can think about this as the difference between ex ante and ex post controls. Voting is the latter: the principal actively chooses whether to reward or sanction the agent. Aligning incentives, in the way I mean it, is more ex ante and relatively passive—it does not entail active participation by the principal once it is set up. Law generally overemphasizes ex post controls while neglecting ex ante ones,[168] and I think the democratic critique does the same.

Finally, it bears noting that democratic procedures bring problems of their own. Most of these I have set aside in considering only a highly idealized version of class action democracy. One deserves special attention, though, because it is intimately connected to the agent’s expertise. When the agent has expertise, voting creates the possibility of pandering: where the agent disregards her superior information and opts to do what the voters prefer in order to win their approval. So, the agent ends up doing something she knows is wrong in the sense that it is not in the voters’ best interests, perversely in order to win their votes.[169] Class actions are ripe for pandering because a key requirement[170] is that the principal is unlikely to learn whether the choices made were the right ones or not. If the agent expects the principal to learn the results of her action before the vote is held, then pandering becomes pointless: in that case, the agent should just choose the right action, confident that truth will out and the voters will reward her accordingly. Pandering might occur, for example, when the effects of a tax plan will not be realized before the next election. In litigation, it will be difficult for any client—class or otherwise—to determine whether the strategies used were the best ones available, especially before the litigation itself has ended. If we doubt that the presiding judge can accurately assess the merits of a case, we should be all the more skeptical about the client’s ability to do so. The possibility of pandering indicates that if elections were instituted in class actions, not only would they likely devolve into rivals presenting unrealistically optimistic plans, but the very presence of the election creates an obstacle for the class as the attorneys pander in lieu of using their expertise.

Where does all this leave class actions? Worries that members of a class cannot control their attorneys have dominated the conversation and profoundly shaped the law. I styled this the democratic critique as the main thrust of it is that class members conspicuously lack ways to meaningfully participate in and steer the litigation that proceeds in their name. They lack a real voice in the litigation; they cannot discipline their attorney the way voters can turn a politician out on their ear, and, as that comparison implies, this democratic deficit sets them apart from other institutions and prompts serious reservations about them. Democracy furnishes legitimacy and, all else being equal, is a good way to make decisions, especially when a large group of people is involved. But, surprisingly, in the context of class actions, democracy does not benefit the class. If given the opportunity to democratically participate in the litigation—if we set aside any practical problems and consider an idealized version of class action democracy—then the members of the class would by and large prefer not to use it. They would rather delegate the decisions to their attorney. This says less about voting and democracy overall than about the distinctive environment of litigation. In this context, voting inevitably results in some loss of information—even if class members can consult with an attorney, they have to take any advice with a grain of salt (how much will vary based on the attorney’s bias). There is a trade-off between expertise and control, and in the case of class actions, this trade-off ultimately favors not taking active control over the litigation. The expertise is too valuable to sacrifice and the attorney’s interests, while not identical to those of her clients, are close enough that this delegation is, in expectation, a good thing for the class.

The foregoing does not imply that class counsel never abuse their position. But the democratic critique operates at a high level of generality and analyzes the institution as a whole. And such abuse is not the product of the incentives and circumstances inherent in class actions.[171] Furthermore, the honest advice scenario reveals that attorney wrongdoing is beside the point; it is not to blame for any of this. Therefore, despite its intuitive force and the impact it has had, the democratic critique misses the mark. The lack of democratic procedures in class actions is not in and of itself a fatal flaw.[172] Nothing said here implies that current class action practice is the best of all possible worlds, though. We should still look for ways to improve and refine class action practice. I propose one in the next Section. My main conclusion, though, is that we should not start from the position that the lack of democracy in class actions renders them inherently illegitimate.

D. The Challenge of Settlement

A key criterion for assessing the trade-off between expertise and control in class actions was Condition (2), that the principal and the agent had interests that were not too different. There is some confidence that this condition is satisfied because both the attorney and the class members want to prevail in the litigation and arrangements like the contingency fee keep their interests aligned—the attorney does better the better the class does. Settlement, which is common in all litigation and especially class actions,[173] can undermine this connection.[174] To borrow an example from Charles Silver and Lynn Baker, suppose that, early in the case, before the plaintiff’s attorney has invested heavily in it, the defendant makes a settlement offer that rewards the attorney, paying her far in excess of her usual hourly rate.[175] It might make economic sense to the attorney to accept the offer, especially since she would be free to take on other work (although the reputational and repeat player effects described previously would counteract this temptation). Along similar lines, there are the infamous coupon settlements, where the defendant offers coupons for purchases of their goods in lieu of money damages while the attorney receives cash.[176] Settlements like this can pull apart the interests of the class and the attorney and threaten Condition (2), breathing new life into the democratic critique.

Courts are very involved in class action settlement. Rule 23 requires judicial approval of a class action settlement, through a “fairness hearing,” designed to address principal-agent problems among other things.[177] However, there is justifiable skepticism about the utility of this process.[178] Judges are “poorly placed to figure out whether proposed deals are good or bad.”[179] For one thing, the legal standard is quite vague, with the Federal Rules only stating that a hearing must be held and that the court must conclude that the settlement is “fair, reasonable, and adequate.”[180] The case law has developed a complicated open-ended rubric, requiring, among other things, an assessment of the merits of the case.[181] While it makes perfect sense that the merits would affect settlement value,[182] settlement removes the adversarial process that courts rely on to make these assessments. When presenting an agreed upon settlement to the court for approval , the plaintiffs and defendants are no longer at odds.[183] In the worst-case scenario, the attorney who has been “bribed” by a preemptive settlement offer has no incentive to argue against it. So, any evaluation of the merits at a fairness hearing, not to mention situating it within the “totality of the circumstances,”[184] is a fraught endeavor.

The analysis called for to ensure that class counsel’s interests do not diverge too much from their clients’ is relatively simple, though. There is no real need for the court to estimate the merits, weigh the costs of continued litigation, and myriad other considerations.[185] The court need only make a much coarser assessment about the proposed settlement—which here ends up being a virtue. Continuing the comparisons with corporate law, settlement raises a real risk of self-dealing, which corporate law addresses through the duty of loyalty. As in that context, a more process-based approach has appeal when assessing class action settlement. A self-dealing transaction by someone with a corporate fiduciary duty is subject to entire fairness review, a thorough investigation of the transaction where the executive bears the burden of establishing that the transaction was fair, i.e., that her conflict of interest did not actually affect it.[186] This broad-ranging inquiry has a lot in common with a fairness hearing, and is equally difficult for a court to perform with any degree of confidence. Alternatively, there is the procedural safe harbor of having the transaction ratified by parties that do not have a conflict of interest.[187] The latter is much easier to assess because all a court needs to do is evaluate the procedures; it does not have to figure out whether this was a good deal or a bad one, merely that the proper procedures were followed.[188]

In similar fashion, courts are more capable of checking to see that a class action fee arrangement or settlement has the right form—namely that it preserves the connection between the attorney’s interests to the class’s—rather than determining that the settlement is the “right” one. That should be the focus and is relatively easy to judge.[189] The interests need not be identical, that would be unrealistic and none of my conclusions depend on the attorney and the class being in perfect harmony. Others have similarly advocated for a “good enough” approach by courts in this area, and the goal should not be an unattainable one of judges ensuring class counsel behaves perfectly.[190]

In weighing the potential for fairness hearings to address the problems created by settlement, a few things are worth keeping in mind. First, we routinely ask courts to do more with less: summary judgment, plausibility pleadings, evaluating regulations drafted by administrative agencies, and preliminary injunctions all require courts to make quite consequential decisions often without the well-developed record present at a fairness hearing. Here, the thoroughness of class certification pays some dividends as the court will have a lot of information—developed when the parties were adverse—available to it.[191] Second, the difference between class members and plaintiffs in more ordinary litigation is really one of degree rather than kind. In the archetypal class action, each member of the class has little at stake and so has correspondingly little interest in carefully reviewing the settlement terms and possibly going to court to object to them. This is simply rational ignorance and apathy emerging again.[192] The same holds true for the named plaintiffs; while they have been willing to play a larger role in the case, there will be serious limits to how extensively they will “double check” their attorney’s work. Yet an individual plaintiff is in a similar, if not the same, boat. Even if they have substantial sums at stake, a client’s capacity, if not their incentive, to monitor their attorney is quite limited. Outside of special circumstances, a client would have to, in effect, solicit a second opinion to evaluate their attorney, and there would have to be a lot at stake to make that worth the expense.[193] Happily, however, in class actions, there is just such a second opinion available—the court. Moreover, the court could order briefing and evidence on the issue as they have for other parts of Rule 23. Overall, though, we should not hold class actions to a far stricter and more unrealistic standard than we apply to other cases.

Finally, it bears underscoring just what is needed for the kind of structural evaluation of class action settlements I propose. To make things a bit more concrete, suppose we have a class action worth $100 million where plaintiffs have a 60% chance of winning, implying an expected value of $60 million. At a fairness hearing, a court would not be able to be nearly this precise, so it could not say whether a $50 million settlement was “fair, reasonable, or adequate.” But even though the settlement is less than the expected value of the case, that does not necessarily mean that the settlement is unfair, unreasonable, or indeed, not in the best interests of the class. That would depend, inter alia, on the relative bargaining power of the plaintiffs and defendants, which determines who can capture the available surplus from trade (i.e., the benefits the parties receive by not having to pay the costs of the full trial), attitudes towards risk, and even factors like the defendant’s solvency.[194] Plaintiffs might accept a $50 million settlement even when they could press for more if a larger figure would drive the defendant into bankruptcy and complicate their recovery. In principle, all of this could be presented to the court with supporting evidence, under seal if need be.[195]

Yet this example with its nuanced analysis of settlement bargaining is not what we really care about. This example is a far cry from Silver and Baker’s collusive settlement or one where class members receive pennies on the dollar for their claims.[196] While there are good reasons to doubt whether a court can effectively judge a $60 million versus a $50 million one, at the scale considered here, those values are fairly close. Plaintiffs would be receiving a significant recovery and there were substantial defenses—a 40% chance that plaintiffs lose is certainly non-negligible. If the settlement in this example were instead for merely $5 million, that seems like the sort of difference that a court could detect. It is this kind of gross disparity that ought to concern us here because that can undermine the connection between the interests of the class and the attorney (Condition (2)).

My proposal echoes Howard Erichson’s “red flags in settlement” that reviewing judges should be especially wary of[197] and for similar reasons: namely the potential principal-agent problems between the class and their attorney.[198] Alongside the much-maligned coupon settlements, Erichson lists elements of settlement that preserve its nominal value—what the lawyer’s take home pay is based on—while reducing the actual cost to the defendant. Cumbersome post-settlement claim procedures are a good example. If the settlement is for $60 million but few class members go through the effort to actually collect, then the true cost to the defendant is far less. Since her payment is based on the nominal amount,[199] the attorney would acquiesce to such procedures in exchange for a higher settlement “value.” This is essentially a subtle collusive settlement. As with coupon settlements, courts should subject the claim procedures to careful scrutiny to ensure they are as convenient as possible.

Crucially, Erichson’s red flags are structural. They do not require judges to evaluate the proper value of the settlement or anything like that. His proposal is somewhat broader than mine, as his red flags encompass things that, inter alia, might induce the judge to accept a settlement like a payment to a cause they favor.[200] I would add that any characteristic of settlement that might undermine the connection between the incentives of the class and those of the attorney should be similarly suspect, including ones that might be idiosyncratic to the settlement, case, or attorneys at hand. Erichson’s red flags are a good rubric to this end, and with something like them firmly in place, settlement would not be such a threat to class action accountability, preserving the well-being of the class by ensuring that Condition (2) is more likely to hold.

Approaching the fairness hearing in this way would be permitted under the current law since the present standard is so open-ended. At most, this represents a small revision of the current practice by refocusing the fairness hearing. Rather than checking to make sure the settlement is the best or the “correct” one, the court just needs to assess its structure, forbidding settlements like the bribery one or its more subtle versions. (Although the front-loaded nature of modern class actions also counteracts a bribery settlement—by the time the class is certified the attorney has usually already put a lot of effort into it.)[201] If their take home pay depends on the class’s award, the attorney’s incentives will take care of the rest. Courts have occasionally taken lawyers’ incentives into account when reviewing settlement.[202] Yet this is not the norm.[203] Moreover, some courts conclude that other considerations trump attorney incentives.[204] These sorts of judicial inquiries should be the rule, not the exception.

Settlement is surely a looming problem for accountability in class actions.[205] At its best, it represents bargaining in the shadow of the law that improves efficiency by dispensing with a costly trial. But it poses significant risks that courts should be vigilant against. There have been a series of proposals to strengthen the review of class action settlements,[206] and my arguments lend further support to robust fairness hearings, especially if they are designed to ensure the attorney is acting as a faithful agent. The advantage of the proposal in this Article—concentrating fairness hearings on the connection between the attorney’s pay and the class’s interests—is that it solves a longstanding problem in fairness hearings by providing clear, actionable criteria for judges to apply.[207]

E. Sophisticated Plaintiffs

To summarize, something like voting is not essential to manage the principal-agent problem in class actions when all their characteristics, like the need for the attorney’s expertise, are taken into account. This defangs the democratic critique, the primary point of this Article, yet substantial agency costs remain. Class members simply cannot do better with democratic participation. The Panglossian scenario is for the expert agent to have interests identical to the principal.[208]

An important premise throughout has been that the class members have little at stake, individually. That was the source of the endemic principal-agent problem in class actions as well as the rational apathy and ignorance that interfered with various mechanisms of accountability. That fits the archetypal class action, but sometimes a principal will have a large stake in the litigation and might approach things differently. Institutional investors like pension funds are one example. Over the past twenty years, they have come to hold the majority of corporate stock, especially in the largest corporations.[209] They thus have a lot more at stake than the average shareholder. This changes their incentives with respect to corporate governance. Unlike a run of the mill shareholder, institutional investors have a far greater incentive to investigate what the firm is doing. They also have enough clout to make a tangible difference; they do not face the same irrationality of voting dilemma—if one represents 30% of the electorate, then the probability of being pivotal skyrockets.

There have been serious attempts to harness this dynamic to aid in class action accountability, replacing the ineffectual figurehead class representative with one that will vigorously monitor class counsel in the same way institutional investors could carefully monitor firms. In securities class actions, the Private Securities Litigation Reform Act (PSLRA) creates a rebuttable presumption that the class representative or “lead plaintiff” will be the one who has the largest financial stake in the case.[210] The hope is that this lead plaintiff will have the incentive to closely monitor the attorney.[211]

Institutional investors have not been a panacea for corporate governance, though, suggesting that their utility for class actions is likely limited. Ronald Gilson and Jeffrey Gordon argue that institutional investors actually create an additional principal-agent problem by interposing a new agent—the institutional investor—between shareholders and executives.[212] Sometimes, an extra agent can be helpful, but other times it can intensify principal-agent problems.[213] Gilson and Gordon identify a number of incentives that would lead institutional investors to rarely exercise their corporate governance power.[214] Among these, they point out that institutional investors possess extremely diversified stock portfolios, so while, in absolute terms, an institutional investor may own large quantities of stock, proportionally, its investment in any one firm is small,[215] limiting how much effort they will put into monitoring any single firm.[216] These are just a few of the explanations in the literature for why institutional investors may not have that great an impact on corporate governance.[217]

The impact of relatively high-stakes lead plaintiffs on class actions is similarly mixed. Courts have shown some reticence in appointing institutional investors to serve as lead plaintiffs,[218] although institutions are taking on an increasing role as lead plaintiffs.[219] In a study of securities class action settlements, Lynn Baker, Michael Perino, and Charles Silver concluded that “[m]any of this study’s findings show that the PSLRA has not worked as hoped.”[220] Not entirely unlike Gilson and Gordon’s analysis, Elizabeth Chamblee Burch points out that these lead plaintiffs will tend to have divergent interests from the rank and file class members and proposes appointing lead plaintiff groups in order to better represent the diversity of interests within the class.[221] For example, the institutional investor may be more intimately connected to the defendant corporation and want to maintain an ongoing relationship with it, giving it an interest in the long-term health of the defendant that other class members do not share.[222] This lead plaintiff would thus be introducing another principal-agent relationship: they are supposed to be acting on behalf of the class as a whole, yet their interests differ substantially, the kernel of a principal-agent problem.[223] As with corporations, then, the benefits of having something like a subset of the principal with a greater stake in the litigation may be limited.

The presence of this kind of high-stakes plaintiff like an institutional investor is important to take into account.[224] For my purposes, though, they do not play that big a role. It comes down to what problem they are thought to solve. The conventional wisdom is that institutional investors, by virtue of having more at stake, will not be subject to something like rational ignorance or apathy, resulting in more active monitoring and engagement. However, the delegation choices detailed above are not caused by rational apathy. Condition (1) does not require that the principal be extremely uninformed, even if that is what we expect to see in the modal class action, just that the agent be sufficiently more informed than the principal.[225] Even very sophisticated parties will not have the same case-specific information as the attorney who has been actually working on it.

This was the case in Lazy Oil Co. v. Witco Corp.,[226] where the class representatives that were both sophisticated and actively involved in the litigation objected to the settlement.[227] In his analysis of the case, Coffee notes that the objectors’ legal reasoning was weak: the odds of winning the case at trial were “formidably stacked against” the class after discovery and “the objectors’ theory for why greater damages were obtainable seemed tenuous . . . .”[228] Even if the principal is sophisticated and has a large stake in the case—like the institutional investor that leads a securities class action or the objectors in Lazy Oil—understanding how the facts map onto the law and the vagaries of litigation is not their bailiwick. Short of the lead plaintiff doing a full investigation itself—thereby obviating the utility of having an agent—the attorney who has been working on the case will know more about it than even a principal with a lot at stake. Put another way, the delegation strategy I have described is not the product of the principal’s cluelessness; rather, it is created by the agent’s expertise. Sophisticated plaintiffs do not solve the problem that motivated delegation of the type discussed in this Article, namely that the realities of strategic communication made the agent’s information hard to fully convey.

All that being said, The PSLRA’s lead plaintiff presumption and similar provisions are not irrelevant, especially if the lead plaintiff takes on an active role. Indeed, they might be especially helpful in negotiating the class counsel’s fee to help ensure she is a faithful agent to the class. Their concern notwithstanding, Baker, Perino, and Silver argue that lead plaintiffs should negotiate the fees and propose a set of arrangements to enable them to do so effectively—a structural improvement that would help mitigate the principal-agent problems in class actions.[229] In part, the benefits come from a lead plaintiff or similar high stakes member of the class being willing to put forth substantial effort, thoroughly scrutinizing the fee arrangement or the attorney’s qualifications. Further, most lead plaintiffs are repeat players, and the attorney recognizing this might work harder and drive a harder bargain at settlement in order to gain more of the lead plaintiff’s business in the future. There is even some possibility of behavioral economics effects: the title of “lead plaintiff,” enshrined in statute and blessed by a judge, might motivate institutional investors to take their position seriously, operating like a nudge. There are many effects to disentangle, which may account for the mixed results observed with lead plaintiffs. But the presence of lead plaintiffs, or high-stakes members of the class in general, do not fundamentally alter the dynamics discussed in this Part.

In terms of this Article, a sophisticated party likely affects the amount of bias that is tolerable for delegation to be a better strategy than democracy for the principal. Whether delegation is the best option depends on a relationship between the amount of bias and the informational gap between the principal and the agent. For rank-and-file class members, this gap will be massive. The phenomenon of rational ignorance implies that class members and shareholders will put virtually no effort into learning about the case. The institutional investor or similarly situated party, on the other hand, could have an incentive to do a thorough investigation. This would mean that delegation is optimal only when the agent has relatively little bias, which a sophisticated plaintiff can reduce by hiring the right attorney, something they also have sufficient incentive to invest in doing. In cases where there is relatively little bias, delegation still makes the principal better off. The agent’s advice is more accurate, giving the principal more information, but delegation performs even better.[230] In short, one should delegate to trustworthy agents. Also, from a social policy perspective, it is probably easier to design institutions to mitigate the agent’s bias—such as the attention to attorney’s fees and settlements I and others recommend—than convincingly bridging the information gap between an attorney and their client.


This Article’s main contribution has been to push back on the democratic critique, answering its overarching challenge to class actions. Along the way, I have clarified the critique, as well. The democratic critique consists of two main ingredients. There is the endemic principal-agent problem—class action attorneys have incentive to act in their own interest, to some degree contrary to the interests of the class—and the class members have little practical capacity to rein them in. Together, these two observations pose a profound question about the legitimacy and efficacy of class actions. But despite democracy’s normative pull, it is not always the best way to address principal-agent problems. Given the inherent trade-offs between expertise and control, additional democratic procedures would do little to help the class, even if they were ready and able to take complete advantage of them.

Accordingly, the fact that class actions lack these sorts of democratic procedures cannot mean they are fundamentally illegitimate. The democratic critique thus loses its force. These observations also suggest that the most promising avenues for class action reform involve aligning the interests of the attorney with their client. One step along this path is to make fairness hearings more process-oriented and ensure that, even at settlement, attorney’s fees have the right structure. Class action law should ensure that the attorney does well if and only if their client, the class, does. Along similar lines, with the basic issue of class action legitimacy resolved, law should focus on other important questions, such as how to sort out potential diversity of interests within a class. Conflicting, or simply just different, interests within the class may necessitate greater participation by class members. Notably, this would matter the least when class members have a small stake, the very circumstances where the democratic critique seemed the most acute.

  1. * Associate Professor of Law, Maurer School of Law, Indiana University – Bloomington. I would like to thank Adeno Addis, Lynn Baker, Charles Barzun, Kristen E. Boon, Guy-Uriel Charles, Michael Coenen, Ryan Doerfler, Adam Feibelman, John Ferejohn, Robert L. Fischman, Pamela Foohey, Robert Force, Luis Fuentes-Rohwer, David Gamage, Charles Gardner Geyh, Michael D. Gilbert, Myiram Gilles, Tim Glynn, Sanford C. Gordon, Catherine Hafer, Andrew Hammond, Ed Hartnett, Samuel Issacharoff, Dawn Johnsen, Dimitri Landa, Leandra Lederman, Ann Lipton, Asaf Lubin, João Marinotti, Troy McKenzie, Geoffrey P. Miller, David Noll, Shu-Yi Oei, Heather E. Payne, Portia Pedro, Teddy Rave, Sally Brown Richardson, Micah J. Schwartzman, Brian Sheppard, Charlie Silver, Gregg Strauss, Charles Sullivan, and Deborah Widiss as well as workshop participants at Chapman University Fowler School of Law, Culp Colloquium, Indiana University Maurer School of Law, New York University, Seton Hall, University of Texas, and University of Virginia for all their insights and feedback. I thank Derek S. Lee, Jackson B. Adams, Madison R. Benoit, Nicholas C. Blum, Charles T. Carr, Nicole E. Chandonnet, Jacob E. Podolsky, Yi Zhang, Dominic A. Ferrante, and Thomas R. Farrell for valuable editing help. All errors and omissions are solely mine.

  2. . In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293, 1300 (7th Cir. 1995).

  3. . Most famously, Brown v. Board of Education, 347 U.S. 483 (1954), was a class action.

  4. . The constitutional concerns cited with respect to class actions vary, including due process related to absent class members, standing, and whether the class action procedure violates the separation of powers by amounting to a judge-made change to substantive law. See, e.g., Hansberry v. Lee, 311 U.S. 32 (1940); TransUnion LLC v. Ramirez, 141 S. Ct. 2190 (2021) (especially at 2209–13); Alexandra D. Lahav, Are Class Actions Unconstitutional?, 109 Mich. L. Rev. 993 (2011).

  5. . See, e.g., Brian T. Fitzpatrick, Do Class Action Lawyers Make Too Little?, 158 U. Pa. L. Rev. 2043 (2010); infra Part I.

  6. . Fitzpatrick, supra note 4, at 2043 (“Class action lawyers are some of the most frequently derided players in our system of civil litigation.”); id. at 2044 (collecting sources).

  7. . Elizabeth Chamblee Burch, Optimal Lead Plaintiffs, 64 Vand. L. Rev. 1109, 1111 (2011); Jonathan R. Macey & Geoffrey P. Miller, The Plaintiffs’ Attorney’s Role in Class Action and Derivative Litigation: Economic Analysis and Recommendations for Reform, 58 U. Chi. L. Rev. 1, 93–94 (1991).

  8. . Deborah R. Hensler, Revisiting the Monster: New Myths and Realities of Class Action and Other Large Scale Litigation, 11 Duke J. Compar. & Int’l L. 179, 180 (2001).

  9. . See, e.g., Judith Resnik, “Vital” State Interests: From Representative Actions for Fair Labor Standards to Pooled Trusts, Class Actions, and MDLs in the Federal Courts, 165 U. Pa. L. Rev. 1765, 1782 (2017).

  10. . Notably the Class Action Fairness Act (CAFA) and the Private Securities Litigation Reform Act (PSLRA) have been justified in these terms, along with revisions to the Federal Rules of Civil Procedure. See, e.g., 28 U.S.C. § 1715(b) ; 15 U.S.C. § 78u-4(a)(3)(B)(iii) ; Figueroa v. Sharper Image Corp., 517 F. Supp. 2d 1292, 1301 (S.D. Fla. 2007) (noting the objection of several attorneys general to the proposed settlement); In re Initial Pub. Offering Sec. Litig., 214 F.R.D. 117, 120–21 (S.D.N.Y. 2002); Howard M. Erichson, CAFA’s Impact on Class Action Lawyers, 156 U. Pa. L. Rev. 1593, 1593 (2008) (“CAFA, like every other major class action development of recent years, was born amidst snide remarks about lawyers’ inventing lawsuits and manipulating the system to enrich themselves at others’ expense.”); see also id. at 1594 (stating the same with respect to the PSLRA, Rule 23 amendments, and Supreme Court class action decisions); Fed. R. Civ. P. 23(e), (g)–(h).

  11. . See Samuel Issacharoff, The Governance Problem in Aggregate Litigation, 81 Fordham L. Rev. 3165, 3182 (2013) [hereinafter Issacharoff, Governance in Aggregate Litigation]; Elizabeth J. Cabraser & Samuel Issacharoff, The Participatory Class Action, 92 N.Y.U. L. Rev. 846, 847 (2017) (“Concern over representational legitimacy permeates the development of modern class action law.”); Samuel Issacharoff, Governance and Legitimacy in the Law of Class Actions, 1999 Sup. Ct. Rev. 337, 340 [hereinafter Issacharoff, Governance and Legitimacy]; Culver v. City of Milwaukee, 277 F.3d 908, 910 (7th Cir. 2002); In re Telectronics Pacing Sys., Inc., 221 F.3d 870, 873–74 (6th Cir. 2000); Mars Steel Corp. v. Cont’l Ill. Nat’l Bank & Tr. Co. of Chi., 834 F.2d 677, 678 (7th Cir. 1987).

  12. . E.g., Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011); Comcast Corp. v. Behrend, 569 U.S. 27 (2013); Am. Express Co. v. Italian Colors Rest., 570 U.S. 228 (2013); Epic Sys. Corp. v. Lewis, 138 S. Ct. 1612 (2018); AT&T Mobility LLC v. Concepcion, 563 U.S. 333 (2011).

  13. . In re Rhone-Poulenc Rorer Inc., 51 F.3d 1293, 1299 (7th Cir. 1995) (noting the feasibility of separate individual trials as a reason to stop the class action from going forward); see also AT&T Mobility LLC, 563 U.S. at 365 (Breyer, J., dissenting).

  14. . E.g., Alexandra Lahav, Fundamental Principles for Class Action Governance, 37 Ind. L. Rev. 65 (2003).

  15. . John C. Coffee, Jr., Class Action Accountability: Reconciling Exit, Voice, and Loyalty in Representative Litigation, 100 Colum. L. Rev. 370 (2000) [hereinafter Coffee, Accountability and Exit].

  16. . Cabraser & Issacharoff, supra note 10, at 847. See especially id. at 866–67.

  17. . Charles R. Korsmo & Minor Myers, Aggregation by Acquisition: Replacing Class Actions with a Market for Legal Claims, 101 Iowa L. Rev. 1323 (2016).

  18. . Tiana Leia Russell, Exporting Class Actions to the European Union, 28 B.U. Int’l L.J. 141, 177 (2010); Hans-Bernd Schaefer, The Bundling of Similar Interests in Litigation. The Incentives for Class Action and Legal Actions Taken by Associations, 9 Eur. J.L. & Econ. 183, 199–201 (2000); see generally Issacharoff, Governance in Aggregate Litigation, supra note 10 (cataloguing responses to the fundamental principal-agent problem in class actions).

  19. . Hostility to class actions is associated with corporate interests and political conservatives. “It should come as no surprise that corporations don’t like class action lawsuits. . . . [B]ig corporations have been trying to get rid of class action lawsuits ever since we put them on the books in 1966. . . . Today, most conservatives seem to want to get rid of class action lawsuits . . . .” Brian T. Fitzpatrick, The Conservative Case for Class Actions 2, 4 (2019). Fitzpatrick resists this position, asserting that class actions, in particular, should be embraced by conservatives, if not by the self-serving corporations that are frequent defendants in them.

  20. . See, e.g., Michael C. Jensen & William H. Meckling, Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure, 3 J. Fin. Econ. 305 (1976); Margaret M. Blair & Lynn A. Stout, A Team Production Theory of Corporate Law, 85 Va. L. Rev. 247 (1999); Simone M. Sepe, Corporate Agency Problems and Dequity Contracts, 36 J. Corp. L. 113 (2010).

  21. . See, e.g., John Ferejohn, Incumbent Performance and Electoral Control, 50 Pub. Choice 5 (1986); Bernard Manin, Adam Przeworski & Susan C. Stokes, Elections and Representation, in Democracy, Accountability, and Representation 40 (Adam Przeworski, Susan C. Stokes & Bernard Manin, eds., 1999). As James Madison memorably put it: “[Y]ou must first enable the government to control the governed; and in the next place oblige it to control itself. . . . A dependence on the people is, no doubt, the primary control on the government . . . .” The Federalist No. 51 (James Madison).

  22. . See infra Section I.B.

  23. . See infra notes 74–80 and accompanying text.

  24. . Such comparisons are very common in this literature. See, e.g., Lahav, supra note 13; Issacharoff, Governance in Aggregate Litigation, supra note 10.

  25. . A non-exhaustive list of papers adopting this approach includes: Edward Brunet, Class Action Objectors: Extortionist Free Riders or Fairness Guarantors, 2003 U. Chi. Legal F. 403, 403–04; Coffee, Accountability and Exit, supra note 14; Macey & Miller, supra note 6, at 3–7; Lahav, supra note 13; Issacharoff, Governance in Aggregate Litigation, supra note 10, at 3183; Howard M. Downs, Federal Class Actions: Diminished Protection for the Class and the Case for Reform, 73 Neb. L. Rev. 646, 659–63 (1994); Alon Klement, Who Should Guard The Guardians? A New Approach for Monitoring Class Action Lawyers, 21 Rev. Litig. 25, 27–28 (2002).

  26. . I am indebted to Charlie Geyh for this observation.

  27. . See infra Section III.D.

  28. . “[E]lections are inherently a blunt instrument of control: voters have only one decision to make with regard to the entire package of government policies. . . . One cannot control a thousand targets with one instrument.” Manin, Przeworski & Stokes, supra note 20, at 49–50; see also Dimitri Landa & Adam Meirowitz, Game Theory, Information, and Deliberative Democracy, 53 Am. J. Pol. Sci. 427, 427 (2009) (“[V]oting is not the best, or at least not the only, political mechanism for ensuring that policy decisions conform to the interests of the citizens.”); Christopher R. Berry & Jacob E. Gersen, The Unbundled Executive, 75 U. Chi. L. Rev. 1385, 1393 (2008); Nicholas Almendares & Patrick Le Bihan, Increasing Leverage: Judicial Review as a Democracy-Enhancing Institution, 10 Q.J. Pol. Sci. 357, 361–62 (2015).

  29. . This overemphasis is fairly common in the law, the jurisprudence related to administrative agency removal being one of the clearest examples. See, e.g., Free Enter. Fund v. Pub. Co. Acct. Oversight Bd., 561 U.S. 477 (2010); Bowsher v. Synar, 478 U.S. 714 (1986); Nicholas Almendares & Catherine Hafer, Beyond Citizens United, 84 Fordham L. Rev. 2755, 2789–94 (2016).

  30. . A number of studies support this conclusion in a variety of contexts. See, e.g., Wouter Dessein, Authority and Communication in Organizations, 69 Rev. Econ. Stud. 811 (2002); Nicholas Almendares, Blame-Shifting, Judicial Review, and Public Welfare, 27 J.L. & Pol. 239 (2012); Sean Gailmard & John W. Patty, Slackers and Zealots: Civil Service, Policy Discretion, and Bureaucratic Expertise, 51 Am. J. Pol. Sci. 873, 874 (2007).

  31. . Dessein, supra note 29, at 812–13, 832.

  32. . The intuitions behind this conclusion are explained in more detail in Section III.B, infra.

  33. . See infra Section III.C.

  34. . I am certainly not the first to suggest class action attorney fees should be looked at carefully, although my particular suggestions differ from much previous work. I take up the issue of settlement in Section III.D. See also Charles Silver & Lynn Baker, I Cut, You Choose: The Role of Plaintiffs’ Counsel in Allocating Settlement Proceeds, 84 Va. L. Rev. 1465 (1998); Christopher R. Leslie, A Market-Based Approach to Coupon Settlements in Antitrust and Consumer Class Action Litigation, 49 UCLA L. Rev. 991 (2002); Fitzpatrick, supra note 4; Myriam Gilles & Gary B. Friedman, Exploding the Class Action Agency Costs Myth: The Social Utility of Entrepreneurial Lawyers, 155 U. Pa. L. Rev. 103 (2006). Howard Erichson’s arguments are the most similar, though he does not connect them to the longstanding critique of class action legitimacy. Howard M. Erichson, Aggregation as Disempowerment: Red Flags in Class Action Settlements, 92 Notre Dame L. Rev. 859, 873 (2016).

  35. . See infra Section III.E.2.

  36. . These could include the recent opioid litigation. E.g., In re Nat’l Prescription Opiate Litig., 976 F.3d 664 (6th Cir. 2020). Parens patriae litigation, where a state stands in the shoes of its citizens asserting a “quasi-sovereign” interest of the citizenry, could be another possible example. See Alfred L. Snapp & Son, Inc. v. Puerto Rico ex rel. Barez, 458 U.S. 592, 607 (1982).

  37. . See supra notes 10–11.

  38. . E.g., Elliott J. Weiss, The Lead Plaintiff Provisions of the PSLRA After a Decade, or “Look What’s Happened to My Baby”, 61 Vand. L. Rev. 543 (2008).

  39. . See infra Section III.D.

  40. . I have in mind here the requirements of commonality and adequacy of representation contained in Rule 23 and the possible utilization of sub-classes. See Fed. R. Civ. P. 23(a).

  41. . Canonical statements of this as the default state of politics can be found in the Federalist Papers and the work of John Rawls. See The Federalist No. 10 (James Madison); John Rawls, A Theory of Justice 340 (rev. ed. 1999); see also Micah Schwartzman, What if Religion Is Not Special?, 79 U. Chi. L. Rev. 1351, 1425 (2012).

  42. . Sean Gailmard, Accountability and Principal-Agent Theory, in The Oxford Handbook of Public Accountability 98 (Mark Bovens, Robert E. Goodin & Thomas Schillemans, eds., 2014).

  43. . See infra Section III.B.

  44. . It is possible to construct circumstances where this generalization does not hold, but they are idiosyncratic. It would take something like a case where there is both a very small number of class members that have vastly disproportionate stakes in the litigation and the other, smaller-stakes class members are not too numerous. Suppose, for instance, that there is a class of 100 people, 99 of which have a claim worth $1 but one of them has a claim $100,000. The amount at stake in the litigation for the attorney and that single high-value plaintiff are comparable. Note, however, if we increase the size of the hypothetical class, the attorney can once again have more at stake than any single class member.

    The closest real-world analogue to this kind of case would be something like institutional investors in securities actions. I address them specifically in Section III.E. Suffice to say, they do not fundamentally change the arguments in this Article.

  45. . Fed. R. Civ. P. 23(a)(1).

  46. . All else may, of course, not be equal. See infra note 48 and accompanying text.

  47. . This scenario is inspired by AT&T Mobility LLC v. Concepcion, 563 U.S. 333 (2011).

  48. . Lynn A. Baker, Michael A. Perino & Charles Silver, Is the Price Right? An Empirical Study of Fee-Setting in Securities Class Actions, 115 Colum. L. Rev. 1371, 1373 n.3 (2015).

  49. . This is a version of risk-averse behavior that does not depend on a particular psychological tendency. It is “rational” risk aversion. See Nicholas Almendares, The False Allure of Settlement Pressure, 50 Loy. U. Chi. L.J. 271, 301–02 (2018).

  50. . See id. at 278–90.

  51. . Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 350 (2011).

  52. . See William H. Wagener, Modeling the Effect of One-Way Fee Shifting on Discovery Abuse in Private Antitrust Litigation, 78 N.Y.U. L. Rev. 1887, 1893 & n.30 (2003).

  53. . Almendares, supra note 48, at 284–87.

  54. . Coffee, Accountability and Exit, supra note 14, at 391.

  55. . See, e.g., Marc Galanter & Mia Cahill, “Most Cases Settle”: Judicial Promotion and Regulation of Settlements, 46 Stan. L. Rev. 1339, 1340 (1994); Judith Resnik, Whose Judgment? Vacating Judgments, Preferences for Settlement, and the Role of Adjudication at the Close of the Twentieth Century, 41 UCLA L. Rev. 1471, 1474–75 (1994).

  56. . Class actions where class members each have substantial individual claims are less common, but not unheard of. High profile examples include In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293 (7th Cir. 1995), and In re Nat’l Football League Players Concussion Inj. Litig., 821 F.3d 410, 420 (3d Cir. 2016), as amended (May 2, 2016).

  57. . Almendares, supra note 48, at 302.

  58. . See generally id.

  59. . Much of the previous work on this issue adopts Albert Hirschman’s tryptic of organizations governance—exit, voice, and loyalty—to categorize mechanisms for controlling agents and then describe what is missing. See, e.g., Coffee, Accountability and Exit, supra note 14, at 376 & n.17.

  60. . See Ellen J. Bennett & Helen W. Gunnarsson, Annotated Model Rules of Professional Conduct §§ 1.1, 1.3 (9th ed. 2019).

  61. . Ann Southworth, Collective Representation for the Disadvantaged: Variations in Problems of Accountability, 67 Fordham L. Rev. 2449, 2452 (1999).

  62. . “Because the class counsel is more than simply a fiduciary, but rather is a joint venturer with the class, a fixation with fiduciary principles alone produces a myopic tunnel vision.” Coffee, Accountability and Exit, supra note 14, at 436.

  63. . Elliott J. Weiss & John S. Beckerman, Let the Money Do the Monitoring: How Institutional Investors Can Reduce Agency Costs in Securities Class Actions, 104 Yale L.J. 2053, 2065 (1995) (footnote omitted).

  64. . Issacharoff, Governance and Legitimacy, supra note 10, at 374–76 (“Nor can a guardian revisit such issues unless the guardian has the full range of information available to counsel—in effect doubling the costs of representation.”).

  65. . Edward B. Rock, Adapting to the New Shareholder-Centric Reality, 161 U. Pa. L. Rev. 1907, 1911 (2013).

  66. . E.g., Coffee, Accountability and Exit, supra note 14, at 376–77; Issacharoff, Governance and Legitimacy, supra note 10, at 374–75.

  67. . See, e.g., 3 William Meade Fletcher, Fletcher Cyclopedia of the Law of Corporations § 837.50 (“A director . . . owes the corporation complete loyalty, honesty, and good faith.”).

  68. . Traditionally, corporate fiduciary duties are divided into two categories, with the duty of care being extremely lax, especially in practice owing to the generous presumptions of the business judgment rule. See, e.g., Aronson v. Lewis, 473 A.2d 805, 812 (Del. 1984); In re Walt Disney Co. Derivative Litig., 906 A.2d 27, 74 (Del. 2006); see also Patricia A. McCoy, A Political Economy of the Business Judgment Rule in Banking: Implications for Corporate Law, 47 Case W. Reserve L. Rev. 1, 7–8 (1996); Stephen M. Bainbridge, The Business Judgment Rule as Abstention Doctrine, 57 Vand. L. Rev. 83, 100 (2004).

  69. . E.g., Kenneth B. Davis, Jr., Judicial Review of Fiduciary Decisionmaking—Some Theoretical Perspectives, 80 Nw. U. L. Rev. 1, 61 (1985); see also Gottlieb v. Heyden Chem. Corp., 91 A.2d 57, 59 (Del. 1952) (“[T]he entire atmosphere is freshened and a new set of rules invoked where formal approval has been given by a majority of independent, fully informed stockholders.”); accord Beard v. Elster, 160 A.2d 731, 738 (Del. 1960); Cohen v. Ayers, 596 F.2d 733, 740 (7th Cir. 1979); Weinberger v. UOP, Inc., 457 A.2d 701, 711 (Del. 1983) (describing the applicable “entire fairness” standard).

  70. . See infra notes 156–61 and accompanying text.

  71. . The conventional wisdom is that compensation like stock options largely achieves the governance goals. E.g., Sepe, supra note 19, at 114.

  72. . Voting can also be used to select the best agents for a position. The main, distinctive issues in class actions is how to ensure the agent takes actions in the litigants’ best interests, essentially a moral hazard problem. Hiring attorneys such as class counsel involves participating in the market for legal services, not all that dissimilar from hiring any attorney. Global class settlements do present unique problems as they can encourage class counsel to engage in a “reverse auction” to settle cheaply. These reverse auctions represent a problem distinct from the general democratic critique of class actions and are best left to their own sustained discussion.

  73. . Christopher R. Leslie, The Significance of Silence: Collective Action Problems and Class Action Settlements, 59 Fla. L. Rev. 71, 76 (2007).

  74. . Coffee, Accountability and Exit, supra note 14, at 422–23.

  75. . E.g., Lazy Oil Co. v. Witco Corp., 166 F.3d 581 (3d Cir. 1999); Cnty. of Suffolk v. Long Island Lighting Co., 907 F.2d 1295, 1325 (2d Cir. 1990).

  76. . Fed. R. Civ. P. 23(c)(2)(B).

  77. . George Rutherglen, Future Claims in Mass Tort Cases: Deterrence, Compensation, and Necessity, 88 Va. L. Rev. 1989, 1995 (2002).

  78. . See, e.g., Theodore Eisenberg & Geoffrey Miller, The Role of Opt-Outs and Objectors in Class Action Litigation: Theoretical and Empirical Issues, 57 Vand. L. Rev. 1529, 1538–40 (2004); Coffee, Accountability and Exit, supra note 14, at 370.

  79. . E.g., Coffee, Accountability and Exit, supra note 14, at 382.

  80. . E.g., Leslie, supra note 72, at 113. While framed differently, Coffee offers a similar version of this argument. See Coffee, Accountability and Exit, supra note 14, at 417, 422.

  81. . Eisenberg & Miller, supra note 77, at 1560 (“In general, one may surmise that class members do not highly value these rights that courts and commentators have so widely praised as essential to the justification for group litigation involving absent parties.”); e.g., Eubank v. Pella Corp., 753 F.3d 718, 728 (7th Cir. 2014); Nick Landsman-Roos, Front-End Fiduciaries: Precertification Duties and Class Conflict, 65 Stan. L. Rev. 817, 847 (2013); Antonio Gidi, Loneliness in the Crowd: Why Nobody Wants Opt-Out Class Members to Assert Offensive Issue Preclusion Against Class Defendants, 66 SMU L. Rev. 1, 18 (2013).

  82. . Eisenberg & Miller, supra note 77, at 1555.

  83. . Transitional governments being the notable exception.

  84. . See Robert Cooter & Thomas Ulen, Law and Economics 301–04 (6th ed. 2016) (detailing how having a definite endpoint can undermine an enduring relationship, also known as the “Endgame Problem”).

  85. . Issacharoff, Governance in Aggregate Litigation, supra note 10, at 3171, 3177.

  86. . Id. at 3169; see also Adam Przeworski, Michael E. Alvarez, José Antonio Cheibub & Fernando Limongi, Democracy and Development: Political Institutions and Well-Being in the World, 1950–1990, at 18–25 (2000).

  87. . The canonical citation is William H. Riker & Peter C. Ordeshook, A Theory of the Calculus of Voting, 62 Am. Pol. Sci. Rev. 25 (1968). See also John Duffy & Margit Tavits, Beliefs and Voting Decisions: A Test of the Pivotal Voter Model, 52 Am. J. Pol. Sci. 603 (2008).

  88. . Riker & Ordeshook, supra note 86, at 26. Riker and Ordeshook’s article is fifty years old, so the U.S. population has increased substantially in the intervening time, but their ballpark estimate for the probability of being pivotal is 10-8. Id. at 25.

  89. . The exception might be if the class is quite small, such as on the order of thirty people. On the matter of small classes, see, for example, William B. Rubenstein, 1 Newberg and Rubenstein on Class Actions § 3:12 (6th ed. 2022) and the sources cited therein.

  90. . E.g., Coffee, Accountability and Exit, supra note 14; Charles R. Korsmo & Minor Myers, Aggregation by Acquisition: Replacing Class Actions with a Market for Legal Claims, 101 Iowa L. Rev. 1323 (2016).

  91. . Inevitably, there are considerations that I have neglected here. One that springs to mind is the danger of divergent interests within the class itself. The focus of this Article is the democratic critique, though, which applies even if the class members are unified in their interests. That is, even class actions with perfect commonality or perfectly crafted subclasses would still be subject to the democratic critique.

  92. . See, e.g., Amchem Prods., Inc. v. Windsor, 521 U.S. 591 (1997); Eubank v. Pella Corp., 753 F.3d 718, 719 (7th Cir. 2014); Issacharoff, Governance in Aggregate Litigation, supra note 10, at 3166 (“Beginning at least with the Supreme Court’s decisions in Amchem Products, Inc. v. Windsor and Ortiz v. Fibreboard Corp., the adequacy of class representation became one of the most promising routes to challenge the appropriateness of class certification.”) (footnotes omitted).

  93. . See Hansberry v. Lee, 311 U.S. 32 (1940); Jay Tidmarsh, Rethinking Adequacy of Representation, 87 Tex. L. Rev. 1137, 1151–85 (2009); Martin H. Redish, Class Actions and the Democratic Difficulty: Rethinking the Intersection of Private Litigation and Public Goals, 2003 U. Chi. Legal F. 71, 125 n.176.

  94. . E.g., Issacharoff, Governance Problem in Aggregate Litigation, supra note 10, at 3166 (referencing Fed. R. Civ. P. 23(a)(2), (b)(3), and adequacy of representation).

  95. . Amchem, 521 U.S. at 622–25

  96. . See, e.g., In re BankAmerica Corp. Sec. Litig., 350 F.3d 747, 751–52 (8th Cir. 2003).

  97. . “The commonality and typicality requirements of Rule 23(a) tend to merge. . . . Those requirements therefore also tend to merge with the adequacy-of-representation requirement . . . .” Gen. Tel. Co. of Sw. v. Falcon, 457 U.S. 147, 157 n.13 (1982); see also Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 349 n.5 (2011) (quoting this passage from Falcon).

  98. . See Fed. R. Civ. P. 23(b) (types of class actions).

  99. . See Wal-Mart Stores, Inc., 564 U.S. at 368 (Ginsburg, J., dissenting in part and concurring in part).

  100. . 527 U.S. 815 (1999).

  101. . Id. at 852, 852 n.30.

  102. . Id. at 852–53.

  103. . 521 U.S. 591 (1997).

  104. . Id. at 627.

  105. . Issacharoff, Governance and Legitimacy, supra note 10, at 366.

  106. . Coffee, Accountability and Exit, supra note 14, at 417.

  107. . See, e.g., Jeremy Waldron, The Wisdom of the Multitude: Some Reflections on Book 3, Chapter 11 of Aristotle’s Politics, 23 Pol. Theory 563 (1995); Kathleen Doherty & Ryan Pevnick, Are There Good Procedural Objections to Judicial Review?, 76 J. Pol. 86, 88–89 (2013); Thomas Christiano, An Instrumental Argument for a Human Right to Democracy, 39 Phil. & Pub. Affs. 142 (2011); Gerald F. Gaus, Justificatory Liberalism: An Essay on Epistemology and Political Theory (1996); see also Dimitri Landa & Ryan Pevnick, Representative Democracy as Defensible Epistocracy, 114 Am. Pol. Sci. Rev. 1 (2020).

  108. . See Issacharoff, Governance and Legitimacy, supra note 10, at 341; Tidmarsh, supra note 92, 1137–38.

  109. . David A. Strauss, Is Carolene Products Obsolete?, 2010 U. Ill. L. Rev. 1251, 1256 (2010). But see Almendares & Hafer, supra note 28, at 2798.

  110. . Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 844 (1984).

  111. . Chevron, 467 U.S. at 865–66; see also Thomas W. Merrill, Judicial Deference to Executive Precedent, 101 Yale L.J. 969, 978–79 (“In addition to its novel framework, Chevron also broke new ground by invoking democratic theory as a basis for requiring deference to executive interpretations.”).

  112. . Almendares, supra note 29, at 239, 285–86.

  113. . Nat’l Fed’n of Indep. Bus. v. Dept. of Labor, 142 S. Ct. 661, 667–70 (2022) (Gorsuch, J., concurring).

  114. . Citizens United v. FEC, 558 U.S. 310 (2010).

  115. . Burwell v. Hobby Lobby Stores, Inc., 573 U.S. 682 (2014).

  116. . In Citizens United, the Court concluded that there was no compelling government interest in preventing shareholders from, in effect, subsidizing political advocacy they might disagree with because corporate democracy and other tools of corporate governance would prevent that from happening. Almendares & Hafer, supra note 28, at 2766; Citizens United, 558 U.S. at 361; see also First Nat’l Bank of Boston v. Bellotti, 435 U.S. 765, 794–95 (1978). The Hobby Lobby majority concluded that corporate governance was up to the task of resolving religious disagreements within corporations, though they only explicitly considered closely held ones. Hobby Lobby, 573 U.S. at 717 (“HHS has not pointed to any example of a publicly traded corporation asserting RFRA rights, and numerous practical restraints would likely prevent that from occurring. . . . [T]he idea that unrelated shareholders—including institutional investors with their own set of stakeholders—would agree to run a corporation under the same religious beliefs seems improbable.”) (emphasis added); see also id. at 718–19 (“State corporate law provides a ready means for resolving any conflicts . . . .”).

  117. . Citizens United considered the First Amendment political speech rights of corporations. Hobby Lobby extended religious protections to corporations, although it actually centered on interpretation of the Religious Freedom Restoration Act (RFRA) rather than the Constitution, a difference courts tend to elide. See Elizabeth Sepper, Free Exercise Lochnerism, 115 Colum. L. Rev. 1453, 1468–69 (2015).

  118. . I consider the stark differences in the legal treatment of class actions and other institutions that engage in collective action, including collective or aggregate litigation, in more detail elsewhere. See Nicholas Almendares, Aggregation by Other Means (2021) (unpublished manuscript) (on file with author).

  119. . I will use the class and the class members interchangeably. But there is arguably a difference between the two. One of the lingering questions in class action law is whether we should understand a class as a combination of individuals, in essence, a complex joinder device. On another view, the class is an entity unto itself, created by the system of civil procedure. That makes it an entity in its own right, something along the lines of the way the law treats corporations and governments. See, e.g., David L. Shapiro, Class Actions: The Class as Party and Client, 73 Notre Dame L. Rev. 913, 939 (1998); Alexandra D. Lahav, Two Views of the Class Action, 79 Fordham L. Rev. 1939, 1943 (2011).

    While this debate has important ramifications for both the theory and practice of class actions, it does not directly impact the central arguments of this Article. The reason is that entities like corporations and governments—and, possibly, the class itself if one adopts that view—constantly confront principal-agent problems. Likewise, a group of individuals that chooses to designate an agent can also be forced to reckon with principal-agent problems. One influential work identifies Congress as a prominent example of this phenomenon. Kenneth A. Shepsle, Congress Is a “They,” Not an “It”: Legislative Intent as Oxymoron, 12 Int’l Rev. L. & Econ. 239 (1992). Whichever view of the class action one adopts—entity or collection of individuals—principal-agent problems can still arise. See Issacharoff, Governance and Legitimacy, supra note 10, at 374 (“Even viewing the class as an entity does little to address the tension that may emerge between the interests and those of their agents.”).

  120. . See, e.g., Tidmarsh, supra note 92; supra notes 104–106.

  121. . Russell M. Gold, Compensation’s Role in Deterrence, 91 Notre Dame L. Rev. 1997, 2003 (2016).

  122. . E.g., In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293 (7th Cir. 1995); In re NFL Players’ Concussion Injury Litig., 307 F.R.D. 351 (E.D. Pa. 2015).

  123. . Russell Gold argues that class interest and deterrence are linked, but for different reasons than those I present here. Gold’s arguments are based on reputation effects and public perceptions of the class device. See generally, Gold, supra note 120.

  124. . The normative account relied on by both deterrence and class interest is broadly utilitarian. See, e.g., Fitzpatrick, supra note 4, at 2047.

  125. . Id. at 2057.

  126. . David Rosenberg, Response, Mandatory-Litigation Class Action: The Only Option for Mass Tort Cases, 115 Harv. L. Rev. 831, 843 (2002); accord Gold, supra note 120, at 2004.

  127. . Gold, supra note 120, at 2005.

  128. . All else may, of course, not be equal. The fraud may come with reputational harms and other difficulties for the defendant. Id. at 2007–09.

  129. . This is not true for all normative positions. See infra note 176.

  130. . See, e.g., Almendares, supra note 48, at 290–91.

  131. . In theory, it is possible that the combination of large litigation costs plus a risk-averse and unconstrainted attorney could just happen to balance out, leading to optimal deterrence in class actions by coincidence. The simple example used in this section does. But, in actuality, this is extraordinarily unlikely because, among other reasons, the attorney’s risk aversion has no correlation with the defendant’s litigation costs.

  132. . “The policy at the very core of the class action mechanism is to overcome the problem that small recoveries do not provide the incentive for any individual to bring a solo action prosecuting his or her rights. A class action solves this problem by aggregating the relatively paltry potential recoveries into something worth someone’s (usually an attorney’s) labor.” Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 617 (1997) (quoting Mace v. Van Ru Credit Corp., 109 F.3d 338, 344 (7th Cir. 1997)).

  133. . E.g., 28 U.S.C. § 1715(b); 15 U.S.C. § 78u-4(a)(3)(B)(iii); In re Initial Pub. Offering Sec. Litig., 214 F.R.D. 117, 120–21 (S.D.N.Y. 2002); Erichson, supra note 9.

  134. . E.g., Coffee, Accountability and Exit, supra note 14, at 377–78; Cabraser & Issacharoff, supra note 10, at 852; Korsmo & Myers, supra note 16, at 1326–27; Schaefer, supra note 17, at 199–201.

  135. . E.g., Amchem Prods., Inc., 521 U.S. 591; Eubank v. Pella Corp., 753 F.3d 718, 719 (7th Cir. 2014).

  136. . Democracy here is treated as an instrumental good, as it often is. See supra note 106.

  137. . Vincent P. Crawford & Joel Sobel, Strategic Information Transmission, 50 Econometrica 1431 (1982).

  138. . See Gailmard, supra note 41, at 97–98 and the citations therein. This is a common phenomenon in principal-agent situations. See, e.g., Almendares, supra note 29, at 255; Gailmard & Patty, supra note 29, at 874.

  139. . Gailmard, supra note 41, at 98.

  140. . I am indebted to Lynn Baker for this example.

  141. . E.g., United States v. Brown, 108 F.3d 863, 864 (8th Cir. 1997); Hannah Fresques, Dollars for Doctors: Doctors Prescribe More of a Drug if They Receive Money from a Pharma Company Tied to It, ProPublica (Dec. 20, 2019, 12:00 PM), [].

  142. . Compare TransUnion LLC v. Ramirez, 141 S. Ct. 2190, 2208–10 (2021) with id. at 2223 (Thomas, J., dissenting).

  143. . When available, bellwether trials and similar procedures can serve these ends, though they have their own challenges. See, e.g., William B. Rubenstein, 4 Newberg and Rubenstein on Class Actions § 11:12 (6th ed. 2022); Edward F. Sherman, Segmenting Aggregate Litigation: Initiatives and Impediments for Reshaping the Trial Process, 25 Rev. Litig. 691, 696­–702 (2006); Peter H. Schuck, Mass Torts: An Institutional Evolutionist Perspective, 80 Cornell L. Rev. 941, 960 (1995); Eldon E. Fallon, Jeremy T. Grabill & Robert Pitard Wynne, Bellwether Trials in Multidistrict Litigation, 82 Tul. L. Rev. 2323 (2008).

  144. . See, e.g., Dessein, supra note 29 (corporate context); Almendares, supra note 29 (administrative agency policymaking context); Brandice Canes-Wrone, Michael C. Herron & Kenneth W. Shotts, Leadership and Pandering: A Theory of Executive Policymaking, 45 Am. J. Pol. Sci. 532 (2001) (elected officials context); see generally Gailmard, supra note 41 (discussing various applications).

  145. . See supra Section I.B.

  146. . There will naturally be things these simple models neglect. But their simplicity clearly illustrates the trade-offs important to the present purposes.

  147. . In theory, the attorney could try to tell class members what their ideal settlement threshold would be if they had all of her expertise. Although it is not clear why she would do so. This strategy still runs into the same problems as the Honest Advice scenario. It would also require the attorney to somehow know the class members’ true preferences and tolerance for risk.

  148. . In a more complicated model, we could include things like reputational effects that could help. However, (1) there are serious challenges to relying on reputational effects in plaintiff class actions, including that learning about the lawyers’ reputations runs into its own rational ignorance and apathy problem, and (2) because the basic dynamic described here is still present, a credible reputation for honesty can only mitigate it to some degree. Such a reputation is a potential response to the trade-off described in this Article, though developing it in the first place raises the same difficulties described here.

  149. . The actors in these models are perfectly rational in the classic economics sense. For a similar effect grounded in behavioral economics and anchoring, see Brian Sheppard & Fiery Cushman, Evaluating Norms: An Empirical Analysis of the Relationship Between Norm-Content, Operator, and Charitable Behavior, 63 Vand. L. Rev. 55 (2010).

  150. . The class members could also “go it alone” and not receive any advice from the attorney. But that would make them even worse off—the attorney’s advice is a noisy signal, but it still conveys some information.

  151. . All of which could be treated as bundled together in the equivalent of a “platform” when the class voted to select an attorney.

  152. . See, e.g., Ariel Rubinstein, Comments on the Interpretation of Game Theory, 59 Econometrica 909, 909 (1991) (“I view game theory as an analysis of the concepts used in social reasoning when dealing with situations of conflict. It is an abstract inquiry into the function and logic of social institutions and patterns of behavior.”); id. at 918 (“Models are not supposed to be isomorphic with reality.”); Debra Satz & John Ferejohn, Rational Choice and Social Theory, 91 J. Phil. 71 (1994).

  153. . See, e.g., Dessein, supra note 29; Almendares, supra note 29; Gailmard & Patty, supra note 29, at 874.

  154. . An implicit condition is that there is a single or primary dimension of interest to the parties involved. That is, there are not cross-cutting considerations. Introducing these would complicate the analysis but would not change the nature of the results.

  155. . Dessein, supra note 29, at 812–13; see also Sean Gailmard, Discretion Rather than Rules: Choice of Instruments to Control Bureaucratic Policy Making, 17 Pol. Analysis 25 (2009).

  156. . Note that while they both entail surrendering some control, the logic here differs from precommitment devices like those contained in constitutions. See generally Jon Elster, Ulysses and the Sirens: Studies in Rationality and Irrationality (1979).

  157. . Dessein, supra note 29, at 832.

  158. . I consider the idea of more sophisticated members of the class below. See infra Section III.E.

  159. . Rock, supra note 64, at 1918–19; see also id. at 1926 (“As interesting as these issues are, they are better characterized as ‘mopping up operations’ than the grand battles against entrenchment and agency costs of the 1980s.”).

  160. . Id. at 1918–19.

  161. . Id. at 1987; Sepe, supra note 19.

  162. . Reinier Kraakman, Hyun Park & Steven Shavell, When Are Shareholder Suits in Shareholder Interests?, 82 Geo. L.J. 1733 (1994); Kent Greenfield, Reclaiming Corporate Law in a New Gilded Age, 2 Harv. L. & Pol’y Rev. 1 (2008).

  163. . “Management compensation can be too high even if its structure is appropriate.” Rock, supra note 64, at 1926.

  164. . See, e.g., John C. Coffee, Jr., Class Wars: The Dilemma of the Mass Tort Class Action, 95 Colum. L. Rev. 1343, 1364–65 (1995).

  165. . Silver & Baker, supra note 33, at 1528–29.

  166. . See Fed. R. Civ. P. 23(a), (g).

  167. . Failure to adequately represent the interests of absent plaintiffs may make the action nonbinding on the absent plaintiff. See Hansberry v. Lee, 311 U.S. 32, 42–43 (1940) (discussing the Due Process Clause); see also Roger C. Cramton, Individualized Justice, Mass Torts, and “Settlement Class Actions”: An Introduction, 80 Cornell L. Rev. 811, 827–28 (1995).

  168. . See supra note 28. But see Baker et al., supra note 47, at 1381–82, 1432 (suggesting ex ante fee arrangements for class actions).

  169. . The canonical model of pandering is found in Canes-Wrone, et al., supra note 143. There are additional complications in their model, such as variation in the expertise of agents (although all agents are more informed than the principal). Id. at 533.

  170. . Another requirement is that the elections are competitive. If the agent has no real chance of winning or losing, then pandering has no utility. Id. at 537.

  171. . For one example, though in the context of multidistrict litigation, so enjoying fewer procedural safeguards than class members, see Elizabeth Chamblee Burch, Life Inside the Mass Tort Machine, Practitioner Insights Commentaries, Aug.11, 2021. See also id. (“Their experiences cannot be chalked up to a few lawyers behaving badly. Two-hundred-ninety-five different lawyers from 145 law firms represented our participants. . . . It is a vast system that fails plaintiffs at every turn.”).

    One of the challenges in assessing these cases is that there are other values than the fairly narrow frame of class welfare that I, following the democratic critique, adopt here. See id.; see generally Victor D. Quintanilla, Human-Centered Civil Justice Design, 121 Pa. St. L. Rev. 745 (2017); Elizabeth Chamblee Burch, Calibrating Participation: Reflections on Procedure Versus Procedural Justice, 65 DePaul L. Rev. 323, 326 (2016).

  172. . There is an alternative kind of democratic critique that could, in principle, be leveled against class actions: that democratic participation is intrinsically good in all contexts. Democracy would then be an end in itself rather than a means to one such as controlling an agent. That would be a fundamentally different critique than the one presented in the academic literature and the law. Coffee has some hints of this kind of argument when he privileges client autonomy, though as discussed above, his definition of autonomy makes his argument one about the instrumental benefits of democratic procedures. See supra note 105 and accompanying text.

  173. . See, e.g., Janet Cooper Alexander, Do the Merits Matter? A Study of Settlements in Securities Class Actions, 43 Stan. L. Rev. 497, 498 (1991).

  174. . It can also ruin the analogy between corporate and class attorney compensation.

  175. . Silver & Baker, supra note 33, at 1527–28.

  176. . See, e.g., Leslie, supra note 33.

  177. . See 7B Charles Alan Wright & Arthur R. Miller, Federal Practice & Procedure § 1797 (3d ed.) and sources cited therein.

  178. . E.g., William B. Rubenstein, The Fairness Hearing: Adversarial and Regulatory Approaches, 53 UCLA L. Rev. 1435, 1481 (2006); Leslie, supra note 33, at 1053–54; Samuel Issacharoff, Class Action Conflicts, 30 U.C. Davis L. Rev. 805, 832–33 (1997).

  179. . Baker et al., supra note 47 , at 1378; see also Weiss & Beckerman, supra note 62, at 2066–67 (1995).

  180. . Fed. R. Civ. P. 23(e)(2).

  181. . See, e.g., In re Agent Orange Prod. Liab. Litig., 597 F. Supp. 740, 761 (E.D.N.Y. 1984), aff’d, 818 F.2d 145 (2d Cir. 1987) (collecting cases).

  182. . See, e.g., Almendares, supra note 48, at 294 and the citations therein.

  183. . For a detailed analysis, see William B. Rubenstein, supra note 177, at 1469. See also Erichson, supra note 33, at 911 (“Thinking critically about settlements always presents a challenge.”).

  184. . In re Agent Orange Prod. Liab. Litig., 597 F. Supp. at 761 (quoting Grunin v. Int’l House of Pancakes, 513 F.2d 114, 124 (8th Cir. 1975)), aff’d, 818 F.2d 145 (2d Cir. 1987).

  185. . See, e.g., In re Agent Orange Prod. Liab. Litig., 597 F. Supp. at 762 (collecting cases).

  186. . See, e.g., Weinberger v. UOP, Inc., 457 A.2d 701, 711 (Del. 1983); Geddes v. Anaconda Copper Mining Co., 254 U.S. 590, 599 (1921); 3 William Meade Fletcher, Fletcher Cyclopedia of the Law of Corporations § 974.

  187. . Gottlieb v. Heyden Chem. Corp., 91 A.2d 57, 59 (Del. 1952); Beard v. Elster, 160 A.2d 731, 738 (Del. 1960); Cohen v. Ayers, 596 F.2d 733, 740 (7th Cir. 1979).

  188. . D. Theodore Rave, Politicians as Fiduciaries, 126 Harv. L. Rev. 671, 704 (2013).

  189. . For a similar suggestion, see Jessica Erickson, The Gatekeepers of Shareholder Litigation, 70 Okla. L. Rev. 237, 271 (2017).

  190. . See Tidmarsh, supra note 92, at 1179.

  191. . Another thing to be avoided at settlement is giving judges perverse incentives to dispose of the cases before them regardless of the merits of the settlement. This would be a serious problem, though the informational challenges are by far the most-cited issue.

  192. . Though there are proposals to counter it. See Coffee, Accountability and Exit, supra note 14, at 423–25.

  193. . I consider this issue again in Section III.E.

  194. . See, e.g., Ken Binmore, Ariel Rubinstein & Asher Wolinsky, The Nash Bargaining Solution in Economic Modeling, 17 RAND J. Econ. 176 (1986).

  195. . Almendares, supra note 48, at 309–10.

  196. . Here, I mean a fraction of the expected value of their claims. It can be misleading to focus on the fraction of alleged damages the class members receive. The case might be quite weak, perhaps because of potential defenses. That would imply a far lower expected value than the damages alleged would indicate.

  197. . Erichson, supra note 33 , at 873. Specifically, Erichson lists ten “red flags.” They include: spurious injunctive relief, non-transferable or non-stackable coupons, overuse of cy pres remedies, burdensome claim procedures, reversions, excessively broad liability releases, expanded class definitions, class representative bonuses, revertible fee funds, and clear sailing agreements where defendants agree not to contest the attorneys’ fees. Id.

  198. . Id. at 863.

  199. . Id. at 880–81.

  200. . Id. at 885. Erichson breaks his ten red flags into three categories: efforts to make the class settlement appear larger than it actually is, features that expand the scope of the class and the settlement, and ones that discourage objections to the settlement or award. Id. at 873.

  201. . See supra notes 49–52.

  202. . E.g., Pearson v. NBTY, Inc., 772 F.3d 778, 783 (7th Cir. 2014); In re Payment Card Interchange Fee & Merch. Disc. Antitrust Litig., 991 F. Supp. 2d 437, 446 (E.D.N.Y. 2014).

  203. . Erichson, supra note 33 , at 870 (“[A] number of recent judicial decisions have looked at class settlements with a more jaundiced eye. . . . These courts have done so, moreover, with blunt commentary on the incentives of class counsel and the powerlessness of class members. It remains to be seen whether these recent cases foretell a broader shift in judicial attitudes about class settlements, but the time seems right.”); Baker, et al., supra note 47 , at 1445–46, 1448.

  204. . Wal-Mart Stores, Inc. v. Visa U.S.A., Inc., 396 F.3d 96, 122–23 (2d Cir. 2005). But see In re Payment Card Interchange Fee & Merch. Disc. Antitrust Litig., 2019 WL 6888488, at *20–21 (E.D.N.Y. Dec. 16, 2019) (“There is . . . commendable sentiment in favor of providing lawyers with sufficient incentive to bring common fund cases that serve the public interest. . . . [I]f attorneys’ fees are routinely set too low, it may create poor incentives to bringing large class action cases.”) (internal citations omitted).

  205. . See, e.g., Bruce L. Hay, Optimal Contingent Fees in A World of Settlement, 26 J. Legal Stud. 259 (1997); Silver & Baker, supra note 33; Leslie, supra note 33; Amchem Prods., Inc. v. Windsor, 521 U.S. 591 (1997).

  206. . E.g., Leslie, supra note 33, at 1087–88; Issacharoff, supra note 177, at 829–30; Rubenstein, supra note 177, at 1474; Baker et al., supra note 47.

  207. . Rubenstein, supra note 177, at 1444–45. Erichson’s red flags provide clearer, more substantial guidance to this end. See supra note 196.

  208. . And expertise is not costless to acquire, further underscoring how unrealistic that situation would be.

  209. . Ronald J. Gilson & Jeffrey N. Gordon, The Agency Costs of Agency Capitalism: Activist Investors and the Revaluation of Governance Rights, 113 Colum. L. Rev. 863, 874 (2013).

  210. . 15 U.S.C. § 78u-4(a)(3)(B)(iii); In re Initial Pub. Offering Sec. Litig., 214 F.R.D. 117, 120–21 (S.D.N.Y. 2002).

  211. . Burch, supra note 6, at 1116; see generally Weiss & Beckerman, supra note 62, at 2054; Erickson, supra note 188, 246–48.

  212. . Gilson & Gordon, supra note 208, at 865.

  213. . Dessein, supra note 29, at 825–27.

  214. . Gilson & Gordon, supra note 208, at 890.

  215. . The tax code and other statutes lead to a maximum of 10% investment for mutual funds. See id. at 890–91.

  216. . Id. at 885–86.

  217. . See generally Jennifer S. Taub, Able but Not Willing: The Failure of Mutual Fund Advisers to Advocate for Shareholders’ Rights, 34 J. Corp. L. 843 (2009).

  218. . “Put simply, some trial judges have gone to considerable lengths to nullify this power to select class counsel.” Coffee, Accountability and Exit, supra note 14, at 414.

  219. . James D. Cox, Randall S. Thomas & Lynn Bai, There Are Plaintiffs and . . . There Are Plaintiffs: An Empirical Analysis of Securities Class Action Settlements, 61 Vand. L. Rev. 355, 366 (2008).

  220. . Baker et al., supra note 47 , at 1380.

  221. . Burch, supra note 6, at 1111.

  222. . Id. at 1124.

  223. . See supra Section I.A.

  224. . Intermediaries can substantially affect the outcomes of an expertise-laden relationship. For examples, see Dessein, supra note 29; Sean Gailmard & Jeffrey A. Jenkins, Agency Problems, the 17th Amendment, and Representation in the Senate, 53 Am. J. Pol. Sci. 324 (2009); see also Cabraser & Issacharoff, supra note 10.

  225. . Exactly how much more informed the agent needs to be depends on Condition (2).

  226. . 166 F.3d 581, 583 (3d Cir. 1999).

  227. . Coffee, Accountability and Exit, supra note 14, at 408–09; Lazy Oil Co., 166 F.3d at 583.

  228. . Coffee, Accountability and Exit, supra note 14, at 408–09.

  229. . Baker et al., supra note 47, at 1381–82, 1432.

  230. . Dessein, supra note 29, at 821 (“[D]elegation is optimal when preferences between agent and principal are not too far apart.”); see also id. at 820–21 (Lemma 1, Proposition 2, Corollary 1).

Published by Nicholas Almendares

Associate Professor of Law, Maurer School of Law, Indiana University – Bloomington.

%d bloggers like this: