Opening Up Intervention to Check Agency Costs

Introduction

Professor Monica Haymond’s Intervention and Universal Remedies is a landmark. Until now, most debate over nationwide injunctions has been about the merits, not the process. Haymond spots the problem of who gets to take part, and she develops a taxonomy of issues and participants. She balances the obvious benefits of intervention by cataloguing the costs. And she does the hard work of building and analyzing an original dataset of nationwide injunctions, shedding empirical light on a poorly illumined, anecdote-driven debate.

More generally, Haymond invites scholars to focus on the distinctive ways that public law litigation plays out in practice. In this Essay, I take up her challenge. In Part I, I flag three common assumptions about structural-reform litigation, the first and third of which are beyond her scope. Most scholars take for granted that structural-reform litigation is valid and here to stay; that federal and state governments represent the public interest well; and that judges already have the tools and perspective to ensure fair results and need only modest nudges or tweaks to focus on the adequacy of representation and the appropriateness of universal remedies.

Part II then questions each of these common assumptions. I reprise the famous debate between Professors Lon Fuller and Abram Chayes. Though Chayes’s polycentric model of structural-reform litigation has won the day among academics, I have some kind words for Fuller’s model of bipolar dispute resolution. In practice, our rules are built to resolve parties’ disputes, and perhaps courts should do more to limit ourselves to such traditional Article III cases or controversies. But when we do try to reform structures, we need different approaches. Here, I part ways with Haymond, whose suggested norms reinforce the traditional norm of party control. That makes far more sense when resolving private disputes than when changing government behavior prospectively.

Consent decrees, settlements, and even broad precedents raise special concerns. The executive branch represents the public interest imperfectly. It may not fight hard, or may even acquiesce, in agreeing to prospective relief. That is a particular danger because there is no strong norm of defending all statutes in court. Executive acquiescence can erode legislative authority by purposefully abandoning some statutes and regulations. It can circumvent the Administrative Procedure Act’s processes for making or repealing regulations, as well as the ordinary hurdles for legislation. It can undermine the adversarial process, keeping courts from hearing the best defenses available. It can reward forum-shopping, letting friendly litigants achieve nationwide victories that might not have lasted if contested. And it can bind future administrations that take very different views. In other words, any one executive-branch administration can impair the judicial process, circumvent the legislative and rulemaking processes, and entrench its own views against future administrations.

To my mind, intervention is an important check on these dangers. So Part III argues for making intervention much more freely available in structural reform cases, limited only as needed to prevent unreasonable delay and complexity. Some kinds of intervention, such as those intended to cure lack of standing or to tack on additional claims, are not needed to combat these dangers. But others are.

In short, I’m far more skeptical of governmental control of public law litigation than Haymond is. Even so, I commend her for raising these thought-provoking questions and prompting overdue debate about party control and the limits of the adversarial process.

I. Three Issues Lurking Beneath Structural-Reform Litigation

Structural-reform litigation, like public law litigation generally, is predicated on three latent assumptions. If they hold true, the adversarial process should work well. Proper representation plus proper incentives should ensure that each side fights vigorously, and judges should know enough to stay in their lane. Any problems should be modest, requiring only tweaks to ensure that each side represents its interests well and that judges have better guidance on when universal remedies are proper.

Most of these assumptions are beyond the scope of Haymond’s article, but they are foundational. It’s worth bringing them to light and critically examining how well they hold up.

First is the assumption that structural-reform litigation is valid and here to stay. Structural reform calls for distinctive remedies and a distinctive remedial role for judges. Plaintiffs sue for relief not just for themselves, but also for all those similarly situated. They seek prospective injunctions, not just damages or declaratory judgments. Those injunctions often go beyond prohibitions to mandate affirmative government conduct, like busing. They ask judges to (re)fashion government behavior or programs and monitor ongoing compliance with court orders, on pain of contempt. That may force governments to spend substantial money, potentially squeezing other priorities.

Second is the assumption that it is reasonable to rely (as Haymond does) on the traditional principle of party control. That principle presumes that, in litigating these cases, federal and state governments represent the public interest well. Government defendants, we assume, will normally treat litigation as zero-sum adversarial combat just as private defendants normally do. They will supposedly raise colorable defenses, litigate thoroughly, and resist remedies that require substantial public spending or compromise other competing priorities. Though the remedies are different, the adversarial process should still work well because adversarial incentives should still motivate defendants.

Third is the assumption that basic principles of a healthy adversarial judicial process still apply. On this assumption, judges have the tools to ensure fair results, skeptically testing each side’s factual and legal proof. They will supposedly limit their relief to the legal merits shown and the plaintiffs and those similarly situated who face harm. They will resist the temptation to legislate from the bench and engage in social engineering. They will keep the separation of powers and fiscal constraints firmly in mind, granting only the minimum relief demanded by law.

II. The Shaky Foundations of Structural-Reform Litigation

These assumptions, though, are questionable. To examine them critically, let’s look back to the famous scholarly debate between Lon Fuller and Abram Chayes. Though today’s scholars occasionally name-drop these giants, they rarely reflect on how these issues remain live.

In brief, Lon Fuller famously understood litigation as well-suited for bipolar dispute resolution. The litigants are two sides, usually unitary private parties on either side. Their interests are diametrically opposed. The litigation is retrospective. It’s about historical alleged wrongs and what damages the defendant owes the plaintiff for them. So it’s zero-sum: every dollar for the plaintiff is a dollar out of the defendant’s pocket.1

On this view, the lawsuit is symmetrical because both plaintiff and defendant are present in court: a win for either side or a settlement will conclude the dispute, precluding relitigation by either party but not rewarding or estopping nonparties, apart from any precedential effect. The damages remedy flows from the right allegedly infringed, limited to making the plaintiff whole or giving expectation damages. The lawsuit is a discrete episode, ending once the court gives judgment and the plaintiff collects on it. And the parties initiate, frame, and control their own dispute, developing the facts and leaving the court to react as a neutral arbiter.2 (I’d add one further point: that’s what most ordinary lawsuits looked like in the eighteenth century, so it may be the case or controversy that Article III presupposes.3 )

Chayes contrasted dispute resolution with structural-reform litigation. In structural-reform litigation, the lawsuit is shaped by the court, the parties, and the lawyers. The dispute is over not private rights but public policy. The parties need not be bilateral but may be “sprawling and amorphous.”4 Because the litigation is over public law, the defendant is usually a governmental body. The facts found are not just historical, but also legislative and predictive.

This forward-looking factfinding leads to forward-looking relief: not (just) damages or a declaration, but an injunction. That injunction often requires the defendant not just to avoid violating rights, but to take affirmative steps. It typically costs the defendant money to comply. It may be negotiated between the parties as part of a settlement or consent decree. Because the affirmative relief will go on for a while, the court must keep supervising and administering it. The judge cannot remain a neutral, passive arbiter, but must help organize and shape the litigation and remedy. Given the many parties, interests, and tradeoffs required, Fuller dubbed such litigation “polycentric,” as opposed to bipolar.5

Scholars tend to assume that the debate is over and that Chayes won.6 And as a descriptive matter, some such litigation is here to stay. But normatively, this practice sits uneasily alongside Article III’s restriction of the judicial power to cases and controversies. In light of the originalist turn of the last few decades, it’s worth asking whether sprawling structural-reform litigation fits with the judicial role authorized by the Framers—and, if not, what we can do to translate their principles into the twenty-first century.

Specifically, the framework of our rules is still built around dispute resolution. And that makes some sense: numerically, most lawsuits still involve resolving bipolar disputes. We treat party control as central, giving each litigant his day in court. We traditionally limit remedies to the parties. We preclude relitigation. And we make it hard for officious intermeddlers to stick their noses into what is fundamentally the parties’ spat.

But structural-reform lawsuits call for different approaches. Though numerically, they are only a small fraction of all cases, they have outsized importance and deserve special treatment. Their remedies typically go beyond damages or prohibitory injunctions to affirmative injunctions, which reach far beyond mandamus and were not traditionally available in equity. Party control is less important because the court’s remedy may benefit those similarly situated to the plaintiffs and may affect many others. For instance, in a criminal justice structural-reform lawsuit, a court may order a town to upgrade its local jail. If the town must commit a huge fraction of its budget over the next five years to comply with the court order, it may not have enough money left over for its schools and hospitals.

Public law litigation also precludes the government asymmetrically.7 A hundred immigration plaintiffs may sue the federal government in a hundred different courthouses, all challenging the same immigration policy. Perhaps the first ninety will lose, but their losses will not preclude other litigants. Once the ninety-first plaintiff wins a nationwide injunction, vacatur, or settlement, the government must stop and remedy the challenged practice. The asymmetry of preclusion and remedies stacks the deck in favor of challengers.

And then there are agency costs. Government officials may not have enough incentives to defend the law vigorously. That can be a problem with damages suits, but the harm to the fisc is limited to each case. It’s a bigger problem with prospective relief that reaches beyond the parties. A sitting executive administration may not like a law passed by Congress or a regulation enacted by a prior administration. Yet it may not have enough support to repeal the law through Congress or the regulation through rulemaking. But as long as the law or regulation is on the books, the administration faces some pressure to apply it.

A lawsuit may be the easy way out. If a friendly interest group sues, the administration may settle on favorable terms. (That is especially true if the group shops for a hospitable forum.) The administration may agree not to enforce the challenged law or regulation or to spend money not appropriated by Congress. The danger of collusion is obvious, which is why such cases are sometimes called “sue and settle.”8 They circumvent the separation of powers and other structural checks that make it hard to pass new laws or regulations, hard to repeal existing ones, and hard to get money appropriated for new policy initiatives. And by doing so, one administration may entrench a policy that will bind future, more hostile administrations.

The EPA has been criticized for suing and settling—even by a sitting EPA Administrator himself. One of many examples is a 2008 dispute about air-quality standards. Medical and environmental groups sued the EPA, alleging that it had failed to issue emissions standards as required by the Clean Air Act.9 In response, the EPA entered into a consent decree agreeing to issue regulations.10 A utility industry group had intervened in the litigation, but the EPA neither notified nor consulted with it before agreeing to that settlement.11 The intervenor-defendant objected, but the district court approved the consent decree. Though the intervenor repeatedly moved for equitable relief from the consent decree, the district court did not respond to them.12

The scope of the problem used to be more limited. Until about a decade ago, there was still a strong norm that the executive branch would vigorously defend any law for which it had a colorable argument.13 Occasionally, an Attorney General or Solicitor General would confess error; mostly, they did so only when they thought they had to as officers of the court. But more than a decade ago, they started declining to enforce laws with which they disagreed, like the Defense of Marriage Act and equivalent state laws. Whatever one thinks of the merits of those laws, this intentional inaction unraveled the adversarial premise on which litigation rested. And state attorneys general, who are often elected and politically ambitious, realized that they could claim credit by opposing laws that they disliked.

As a result, courts can no longer count on the executive to vigorously defend the law or the fisc. These settlements may be undemocratic, tying successors’ hands and preventing the repeal of settlements that voters dislike. The same concerns surround the government’s discretion to appeal.

An agency-cost perspective makes me much more skeptical of trusting the government’s vigor and incentives. When aggrieved plaintiffs just seek damages, as in most cases, the government may do a tolerably good job of resisting liability. But when they seek injunctions to change the law in practice going forward, judges should worry about how vigorously the government will resist them to uphold the law. It is the President’s job to take care that the law be faithfully executed, but that does not absolve judges from checking their work.

Though these problems are more obvious in cases seeking universal remedies, they extend beyond those cases. Ordinary settlements and consent decrees can raise many of the same problems. Even broad precedential rulings can have some of these effects. But the stakes are higher, and the agency costs more costly, when the relief is so vast.

III. Strengthening Checks on Structural-Reform Suits

We can’t (and probably shouldn’t) wave a magic wand and wish we were back in the eighteenth century, with almost exclusively bipolar litigation. Structural-reform litigation is here to stay. But we can recognize that it is exceptional, fraught with danger, and needs special safeguards. To my mind, it calls for nearly the opposite of Haymond’s reforms. She thinks that intervenors currently play outsized roles in controversial cases, to the detriment of judicial processes. Stressing the traditional principle of party control, Haymond fears that intervenors can be officious intermeddlers, mucking up the government’s business. She offers two paths forward to address this: changing the doctrine or changing the Federal Rules to limit intervention.14 I disagree with her premise.

I think that excluding intervenors creates more of a threat to the adversarial process than including them. The executive branch’s interests sometimes diverge from the public interest, especially on contentious issues. That calls for measures to protect the vigor of the adversarial process. The danger of structural-reform litigation is that the executive branch may not vigorously defend the existing structure. So we need more intervention, not less.

I don’t think that every public law suit needs intervenors. Some kinds of intervention are not needed to limit injunctive remedies. Some intervenors, Haymond notes, seek to cure a plaintiff’s dubious standing. But if there is no standing, there will be no injunction. Other intervenors, she notes, seek to add claims. Again, though, omitting a claim will not broaden the relief granted; if anything, it will narrow relief. In neither case do courts need intervention to ensure adversarial testing. Other suits can finish the job.

But some public law suits do need intervenors—especially highly charged ones with political valence. I agree with Haymond that judges could alter intervention either through case law or by amending the Federal Rules of Civil Procedure. One way or another, judges should create a presumption in favor of mandatory intervention for intervenors seeking to raise colorable arguments or defenses the executive branch has abandoned to score political points. Judges should limit that intervention only as needed to cap delay and complexity. For instance, when a bevy of intervenors comes knocking, the court might let in the one who is best lawyered and best positioned to raise the full range of interests and defenses. That would protect judicial efficiency, which Haymond notes as a concern, without compromising the integrity of the judicial process.

This proposed limit on judges’ discretion is justified to ensure better adversarial testing. The reason intervention should be mandatory, not just permissive, is so that the intervenor (unlike an amicus) can raise proper defenses, check settlements, and appeal. I tend to agree with Haymond that too much judicial discretion in this area lets universal remedies run amok; a presumption would curb them.

The government could rebut the intervention presumption as long as it contested the case vigorously, raised all substantial defenses, and resisted prospective relief. Conversely, if the government fails to defend a law robustly, judges should be leery of granting broad relief.

Introducing intervenors at the remedy stage, as Haymond proposes, is not enough. She views intervention at that stage as sufficient because her main concern for intervenors is to avoid depriving them of their day in court, and she worries about protecting judicial efficiency earlier on. This proposal does not address my concern, because the government can concede liability for a lawful practice well before the remedy stage. True, courts may decline to impose a nationwide injunction at the remedy stage, limiting the damage that I’m most concerned about. But this still creates precedent that can be used in similar cases in the future. I’ve seen too many poorly lawyered cases where neither side raises the best arguments. With that little help, judges make mistakes. And when the remedy is as wide-reaching and enduring as a nationwide injunction, the repercussions can be enormous, forcing the Supreme Court to step in before it has the benefit of lower-court percolation. Better to allow intervention early on, to balance the process, than to force the Court’s hand too early.

Intervention is a front-end solution to this problem. But back-end solutions are possible too. Other litigants can come forward later if they can show that they were not adequately represented. That works, though, only if those third parties have standing. And later administrations might be able to move to reopen cases under Rule 60(b)(5). But there are two difficulties. First, the standard for reopening is high. To address this concern, courts should recognize that a later administration’s doubts about the continuing wisdom of a settlement should be enough. Second, settlements are contracts that bind later administrations. The solution would be sunset provisions, recognized either by precedent or legislation, letting administrations periodically revisit whether prospective relief is still warranted.

In sum, structural-reform litigation falls outside the traditional bipolar model. Our court system is not set up for it because it lets the government change national policy by settling a single lawsuit with a single plaintiff. Third-party intervenors need to remain as important checks on the government’s duty to defend the law in court.

Though I see things somewhat differently from Professor Haymond, that different perspective should not obscure the bigger picture. Her article is a welcome invitation to look at how nationwide injunctions play out in the real world and to redesign procedures accordingly. I applaud her empirical and normative contributions on this crucial topic and look forward to learning much more from her follow-up work.

*  *  *

Stephanos Bibas is a judge on the U.S. Court of Appeals for the Third Circuit.

• 1See generally Lon L. Fuller, The Forms and Limits of Adjudication, 92 Harv. L. Rev. 353, 393–405 (1978) (asserting that polycentric problems are ill-suited for traditional adjudication).
• 2See id. at 381–405; Abram Chayes, The Role of the Judge in Public Law Litigation, 89 Harv. L. Rev. 1281, 1282–83 (1976).
• 3See Samuel L. Bray, Multiple Chancellors: Reforming the National Injunction, 131 Harv. L. Rev. 417, 427 (2017). But see generally James E. Pfander, Cases Without Controversies: Uncontested Adjudication in Article III Courts (2021) (surveying exceptions such as bankruptcy, naturalization, and some admiralty and equity proceedings). Unlike public law injunctions, most of Pfander’s exceptions affect only those who were parties or could have been parties to the suit.
• 4Chayes, supra note 2, at 1302.
• 5Fuller, supra note 1, at 394 (citing Michael Polanyi, The Logic of Liberty: Reflections and Rejoinders 171 (1951)).
• 6See, e.g., Doug Rendleman, Preserving the Nationwide Government Injunction to Stop Illegal Executive Branch Activity, 91 U. Colo. L. Rev. 887, 939–40 (2020) (rejecting out of hand that anyone still accepts Fuller’s view).
• 7See Zachary D. Clopton, National Injunctions and Preclusion, 118 Mich. L. Rev. 1, 3 (2019).
• 8See, e.g., Michael W. McConnell, Why Hold Elections? Using Consent Decrees to Insulate Policies from Political Change, 1987 U. Chi. Legal F. 295; Michael T. Morley, Consent of the Governed or Consent of the Government? The Problems with Consent Decrees in Government-Defendant Cases, 16 U. Pa. J. Const. L. 637 (2014).
• 9Complaint at 2, Am. Nurses Ass’n v. Jackson, No. 1:08-CV-02198 (D.D.C. Dec. 18, 2008).
• 10Consent Decree, Am. Nurses Ass’n v. Jackson, No. 1:08-CV-02198 (D.D.C. Apr. 15, 2010).
• 11Am. Nurses Ass’n v. Jackson, 2010 WL 1506913, at *2 (D.D.C. Apr. 15, 2010).
• 12Second Motion for Relief from Judgement by Utility Air Regulatory Group, Am. Nurses Ass’n v. Jackson, No. 1:08-CV-02198 (D.D.C. Nov. 7, 2011).
• 13Daniel J. Meltzer, Executive Defense of Congressional Acts, 61 Duke L.J. 1183, 1198 (2012).
• 14Monica Haymond, Intervention and Universal Remedies, 91 U. Chi. L. Rev. 1859, 1931–40 (2024).