Over the last several years, a debate has flared up over universal injunctions, court orders that purport to benefit individuals across the nation, including vast numbers of people not party to the litigation from which the injunction issues. Critics on the left decry injunctions seeking to shut down executive action by the Obama Administration, while those on the right decry the mirror image injunctions against programs of the Trump Administration. To these actions, a third round of injunctions against immigration policies of the Biden Administration can now be added. All the while, a solution to these controversies remains hiding in plain sight in Federal Rule of Civil Procedure 23. Subdivision (b)(2) of the rule allows class actions when “final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” Subdivision (b)(1) goes into even greater detail in identifying when class actions should be certified because individual actions would prejudice class members or parties party opposing the class. Both subdivisions speak to the need for injunctions whose benefits go beyond the named parties to the litigation. Why don’t these provisions solve the problem of universal injunctions? If a class can be certified, then the injunction can reach beyond the named parties. If it cannot, then the injunction must be limited to the named parties.
The scholarly commentary on universal injunctions has recognized the relevance of these provisions, particularly Rule 23(b)(2), but then inexplicably pushed them to the margins of analysis. Failure to certify a class has been identified as a precondition of the problem but not as a solution to it. One author has simply excluded certified national classes from the analysis, while another has pronounced the terms of Rule 23 to be “formalistic.” A sophisticated history of universal injunctions in equity puzzles over the marginalization of the rule, but then moves on to formulate a different set of constraints on universal injunctions. As this article notes, “the need for and value of this class action provision is greatly diminished if plaintiffs can get the same relief in an individual suit that they can in a class action.” Still another article delves deeply into the history of equitable remedies but stops abruptly in the middle of the twentieth century, before the current version of Rule 23 began to take shape. If the rule could so easily answer the problem they have posed, perhaps they fear that it would be dismissed as merely procedural, rather than a matter of constitutional dimensions concerning the remedial power of the federal courts. Conversely, those who favor universal injunctions, and who are less concerned with limits upon them, might find the rule too restrictive because it requires certification of a national class to support a universal injunction.
This Essay argues that debates over these apparently binary choices are misconceived. Analysis under Rule 23 does not displace, but instead incorporates, fundamental issues of constitutional law and federal judicial power. So, too, it does not dictate an all-or-nothing answer to the question whether universal injunctions are ever appropriate. The rule frames the appropriate terms in which such questions can be answered, without the addition of tests and factors that would further restrain or enhance the availability of universal injunctions. This Essay advances this argument in three parts. The first analyzes the prominence of constitutional issues in the current debate. The second recounts the history of universal injunctions from Equity Rule 38 to Federal Rule 23. The third responds to concerns that certification of class actions is too “formalistic” and argues that it should be seen instead as a necessary precaution related to the merits of the plaintiffs’ claims and the risk of conflicts of interest within the proposed class. All of these considerations yield the simple conclusions that universal injunctions must be preceded by certification of similarly broad class and that there is no need to address the power of courts to issue this remedy if this prerequisite is not met.
I. The Current Debate and Constitutional Issues
The current controversy over universal injunctions gained national prominence with litigation over initiatives in both the Obama and the Trump Administrations. In Texas v. United States, the district court issued a preliminary injunction against the Obama policy of not enforcing the immigration laws against “Dreamers,” adults who had immigrated as children without documentation as children. A few years later, universal injunctions were issued against Trump’s executive orders creating the “travel ban,” restricting entry of aliens from identified countries with predominantly Muslim populations. These injunctions were reversed by the Supreme Court on the merits. Justice Thomas filed a concurring opinion also disapproving of the universal injunctions as an inappropriate remedy insofar as it extended to nonparties. He expressed doubt that such injunctions conformed to the “case or controversy” requirement of Article III.
He defined the problematic injunctions as those “that prohibit the Executive Branch from applying a law or policy against anyone—often called ‘universal’ or ‘nationwide’ injunctions.” These injunctions have the highest profile and therefore generate the most debate, but the same problems arise with respect to any injunction that extends broadly beyond the parties to the litigation. For instance, an injunction that protects enforcement of a state statute against anyone raises the same basic issues. As Howard Wasserman has pointed out, the problem is not with the geographic scope of the injunction but with those who can benefit from it. At the opposite extreme, the same problems do not arise with respect to injunctions for the benefit of existing parties that have incidental effects on nonparties. Orders to abate a nuisance are the standard example. A nuisance action by one landowner to enjoin a nearby factory from polluting the air works to benefit of adjoining landowners, but only because full relief to the actual plaintiff requires the factory to reduce pollution to all the landowners. This necessary incidental effect is a far cry from the wholesale extension of an injunction to reach nonparties all across the state or nation.
But is it fundamentally a constitutional problem? Anyone acquainted with the legal doctrine surrounding the “case or controversy” requirement under Article III knows that the definition of its scope and limits has proved elusive. It usually raises more questions than it answers. Yet most of the commentary on universal injunctions has sought a definitive resolution of their validity in constitutional law. Perhaps, given the vicissitudes of the decisions defining a “case or controversy,” this question is better avoided.
The leading article critical of universal injunctions, by Samuel Bray, has given this constitutional issue an historical dimension. He argued, and Justice Thomas agreed, that universal injunctions have become common only since the 1960s. According to Bray, universal injunctions might have made sense in England in the eighteenth century, with completely unified courts of equity under the control of a single chancellor, but they were not suited to a federal system of government with judicial power widely distributed among many state and federal judges. The disjunction between a unitary chancellor and federal judicial system became problematic in the 1960s as courts moved away from traditional rules of standing, according relief primarily to the parties before them, to an emphasis on declaring what the law is, based on “facial” challenges to statutes and other forms of government regulation independent of the facts of a particular case.
The key decision for Professor Bray is Frothingham v. Mellon, usually regarded as a case barring taxpayer standing under Article III, but one heavily dependent on limited equitable remedies. As the Court reasoned:
The party who invokes the power must be able to show not only that the statute is invalid but that he has sustained or is immediately in danger of sustaining some direct injury as the result of its enforcement, and not merely that he suffers in some indefinite way in common with people generally.
The implicit contrast is with Flast v. Cohen, which upheld taxpayer standing to challenge a statute that disbursed funds to religious schools allegedly in violation of the Establishment Clause. The plaintiffs sought to prevent any expenditure at all under the statute, relief that went far beyond any immediate benefit to the plaintiffs.
The leading response to Bray’s argument, and similar attempts to invoke historical limitations upon equitable remedies, is Mila Sohoni’s exhaustive examination of equity practice through the first half of the 20th century. She found a wide range of cases, in both the Supreme Court and the lower federal courts, that awarded or approved of universal injunctions. She took the position that, regardless of the policy questions raised by universal injunctions, the history of equitable remedies demonstrates that it is not a constitutional problem under Article III. She, however, cuts off her historical inquiry in the middle of the 20th century.
The policy problems that surround universal injunctions, to the extent they are independent of constitutional problems, revolve around the disproportionate power they allow a single district judge to exercise. A universal injunction that inures to the benefit of nonparties elevates the status of a single judge’s decision to a level comparable to a decision of the Supreme Court. It truncates the development of different lines of authority and forces the case onto the agenda of the Supreme Court. It also creates a risk of inconsistent decisions by different lower courts and invites the plaintiff to go forum shopping for a judge likely to be favorable. By contrast, where parties seek to consolidate multi-district litigation through a change in venue, the choice of the transferee district lies with the panel on multidistrict litigation. The district judge becomes an overseer of government at every level, regardless of the district or circuit boundaries that circumscribe the precedential effect of decisions by the lower federal courts.
Some argue for universal injunctions based on judicial review of administrative action under the Administrative Procedure Act (APA). It does create a kind of parity between the actions of the executive branch and the remedy available in the judicial branch. National or regional measures taken by the executive can be met by remedies of equal scope from the judiciary. A gap remains, however, between invalidating administrative action by depriving it of any force in the proceedings before the reviewing court and enjoining reliance upon it in any other proceedings. The APA authorizes courts to “set aside” agency action in the first sense by disregarding it, but it does not authorize injunctions to prevent reliance upon it generally. More generally, the limitation on federal jurisdiction to “cases and controversies” negates any implied principle of parity between executive and judicial action. Federal judges hand down their decisions only within the confines of a concrete case or controversy.
The preclusive effect of universal injunctions also contravenes the limits on non-mutual issue preclusion against the federal government, an extension of preclusion that the Supreme Court has explicitly prohibited. Indeed, to the extent the decision represents a binding judgment, it is binding only upon the defendant. The nonparties who benefit from the injunction are bound by an unfavorable decision only if they are in privity with the plaintiff. As will be discussed in Part II, a universal injunction revives the practice of “one-way intervention,” rejected in amendments to Federal Rule 23 in 1966.
These problems have elicited ad hoc responses, such as inquiring whether the injunction is necessary to secure equal treatment of nonparties; or limiting the geographical scope of the injunction to a federal judicial district or circuit; or requiring decisions from at least three federal circuits as evidence of settled law; or barring injunctions that resulted from forum shopping for a favorably inclined judge. Professor Sohoni tentatively suggests reinstating the practice of constituting three-judge district courts, allowing only those courts to issue universal injunctions with direct appeal to the Supreme Court. In making this suggestion, she neglects the complications that arose when three-judge district courts were widely available, raising questions about whether the court was properly convened and the effect of summary affirmances by the Supreme Court. This back-to-the-future approach accords with her reliance on past equity practice in issuing universal injunctions for the benefit of nonparties. But as Part II discusses in detail, much has happened since the Federal Rules displaced the Equity Rules in 1938 and altered the procedures that apply in equity to unite with actions at law. While Professor Sohoni finds no ironclad rule that equitable relief must be limited to the parties, she does not examine the question of who can be made parties under modern procedure.
II. From Rules of Equity to Rules of Civil Procedure
Before 1938, federal equity practice supported a range of representative suits. Equity Rule 38 codified this practice:
When the question is one of common or general interest to many persons constituting a class so numerous as to make it impracticable to bring them all before the court, one or more may sue or defend for the whole.
A few of these provisions have survived in Federal Rule 23, such as numerosity—“the class is so numerous that joinder of all members is impracticable”—and commonality—“there are questions of law or fact common to the class.” Yet most of the provisions in the current version of Rule 23 have no antecedent in Equity Rule 38, such as the provision for certification of a class action “[a]t an early practicable time.” In contrast to the short paragraph of the equity rule, the current version of the rule has eight subsections, most with subdivisions, many of which often are further subdivided. It is also a long way from the equity rule to the complexity of class action practice today.
The process of elaboration began with the drafting and approval of the original Rule 23. This process changed the language of Equity Rule 38 and made it into a separate subsection (a), adding subsection (b) on derivative actions, and subsection (c) on notice. The most controversial change was the addition of three subdivisions to subsection (a), spelling out commonality in terms of “the character of the right sought to be enforced.” As the terminology evolved, class actions could be “true,” where the right is “joint or common”; “hybrid,” where the right is “several” and involves “specific property”; and “spurious,” where the right is “several, and there is a common question of law or fact affecting the several rights and a common relief is sought.” Dividing class actions along these lines attracted criticism for relying on “outworn categories of rights,” not fitting the class actions recognized in current practice, and failing to “correspond to any essential differences in the handling or effect of class suits.” Zechariah Chafee, a prominent critic of the original Rule 23, would have reduced subsection (a) to a slight variant of Equity Rule 38, jettisoning the three subdivisions entirely.
The amendments to Rule 23 took the opposite course, driven by concerns over notice to class members in spurious class actions. The Advisory Committee in 1966 focused on the problems created by “one-way intervention” in those class actions, where the absent class members receive notice only after judgment has been entered. Although the committee hesitated to address preclusion, it recognized that it would be difficult to bind class members to a losing judgment in the absence of notice. But if they could still take advantage of a favorable judgment, the party opposing the class was left at a glaring tactical disadvantage. That party could not assert preclusion against absent class members based on a judgment unfavorable to the class, but absent class members could assert preclusion against the opposing party based on a favorable judgment. And instead of eliminating the subdivisions that identified different forms of class actions, the revisers preserved and altered them, moving them to a new subsection (b). That subsection contained the now familiar division of class actions by necessity under (b)(1), in which individual actions would work to the prejudice of the class or the party opposing the class; (b)(2) for class actions for injunctive or declaratory relief for the benefit of the class as a whole; and (b)(3) for class actions for damages and other forms of individual relief.
To eliminate the problem of one-way intervention in all class actions and to clarify the basis for preclusion by a class action judgment, the revised rule required an early decision on certification and a description of the class in any resulting judgment. The current provisions elaborate upon those adopted in 1966, but they take the same basic form, which is worth quoting at length:
(c) CERTIFICATION ORDER; NOTICE TO CLASS MEMBERS; JUDGMENT; ISSUES CLASSES; SUBCLASSES.
(1) Certification Order.
(A) Time to Issue. At an early practicable time after a person sues or is sued as a class representative, the court must determine by order whether to certify the action as a class action.
(B) Defining the Class; Appointing Class Counsel. An order that certifies a class action must define the class and the class claims, issues, or defenses, and must appoint class counsel under Rule 23(g). . . .
(3) Judgment. . . .
(A) for any class certified under Rule 23(b)(1) or (b)(2), include and describe those whom the court finds to be class members; and
(B) for any class certified under Rule 23(b)(3), include and specify or describe those to whom the Rule 23(c)(2) notice was directed, who have not requested exclusion, and whom the court finds to be class members.
These provisions rejected practice under the old rule, which closely resembled the current practice of issuing universal injunctions. An early determination of who was a party to the class action was required, as opposed to the late inclusion of nonparties by one-way intervention or at the remedy stage of the litigation. Those who might benefit from, and be bound by, the class action had to be made known early and had to be specified in the judgment; a general injunction issued at the end of the case would not do.
The crucial provision is in subdivision (c)(1)(A), stating that “the court must determine” whether the case proceeds as a class action “[a]t an early practicable time after a person sues or is sued as a class representative.” A plaintiff who seeks a universal injunction is suing “as a class representative.” This way of formulating the plaintiff’s role is not some recent invention. Chafee devotes two whole chapters of his book, Some Problems of Equity, to class actions under the heading of “Representative Suits.” He published this book in 1950 and it served as a resource for revisers of Rule 23 in 1966, with prominent citations in their advisory committee notes. They adopted his critique of the original rule, but not his proposal to return to a variation on the old equity rule.
Advocates of universal injunctions without class certification might point to the introductory phrase in Rule 23, which seems to contain permissive language that does not require a class action but allows one: “One or more members of a class may sue or be sued as representative parties on behalf of all members only if” the requirements in the rule are met. But this resort to an isolated phrase neglects the logical structure of the rule, which sets out the necessary and sufficient conditions for maintaining a class action. Replacing “may” with “must” would seemingly command named plaintiffs to commence litigation “as representative parties,” when they might prefer to bring individual actions and seek only individual relief. This conclusion is confirmed by the use of “may” in the introductory phrase in subdivision (b): “A class action may be maintained if Rule 23(a) is satisfied and if” the requirements of one of the subdivisions of (b) is satisfied.
The Supreme Court has made clear that there is nothing permissive about the duty to make a decision on certification imposed by subdivision (c)(1)(A). In East Texas Motor Freight System Inc. v. Rodriguez, one reason the Court gave for reversing certification of a class action was the plaintiffs’ failure to move for certification prior to trial. Even if the district court had a duty sua sponte to make the certification decision, the plaintiffs’ failure to do so established that they were not adequately representing the class. Transposed to the context of universal injunctions, plaintiffs cannot engage in artful pleading to refuse to seek class certification or to wait until the remedy stage to request a universal injunction. To delay in this manner is equivalent to delaying a request for class certification and it demonstrates that the case should be treated as an individual action with an individualized remedy.
To dispense with certification is equivalent to dispensing with all of the detailed requirements for maintaining a class action under Rule 23. Plaintiffs who seek a universal injunction without certification simply invite the court to ignore those requirements. The current version of the rule, and its predecessor in 1966, could not have been drafted with this option for wholesale evasion in mind. Here again, the Advisory Committee in 1966 deviated from Chafee’s comment on earlier equity practice: “The very identity of interests which made it easy to bring everybody in, also made it somewhat superfluous to do so.” Instead of going back to equity, the Advisory Committee elaborated at length on the provisions for class actions under Rule 23.
Critics of certification as a prerequisite to universal relief might appeal to the prohibition in the Rules Enabling Act that the “rules shall not abridge, enlarge or modify any substantive right.” If equity authorizes courts to issue universal injunctions, so the argument goes, then it does so as a matter of substantive law and the Federal Rules cannot infringe upon the plaintiffs’ right to obtain such an injunction. An argument along these lines, however, misconceives the relationship between substance and procedure. Certification under Rule 23(b)(2) presupposes “that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” If such equitable relief could not possibly be issued, certification under subdivision (b)(2) has to be denied. The rule no more limits the equitable power than Rule 8(a)(2), requiring “a short and plain statement of the claim showing that the pleader is entitled to relief,” limits the substantive claim asserted by the plaintiff. Both of these provisions, and many others in the Federal Rules, define the conditions under which substantive rights can be asserted in litigation. This is precisely the function of procedural rules: to regulate the process for enforcing substantive rights.
III. The Burdens and Benefits of Certification
If the argument for certifying a class is so compelling, how did courts come to dispense with it? The answer returns to the history of Rule 23 and the prior equity practice of issuing universal injunctions. That practice had to be reconciled with the division of class actions under the original version of Rule 23 into true, hybrid, and spurious. Absent a joint right shared by the entire class or litigation concerning a common question with regard to a particular piece of property, the first two categories would not apply at all. That left most litigation over universal injunctions in the category of spurious class actions.
Yet, certifying a spurious class action did not yield much in the way of benefits. The court still had to work its way through the three-part division of class actions in an overly conceptual framework. And if the class were certified as spurious, it still permitted one-way intervention by class members. They could take advantage of a judgment favorable to the class and avoid being bound by an unfavorable judgment. The same would be true of a universal injunction in the absence of certification. Nothing much seemed to be gained by working through the complications of the original Rule 23.
It comes as no surprise that courts avoided certification and the precedential force of the prior equity practice retained its strength. The amendments to Rule 23 in 1966 should have altered the balance between the rule and equity practice, but they did not. The momentum of established precedent has carried over in several circuits, imposing a requirement of “necessity” as a preliminary step in deciding whether to certify a class action. Only the Seventh Circuit has unequivocally rejected this approach. The changes made by the 1966 amendments, as discussed earlier, disapproved of one-way intervention, required early determination of certification, and created a special subdivision for class actions seeking declaratory and injunctive relief. Perhaps the uncertainty surrounding the new version of the rules led litigants and courts to avoid it. After several decades of practice under the rule, those concerns should have dissipated.
Some critics of required certification pronounce it to be “formalistic,” and most lower federal courts have dispensed with certification when it is not needed. This label calls attention to the burdens of maintaining a class action, in terms of certifying the class, assuring fairness to class members, selecting class counsel, managing the class action, and approving any settlement. These costs are borne by the parties, their attorneys, and the judge. Yet they yield benefits in terms of defining the class affected by the judgment, determining the scope of the judgment itself, and protecting against inadequate representation and collusive settlements. These benefits often go to the form that a certification order and a judgment take, but they are not limited to matters of form.
The foundational case on adequacy of representation under the Due Process Clause, Hansberry v. Lee, illustrates the need for careful attention to the certification and management of class actions. There, white homeowners tried to bind prospective Black homeowners and those who would sell to them by a judgment upholding a racially restrictive covenant. The interests of these two groups were directly adverse. The first group wanted segregation; the second wanted integration. Several of the requirements of Rule 23 are directed to the same end of protecting the class. Transposed to recent cases on universal injunctions, adequacy of representation appears to be a significant constraint on judicial power. In Texas v. United States, for instance, it is hard to believe that every state would have followed Texas in opposing the Obama Administration’s immigration policy with respect to “dreamers.” An injunction for the benefit of Texas, or perhaps limited geographically to Texas, has far more plausibility than one that applied nationwide. Just over half the states joined Texas in claiming that the federal immigration policy had a net adverse effect upon them. The other half did not want the injunction and a third opposed it,, yet it applied in their territory to the same extent as in Texas.
Class actions for injunctions under subdivision 23(b)(2) impose significantly lighter burdens than those, usually for damages, under subdivision 23(b)(3). The prerequisites for certification are simpler and less onerous under subdivision (b)(2), which requires only that “the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” By contrast, subdivision (b)(3) requires a detailed inquiry into whether the questions common to the class predominate over individual issues and that a class is superior to other means of adjudication. Likewise the provisions for notice to the class are simply discretionary under subdivision (b)(2), while they are mandatory for all identifiable class members under subdivision (b)(3). For these reasons, plaintiffs usually prefer certification under (b)(2) to certification under (b)(3), as in the well-known case of Wal-Mart Stores, Inc. v. Dukes. Given the more lenient standards for certification under (b)(2), there is no need to give plaintiffs the further option of not seeking certification at all.
Some decisions have dispensed with the need for certification for reasons entirely independent of the requirements of Rule 23. They have relied on the agreement by the party opposing the class to be bound by a universal injunction. This concession goes to the remedy stage of the litigation, to the acceptance that any relief awarded to the plaintiff extends to everyone similarly situated. Judge Friendly took this approach in his influential opinion in Galvan v. Levine, a case challenging state limits on unemployment benefits. The state had already withdrawn its policy of denying benefits to workers from Puerto Rico before judgment was entered. In that posture of the case, Judge Friendly concluded, certification of a class was “largely a formality, at least for the plaintiffs.” But so was entry of a statewide injunction, since the state no longer contested eligibility for unemployment benefits to individuals in the plaintiffs’ position. By the time the case came before Judge Friendly on appeal, the state had effectively waived any objection to the scope of the injunction, making it difficult to reverse the district court’s judgment on this ground.
Nevertheless, even in cases where the defendant does not contest the scope of the relief requested, certification has its uses. It prevents the party opposing the class from changing positions, for instance, if a new government comes into office. There is no apparent reason to postpone resolution of such issues to the enforcement stage of the case, when they can be resolved at the outset. They would also be easily resolved if the party opposing the class agrees that the case should proceed as a class action. The decision to certify the class and issues of managing the class action become much easier to resolve once the parties agree on the scope of the action—if, indeed, these issues are contested at all.
Requiring certification forces an early decision on who can benefit from the litigation. The requirement of a decision “[a]t an early practicable time,” comes long before the remedy stage of litigation, in which the presumption that the scope of the remedy should match the scope of the wrong might exert undue influence. Rule 23 creates a more systematic structure for determining the scope of the injunction than the various ad hoc factors that have been proposed as limits: geographical restrictions based on the limits of the federal district or federal circuit; precedent in at least three circuits that supports the injunction; assuring equal treatment of all those who might benefit from the injunction. Other proposals look to expanding the jurisdiction of three-judge district courts to handle universal injunctions, with direct review by the Supreme Court; and to relying upon judicial review under the Administrative Procedure Act, so that the scope of any court order matches the scope of executive action.
The terms of Rule 23 retain enough flexibility to accommodate these considerations, assuming they are relevant, or if Congress acts to amend the relevant statutes. The rule itself does not prevent certification of class actions on a national, regional, or state-wide scale. The Supreme Court, for instance, upheld a nationwide class action in Califano v. Yamasaki. Whether other cases can be certified on such a large scale depends on whether the requirements of the rule are met. As the Supreme Court has emphasized, “careful attention to the requirements of Fed. Rule Civ Proc. 23 remains nonetheless indispensable.” The availability of universal injunctions need not be addressed as a question of all or nothing. Indeed, even if a class action is certified, the court still must address the question of appropriate relief, which might, or might not, result in issuance of a broad injunction. As an initial matter, however, the scope of an injunction must be addressed for what it is: a question of joinder of parties.
The historical argument for universal injunction has, paradoxically, a curiously anachronistic quality. A detailed look at the historical record establishes the practice of issuing such injunction as a matter of equity. A detailed look at the Federal Rules of Civil Procedure makes this practice subject to procedural rules on joinder. Perhaps after 1938, but certainly after 1966, Rule 23 changed the procedural landscape surrounding equity practice, no matter how much it previously favored universal injunctions without joinder. It is time to follow the rule.