This is the fifth and last in in a series of posts summarizing an article titled Remand Without Vacatur and the Ab Initio Invalidity of Unlawful Regulations in Administrative Law, which is forthcoming in the BYU Law Review. The current draft is available on SSRN.
These posts, and the article on which they are based, criticize the doctrine of remand without vacatur. This last post discusses some implications of those criticisms for the debate about universal relief against the government – remedies in suits against the government that provide relief to everyone affected by agency action, not just the parties.
First, the ab initio invalidity of unlawful regulations shows that one leading argument in favor of universal relief against regulations is unsound. According to that argument, universal relief against unlawful regulations comes within the generally accepted principle that benefits to non-parties are permissible when they are indivisible from relief to parties. Relief is indivisible when vindicating the rights of parties inevitably entails benefiting non-parties. For example, an injunction against making excessive noise may provide benefits to neighbors who are not plaintiffs that cannot be separated from relief to plaintiffs.
Vacatur of unlawful regulations, the argument goes, is necessary to relieve the parties from their obligations under the regulation, and vacatur provides benefits to non-parties that are indivisible from the benefits to parties. Vacating a regulation renders it wholly inoperative, the way rescission by the agency would. A regulation that has been made wholly inoperative no longer binds anyone, party or not. Giving the parties relief by eliminating the rule’s binding force inevitably and permissibly benefits everyone subject to the rule.
Because unlawful regulations are void ab initio, however, the reviewing court does not need to bring about invalidity in order to give the plaintiff a remedy. Rather, party-specific relief, like injunctions against enforcement against the plaintiff, or a declaration that the plaintiff has no duty to comply, are enough. Those party-specific remedies reflect the conclusion that the regulation was void when adopted, a situation the courts recognize but do not bring about. Vacatur is not needed to relieve the parties from a regulation that is already invalid. Injunctions against enforcement and declaratory judgments, as opposed to vacatur of a rule as such, are divisible relief. A court can enjoin proceedings against A without enjoining proceedings against B. A court can also declare A’s legal relations with the government without declaring B’s.
Second, recognizing ab initio invalidity reveals serious questions concerning the supposed remedy of vacatur, which operates on regulations as such and not only on parties. Ab initio invalidity might seem to rule out vacatur. How can a court invalidate an already-invalid regulation? Despite that difficulty, invalidation of an invalid regulation is in a sense possible. Whether that effect can be brought about by a court, however, is not clear.
Strange as it may seem, an invalid regulation can be invalidated. That odd possibility arises because regulations, like statutes, have more than one necessary condition for validity. To be binding, a statute must have been adopted pursuant to the Constitution’s law-making process, its content must be consistent with the Constitution, and it must not have been repealed. Repeal deprives a statute of a necessary condition for validity that is independent of substantive constitutionality. For that reason, when Congress repeals a statutory provision that is wholly unconstitutional, a meaningful legal event takes place. In similar fashion, a duly-promulgated regulation that is wholly invalid because not authorized by statute, for example, can in a meaningful sense be repealed. Congress can pass new legislation that refers to the regulation and eliminates its legal force, and the agency can rescind it. A regulation, like a statute, can be deprived of one necessary condition for validity even if it already lacks another.
Both legislative and executive power can operate on rules as such, whereas the core operation of judicial power is on parties to lawsuits. Whether judicial power can operate on regulations the way legislative and executive power can is not clear. Courts cannot repeal statutes, so maybe they cannot rescind rules. A court can order an agency to rescind a rule, but an injunction directing rescission is an order to a party, not an act of law-making. Whether Congress has authorized the federal courts to deprive regulations of legal force is also not clear. In describing proceedings in which judicial review takes place, section 703 of the APA mentions suits for injunctions and declaratory judgments, which operate on parties. Section 703 does not list proceedings for vacatur. That provision does mention special statutory review proceedings; whether any of them contemplates that a reviewing court will directly change the content of the law depends on the meaning of those provisions.
The analogy between court-court and court-agency review obscures the question whether courts can change the content of regulatory law and the question whether Congress has empowered them to do so. Agencies and courts exercise different kinds of power, whereas appellate courts exercise judicial power just as lower courts do. An appellate court can displace the legal effects of a lower-court order by vacating it, because lower courts and appellate courts work together in deciding cases through the exercise of judicial power.
Agencies are not courts. The regulations agencies produce are more like statutes than like lower-court judgments, in that unlawful agency regulations are in general invalid, just as unconstitutional statutory rules are invalid. Courts should not tell regulated parties that they are obliged to comply with unlawful regulations while agencies seek to repair the regulations’ flaws. The analogy between agencies and lower courts on which remand without vacatur rests is no more than an analogy, and often is misleading.