Seila Law and the Law of Judicial Review

John Harrison1John Harrison is the James Madison Distinguished Professor of Law and Thomas F. Bergin Teaching Professor at the University of Virginia. Professor Caleb Nelson provided helpful comments.

* * *

A part of the Seila Law and the Roberts Court series.

The Court in Seila Law LLC v. Consumer Financial Protection Bureau did not hold that the restriction on presidential removal of the Consumer Financial Protection Bureau (CFPB) director was unconstitutional. At least, it did not do so according to standard principles of stare decisis and the orthodox account of the law of judicial review—the legal principles under which courts implement the hierarchical superiority of the Constitution to all other legal norms. The justices’ belief that they held the removal restriction unconstitutional reflects a departure from the judicial-review orthodoxy. 

That belief about Seila Law‘s holding rests on the assumption that the Court’s decision caused the removal restriction to become inoperative, and that previously it had been legally effective. The Court made it invalid. But according to the standard explanation of judicial review, courts find invalidity, and do not make it. Invalidity figures in courts’ reasoning, as set out in their opinions. On the basis of that reasoning, courts give judgments that concern the legal relations of the parties before them and do not change the law in the abstract. The justices’ tendency to think that courts bring about invalidity reflects a common mistake: taking too literally the metaphor of judicial lawmaking. 

That misstep is especially easy to take in constitutional law because of another common misunderstanding: confusing rules about power with rules about duty. In central instances of judicial review, like Seila Law, the Constitution constrains the powers of legislatures. It provides criteria for the validity of statutory rules; rules that fail those criteria are inoperative because the Constitution makes them so. In such cases, the Constitution does not impose duties on legislatures, executive officials, or courts—duties they may fail to perform as a tortfeasor might. Rules about power thus differ from rules that impose duties. Failures to comply with those different kinds of rules have different consequences. Failure to comply with a rule about power automatically leads to invalidity, whereas violation of a duty calls on courts to remedy the violation with a decree that will bring the world into better conformity with the right that was violated. Confusing those categories leads judges to treat unconstitutional legal rules as if they were torts that impose harm and call for a remedy of invalidation that will alleviate the harm.

I. What Was the Holding of Seila Law?

The Seila Law case arose out of a Civil Investigative Demand (CID) that the CFPB issued to Seila Law. When Seila refused to comply, the agency sued to enforce the CID. Seila argued that it did not have to comply because the statutory provision insulating the CFPB’s director from removal by the president was unconstitutional. The district court rejected the constitutional argument and directed Seila to comply. The Ninth Circuit affirmed. The Supreme Court vacated that judgment and remanded with instructions to inquire into the effects of subsequent ratifications of the CFPB’s earlier actions.

Chief Justice John Roberts, writing for himself, Justice Samuel Alito, and Justice Brett Kavanaugh concluded that the removal restriction was unconstitutional but that the CFPB’s power to issue CIDs was severable from it. Because the power was severable, Chief Justice Roberts could not find that the CID was not binding and simply reverse the Ninth Circuit. Later ratification, he concluded, may have made the CID enforceable. The chief justice found that ratification should be addressed in the first instance by the lower court on remand.

Justice Clarence Thomas, joined by Justice Neil Gorsuch, agreed with the part of Chief Justice Roberts’s opinion that found the removal restriction unconstitutional. He found that the CID was ineffective regardless of whether the power to issue it was severable from the removal restriction, and that in any event the two were not severable. The appropriate disposition, he concluded, was to reverse and remand the case with instructions that the district court should deny the CFPB’s petition to enforce the CID.

Justice Elena Kagan, joined by the remaining three members of the Court, thought the removal restriction constitutional. She also said that if the restriction were unconstitutional, it would be severable from the power to issue CIDs. In her view, the Court should have affirmed.

According to the Court’s standard principles governing the precedential force of the justices’ opinions, the conclusion that the removal restriction is unconstitutional does not bind either the lower courts or the Court in the future. Binding precedent is established when “a single rationale explaining the result” of a Supreme Court case “enjoys the support of five Justices.”2In calling the Marks formula the Court’s standard principle, I do not mean to suggest that the formula perfectly describes the principles the federal courts follow in deciding what force to give Supreme Court opinions.  Many lower-court judges no doubt would simply count the stated positions of justices in Seila Law, find five who believe that the removal restriction is unconstitutional, and take that as the holding of the case. Marks is often cited and followed, however, and for good reason: it locates the binding force of opinions in their connection to the result, and thereby respects the principle that the secondary function of setting precedent arises out of the courts’ primary function of deciding cases. The result in Seila Law was a remand to decide whether the CFPB’s initial decision to issue a CID had later been validly ratified. No rationale explaining that result had the support of five justices. No group of five justices even agreed that a remand for further proceedings was the correct disposition. Two justices would have finally disposed of the case by reversal and remand with instructions to dismiss, and four would have affirmed. 

How the Court was able to decide the case at all is thus a puzzle.3The Court has a practice for deciding cases in which no majority supports any disposition. See, e.g., United States v. Screws, 325 U.S. 91, 134 (1945) (Justice Rutledge voting to remand for new trial, although he believed the judgment should have been affirmed, because otherwise the Court would have no majority for a disposition). No justice in Seila Law explicitly cast such a vote. A more fundamental and related puzzle is why the justices apparently believe that the Court held the removal restriction to be unconstitutional. The parts of Chief Justice Roberts’s opinion taking that position are styled the opinion of the Court, the term used to identify the parts that have majority support and are binding precedent.             

A pointer to the explanation appears in the chief justice’s discussion of ratification. The CID was issued when Richard Cordray was director of the CFPB. After his resignation, Mick Mulvaney for a time was acting director. When the case was decided, Kathleen Kraninger had been appointed director. Discussing severability and ratification, Chief Justice Roberts referred to “Director Kraninger’s activities in litigating the case . . . before becoming removable at will herself in light of our decision.”

Chief Justice Roberts, and most of his colleagues, believe that the Court in Seila Law engaged in a jural act that caused the legal rules governing removal to change. Before the case was decided, CFPB directors were not freely removable. Now they are. The Court did something that had an effect like that of an act of Congress amending the statute. That understanding matches the chief justice’s statement at the beginning of the part of his opinion about severability and ratification, “[W]e now turn to the appropriate remedy.” Remedies change legal relations, for example by creating an obligation to pay damages that did not exist before. The Court was going to issue an order causing the statute to become ineffective to some extent, and the question was, to what extent.

That understanding explains Justice Kagan’s characterization of her own position. Her opinion says that she concurs in the judgment with respect to severability and dissents in part. What judgment with respect to severability was she talking about, and what was she dissenting from? Justices do not have to dissent from opinions that are not those of the Court; they can simply disagree. Justice Kagan likely thought that she was dissenting from the Court’s decision to cause the removal restriction to become ineffective, while concurring in its decision not to cause the power to issue CIDs to become ineffective. It was as though she agreed with one component of an injunction and not another. 

This way of thinking is inconsistent with the Court’s standard explanation of judicial review of statutes. According to that explanation, the Constitution’s hierarchical supremacy in the legal system entails that lower-level rules that are inconsistent with it are inoperative. They are void, and for that reason, courts do not decide cases according to them.  The courts have no power to cause previously operative statutory rules to become inoperative, nor do they need any such power to engage in judicial review. They have the power to decide cases, the obligation to decide them according to law, and the consequent obligation to say what the law is when the Constitution may conflict with another source of law. In case of a conflict, they decide according to the Constitution, not the lower-level law.

Invalidity comes from the Constitution, not the courts, which recognize but do not cause it. Because the Constitution causes rules that conflict with it to be inoperative, they never come into effect. Congress can adopt a statute that has an unconstitutional provision, but in an important sense it cannot make an unconstitutional law.  Statutes that purport to do so fail to. Hence, judicial findings of unconstitutionality are in a manner of speaking retroactive: they can apply to facts that already have occurred.

The difference between the roles of the Constitution and the courts as to invalidity manifests itself in the difference between opinions and judgments. An opinion is an explanation for a judgment. That explanation may reflect the conclusion that a statutory rule is unconstitutional and therefore ineffective. Based on that conclusion, the court will issue a judgment dealing with the parties before it. Conclusions about constitutionally induced voidness are found in opinions, not judgments. The Court in Marbury v. Madison did not give a remedy that caused Section 13 of the Judiciary Act to become partly inoperative. It found that the provision already was partly inoperative, and decided the case accordingly, by providing no relief to any party.

The justices’ understanding of what they did in Seila Law is hard to reconcile with this standard account. The orthodox account underlies, for example, the Court’s leading removal restriction case, Myers v. United States. President Woodrow Wilson directed the postmaster general to remove Frank Myers as postmaster for Portland, Oregon, contrary to a statutory restriction on removal. When Myers came before the Court, Frank Myers had been removed by a power more implacable than the president.  He had died, and the suit was for back pay only, not reinstatement. The Court found the removal restriction unconstitutional and denied the claim for back pay.  But if postmasters like Myers became removable only once Myers had been decided, then he was non-removable when the president sought to remove him. Myers’s removal having been ineffective the day it happened and throughout the remainder of Myers’s life, his estate should have been entitled to back pay. The Court in Myers, understanding that statutory rules inconsistent with the Constitution are void from the beginning, denied the claim.

Invalidity is made by the Constitution and found by the courts.

II. Cashing Out the Metaphor of Judicial Law-Making

According to the orthodox account of judicial review, courts find invalidity and do not make it. So much the worse for orthodoxy, perhaps. Courts make law, and change it too—what else does it mean to refer to novel holdings as “new law?” Admitting that courts make law is simply acknowledging reality. Putting aside the myth that courts merely find law is one reason legal realism was realistic.

That response is wrong. Treating courts as making and changing law the way legislatures do is itself unrealistic. It rests on a metaphor, and metaphors are not realistic. They are figures of speech that are sometimes useful, including as shorthand summaries, but they are not entirely accurate. They take similarity for identity and thereby disregard difference. Metaphors like judicial law-making must be cashed out with accurate descriptions. Cashing out the metaphor of judicial law-making in Seila Law is especially instructive. It shows how the justices have failed to see an important implication of a leading precedent concerning removal provisions, Bowsher v. Synar.

The first step is to replace the figure of speech of judicial law-making with the reality it summarizes. Courts do not change the content of written law, the way legislatures do. When they set authoritative precedents, however, courts do shape the way judges and other government officials act. Judges treat binding precedents as correct statements of the law. Executive officials regularly take judicial precedents as authoritative, adjusting their understandings and actions accordingly. Private parties, in turn, adjust their expectations and their conduct in light of what the courts have said and what executive officials will make of it. Through this chain of effects on conduct and expectations, judicial decisions have effects much like those of legislation. That is one way courts make law.4They also make law in the sense of deciding on policy grounds, but that is a different aspect of the metaphor.

This more detailed understanding of judicial lawmaking justifies sometimes treating judicial decisions as if they made or changed written law even though they really do not. Treating judicial decisions as making or changing law makes sense when understandings and expectations about the law, and not just the law itself, matter. Courts operate on the former and not the latter. Qualified immunity, for example, takes into account understandings and expectations. Its purpose is to avoid penalizing official conduct that reflected a reasonable understanding of the law, as shaped by judicial decisions, whether that understanding was correct or not. That is why judicially-made “new law” does not impose personal liability for events that took place before the doctrine was announced and could influence conduct.

Whether courts should be treated as making or changing law thus depends on whether the appropriate inquiry is into the content of the law, or into understandings of it. When only the content of the law matters, courts treat their decisions as finding it. When subjective understanding matters, they have reason to regard themselves as making and changing it.

That distinction is especially relevant in cases like Seila Law because such cases depend in part on the ripeness holding of Bowsher. Bowsher was like Seila Law in that it involved the constitutionality of a removal provision but not the removal of an officer. In the Gramm-Rudman-Hollings Balanced Budget and Emergency Deficit Control Act, Congress had given the comptroller general authority to impose spending reductions to achieve budgetary goals. The statute creating the office provided that the comptroller might be removed by a congressional joint resolution—an act of Congress. Parties whose federal funding had been reduced by Comptroller Charles Bowsher sought an injunction against the imposition of the reductions. They argued that the comptroller was a congressional officer impermissibly performing an executive function, pointing to the removal power to show that the comptroller worked for Congress.

As in Seila Law, no removal had been attempted. That aspect of Bowsher raised the question whether the constitutional objection to the removal provision was properly before the Court—whether it was ripe, as the Court put it. The Court found that the issue was ripe for consideration by pointing to the here-and-now effect of the removal power, which might influence the comptroller’s decisions. Removal was not before the Court, but a decision was, and the issue was ripe.

Whether Bowsher calls for an inquiry into subjective expectations is not clear from the Court’s opinion. The here-and-now effect might be actual influence on the comptroller’s decisions of the possibility of removal by Congress. Or the here-and-now effect might matter merely because it is possible; the constitutional rule might be a prophylaxis.5I think the latter view is the correct reading of Bowsher. The Supreme Court found the removal power ripe for consideration, as had the three-judge district court below, and both pointed to here-and-now effects, but neither court found any facts about Comptroller Bowsher’s state of mind. As far as I know, there was no indication that he was influenced by the very remote possibility that Congress would pass a statute removing him. The constitutional rule is prophylactic, and Bowsher’s beliefs were irrelevant. If the constitutional rule is a prophylaxis, the Constitution invalidates certain combinations of powers and removal provisions, which may not coexist.6By invalidating the combination but not either provision alone, the Constitution raised a severability-type question: given that the law was not what the statute provided, what was it? In Bowsher, Congress had explicitly provided a fallback system to implement the spending reductions the statute called for. See Bowsher, 478 U.S. at 735–36 (describing the fallback provision). Once the Court concluded that the law was not what Congress had adopted, because the Constitution invalidated it, the Court implemented the applicable law, found in the fallback system.

Later decisions, including Seila Law, take the latter view: the actual, subjective understandings of officials concerning their own removability do not matter. The issue came up in Seila Law through an argument offered by Court-appointed amicus Paul Clement. He maintained that Seila could not rely on any infirmity in the removal restriction because Seila was not affected by it. Director Kraninger had stated her agreement with the solicitor general’s position that she was freely removable by the president, so the removal restriction had no actual here-and-now effect on her decisions. Those decisions were not traceable to the removal restriction, and Article III standing requires traceability. The chief justice rejected the traceability argument. He cited his opinion in Free Enterprise Fund v. Public Company Accounting Oversight Board, which said that no counterfactual inquiry into how an official would have behaved under a different legal rule was appropriate. The important question was not what officials think, but what the law is.

That is, I think, a correct application of Bowsher. It is also, however, inconsistent with other aspects of Chief Justice Roberts’s reasoning. As noted above, he described the director as having become removable, relying on the metaphor of judicial law-making. Treating courts as changing law is appropriate when subjective understandings matter, but the chief justice had rejected inquiry into the directors’ beliefs about their own removability. Given his reading of Bowsher, beliefs did not matter. But courts change the law by changing beliefs, so if beliefs do not matter, all that does matter is the law, which courts do not affect.7Another illustration of this confusion is the disposition of Free Enterprise Fund.  The Court said that although the plaintiffs were not entitled to an injunction against PCAOB actions regulating them, they were entitled to “declaratory relief sufficient to ensure” that they are regulated “only by a constitutional agency accountable to the Executive.” The Court reasoned as if the declaratory judgment would cause a previously unconstitutional, because unaccountable, agency, to become accountable and constitutional. Declaratory judgments are so called because they declare the law and do no more. The purpose of declaratory relief is to change the parties’ expectations and beliefs about the law by stating their legal relations authoritatively. If the Court in Free Enterprise Fund was correct in its application of Bowsher to the question of traceability, subjective beliefs about removability did not matter, and declaratory relief would have no significant effect.

Realism about the activities of courts and legislatures requires understanding how those institutions are similar and how they are different. Relying on the concept of judicial law-making, without a more detailed inquiry, suppresses the differences.

III. Rules About Power and Rules About Duty

The metaphor of judicial lawmaking almost certainly influences judges, and cashing that metaphor out in the reality of law-clarification would clarify their reasoning. The inadequacy of that metaphor has received attention recently, both on and off the bench. Taking it too literally is probably especially tempting to judges in constitutional cases because of another common mistake: confusing the operation of rules about power with the operation of rules about duty. That mistake gives judges another reason to think that their decisions cause statutory rules to become invalid, even though that result is actually brought about by the Constitution and just recognized by courts. They think they are undoing constitutional violations by applying the remedy of invalidation, but that way of thinking is incorrect.  Courts do remedy violations of rules about duty. Rules about power produce invalidity of their own force.  If invalidation is thought of as a remedy, the Constitution supplies the remedy itself, leaving the courts to decide when it has done so..

The Constitution provides criteria that determine the validity of lower-level rules. It thereby constrains the content of the lower-level law. The Constitution does that because many of its rules are about the powers of government. A power is an ability to change legal relations through a deliberate act directed at doing so. Constitutions confer power on legislatures, as the law of contract gives private people the power to enter into contracts.8Power in this sense is one of four interrelated concepts. If one person has a power, some other person has a liability to the exercise of that power. Someone who does not have a power, as for example I do not have the federal legislative power, is said to have a disability in that respect. If one person has a disability, someone else has what is called an immunity from exercises of power by the person with the disability. Congress has power to legislate within its enumerated authority, provided it legislates consistently with other restrictions on that power, and Americans have liability relative to Congress: it can change their legal positions. Where Congress lacks enumerated power, it has disability, and Americans are immune from purported acts of legislation. Constitutions also affirmatively restrict the powers they grant.

Rules about power, like the Constitution and the law of contract-making, determine whether purported jural acts are effective. Acts that satisfy those rules, like statutes pursuant to an enumerated congressional power that are not contrary to any restriction on congressional power, are valid. Acts that do not satisfy those rules, like an oral agreement when only a written one will work, are invalid. The consequences of compliance and non-compliance with rules about power are validity and invalidity. An attempted exercise of a power the actor does not have is ineffective. If I were to promulgate a federal statute, nothing would happen. 

Another kind of legal rule works differently. Rules about duty tell people what they should and should not do. When parties make a valid contract, the contract imposes duties on the promisor. Tort law also imposes duties.9Duty, too, is one of four interrelated concepts. When one person has a duty, another person has a right that the duty-bearer comply with the duty. The absence of duty is called liberty or privilege. Someone who has no duty concerning some act has liberty with respect to it. If one person has liberty, someone else has what is called a no-right, which is the opposite of a right. See Wesley N. Hohfeld, Some Fundamental Legal Conceptions as Applied in Judicial Reasoning, 23 Yale L.J. 16, 30–33 (1913) (setting out relations among those four conceptions). Duties run to parties who have rights that correlate with the relevant duty. Sometimes people act contrary to their duties. People breach their contracts, and they negligently back into one another’s cars. When they do so, the courts supply remedies to rights-holders. Those remedies cannot cause breaches of duty that have already taken place never to have happened, although remedies like preventive injunctions can keep breaches from occurring. Retrospective remedies like damages can, however, bring the actual world into better alignment with the rightful positions of rights-holders, the positions they would have occupied if the duty-bearers had complied with their duties. Damages can restore the wealth level of the right-holder.

Rules about power and rules about duty are thus fundamentally different in their associated consequences. Failure to comply with rules about power produces nullity. Nullity does not compensate for injuries that have happened, nor does it keep an injury that might happen from occurring. Nullity entails that changes in legal relations contrary to the rules about power are impossible. They cannot and do not happen.

In concluding that Seila Law holds the removal restriction unconstitutional, the justices seem to have treated constitutional rules as if they impose duties rather than limiting powers. They reasoned as though Congress, when it passed the Dodd-Frank Act, made a rule that insulated the director from removal. That statute was legally effective, but it violated the Constitution, much as negligently backing into someone else’s car violates the law of tort. Congress had power to make that rule, but had a duty not to—a duty it breached.

When Congress passes a law that violates the Constitution, the reasoning goes, courts can undo that harm with the remedy of invalidation. Like damages, the remedy cannot change the past; it can affect only the future. Invalidation results when a majority of justices concludes that a statutory rule is unconstitutional. By agreeing to an opinion to that effect, the majority invalidates the rule, causing it to become invalid where it had previously been valid.  

This line of reasoning explains how Seila Law can be seen to have held that the removal restriction was unconstitutional. Holdings are found in opinions that explain a result and are joined by a majority. The result in Seila Law was invalidation, and parts of Chief Justice Roberts’s opinion had the support of a majority and so explained the result. The fact that no majority formed as to whether to affirm, reverse, or vacate, did not undo the result of invalidating the removal restriction. The Court took that step, and then fragmented as to the next step.

The justices in Seila Law appear to have analogized the Constitution’s rules about statutory validity to tort law, treating those constitutional rules as imposing duties that can be violated. That was an easy misstep because the Court regards some constitutional provisions as imposing duties. The Fourth Amendment, in the Court’s view, imposes duties on people acting under color of federal law. Those duties correlate with rights that are defined in terms of kinds of injuries. The rights the Fourth Amendment creates trigger the law of remedies, entitling those whose rights have been violated to compensation, and to exclusion of evidence. The Fourth Amendment may work that way, but the rules that grant and limit power do not. They work through invalidity, not through the remedies that go with violations of duties.

This Essay has treated the theory set out in Marbury as orthodox, and has measured the justices’ theories against it. At some point, however, the thinking of lawyers and judges becomes orthodoxy. I do not believe that the justices have consciously embraced the assumption that their decisions genuinely cause statutory rules to become ineffective, or that the Constitution itself does not do so. Rather, I think they would agree that the metaphor of judicial law-making is a metaphor and must not be taken as if it were strictly accurate, and that the difference between powers and duties matters. The justices do not mean to overrule Marbury, although they sometimes write opinions that do not make its doctrine clear.

* * *

John Harrison is the James Madison Distinguished Professor of Law and Thomas F. Bergin Teaching Professor at the University of Virginia. Professor Caleb Nelson provided helpful comments.

* * *

Click here to return to the Seila Law series main page.


Leave a Reply