David A. Strauss1David A. Strauss is the Gerald Ratner Distinguished Service Professor of Law at the University of Chicago.
* * *
A part of the Seila Law and the Roberts Court series.
The majority and dissenting opinions in Seila Law LLC v. Consumer Financial Protection Bureau disagreed about the interpretation of previous Supreme Court decisions that considered the president’s power to remove executive branch officials: Myers v. United States, Humphrey’s Executor v. United States, Morrison v. Olson, and Free Enterprise Fund v. Public Company Accounting Oversight Board. But in many ways the more important disagreement was about historical practice. The majority and the dissent disagreed about what the historical practice was; they also disagreed about the role historical practice should play in determining what the Constitution requires. The majority insisted that the Consumer Financial Protection Bureau (CFPB) “lacks a foundation in historical practice and clashes with constitutional structure.” The CFPB, according to the majority, was historically anomalous because the director of the CFPB is a single individual whom the president cannot remove except for cause.2See, e.g., Seila Law, 140 S. Ct. at 2191 (“In organizing the CFPB, Congress deviated from the structure of nearly every other administrative agency in our history.”); id. at 2201 (“An agency with a structure like that of the CFPB is almost wholly unprecedented.”). According to the majority, no agency has ever been structured like that, leaving aside a few peculiar (in the majority’s view) examples: when an agency has a single head, that individual is always removable at will; for-cause protection has been extended only to multimember agencies. Related, the CFPB was inconsistent with the “constitutional structure,” the majority said, because it contravened the principle that “the President’s removal power is the rule, not the exception.”
Justice Elena Kagan, in dissent, argued that the other instances of a single-headed agency with for-cause removal protection were not so peculiar, that the difference between multimember and single-headed agencies has never been thought to matter much and, in fact, that it should not matter much. But her primary concern was not that aspect of the majority’s reasoning,3In fact, after noting that “[t]he majority focuses on one (it says sufficient) reason [for its conclusion]: The CFPB Director is singular, not plural,” Justice Kagan remarked: “I’m tempted at this point just to say: No. All I’ve explained about constitutional text, history, and precedent invalidates the majority’s thesis.” but the majority’s more far-reaching claim that the “structure” of the Constitution made for-cause removal provisions constitutionally suspect in general. And the primary basis of her attack on that claim was not the text of the Constitution or the handful of Supreme Court precedents, although she discussed both of those. Rather, Justice Kagan argued that that majority’s assertions about the structure of the Constitution could not be reconciled with “the broad sweep of history.”
In this way, both opinions invoked, although in different ways, what might be called non-judicial precedent: past decisions made not by courts but by other branches of government. In particular, they invoked the decisions, reflected in legislation enacted by Congress, about the extent to which agency heads should be removable at will by the president. Arguments based on past practice, or non-judicial precedent (I will use those terms interchangeably), were at least as important in Seila as the more familiar kind of arguments based on judicial precedents.
This is part of a pattern. In cases about the separation of powers, non-judicial precedent often seems to count for a lot. Trump v. Mazars USA, decided ten days after Seila, dealt with the power of committees of the House of Representatives to subpoena the private papers of the president from a third party. The Court noted that it had “never addressed a congressional subpoena for the President’s information” and that “[h]istorically, disputes over congressional demands for presidential documents have not ended up in court.” In the absence of judicial precedent, the Court extensively discussed non-judicial precedent: how, in the past, Congress and the executive branch had dealt with demands for documents. “Such longstanding practice,” the Court said, “is a consideration of great weight in cases concerning the allocation of power between [the] two elected branches of Government” (internal quotation marks omitted).
Mazars, in turn, relied on two other separation-of-powers cases: NLRB v. Noel Canning, the 2015 decision dealing with the president’s power to make recess appointments, and The Pocket Veto Case, which addressed questions about when the president’s failure to return an enrolled bill to Congress constituted a veto. Both Noel Canning and The Pocket Veto Case extensively cited past practice and emphasized its importance. Similarly, Zivotovsky v. Kerry relied on “[t]he weight of historical evidence” in concluding that “Congress has accepted that the power to recognize foreign states and governments and their territorial bounds is exclusive to the Presidency.” And in Youngstown Sheet & Tube Co. v. Sawyer, the foundational decision about the president’s power to act without congressional authorization, Justice Frankfurter’s influential concurring opinion asserted that “[d]eeply embedded traditional ways of conducting government,” while they “cannot supplant the Constitution or legislation,” are important because they “give meaning to the words of a text or supply them.”
This consistent reliance on non-judicial precedent in separation-of-powers cases raises a number of questions about the nature of U.S. constitutional law—questions that Seila can help answer. For example: Why is reliance on non-judicial precedent so common in separation-of-powers cases, in particular? Is it justifiable to rely on non-judicial precedent in that way? Or does relying on past practice reflect a misunderstanding of constitutional law—as Justice Antonin Scalia once suggested, “an adverse-possession theory” of the Constitution?
I. Non-Judicial Precedent and the Separation of Powers
Non-judicial precedent may play a larger role throughout constitutional law than the Supreme Court’s opinions acknowledge, but the most explicit invocations of non-judicial precedent occur in cases about the separation of powers.4Separation-of-powers decisions have been the primary focus of the extensive and excellent scholarship on the use of past practice. See, e.g., Curtis A. Bradley & Neil S. Siegel, Historical Gloss, Constitutional Conventions, and the Judicial Separation of Powers, 105 Geo. L. J. 255 (2017); Curtis A. Bradley, Doing Gloss, 84 U Chi L Rev 59 (2017); Shalev Roisman, Constitutional Acquiescence, 84 Geo. Wash. L. Rev. 668 (2016); Curtis A. Bradley & Neil S. Siegel, After Recess: Historical Practice, Textual Ambiguity, and Constitutional Adverse Possession, 2014 S. Ct. Rev. 1; Curtis A. Bradley & Trevor W. Morrison, Historical Gloss and the Separation of Powers, 126 Harv. L. Rev. 411 (2012); Michael J. Glennon, The Use of Custom in Resolving Separation of Powers Disputes, 64 B.U. L. Rev. 109 (1984). See also Curtis A. Bradley & Neil S. Siegel, Historical Gloss, Madisonian Liquidation, and the Originalism Debate, 106 U. Va. L. Rev. 1 (2020); Richard H. Fallon, Jr., The Many and Varied Roles of History in Constitutional Adjudication, 90 Notre Dame L. Rev. 1753 (2015). The Court in Mazars suggested one reason why: courts are reluctant to intervene in disputes between the other two branches, so sometimes there is just not very much relevant judicial precedent on the books. That might explain why there are not a lot of decided cases in which one branch of the government sued another. But separation-of-powers issues can, of course, arise in suits brought by or against private parties, as in Seila itself and in the cases discussed in the Seila opinions—whenever the resolution of a separation-of-powers issue will affect a private party’s interests in a way that gives that party standing to sue or to raise the separation-of-powers issue defensively. Courts are not reluctant to entertain those kinds of lawsuits. So a paucity of judicial decisions resulting from courts’ reluctance to get involved cannot alone explain why non-judicial precedent seems so important in separation-of-powers cases.
A big part of the answer, I think, can be found in a passage in Chief Justice John Marshall’s famous opinion in McCulloch v. Maryland. Marshall wrote:
[A] doubtful question, one on which human reason may pause, and the human judgment be suspended, in the decision of which the great principles of liberty are not concerned, but the respective powers of those who are equally the representatives of the people, are to be adjusted; if not put at rest by the practice of government, ought to receive a considerable impression from that practice.
Chief Justice Marshall’s distinction is between cases concerning individual rights—“the great principles of liberty”—and cases concerning “the respective powers of the branches”; he thought that past practice was important in the latter, not the former. He did not explain why, but here is a possible reason: in dealing with issues about individual rights, the courts (like the rest of us) have a clearer idea of what a fair or morally right answer is. It is never acceptable for a government to arrest the political opposition, persecute religious dissidents, discriminate against racial or ethnic minorities, or imprison people without a fair trial. Those principles are reflected in (or are read into) the provisions of the Constitution that protect individual rights. Of course, there is still litigation about disputed issues. But in resolving those issues, the courts can draw on widely shared views about the principles that any decent society will respect.
That is much less true of the separation of powers. Political systems that reject the basic principles of the U.S. version of separation of powers can still be perfectly acceptable. They might even be better. James Madison, “in language that has become a maxim of structural constitutional law,” said that “[t]he accumulation of all powers, legislative, executive, and judiciary, in the same hands . . . may justly be pronounced the very definition of tyranny.” But, as it happens, that’s not true. To quote Professor Daryl Levinson:
[W]hat, exactly, would be so bad about a system in which one political institution possessed all of the power? Perhaps it would help to imagine what would happen to the American constitutional system if the presidency and judicial review were eliminated altogether. For good measure, let us also imagine doing away with any constitutional limitations on Congressional power or other guarantees of state sovereignty, effectively eliminating constitutional federalism and concentrating all the power of the American state in Congress. Would this be “the very definition of tyranny”?
Actually, it would be England. This is roughly how the British Westminster model of government works, vesting nearly unconstrained power in Parliament, and therefore in the party that controls Parliament at any given time. Political scientists and comparative constitutionalists have long debated the advantages and disadvantages of the Westminster system as compared to the U.S. presidential system.
Because a radically different approach from ours to the separation of powers might be completely fine—unlike a radically different approach from ours to, say, issues of religious toleration or racial discrimination—judges in separation-of-powers cases do not have the same kind of normative guidelines about what they are trying to achieve (or avoid). So they have to look elsewhere. A natural place to look—and, I will suggest, an appropriate place to look—is at what our system actually is. We can’t say which approach to separation-of-powers issues is better, but we can say something about which approach is ours. That means that past practice will be centrally important in separation-of-powers cases in a way it might not be elsewhere.
Related, Justice Louis Brandeis’s maxim—“in most matters it is more important that the applicable rule of law be settled than that it be settled right”—is more likely to be true in separation-of-powers cases than in cases about individual rights. That is so partly for the same reason: we have a harder time, in separation-of-powers cases, deciding which outcome is better in the abstract, so we might as well just get things settled. It is also because institutions operate better when issues about their composition are settled. When people know the rules of the game will not change, they can adjust their behavior accordingly. That, too, suggests that in separation-of-powers cases, what we usually want to do is to figure out what our system is, and to stick with it.
The disagreement between the majority and the dissent in Seila illustrates much of this. Some passages in the majority opinion condemned the CFPB in terms that echoed Madison’s apothegm about “very definition of tyranny”:
The CFPB’s single-Director structure contravenes [the Constitution’s] carefully calibrated system by vesting significant governmental power in the hands of a single individual accountable to no one. . . . With no colleagues to persuade, and no boss or electorate looking over her shoulder, the Director may dictate and enforce policy for a vital segment of the economy affecting millions of Americans.
Yet it is not hard to imagine an opinion relying on comparable abstractions but upholding the CFPB’s structure. It might go something like this:
Congress established the CFPB with a single head to make sure that the agency’s policies would be coherent and focused – not the product of intra-agency logrolling. And Congress believed that the Director’s insulation from political pressure was crucial to her accomplishing the agency’s mission: to defend the interests of ordinary consumers, who are constantly at risk of being outspent and out-lobbied by the well-organized, well-funded financial services industry. The industry’s influence may easily extend to people who have the ear of the President and who could, in the absence of for-cause protection, rein in a Director who was doing her job too well.
At the same time, the Director’s power is very far from unlimited. Her actions are subject to judicial review to make sure they conform to the law, and she can be removed for malfeasance or misfeasance.
It is not obvious that that defense of the CFPB is any less plausible than the majority’s attack. The point is not that there is no right answer in principle to the question of how the CFPB should be constituted; a conscientious member of Congress would have to answer that question, as best she could. But the answer is mired in normative and empirical uncertainty, and it cannot be found by invoking Madisonian abstractions about the nature of power.
Once you accept that, it is natural to change the question. Instead of asking which structure better preserves liberty or avoids tyranny, we can ask which arrangement is consistent with how we usually do things. That will avoid any upheavals; it will keep things settled, in Justice Brandeis’s terms. And how bad can an arrangement be if it conforms to the way we have done things all along, given that when we are dealing with the separation of powers—unlike when we are dealing with questions of individual rights—we do not have a good sense of what makes things better or worse? Courts’ instinctive reliance on non-judicial precedent in separation-of-powers cases suggests that they understand this.
II. Non-Judicial Precedent and Judicial Precedent
There is a more straightforward justification for invoking non-judicial precedent: in important ways, non-judicial precedent is not that different from judicial precedent. That justification extends beyond separation-of-powers disputes to much of constitutional law. In resolving constitutional issues, courts (and other political actors, for that matter) routinely rely on judicial precedent. But many of the reasons for relying on judicial precedent apply to non-judicial precedent, too.
Those reasons constitute what might be described as the ideology of the common law: a set of arguments that justify the practice of following precedent. They are commonly identified with Edmund Burke, the late eighteenth-century British statesman. One Burkean reason for adhering to what has been done before is a counsel of humility.5See Edmund Burke, Reflections on the Revolution in France 74 (Frank M. Turner ed.) (2003) (“We are afraid to put men to live and trade each on his own private stock of reason; because we suspect that this stock in each man is small, and that the individuals would do better to avail themselves of the general bank and capital of nations and of ages.”). If others, over time, have settled on a way of doing something, it is an unjustified act of intellectual arrogance to believe that you know better than the accumulated judgments of many predecessors. Another is the risk of underestimating the complexity of real-world arrangements. If an institution has been sustained over time, it reflects practical, trial-and-error adjustments to complicated circumstances, and a dramatic change in that institution is likely to make things worse.6“The science of constructing a commonwealth, or renovating it, or reforming it, is, like every other experimental science, not to be taught a priori. . . . The science of government being therefore so practical in itself, and intended for such practical purposes, a matter which requires experience, and even more experience than any person can gain in his whole life, however sagacious and observing he may be, it is with infinite caution that any man ought to venture upon pulling down an edifice, which has answered in any tolerable degree for ages the common purposes of society.” Burke, Reflections, at 52–53.
These are reasons for courts to follow both judicial and non-judicial precedent. In fact, the common description of these ideas as “Burkean” demonstrates the connection between judicial precedent and past practice in other institutions. In the tract in which he advanced these ideas, Burke attacked the too-radical changes in government and society that the French Revolution brought about. But Burke’s template for the way government and society should change was the common law,7See, e.g., Burke, Reflections 27–28; J.G.A. Pocock, Burke and the Ancient Constitution: A Problem in the History of Ideas, in J.G.A. Pocock, ed., Politics, Language, and Time: Essays on Political Thought and History 206–32 (1989). and arguments anticipating Burke’s were made by common lawyers.8For a discussion, see David A. Strauss, Common Law Constitutional Interpretation, 63 U. Chi. L. Rev. 877, 893–94 (1996).
Justice Kagan’s dissent in Seila implicitly criticized the majority opinion as a direct affront to these Burkean themes. She discussed the text of the Constitution and the founding-era history, and her specific argument was that Congress, not the courts, should in general determine the scope of the president’s removal power. But some of the most powerful parts of her opinion condemned the majority for elevating its theoretical constructs over the variety and complexity of U.S. history. Justice Kagan identified in the majority opinion a kind of abstract reasoning—a “civics class version of the separation of powers”—that, she said, disregarded the actual institutions of the U.S. government. The majority had asserted that there is a “‘general rule’ of ‘unrestricted [presidential] removal power’ with two grudgingly applied ‘exceptions.’” But that view of the Constitution “repudiates the lessons of American experience, from the 18th century to the present day.” “The majority’s general rule does not exist [and i]ts exceptions . . . are made up for the occasion.”
Justice Kagan described how Congress had repeatedly insulated agencies from for-cause removal. Her language reflected the Burkean emphasis on the practical, adaptive nature of institutions that are established over time: “Confronting new economic, technological, and social conditions, Congress—and often the President—saw new needs for pockets of independence within the federal bureaucracy.” She noted that many of the agencies over which the president had only limited removal power engaged in financial regulation, like the CFPB—suggesting that one of the lessons of experience is that independence is especially important when that is an agency’s mission. Justice Kagan cited examples from the founding era, the antebellum period, and the Civil War; then, she noted, beginning in the late nineteenth century, independent agencies became common. “[Y]ear by year by year,” she concluded, “the broad sweep of history has spoken to the constitutional question before us” and has validated Congress’s power to limit the president’s authority to remove executive branch officials.
Once we step back from Justice Kagan’s arguments, the puzzle may be not why non-judicial precedent matters in separation-of-powers cases; the puzzle is why, when there is a wealth of non-judicial precedent, the Supreme Court pays so much attention to judicial precedent. Although she does not put it in these terms, Justice Kagan’s opinion invites us to contrast the relative handful of cases in which the Court has addressed the removal power (some featuring notably overstated or obscure opinions) with long-standing, generally accepted institutions of government that have made highly significant decisions, day by day, for decades. If you were concerned with precedent in a general sense—with maintaining continuity with the past and learning its lessons—it is difficult to see why you would prefer a close parsing of the language of a few scattered Supreme Court opinions to that enormously important institutional practice.
Another way to put the point is that past practice does, in fact, embody a form of judicial precedent, after the fashion of the proverbial (and clichéd) dog that did not bark. Established past practice reveals all the lawsuits that no one thought were plausible enough to bring. Over the years, countless firms and individuals have undoubtedly been aggrieved by the decisions of agencies that had for-cause removal protections. Had they thought that they might have a plausible constitutional challenge to those agencies’ actions, based on the agency heads’ insulation from removal by the president, they would have sued. The relative lack of those lawsuits tells us something: it reflects a widespread understanding in the legal culture that agency heads could be protected from removal. If a legal proposition is so well established that people with an incentive to sue don’t think it’s worthwhile to challenge that proposition, that should count for something. If we are evaluating the weight of precedent, shouldn’t it count for more than a few hard-to-reconcile decisions, spread out over decades? That is the question that the use of non-judicial precedent, in Seila and elsewhere, invites us to consider.
III. Non-Judicial Precedent and the Constitution
There is, on the other hand, a straightforward objection to the assimilation of non-judicial precedent to judicial precedent. Judicial decisions are, often enough, the product of a deliberate, good faith effort by judges to figure out what the Constitution requires. We cannot, in general, say that about historical practice—not because of a lack of good faith, but because historical practice normally develops without explicit consideration of the constitutional issues. Administrative agencies with for-cause removal protection were established by politicians voting in Congress (and White House officials advising the president on the positions he should take). Maybe some of those agencies were to some extent the product of a deliberate assessment of the constitutional merits, but that is not the norm for political decisions, and politicians generally do not have an incentive to engage in that kind of reflection. So (the argument goes), it is a mistake to equate judicial and non-judicial precedent, at least unless there is evidence that the non-judicial precedent was the result of a deliberate consideration of the constitutional issues.9See, in this connection, Shalev Roisman, Constitutional Acquiescence, 84 Geo. Wash. L. Rev. 668 (2016).
This objection obliquely identifies a practical difference between judicial and non-judicial precedent: because judicial decisions are accompanied by opinions that explain their basis, it is easier to figure out what judicial precedents stand for. Historical practice is a set of outcomes that impliedly claim to be constitutional but don’t explain why they are. We know what was done, but we can only infer what justifies it. It is a little like working with judicial precedent in a system in which every court simply issued a holding without any reasoning and without any assurance that the decisionmaker actually thought about the constitutional issues.
In Seila, this problem is illustrated by the majority’s most important use of past practice: its assertion that the CFPB presented a “’severe constitutional problem’” because “[a]n agency with a structure like that of the CFPB is almost wholly unprecedented.” Specifically, there are, the majority claimed, “‘only a handful of isolated’ incidents in which Congress has provided good-cause tenure to principal officers who wield power alone rather than as members of a board or commission.”
The majority did not actually say why that history should matter, but it evidently believed that the CFPB’s outlier status supported its theory about the Constitution’s allocation of power: that the Constitution “divide[s] power everywhere except for the Presidency, and render[s] the President directly accountable to the people through regular elections.” The “CFPB’s single-Director structure contravenes this carefully calibrated system by vesting significant governmental power in the hands of a single individual accountable to no one.”
Maybe the majority is correct about the implications of the history. Maybe Congress has not established a lot of agencies like the CFPB because it agreed with the majority’s views about the “carefully calibrated system.” But we do not have, from Congress, anything comparable to a judicial opinion explaining that this was the reason for the historical pattern. Congress’s decisions about the composition of agencies could have reflected its views about the most effective ways to implement different regulatory regimes at different times, the need for legislative compromise, the political acceptability of different arrangements in varying contexts, or any number of other things. The majority just imputed to the historical pattern a reason that supported its particular view of the Constitution. And, as the dissent argued, it is not even clear that the difference between single-headed and multimember agencies has anything to do with the “structural” concerns that the majority invoked.10Justice Kagan’s response to the majority’s use of history on this point (in addition to disputing that the history was clear) was to say that the differences between an agency with a single head and a multi-member agency are not significant enough to support any across-the-board judgment that the latter arrangement is preferable. See, e.g., 140 S. Ct. at 2243 (Kagan, J., dissenting) (“Why, then, would anyone distinguish along a simple commission/single-director axis when deciding whether the Constitution requires at-will removal?”).
Having said that, though, in principle it is not indefensible to impute a rationale to a historical pattern in the way the majority did. In fact it is unavoidable, if past practice is going to be used at all outside of the rare case in which the historical developments are somehow accompanied by an explicit explanation. What’s more, the difference with judicial precedent can be overstated. In dealing with a troubling precedent, courts will sometimes accept the outcome as correct but impute a new justification; they will effectively rewrite the opinion to provide what seems to them a more plausible basis for that outcome and use the rewritten version of the opinion as the governing precedent.11For an argument that then-Judge Benjamin Cardozo’s celebrated opinion in MacPherson v. Buick Motor Co., 111 N.E. 1050, 217 N.Y. 382 (1916), is an example of this use of precedent (a widely although not universally held view of MacPherson), see David A. Strauss, The Living Constitution 80–85 (2010). That is, more or less, what we do, of necessity, with non-judicial precedent.
Finally, the relationship between non-judicial precedent and the Constitution raises one of the most fundamental issues about U.S. constitutionalism. It is the question whether the meaning of the Constitution was fixed at one time, presumably by some combination of the text and original understandings, or continues to evolve. If it was fixed, then it is hard to see why, in resolving constitutional issues, we care about any subsequent developments, judicial or otherwise, unless they are efforts to discern that fixed meaning. But if constitutional law is an evolutionary enterprise (which, it seems to me, it is12For a defense, see, e.g., Strauss, The Living Constitution.), then there is no obvious reason for judges to have a monopoly on the direction and pace of the evolution. The other branches should have their say, too, in the way they actually go about building the government. If you accept Burke’s arguments, the practical, evolutionary process of institution-building by the other branches of government is entitled to deference. That was the process described in the dissenting opinion in Seila. Those competing ideas about the very nature of the constitutional law may be the most important question raised by the dispute over the removal power in Seila.
* * *
David A. Strauss is the Gerald Ratner Distinguished Service Professor of Law at the University of Chicago.
* * *