Seila Law as an Ex Post, Static Conception of Separation of Powers

Commentators have explored many important questions in the wake of Seila Law LLC v. Consumer Financial Protection Bureau. Do Myers v. United States and Humphrey’s Executor v. United States still stand for the proposition that Congress can impose limitations on the president’s removal authority for agency heads as long as it does not retain a role for itself? If Humphrey’s Executor and Morrison v. Olson are “exceptions” to the view that Congress cannot impose limitations on the president’s removal authority, what is the scope of these exceptions? Is there really a conceptually relevant difference between agencies with one head and those with multiple heads? Is the modern-day FTC now vulnerable? What about the Federal Reserve?

But less discussed is the Seila Law Court’s failure to consider the realities of interbranch relations over time. We argue that the Court’s analysis takes an ex post, static conception of interbranch dynamics. The conception is ex post because it does not ask the question of whether, going forward, presidential power will in fact be expanded. And it is static because it assumes that, with respect to the Consumer Financial Protection Bureau (CFPB), the president will now be able to remove the director at will and all other institutional relationships will continue unchanged. In concluding that the removal provision for the CFPB director put impermissible constraints on the president, the Court apparently assumes that, if that provision is struck down, presidential power will be increased. This is like saying that if water is flowing in through a hole in a levee, one need only plug that hole and the problem will be solved. It doesn’t take a lot of hydraulic expertise to know that the levee is likely to be breached at the next weakest point.

This short Essay has both a positive and a normative goal. The positive goal is to consider the downstream impacts of the Court’s decision. And the normative goal is to provide a roadmap for a consequentialist criticism of the Seila Law majority. As it is often said, the Constitution is not a “suicide pact.”1 Nor, we believe, is it a “walk-around-blindfolded pact.” A decision that has potentially negative consequences—ones that the Court had no interest in examining—might not be what the Constitution requires.

I. Formalism and Functionalism

The perennial debate in separation of powers doctrine is whether it should be conceptualized formally or functionally. Did the Framers command “high walls” of clear separation between the three branches (formalism), or rather a system of “blended powers” (functionalism)?

In his majority opinion in Seila Law, Chief Justice John Roberts embraces formalism, arriving at an apparently bright-line rule that a for-cause removal restriction on a single-headed agency with executive power violates Article II.2 In dissent, Justice Elena Kagan argues that the majority opinion’s formalism amounts to a “Schoolhouse Rock” version of the doctrine. The split between the chief justice’s formalism and Justice Kagan’s functionalism tracks the split ten years ago in Free Enterprise Fund, when the chief justice struck down a for-cause removal provision in a majority opinion sounding in formalism while Justice Stephen Breyer dissented on functional grounds.3

But missing from both formal and functional analyses is a deep engagement with changing institutional relationships over time. After all, it is one thing to insufficiently analyze the functional realities of agency design in a given moment. But it is a distinct, if related, problem to insufficiently consider and engage with how interbranch arrangements will likely change over time.

Thus, even if the Court is correct as to (1) the location of the formal constitutional boundary that it must enforce and (2) its current empirical analysis of which features of agency design currently overstep that boundary, the decision might nonetheless lead to unintended consequences that are the same or worse than the current arrangement. The majority’s failure to analyze the dynamic aspects of separation of powers casts doubt both on whether the results of the decision will be salutary from a governance perspective and on whether the outcome will effectuate the formal constitutional goals the Court is trying to achieve.

II. The Challenges of Regulating Separation of Powers Ex Post

The CFPB was created against a backdrop of a gradual increase in presidential control of the regulatory state, through both judicial decisions and executive creation and expansion of offices such as the Office of Information and Regulatory Affairs. For at least four decades prior to the CFPB, presidents of both parties have sought and wielded this control.4 In 2009, President Barack Obama wanted broad authority to address the global economic crisis, and it seems likely that he would have preferred this authority to come with fewer (or no) strings attached. Yet the birth of any agency necessarily requires interbranch negotiation to arrive at an institutional arrangement that pleases both sides. And so, when the CFPB was created, it cut back on this trend toward consolidated presidential control over the administrative state.5

Ten years later, Seila Law announced ex post that this institutional settlement was constitutionally problematic. Of course, in a simplistic sense, all decisions in federal court are ex post, because federal courts have jurisdiction only over live cases and controversies. If courts can issue rulings only as to parties that are adverse over an already-existent arrangement, how could they avoid issuing ex post decisions? In a literal sense, a court must engage in ex post decision-making because to do otherwise would violate the ban on “advisory opinions.”

Yet ex post decision-making in complex separation-of-powers disputes poses difficult remedial questions. Seila Law held that the unconstitutional for-cause removal provision protecting the CFPB director was severable—that is, the unconstitutional provision had to fall but the rest of the law could remain in effect. Perhaps this was the correct conclusion simply on the basis of the fact that Dodd-Frank included a severability clause, which, on its terms, manifested congressional intent that the rest of the Act should survive if a part were invalidated. The Court admitted that its only instrument was “a blunt one,” and this conclusion on severability was far less invasive than a conclusion that the provision was not severable and the whole law had to fall.

But even if the Court’s decision is correct given the statutory text, it is at least puzzling. To the extent that the agency’s design resulted from careful negotiation and institutional settlement, the choice when Congress was drafting the Dodd-Frank Act was not likely to be between a CFPB with a for-cause removal provision (the position reached by seven justices) and no CFPB at all (see the concurrence of Justices Clarence Thomas and Neil Gorsuch). Instead, Congress would likely have placed other constraints on presidential power. The Court lacked the capacity to reconstruct this alternative interbranch accommodation.

To be sure, these difficulties arise in many constitutional disputes, and perhaps they should not affect the merits. But federal courts do—and must—consider the consequences of their decisions and their chosen remedies, even in constitutional cases. This term, Chiafalo v. Washington presented the Supreme Court with the question of whether the Constitution permitted state laws that fined or otherwise sanctioned presidential electors who did not vote for the candidate to whom they were pledged. Both Justices Samuel Alito and Brett Kavanaugh suggested at oral argument that, if a constitutional case involves a “close call or a tiebreaker, that we shouldn’t facilitate or create chaos” and stated that, even if a bad outcome has not yet occurred, “we have to look forward.”  Similarly, in its landmark case Casey v. Planned Parenthood, the Court explained how it determines when it should change a constitutional rule. Casey held that the Court must analyze whether the existing constitutional rule is “unworkable.” Such an analysis necessarily requires considering and comparing the practical consequences of different constitutional interpretations.

Moreover, the Court sometimes concludes that it lacks jurisdiction in part because of anticipated results of a ruling. In Rucho v. Common Cause, Chief Justice Roberts wrote for the Court that partisan gerrymandering was a nonjusticiable political question because he concluded there were no “judicially manageable standards” to guide the courts in intervening in “one of the most intensely partisan aspects of American political life.” Whether this conclusion is right or wrong, any analysis of whether an issue is a political question beyond the scope of the courts clearly involves thoughtful consideration of anticipated consequences.6 In some cases at equity, the Court has held that remedial difficulties might altogether deny standing to a plaintiff, at least where the Court predicts jurisdiction would require “future intervention that would be so intrusive and unworkable” or would “displace the normal principles of equity, comity, and federalism.”

In Seila Law, the Court apparently did not look forward: it applied a formal rule and remedy and considered the matter settled. But it might have grappled with the dynamic nature of separation of powers and asked whether Congress would indeed leave its relationship to the agency and to the president unchanged. If the institutional arrangement was likely to adjust dynamically, those anticipated consequences perhaps should have led the Court to reach a different decision.

III. Interbranch Dynamics and the Legislative Veto

As political scientist Robert Jervis has explained, interaction effects in complex systems can complicate goal-oriented decision-making. In an interconnected system, not only will an action cause counteractions, but “multiple parties and stages permit many paths to unanticipated consequences.” The three branches of government constitute a complex system like any other, and Congress and the presidency adjust dynamically to Supreme Court decisions, often in ways that the Court neither anticipates nor desires. The saga of the legislative veto clearly illustrates this point.

The legislative veto was a product of institutional settlement between Congress and the presidency. Beginning in the New Deal, Congress passed hundreds of laws that authorized one or both houses of Congress to veto final decisions made by an executive agency. At the time, Congress was delegating expansive new powers to agencies to address the Great Depression. The two were inextricably linked: exigent circumstances required Congress to delegate its power to the executive, so Congress designed the veto to maintain some control. President Franklin Roosevelt found this out the hard way. In 1937, Congress enacted a law enabling President Roosevelt to reorganize the executive branch, subject to a legislative veto. After initially rejecting the legislative veto as unconstitutional, Roosevelt quickly backtracked a few weeks later when he realized the House of Representatives would not grant him the reorganization power unless it came with the veto. Similar negotiations took place during the Truman and Eisenhower administrations. The use of the legislative veto grew in the 1970s—and grew more contentious. But the two branches largely settled the matter, depending on the relative bargaining positions of the president and Congress, the breadth of the delegation, the need for administrative flexibility, and the availability of other alternative forms of congressional oversight.7

But, in 1983, the Supreme Court decided INS v. Chadha, which held unconstitutional a one-house veto over the attorney general’s decision to suspend an immigrant’s deportation. In a sweeping decision, the Court explained that the congressional veto of the attorney general’s suspension decision was fundamentally “legislative,”8 so it violated the proper procedures for enactment of law—bicameralism and presentment to the president—required by the text of Article  I, Section 7 of the Constitution.

Justice Byron White’s dissent in Chadha argued that the majority’s invalidation of the legislative veto would disrupt the separation of powers and put a thumb on the scale in favor of the president.9 After all, Chadha stripped Congress of a key tool it had in maintaining leverage over executive decisions, leaving Congress with a “Hobson’s choice.” Because it is cumbersome to pass a law, Congress has often delegated broad discretion to agencies. But to ensure agency accountability, Congress used various formal and informal tools to ensure that agencies did not exercise that discretion improperly. Chadha upset the balance in hundreds of laws, including many that are relevant today, such as whether President Trump has authority to declare an emergency to build a border wall.10

Yet separation of powers is dynamic, and branches engage in what David Pozen has called “self-help” to accomplish their goals. Assuming the Chadha Court was correct in its sweeping conclusion that the legislative veto constituted congressional lawmaking outside of the textually prescribed constitutional procedures, it is not clear that the equilibrium that evolved post-Chadha was less problematic.

In the first place, the Court faced a compliance problem. The other branches, which had arrived at their own settlement, did not eagerly embrace the Court’s new rule. As Louis Fisher has demonstrated, Congress continued using legislative vetoes. In the nine years following the decision, Congress passed—and the president signed into law—more than 200 bills that included a legislative veto. Even if the president and his legal advisers could easily argue in court that they were unenforceable, agencies, bound as they are to appropriations and oversight by the relevant committee, often treated them as if they had force. And many of these provisions cannot get to court because private litigants do not have standing to bring a challenge. The Chadha majority apparently did not consider these problems, predictable though they were.

More importantly, compliance with Chadha led Congress to devise new methods to achieve the same ends. The Chadha ruling shifted legislative vetoes underground into a world of informal agreements to influence executive decisions—and often with less transparency. For example, in the early months of the George H.W. Bush administration, Secretary of State James Baker agreed to give veto power to four committees over funding the Nicaraguan Contras. Robert Bork regarded the so-called “Baker Accord” as “even more objectionable” than the legislative veto struck down in Chadha. This phenomenon holds true in the states: empirical research has shown that, when courts invalidate state legislative vetoes, state legislatures respond by augmenting other methods to control the executive that may be even more intrusive.

The Chadha Court did not anticipate these shifts in interbranch dynamics. But with the benefit of hindsight, these developments suggest that invalidating the legislative veto may have simultaneously made governance more difficult (or at least more complicated) and failed to effectively police the constitutional line it intended to police. Not only did the Court’s decision have arguably negative consequences, but those consequences may have undermined the goals the Constitution purportedly mandated.

IV. The Future of Agency Design

Like Chadha, Seila Law upset an arrangement that Congress and the president found to serve their joint interests. The branches are unlikely to leave this relationship unchanged. With a president in more direct control of agencies, Congress might want to influence them more aggressively through its oversight function. Maybe Congress will amend the CFPB statute to limit its authority. Maybe the Senate will scrutinize director nominees more extensively and even refuse to approve some of them. Maybe the House and Senate will take a more intrusive role in oversight hearings. The Seila Law holding was motivated by the intuition that the president must have more control over the CFPB in order to “take Care that the Laws be faithfully executed.” But it is not obviously true that, once the branches have reached a new arrangement, the president will have more control. That is an empirical question involving the interaction of many complex institutional mechanisms—a question the Court was not interested in asking.

The post-Seila Law equilibrium may have little effect on—and may even reduce—the president’s executive power. Various political science studies illustrate that there is little systematic difference in the responsiveness of agencies to the president’s preferences depending on whether their heads are removable at will. Empirically, the removal power does not appear to be particularly useful to the president at getting his way, at least compared with other forms of control. Furthermore, now that Seila Law forbids one kind of limitation on the removal power, Congress will likely substitute alternate methods of influencing agencies, either in amending the CFPB’s structure or when designing future single-headed agencies. If these alternative forms are more effective at aggrandizing congressional power than limits on the removal power, the president’s power is likely to decrease.

Ironically, whatever actual effect Seila Law has on executive power, it likely creates the belief that the president has more power. At least five justices believe this, and much of the public likely does as well. Thus, if the president’s power declines, he may be doubly weakened: he will be held more accountable even as he in fact has less control over the agency.

In addition to adversely affecting presidential control over the agency, Seila Law might also reduce the aggregate amount of power Congress delegates to the executive branch. What will happen the next time the president wants new broad authorities to be carried out by a new agency? In some instances, Congress will constrain the president’s power in other ways, as discussed above. But in others, Congress might refuse to give the president power at all—exactly the prospect that President Roosevelt faced in 1937 when he fought the legislative veto. But where Roosevelt could backtrack from his earlier position and compromise, Seila Law forecloses the possibility of compromise. There simply may be no available institutional arrangement that satisfies both sides, so Congress may simply refuse to empower the president at all.

Beyond the dynamic adjustment of the relative levels of legislative and presidential control and the breadth of delegations to the executive branch, Seila Law may have other downstream effects. Most obviously, it may reduce agency effectiveness and increase the federal courts’ separation-of-powers docket. The Senate might refuse to confirm the president’s nominees to head agencies, leaving agencies in the hands of acting heads—weakening agency effectiveness and spurring more legal challenges under the Federal Vacancies Reform Act and the Appointments Clause.11 Also, Congress might use its oversight authority to investigate agencies and discourage presidential control, for example by making more burdensome requests for documents and for the testimony of agency heads. If the president objects to these strategies, the Court will likely face still more separation-of-powers disputes.

One might argue that the above consequences are simply irrelevant. Seila Law may have a costly effect on governance, and new congressional mechanisms may curtail executive power even more than the for-cause removal provision ever did. But, the argument would go, if the Constitution requires a formal rule as to removal of single agency directors, any suboptimal second-order effects on institutional arrangements are beside the point. Perhaps, on this view, the Constitution mandates that the Court look at separation of powers from a static perspective. After all, the Court is not a body of political scientists, much less clairvoyant ones.

But it is strange to conclude that the Constitution mandates a clear answer here, much less an inflexible methodology. After all, the constitutional text is completely silent on removal and past cases do not obviously settle the question. The text that the president must “take Care that the Laws be faithfully executed” is far too vague to decide the case. Rather, when text and doctrine do not set out a clear answer, effects should be considered.12 And if a decision may lead to unintended consequences—especially ones that work against the intended goal—the Court should consider a more judicially modest approach. Such modesty may also be warranted if the consequences are deeply uncertain. The Constitution should not be understood to require a formalistic game of whack-a-mole. It should not require us to walk around blindfolded.

V. Conclusion

Seila Law may not lead to these unintended consequences, but it very well might. In restricting itself to an ex post, static conception of separation of powers, the Supreme Court showed no interest in examining this question. The Court cannot tell the future, but it should not be blind to the dynamic nature of separation of powers.

• 1This oft-quoted line is attributed to Justice Robert Jackson, but its most common formulation comes from Justice Arthur Goldberg’s 1963 decision in Kennedy v. Mendoza-Martinez. See Linda Greenhouse, The Nation; ‘Suicide Pact’, N.Y. Times (Sept. 22, 2002).
• 2Arguably, the rule is not entirely bright-line, as the Court explains that the CFPB’s budgetary independence and the length of the term of the director are also potentially problematic. But the opinion seems to reach its conclusion without relying on these elements as factors.
• 3The formalism/functionalism framing may often obscure more than it clarifies. See William N. Eskridge, Jr., Relationships Between Formalism and Functionalism in Separation of Powers Cases, 22 Harv. J.L. & Pub. Pol’y 21, 21 (1998) (noting that the formalist/functionalist dichotomy can be understood “no fewer than three different ways”). Consider, for example, the dueling opinions in Stern v. Marshall concerning how and to what extent Congress may assign decision-making power to non-Article III tribunals. The formalist majority opinion appears to arrive at a bright-line rule: a state-law counterclaim is not a public right and therefore cannot be heard in bankruptcy court. But Stern’s rule is more ambiguous than it appears, as it turns on how the Court applied complex doctrine to construe the term “public right” and how it understood the particular state-law counterclaim at issue.
• 4As then-Professor Kagan noted, many commentators expected the Clinton administration to curtail presidential control of the regulatory state—given longstanding Democratic criticisms of the Reagan-era practice—but in fact “the very opposite occurred.” Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2281 (2001).
• 5In fact, President Obama’s Treasury Department’s June 2009 proposal concerning the CFPB already proposed that the agency would be independent, but this may have been the result of discussion and compromise with institutional actors in Congress.
• 6Indeed, in part because the consequences of decisions concerning for-cause removal provisions in agency design are so unpredictable, Professor Aziz Huq has argued compellingly that it is a political question.
• 7A few presidents, when signing a bill with a legislative veto, sometimes issued a signing statement explaining their belief that the legislative veto was unconstitutional. A signing statement, however, has no formal legal significance. Either the president has signed the bill into law, or he has not. To give legal significance to these statements would be akin to a line item veto, which the Court has held unconstitutional. And giving legal significance to signing statements also deprives Congress of its right to override a presidential veto.  Of course, signing statements may still have a useful function for a president in the process of negotiating with Congress. Oddly, the Seila Law Court seems to suggest that veto threats or arguments in a signing statement may have some relevance to the legality or constitutionality of a feature of agency design, but it is not clear why this would be true.
• 8Here we encounter one of the many problems with formalism: labels often lack clear and consistent meaning—even when application of those labels is dispositive. The Chadha Court believed the same exact decision concerning suspension of deportation, if made by the attorney general, was “executive.” As Seila Law itself acknowledges, separation-of-powers doctrine has been plagued by inconsistent use of the terms “legislative” and “executive.” Nonetheless, formalist decisions—including Seila Law—continue to turn on insufficiently theorized definitions.
• 9See also,e.g., Peter L. Strauss, The Place of Agencies in Government: Separation of Powers and the Fourth Branch, 84 Colum. L. Rev. 573, 634–38 (1984) (critiquing the majority opinion in Chadha).
• 10Because Chadha invalidated an element of a careful interbranch arrangement, it implicated identical remedial difficulties to those discussed in Part II in the context of Seila Law. Like the Seila Law majority, Chief Justice Warren Burger’s majority opinion in Chadha simply severed the veto and ruled it unconstitutional. But in dissent, then-Justice William Rehnquist argued that the broad grant of power to the attorney general to suspend deportation proceedings was so interlocked with the legislative veto that if the latter fell, the former had to also—despite a savings clause. Justice Thomas took the same position in his Seila Law concurrence when he argued all of Dodd-Frank must fall, despite the severability clause.
• 11Notably, Congress is already considering amendments to the Federal Vacancies Reform Act to limit the president’s use of acting officials. The Accountability for Acting Officials Act, a bill proposed by Representative Katie Porter of California, would limit who may be appointed as an acting official, restrict how long an acting official may serve, and require each acting official to testify before Congress at regular interviews throughout their tenure.
• 12See Daniel A. Farber, Legal Pragmatism and the Constitution, 72 Minn. L. Rev. 1331 (1988) (arguing for more pragmatic approaches to constitutional questions); cf. William N. Eskridge & Philip P. Frickey, Statutory Interpretation as Practical Reasoning, 42 Stan. L. Rev. 321, 322 (1990) (contending that interpretation of statutes should be grounded on “practical reason”).