Agency Design and the Zero-Sum Argument

Essay — Volume 104, Issue 5

104 Va. L. Rev. Online 136
Download PDF

Zero-sum arguments are common in discussions of the administrative state. Such an argument was forcefully presented in Free Enterprise Fund, where the Court wrote, “In a system of checks and balances, ‘[p]ower abhors a vacuum,’ and one branch’s handicap is another’s strength.”[1] This zero-sum argument has gained considerable force in recent challenges to agency design. Under the zero-sum framework, one branch’s diminished control over the administrative state necessitates a gain to the other branches. Typically, the President’s loss is Congress’s gain, although this is not always the case.[2] Empowered by this baseline understanding, opponents of a particular structure argue that limitations on presidential control disrupt the separation of powers, impermissibly altering the balance required by the Constitution.


But the zero-sum argument is not the only one to appear in agency-design case law. In Free Enterprise Fund, for instance, the Supreme Court considered whether two layers of for-cause removal protection violated Article II’s Vesting Clause.[3] In finding the protection unconstitutional, the Court expressed concern that “the diffusion of power carries with it a diffusion of accountability.”[4] Insulating officers through two layers of removal protection “subverts . . . the public’s ability to pass judgment” on the President.[5] In other words, the agency’s design was impermissible because it limited political accountability. This concern exists without regard to a corresponding gain by Congress.

It is worth noting that these two concerns are not identical. In the first situation—commonly known as “aggrandizement”—power had passed from the President to Congress, compromising the balance between politically accountable actors. In the second situation—the context of “diffusion”—power had passed from the President to unaccountable hands, beyond the reach of the electorate altogether.[6] These two arguments are frequently conflated under the broad rubric of separation of powers,[7] and the zero-sum argument is the mechanism through which this happens. For example, the Fifth Circuit recently ignored diffusion while holding that the Federal Housing Finance Agency is unconstitutionally structured because “when one branch tries to impair the power of another, this upsets the co-equality of the branches and degrades the Constitution’s deliberate separation of powers.”[8]

This Essay contends that the zero-sum argument is misplaced and not required by existing law. A decrease in the executive’s control over the administrative state does not always correspond with a gain to Congress or the judiciary. Zero-sum rhetoric groups ideas that are best left distinct, and this rhetoric comes with a cost. First, it masks the new and powerful role that diffusion arguments are playing in agency design cases. Second, it leads to confusion about the issues at stake in assessing agency structure. Third, it prevents the courts from looking to more useful facts in understanding diffusion as an independent constitutional harm.

Stripping away zero-sum language reveals that many cases that purport to be about the separation of powers, in fact, are not—at least not in the way the label is frequently employed. The challenge for a modern court is assessing not the balance between constitutional actors, but the point at which federal power is exercised beyond the reach of any accountable actor at all.[9] The harm that comes from the former defect—aggrandizement—is distinct from the harm that comes from the latter—diffusion.

Since Free Enterprise Fund, lower courts have grappled with how to identify when diffusion rises to an unconstitutional level. This essay does not speak for or against this project. The specific question in Free Enterprise Fund, the removal power, has long been a topic of academic attention.[10] Likewise, the broader question of how best to understand the constitutional allocations of powers between branches of government has been well developed by many able scholars.[11] Instead, this Essay takes the law as it exists now and critically examines what it requires from lower courts tasked with applying it. This examination leads to two conclusions. First, the courts should look beyond the Supreme Court’s precedents involving aggrandizement to identify when diffusion becomes unconstitutional. Free Enterprise Fund requires the courts to draw a new constitutional line, and these precedents do not offer any guidance on where it is. Second, any attempt to draw this line will require a careful examination of how the precise statute in question affects political accountability as a factual matter. Judge Griffith’s recent opinion in the CFPB litigation, discussed below, demonstrates this analytical approach. In brief, moving beyond zero-sum arguments helps us find the right answers by identifying the right questions.

I. The Government is not Zero-Sum

Lawyers, scholars, and judges often employ zero-sum language to undergird larger arguments about the proper “balance of power” in the constitutional system.[12] While disagreeing about the proper balance, they all seem to share a premise that what one branch gains, another must lose. As this Part argues, “balance” is not a unitary concept. Since power is not zero-sum, balance can be upset by a limit on one constitutional actor (diffusion), even without a corresponding expansion in another (aggrandizement). These are discrete, and at times contradictory, harms.

This Part begins by briefly summarizing the Supreme Court’s case law on agency design, arguing that prior to Free Enterprise, diffusion did not play an outcome-determinative role.  In fact, cases from before 2011 fit neatly into three groups: (1) cases interpreting a specific structural provision of the Constitution, like the Appointments Clause, (2) cases that upheld the statute in question against a broader separation of powers challenge,[13] and (3) cases that invalidated a statute because the Court found that one branch aggrandized itself with the power of another.[14] To be sure, the court often suggested broader principles were at work, but these theories were not essential to resolving the cases.  This Part then explores how Free Enterprise Fund changed the Court’s agency design doctrine,[15] before turning to recent attempts to apply Free Enterprise Fund to financial regulatory institutions.

A. The Law Before Free Enterprise

For many years after the New Deal, the Court’s jurisprudence regarding legislative incursions on the executive’s control of government proceeded on two tracks, which the Supreme Court recognized.[16] On the first track, the Court interpreted the specific constitutional provisions that assign responsibility among branches of government. For instance, the Court invalidated attempts to place government officers in a manner not consistent with the Appointments Clause,[17] appoint officers in violation of the Recess Clause,[18] or pass laws in a manner that did not conform to the Presentment Clause.[19] In these cases, the Court simply discerned the meaning of the constitutional provision at issue.[20] When a specific structural provision was not implicated,[21] the court adopted a balancing approach,[22] asking if a law “prevents the Executive Branch from accomplishing its constitutionally assigned functions.”[23] The decisions along this second track fell into a predictable pattern. While some of the opinions employed zero-sum rhetoric,[24] the Court struck down only statutes that involved aggrandizement. In other words, agency design was unconstitutional if it put members of Congress (or their agents) in control of the administrative state or put executive function inside the legislative branch.

The line between the two tracks was not always clear—some opinions, for example, reach the same conclusion along both routes[25]—but the outcome was. Statutes could limit the President’s ability to control personnel or agency action, as long as another constitutional actor was not stepping in to fill the void. Currently, the statutes at large are full of organic statutes and general management laws that structure the President’s ability to manage the federal government.[26] These laws fit neatly into the Supreme Court’s then-existing doctrine, which permitted the administrative state to serve as the repository of managerial functions and to disperse power among different executive branch actors, but prevented inter-branch encroachment. Past Court holdings treated aggrandizement as more problematic than diffusion; it did not conflate them into a single zero-sum harm.

In this respect, the Court’s doctrine around agency independence was not unique. Other areas of administrative law support the premise that although agencies are within the executive branch, they are analytically separate from the President and the Congress. The analysis in Nixon v. General Services Administration is instructive. Under the Presidential Recording and Material Preservation Act, an executive agency (GSA) is required to take possession of presidential records and screen them for preservation. The case gave rise to a separation of powers challenge on the grounds that “the Act encroaches upon the Presidential prerogative to control internal operations of the Presidential office and therefore offends the autonomy of the Executive Branch.”[27] On the challenger’s view, this reflects “an impermissible interference by the Legislative Branch into matters inherently the business solely of the Executive Branch.”[28] After stating the applicable test, based on the ability to carry out “constitutionally assigned functions,” the Court rejected the claim, noting that

[i]t is therefore highly relevant that the Act provides for custody of materials in officials of the Executive Branch and that employees of that branch have access to the materials only for ‘lawful government use, subject to the Administrators regulations. For it is clearly less intrusive to place custody and screening of the materials within the Executive Branch itself than to have Congress or some outside agency perform the screening function.[29] 

This analysis requires a few assumptions. First, authority in the GSA is not the same as authority in the President. Second, authority in the GSA is not the same as authority in Congress. Finally, the proper way to frame the question is to look at how much the President is impaired by the agency’s authority, irrespective of Congress. In other words, assess diffusion as distinct from aggrandizement.

Stepping back, the law before Free Enterprise Fund is easily summarized. As a matter of agency design, one branch of government could not encroach on another. The Court acknowledged that some limitations on the President would be problematic in and of themselves,[30] but the Court did not draw a constitutional line.

B. The Law After Free Enterprise Fund

The legal landscape changed with Free Enterprise Fund v. PCAOB, where the Court found that two levels of for-cause removal protection from the President amounted to an unconstitutional insulation from executive authority.[31] This case was the first to invalidate a personnel restriction that did not infringe on a specific textual provision or directly increase congressional control.[32] Even Myers v. United States,[33] commonly understood to represent a robust view of presidential authority, involved a statute that gave Congress itself the authority to remove officers.[34]  Rather than proceeding along the two-track tradition, the Court instead found the scheme invalid under the Article II Vesting Clause.[35]

It is difficult to see the structure at issue as an aggrandizement of Congress. The case involved the provision of the Sarbanes–Oxley Act that sets up the Public Company Accounting Oversight Board (“PCAOB”), a body to oversee accounting firms that audit public companies. The Board was created to supervise, investigate, and sanction firms in this industry, under the oversight of the SEC. The members of the PCAOB are selected by the SEC, not Congress, to five-year terms and are protected from at-will removal by the SEC. This structure represents a fair amount of insulation from the President, but it puts the members no closer to Congress, which has no role in their selection, supervision, or removal.[36] The relevant harm is thus a far cry from the prior cases that either placed the legislative branch directly in charge of executive functions or made officers removable only with the consent of the Congress.

Instead of looking to Congress’s gain, then, the opinion rightly focused on the President’s loss of control. After reviewing the prior cases involving personnel independence,[37] all of which could be made to fit the two-track approach, the Court articulated a new boundary. As the majority understood the issue, two-layers of for-cause removal protection meant the President lacked the authority to “oversee the Board” or check the “dispersion of responsibility.”[38] While the Court concluded that the “Act’s restrictions are incompatible with the . . . separation of powers,”[39] the opinion does not speak to other branches. Instead, the real fear is that the legislation “reduce[d] the Chief Magistrate to a cajoler-in-chief.”[40] The opinion focused on the balance between the agency and the President, not the President and the other branches. Zero-sum rhetoric makes an appearance,[41] but there is no real suggestion that Congress was grabbing control of the accounting industry through the PCAOB.

While the Court referenced its aggrandizement decisions, its analysis was motivated by diffusion. As such, the case started the law down a new path. This is not to say that the Court ignored its precedents or fundamentally undermined the existing course of the law. To the contrary, many prior cases suggested that such a limit might exist,[42] and in Free Enterprise, the government itself admitted that constraints on the President’s removal authority could present constitutional defects.[43] On its own terms, the opinion suggests that it was the novelty of the agency design, not a change in the law, that drove the change in outcome.[44]

The opinion is not remarkable for its introduction of the diffusion argument. It is notable because the argument finally carried the day. Since then, the challenge has become drawing the new line that the diffusion rationale requires. With the removal power, where diffusion and the zero-sum argument point in the same direction, this challenge might not seem so difficult. In fact, that is likely the best way to understand the zero-sum argument in Free Enterprise Fund itself. The earlier decisions involving aggrandizement all limited the president without a corresponding reduction in congressional control, so it was appropriate to conclude that “Congress’ political power . . . necessarily increase[d] vis-à-vis the President.”[45] When litigants challenge an agency for its independence from both the President and Congress, however, diffusion and aggrandizement point in opposite directions. In these situations, the zero-sum argument loses its force and the Court’s task becomes more difficult.

C. The CFPB and FHFA Litigation

The financial industry would soon provide two prominent examples of agencies challenged because they are independent of both the White House and Congress. First, the Housing and Economic Recovery Act of 2008 created the Federal Housing Finance Agency to oversee Fannie Mae and Freddie Mac.[46] Several years later, the Dodd–Frank reform legislation created the Consumer Financial Protection Bureau (“CFPB”), consolidating formerly scattered authority in this area and providing new regulatory tools to the agency.[47] Given that Free Enterprise Fund suggested a willingness to accept longstanding regulators,[48] the novel structures of these two agencies were fertile ground to test the boundaries of permissible diffusion. These challenges have given rise to two major circuit court opinions, both of which demonstrate the trouble with zero-sum thinking. Specifically, litigants and judges have struggled with assessing features of agency design that limit both congressional and presidential control. Specifically, there is confusion over how to weigh the agencies’ independence from the budget process, which would otherwise afford Congress an opportunity each year to affect agency policy.

 In PHH Corp. v. CFPB, litigants brought a challenge to the CFPB’s design under the Vesting Clause of Article II.[49] As structured by Dodd–Frank, the agency enjoys significant independence from the President. The CFPB is headed by a single administrator, not a multi-member commission. This director is appointed to a five-year term, with for-cause removal protection during that tenure. The agency also enjoys other structural features that insulate it from political influence. For example, since it receives funding directly from the Federal Reserve, the CFPB does not depend on congressional largesse during the annual appropriations cycle.[50]

The litigation ultimately focuses on whether the removal provision, either standing alone or in combination with the budgetary process, is unconstitutional.[51] Advocates of the CFPB see the agency as nothing special, at least in a constitutional sense. Congress has long employed single administrators, for-cause removal protections, and funding outside the annual appropriations cycle. The combination, they claim, is no more problematic than any discrete part. To its detractors, however, the aggregation of these features in a single entity is both novel and threatening, pulling the agency well outside the normal push and pull of partisan politics. While the challengers prevailed in the initial hearing before the D.C. Circuit, the agency prevailed en banc.

If the Supreme Court is one day tasked with resolving the question, one thing should be clear: The zero-sum conception of constitutional power is wholly inapplicable to a structure like that of the CFPB. If all adjustments to the administrative state amount to different ways of distributing the pie among branches of government, as the zero-sum concept would require, then the features of the CFPB actually mitigate one another. If the government is zero-sum, the removal restrictions of the CFPB Administrator limit the President and aggrandize Congress. Likewise, the budget autonomy limits the Congress and necessarily aggrandizes the President. If the zero-sum logic is followed to its conclusion, an agency can avoid constitutional defect by simply applying additional limits to each branch in equal measure, a result that is clearly contradictory to the accountability rationale of Free Enterprise Fund

Given this implication, it is unsurprising that zero-sum language did not make a prominent appearance in the briefing. The suggestion that “power abhors a vacuum” is nowhere to be found. And the challengers even argue that Congress abdicated its own authority in the creation of the agency, a sort of anti-aggrandizement. As they saw it, the decision to place the operating budget outside the annual appropriations process deprives Congress of an “important check . . . over the CFPB,” while “limit[ing] [the] accountability to the President too.”[52] By arguing that this independence limited the authority of both branches, the petitioners tacitly accepted that a zero-sum argument did not help their case.

Nonetheless, the briefs filed in advance of the en banc rehearing suggest the parties did not distinguish between the types of harms reflected in the case law. Their citations demonstrate that the litigants are fighting over diffusion while talking about aggrandizement. The petitioner’s argument on the constitutional merits cites Chadha, Myers, Free Enterprise, Bowsher, Morrison, Humphrey’s Executor, Noel Canning, and Freytag. The only case involving a diffusion concern, Association of American Railroads, was cited for a different proposition.[53] In defending the statute, the CFPB followed a similar course, citing to Nixon, Humphrey’s Executor, Morrison, Noel Canning, Bowsher, Mistretta, and Free Enterprise.[54] As discussed in Part I, these cases (with the exception of Free Enterprise) have little to say about the line between appropriate independence and unconstitutional diffusion.

The FHFA litigation has similarly highlighted the difficulty of fitting budget independence into the zero-sum framework. Unlike the D.C. Circuit in PHH Corp., which ultimately upheld the CFPB, a Fifth Circuit panel applied Free Enterprise Fund to find that the FHFA was unconstitutionally structured.[55] In doing so, the Court engaged in a thorough analysis of both the case law on removal and the literature suggesting that removal is not the only source of agency independence.[56] Much of the opinion focuses on diffusion, noting the harm that emerges from having government actors too isolated from political accountability. Focusing on this harm leads the Court to correctly conclude that the question before them is one of degree: “Ultimately, ‘an agency’s practical degree of independence from presidential influence depends’ on the combined effect of these (sometimes mutually reinforcing) structural features.”[57] This is exactly the sort of analysis that diffusion as a separate harm requires.

The Court’s reasoning is weakened, however, by its need to square this fact-intensive and practical analysis with zero-sum rhetoric around agency design. For instance, the opinion quotes Free Enterprise Fund for the proposition that “excessive insulation allows Congress to accumulate power for itself.”[58] On this view, Congress’s “control over the salary, duties, and even existence of executive offices” goes unchecked when an agency is isolated from the President.[59] The problem with this type of argument becomes plain when the court later discusses the FHFA’s funding source. Much like the CFPB, the FHFA receives funding outside the annual appropriations cycle.[60] As previously mentioned, the CFPB litigants saw this feature as weakening Congress’s control. In forcing its analysis into the zero-sum box, the Fifth Circuit argues that the FHFA’s funding source weakens the President: “By placing an agency outside the normal appropriations process, the President loses ‘leverage’ over the agency’s activities. . . . The FHFA stands outside the budget . . . and is therefore immune from presidential control.”[61]

Compare this argument to the earlier assertion that agency independence empowers Congress through its power to control salaries, duties, and offices. These levers are the product of Congress’s power to enact statutes. So is the annual appropriations process. Both require bicameralism and presentment. Both can be vetoed, and both could see a veto overridden. Yet the Fifth Circuit frames the legislative power to set duties, salaries, and offices (by statute) as a reason for congressional dominance, while framing the absence of the power to set funding levels (by statute) as an inhibition on the President. The attempt to put the budget provision into the zero-sum framework of earlier cases is understandable. Once the unique contribution of Free Enterprise Fund is recognized, however, it is unnecessary and detracts from the core accountability concern that is well developed in other parts of the opinion. 

The CFPB and FHFA cases both confronted agencies free from the annual appropriations cycle. As a practical matter, removing an agency from the pressures of annual appropriations inhibits both branches. The power of the purse is one of Congress’s primary checks on the actions of the executive branch.[62] Similarly, the President, acting through the Office of Management and Budget, exerts tremendous pressure on agencies through the budget request process.[63] Although this form of independence could be relevant to the diffusion of power, it does not tell us much about the balance of power between the branches. But this reality is difficult to square with the reliance on the zero-sum argument in earlier cases, which sees the two as one in the same. In the CFPB case, the litigants accepted as much and focused solely on diffusion. In the FHFA litigation, however, the court examined the feature in a way that fit the zero-sum framework, grouping its analysis under a single “separation of powers” idea.

II. Understanding Diffusion as a Separate Harm

The theory of the Vesting Clause found in Free Enterprise Fund is applicable to any feature of agency design that could plausibly limit political accountability. It can therefore be deployed to a range of statutes affecting financing, litigating authority, officer qualifications, and direct reporting, among others.[64] While some methodological approaches might lead to these statutes being invalidated wholesale, the Court’s approach in Free Enterprise Fund was more fact intensive and functional, as the lower courts have recognized.[65] Moreover, while interpretive techniques like a “presumption against novelty” in agency design may help the Court reach decisions involving new structures, they tell us little about the ones we already have.[66]

This Part briefly notes two potential analytical approaches, both of which move beyond the zero-sum framework. The first is to consider other areas of the law that implicate a diffusion harm, such as privatization and federalism. While these areas may share certain qualities with agency design, they ultimately do not offer much guidance for lower courts. The second is to develop an agency-specific understanding of diffusion, focused on political accountability. This option requires the courts to think more about what accountability means and how it is measured; a project that may prove difficult to square with the current focus of the law on personnel alone. This approach requires that lower courts engage in a more fact-bound assessment of statutes that structure agencies until the Supreme Court provides more complete guidance, especially because other areas of law do not provide workable standards.

A. Analogies Outside the Separation of Powers

The diffusion harm in Free Enterprise is grounded in political accountability concerns. The Supreme Court has explored political accountability in both its federalism decisions and its cases involving the delegation of power to private actors, so these are natural starting points for trying to develop a workable constitutional limit on diffusion.

Modern American federalism is characterized by sovereigns acting in overlapping domains, which allow for cooperation and contestation.[67] The current doctrine does not attempt to carve out separate spheres of federal and state action. Instead, it ensures the federal government cannot avoid accountability for its actions by commandeering the states. Under New York[68] and Printz[69]—the two most significant cases in this area—the federal government cannot use the states to enforce its policies, at least not directly. The role of political accountability in these cases is therefore straightforward and defensible. Since the Constitution divides power between two elected sovereigns, one cannot conscript the other and thereby distort the public’s assessment of credit and blame.[70] The Court recently applied this logic to coercion, finding that indirect mandates in federal grants can have the same distorting effect in extreme cases.[71] In the context of federal–state relations, therefore, how a policy is implemented matters.[72] The federal government may have the power to impose its will on local subjects, but it must do so directly and on fair terms.

Political accountability is also central in a line of cases involving delegations of power to private actors.[73] When Congress delegates discretion to public actors within the executive branch, that practice aligns with the structure of the aggrandizement cases; one branch loses the ability to fill gaps in the law while the other branch gains that power. If there is a constitutional defect, it is the transfer of legislative power from Congress to the President.[74] The harm that stems from delegating power to private actors, however, is different and warrants greater scrutiny.[75] Notably, this scrutiny is framed in terms of accountability. As the D.C. Circuit recently stated, “delegating the government’s powers to private parties saps our political system of democratic accountability. This threat is particularly dangerous where both Congress and the Executive can deflect blame for unpopular policies by attributing them to the choices of a private entity.”[76] In other words, the harm is not the balance between political actors, but rather shifting the blame outside the government altogether. To support this conclusion, the opinion cited not only to prior cases involving delegations to private actors, such as the New Deal–era Carter Coal decision, but also to the federalism cases.[77] This makes sense to the extent that the cases share a common premise—that federal actors have empowered (or coerced) actors outside their control to implement national policy, thus compromising the public’s ability to hold the proper official accountable. This accountability concern is central to the limits on diffusion identified in both areas.

The value of these cases to questions of agency design is limited. While these areas are concerned with maintaining a distinction (federal/state, public/private), the Constitution does not reflect a similar concern for administration. To the contrary, political control over administration is explicitly designed to make many individuals accountable for any given action.[78] Branches share responsibility for appointing personnel, financing government operations, and constraining incursions by administrators on individual liberty. Under a structure where both political branches are supposed to share blame and credit for federal action, what work is political accountability doing? Moreover, the privatization cases suggest that diffusion concerns are heightened for private actors, even relative to independent agencies.[79] In brief, whereas the other two areas can look to political accountability as a way to draw a useful and enforceable distinction, it is not clear that idea can do the same work here. Instead, the courts will likely need to try something new; they will need to develop a framework for understanding the unique relationship between agency design and political accountability. 

B. Independent Harms in the Administrative State

An agency’s design clearly has some effect on its political responsiveness.[80] While Free Enterprise Fund identifies one point at which this effect is too limiting, it does little to provide a way forward.  To apply its holding, the Court must accept two premises. To note these premises is not to resolve them, but it is a necessary step in developing a workable doctrine of diffusion.

First, the extent to which a particular agency design diffuses power presents an empirical and measurable question: How much does the structure actually limit political accountability?[81] In measuring that harm, the zero-sum argument, which frames the question as one of the separation of powers, offers little guidance. As Adrian Vermuele has noted, the separation of powers is not an unassailable “idol”: “[I]t is not obvious that what are, after all, merely institutional arrangements could ever be the sort of things that could be ‘contaminated,’ even in principle. The language of the sacred is simply misplaced as to such highly contingent matters of institutional design.”[82]

Instead, the courts must examine a core principle—democratic accountability—as a factual matter. Ultimately, this question is not a “vague and slippery” search for balance between the branches.[83] It requires a factual assessment of what the terms of the restriction on presidential authority mean.[84] Labels like “for-cause removal” or “financial independence” are too broad, and courts must parse more critically the provisions before them. As the financial regulators demonstrate, not all features of agency design will necessarily empower one branch over the other. This time Congress chose to limit its control through annual appropriations. Next time it might involve an agency free of congressional subpoenas or oversight hearings. In either event, the courts will be asked to assess the effect of such a feature, even though there is no aggrandizement.

The D.C. Circuit’s recent assessment of the CFPB demonstrates how such an analysis might look. The majority[85] and dissents[86] offer thorough and persuasive accounts of how existing case law speaks to the permissibility of the CFPB’s design. While disagreeing on the outcome, the majority and dissent ask general questions about how the agency’s design fits into existing precedent. As discussed in Part I, however, this precedent does not tell us much about the dispositive question under Free Enterprise Fund: at what point does the diffusion of power to an agency, and the attendant loss of political accountability, become unconstitutional?

Answering requires knowing the actual limits placed on the President. And for that, Judge Griffith’s solo concurrence on the removal question makes an important contribution to the debate. He explains the reason for his separate opinion at the outset: “My colleagues debate whether the agency’s single-Director structure impermissibly interferes with the President’s ability to supervise the Executive Branch. But to make sense of that inquiry, we must first answer a more fundamental question: How difficult is it for the President to remove the Director?”[87] Judge Griffith goes on to describe the language of the removal provision—allowing for removal in cases of “inefficiency, neglect of duty, or malfeasance in office”—and finds that it allows for removal based on policy disagreement. Judge Wilkins’s concurring opinion employs a similar inquiry, ultimately finding Judge Griffith’s conclusion contestable.[88] But the result is not as important as the question, which rejects the idea that all removal restrictions should be treated equally or raise the same constitutional concerns. As Judge Griffith put it, “agency independence is not a binary but rather a matter of degree.”[89]

The new diffusion doctrine’s second necessary premise is that individual limits on presidential authority cannot be viewed in isolation. As the Fifth Circuit noted, the Supreme Court’s holding in Free Enterprise Fund requires the court to “look at the aggregate effect of the insulating mechanisms to determine whether an agency is excessively insulated.”[90] While the court then went on to consider a range of statutory provisions applicable to the FHFA, there is reason to think that the inquiry should have been even broader. The insulation of a given agency can turn on its relationship to other agencies, state governments, and political actors in Congress and in private life.[91] Even though these non-traditional features of independence are hard to measure and not readily susceptible to easy labels (like “for-cause removal”), they are powerful.

Analyzing diffusion requires a fact-intensive assessment of the limits on presidential control, considered in their full legal and political context. If this seems like a functional inquiry, it is because the law requires one. The court in Free Enterprise Fund drew a line in the sand at two layers of for-cause removal from the President. The defect in that statutory arrangement is clearly distinct from those in prior aggrandizement cases. This diffusion harm is rooted in a loss of political accountability, which only a careful factual examination can measure.


*    *    *

An agency’s design can implicate two discrete harms. The first—aggrandizement—involves one branch encroaching on the domain of another. This encroachment upsets the separation of powers by placing the responsibilities of one constitutional actor under the control of another. The second—diffusion—does not implicate the balance of power between the branches; it instead places the exercise of federal power beyond the reach of an accountable official. The reliance on zero-sum rhetoric masks this distinction. In doing so, courts attempt to fit facts into a theoretical framework ill-suited to answer the question required by current law. Instead of relying on the unitary concept of agency independence expressed in the zero-sum argument, courts should take on the difficult task of defining political accountability and setting the limit at which it has been impermissibly diffused. These tasks may be unfamiliar, but under current law they are unavoidable.


    [1] Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 561 U.S. 477, 500 (2010) (citing Judge Kavanaugh’s opinion for the D.C. Circuit below); see also Mistretta v. United States, 488 U.S. 361, 382 (1989) (“We have not hesitated to strike down provisions of law that . . . undermine the authority and independence of one or another coordinate Branch.”); Nixon v. Gen. Servs. Admin., 433 U.S. 425, 443 (1977) (“Rather, in determining whether the Act disrupts the proper balance between the coordinate branches, the proper inquiry focuses on the extent to which it prevents the Executive Branch from accomplishing its constitutionally assigned functions.”).

    [2] See, e.g., Stern v. Marshall, 564 U.S. 462 (2011) (finding that an adjudication by the bankruptcy courts violated Article III); Indus. Union Dept., AFL-CIO v. Am. Petroleum Inst., 448 U.S. 607, 685–88 (1980) (Rehnquist, J. concurring in the judgment) (arguing that the OSH Act of 1970 violated the non-delegation doctrine).

    [3] 561 U.S. 477.

    [4] Id. at 497.

    [5] Id. at 498.

    [6] The labels “aggrandizement” and “diffusion” for these two defects are not new. The first is commonly used in the court’s opinions, see, e.g., Buckley v. Valeo, 424 U.S. 1, 122 (1976), while the latter has been invoked both in Free Enterprise Fund itself and subsequent scholarship on the case. See Edward H. Stiglitz, Unitary Innovations and Political Accountability, 99 Cornell L. Rev. 1133, 1133 (2014).

    [7] See John F. Manning, Separation of Powers as Ordinary Interpretation, 124 Harv. L. Rev. 1939, 1961–71 (2011) (describing the formalist concern with “encroachment,” whereby specific limits on presidential control are invalidated because of the general separation of powers principle).

    [8] Collins v. Mnuchin, 896 F.3d 640, 659 (5th Cir. 2018).

    [9] While the court in Free Enterprise used “diffusion” to signal a particular kind of constitutional defect, the diffusion of power in many executive branch actors can arguably mitigate, not create, constitutional defects. See M. Elizabeth Magill, Beyond Powers and Branches in Separation of Powers Law, 150 U. Pa. L. Rev. 603, 605–06 (2001); Manning supra note 7, at 1947 (“[R]ather than embracing an overarching separation of powers principle, the [Constitution] . . . reflects countless context-specific choices about how to assign, structure, divide, blend, and balance federal power.”).

    [10] Compare Neomi Rao, Removal: Necessary and Sufficient for Presidential Control, 65 Ala. L. Rev. 1205 (2014) (arguing that presidential removal of officers is constitutionally required), with Adrian Vermeule, Conventions of Agency Independence, 113 Colum. L. Rev. 1163 (2013) (arguing that removal restrictions, both in statute and prevailing convention, are not critical to determining constitutional questions of agency independence). See also Aziq Z. Huq, Removal as a Political Question, 65 Stan. L. Rev. 1 (2013) (arguing that courts should consider this entire area non-justiciable under the political question doctrine); Saikrishna Prakash, Removal and Tenure in Office, 92 Va. L. Rev. 1779 (2006) (presenting a theory of the removal power distributed across all three branches of government).

    [11] An abbreviated survey includes Steven G. Calabresi & Christopher S. Yoo, Remove Morrison v. Olson, 62 Vand. L. Rev. En Banc 103, 107-11 (2009) (providing a concise presentation of the Unitary Executive theory, which the authors presented more fully elsewhere); Peter L. Strauss, The Place of Agencies in Government: Separation of Powers and the Fourth Branch, 84 Colum. L. Rev. 573, 581 (1984) (arguing for a “framework for understanding the scope of Congress’s authority to structure American government that . . . require[s] that those who do the work of law-administration have significant relationships with [Congress and the President]”); Manning, supra note 7 (arguing for a less general approach to separation of powers questions that would allow for more flexibility in areas where the Constitution is less clear about institutional arrangements); Vermeule, supra note 10, at 1231 (arguing that separation of powers law should account for “unwritten rules of the game, or conventions,” which lie “[b]etween ‘politics’ on the one hand and formal written law on the other”).

    [12] Eric A. Posner, Balance-of-Power Arguments, the Structural Constitution, and the Problem of Executive “Underenforcement”, 164 U. Pa. L. Rev. 1677, 1678 (2016).

    [13] See, e.g., Loving v. United States, 517 U.S. 748 (1996) (finding that Congress can delegate authority to the President to define aggravating factors for military capital cases); Mistretta v. United States, 488 U.S. 361 (1989) (upholding a delegation to the judiciary to promulgate the Sentencing Guidelines); Nixon v. Gen. Servs. Admin, 433 U.S. 425 (1977) (upholding the Presidential Recordings and Materials Preservation Act).

    [14] See, e.g., MWAA v. Citizens For The Abatement of Aircraft Noise, Inc., 501 U.S. 252 (1991).

    [15] The notion that Free Enterprise Fund represents a fundamental break from past practice is found in other scholarship. See Huq, supra note 10, at 14.

    [16] See MWAA, 501 U.S. at 274 (“To forestall the danger of encroachment ‘beyond the legislative sphere,’ the Constitution imposes two basic and related constraints on the Congress. It may not ‘invest itself or its Members with either executive power or judicial power. And, when it exercises its legislative power, it must follow the ‘single, finely wrought and exhaustively considered, procedures’ specified in Article I.” (citations omitted)); see also Public Citizen v. U.S. Dep’t of Just., 491 U.S. 440, 484–86 (1989) (Kennedy, J. concurring in the judgment) (“In some of our more recent cases involving the powers and prerogatives of the President, we have employed something of a balancing approach, asking whether the statute at issue prevents the President from accomplishing his constitutionally assigned functions. . . . In a line of cases of equal weight and authority, however, where the Constitution by explicit text commits the power at issue to the exclusive control of the President, we have refused to tolerate any intrusion by the Legislative Branch. . . . The justification for our refusal to apply a balancing test in these cases, though not always made explicit, is clear enough. Where a power has been committed to a particular Branch of the Government in the text of the Constitution, the balance already has been struck by the Constitution itself.” (citations and emphasis omitted)).

    [17] Buckley v. Valeo, 424 U.S. 1 (1976).

    [18] Noel Canning v. NLRB, 134 S. Ct. 2550 (2014).

    [19] INS v. Chadha, 462 U.S. 919 (1983).

    [20] See, e.g., Freytag v. Comm’r of Internal Revenue, 501 U.S. 868 (1991) (defining “officers” for the purpose of applying the analysis of Buckley).

    [21] This Essay assumes that the Vesting Clauses themselves are not a specific structural provision, at least not one comparable to the Appointments Clause. This assumption is based on the Court’s reluctance to view the Vesting Clauses as a source of rigid structural rules. As discussed in Section I.B, even recent decisions based on the Vesting Clause, like Free Enterprise Fund, did not import specific rules through the Vesting Clause, instead using the provision as the launching point for a more balanced inquiry involving political accountability.

     [22] See, e.g., MWAA v. Citizens for the Abatement of Aircraft Noise, 501 U.S. 252 (1991).

    [23] Nixon v. Gen. Servs. Admin., 433 U.S. 425, 443 (1977).

    [24] See, e.g., Bowsher v. Synar, 478 U.S. 714, 721–22 (1986) (explicitly mentioning the separation of powers and diffusion in a case involving aggrandizement); see also Loving v. United States, 517 U.S. 748, 757 (1996) (“Even when a branch does not arrogate power to itself . . . the separation-of-powers doctrine requires that a branch not impair another in the performance of its constitutional duties.”).

    [25] See MWAA, 501 U.S. at 277, n. 23 (finding that because the scheme was invalid as aggrandizement, potential objections based on specific textual provisions were unresolved); see also Bowsher, 478 U.S. 714. While the majority in Bowsher invalidated the scheme on the grounds of congressional aggrandizement, Justice Stevens would have reached the same conclusion by finding the scheme a violation of Article I’s requirement of bicameralism and presentment. Id. at 737 (Stevens, J. concurring in the judgment).

    [26] See, e.g., Federal Advisory Committee Act, Pub. L. No. 92-463 (1972) (codified at 5 U.S.C. Appendix); 12 U.S.C. § 250 (2012); see also Kirti Datla & Richard L. Revesz, Deconstructing Independent Agencies (and Executive Agencies), 98 Cornell L. Rev. 769, 784-812 (2013) (surveying statutory provisions).

    [27] Nixon, 433 U.S. at 439–40 (1977).

    [28] Id. at 440.

    [29] Id. at 443–44. Earlier in the opinion, the Court noted that the Administrator of GSA is appointed by the President and the staff who conduct the record review are executive employees, both of which the Court saw as relevant to the constitutional question. Id. at 441.

    [30] This lay of the land is summarized in Morrison v. Olson, 487 U.S. 654, 685–97 (1988). By framing the Court’s holding in Myers v. United States as involving an aggrandizement concern, the Court in Morrison fit the case into the existing framework, even though the language of the opinion extended well beyond that rationale. Id. at 686–87 (discussing Myers). Significant portions of Morrison have been challenged by later cases, but Free Enterprise Fund did not directly conflict with Morrison’s holding on the removal question and its vitality remains an open question.

    [31] Free Enter. Fund v. PCAOB, 561 U.S. 477 (2010).

     [32] At the time of the opinion, there was uncertainty as to whether the holding would work a major change in separation of powers law, or was instead a more incremental or “boundary enforcing” decision. See, e.g., Richard H. Pildes, Free Enterprise Fund, Boundary-Enforcing Decisions, and the Unitary Executive Branch Theory of Government Administration, 6 Duke J. of Const. L. & Pub. Pol’y 1, 9 (2010).

    [33] 272 U.S. 52 (1926).

    [34] This had been the previous rationale for harmonizing the case with later, more permissive holdings. See Morrison, 487 U.S. at 686 (1988) (“Unlike both Bowsher and Myers, this case does not involve an attempt by Congress itself to gain a role in the removal of executive officials. . . .”). The majority opinion in PHH Corp. also distinguished the case on these grounds. PHH Corp. v. CFPB, 881 F.3d 75, 78 (D.C. Cir. 2018) (en banc).

    [35] Free Enter. Fund, 561 U.S. at 484.

    [36] See id. at 485.

    [37] All of the cases cited on the personnel question either involved aggrandizement (Myers, Bowsher) or resulted in the agency design being upheld (Humphreys, Perkins, Morrison).

    [38] Free Enter. Fund, 561 U.S. at 495–98.

    [39] Id. at 498.

    [40] Id. at 502.

    [41] Id. at 500.

    [42] See, e.g., Nixon v. Gen. Servs. Admin., 433 U.S. 425, 443 (1977).

    [43] Free Enter. Fund, 561 U.S. at 502 (“The United States concedes that some constraints on the removal of inferior executive officers might violate the Constitution.”).

    [44] See id. at 496.

    [45] In re Sealed Case, 838 F.2d 476, 508 (D.C. Cir. 1988). Judge Kavanaugh discussed the relationship between the two at length in his dissent in Free Enterprise Fund, which argued that the PCAOB was unconstitutional. See Free Enter. Fund v. PCAOB, 537 F.3d 667, 694 n.4 (D.C. Cir. 2008).

    [46] Housing and Economic Recovery Act of 2008, Pub. L. No. 110-289, § 1101, 122 Stat. 2654 (2008).

    [47] Dodd–Frank Wall Street Reform and Consumer Protection Act, Pub. L. No. 111-203, Title X, 124 Stat. 1383 (2010).

    [48] See Free Enter. Fund, 561 U.S. at 496, 505–06.

    [49] PHH Corp. v. CFPB, 839 F.3d 1 (D.C. Cir. 2016) (order vacated, rehearing en banc granted Feb. 16, 2017).

    [50] PHH Corp. v. CFPB, 881 F.3d 75, 81 (D.C. Cir. 2018) (en banc). The agency is also empowered to communicate to Congress without White House approval. See 12 U.S.C. § 250.

    [51] PHH Corp., 881 F.3d at 92–101.

    [52] Opening En Banc Brief for Petitioners at *26–27, PHH Corp. v. CFPB, 881 F.3d 75 (D.C. Cir. 2018) (en banc)  (No. 15-1177), 2017 WL 947733.

    [53] Id. at *28.

    [54] Brief on Rehearing en Banc of Respondent at *17–32, PHH Corp. v. CFPB, 881 F.3d 75 (D.C. Cir. 2018) (en banc) (No. 15-1177), 2017 WL 1196119.

    [55] Collins v. Mnuchin, 896 F.3d 640, 640 (5th Cir. 2018).

    [56] Id. at 660–61.

    [57] Id. at 661.

    [58] Id. at 661.

    [59] Id. at 662.

    [60] Id. at 667–69.

    [61] Id. at 669. (emphasis added).

    [62] See generally Josh Chafetz, Congress’s Constitution: Legislative Authority and the Separation of Powers 45–77 (2017) (discussing the power of the purse).

    [63] See generally Eloise Pasachoff, The President’s Budget as a Source of Policy Control, 125 Yale L.J. 2182 (2016).

    [64] See Datla & Revesz, supra note 26, at 784–812.

    [65] See Collins v. Mnuchin, 896 F.3d 640, 664 (5th Cir. 2018).

    [66] Professor Leah Litman’s recent work makes a persuasive case against such a presumption in constitutional law. Leah M. Litman, Debunking Antinovelty, 66 Duke L.J. 1407 (2017). The phenomenon can be observed in many Supreme Court and appellate court opinions that would strike down legislation on separation of powers grounds. See Bank Markazi v. Peterson, 136 S. Ct. 1310, 1333 (2016) (Roberts, C.J.  dissenting); PHH Corp. v. CFPB, 839 F.3d 1, 7 (D.C. Cir. 2016); Assoc. of Am. Rrs. v. U.S. Dep’t of Transportation, 721 F.3d 666, 673 (D.C. Cir. 2013).

    [67] See generally Jessica Bulman-Pozen & Heather K. Gerken, Uncooperative Federalism, 118 Yale L.J. 1256, 1260–65 (2009) (describing the prevailing theories of federalism).

    [68] New York v. United States, 505 U.S. 144, 182–83 (1992) (discussing the possibility of “shifting responsibility” to avoid accountability).

    [69] Printz v. United States, 521 U.S. 898, 929–31 (1997).

    [70] Id.

    [71] NFIB v. Sebelius, 567 U.S. 519 (2012). The opinion was not framed in Tenth Amendment terms, but did cite to these cases as a means of limiting Congress’s Article I power to tax and spend. Id. at 559. The reach of this decision is still largely undetermined. See, e.g., Eloise Pasachoff, Conditional Spending After NFIB v. Sebelius: The Example of Federal Education Law, 62 Am. U. L. Rev. 557 (2013).

    [72] To see this point in action, consider that the majorities in both New York and NFIB rejected the notion that the laws in question were appropriate because a more aggressive exercise of power would be appropriate—the “greater includes the lesser” objection. See NFIB, 567 U.S. at 624 (2012) (Ginsburg, J., concurring in part, concurring in the judgment, and dissenting in part) (“Congress could have recalled the existing legislation, and replaced it with a new law making Medicaid as embracive of the poor as Congress chose. The question posed by the 2010 Medicaid expansion, then, is essentially this: To cover a notably larger population, must Congress take the repeal/reenact route, or may it achieve the same result by amending existing law?”); Printz, 521 U.S. at 959 (1997) (Stevens, J. dissenting) (“Perversely, the majority’s rule seems more likely to damage than to preserve the safeguards against tyranny provided by the existence of vital state governments. By limiting the ability of the Federal Government to enlist state officials in the implementation of its programs, the Court creates incentives for the National Government to aggrandize itself. In the name of State’s rights, the majority would have the Federal Government create vast national bureaucracies to implement its policies.”). This objection does not carry force when the concern is political accountability between the state and federal governments, not simply the substantive limits of Congress’s power. 

    [73] In practice, there are not many cases in this area because courts have long accepted that privatization, even of significant tasks, does not automatically equate to a private exercise of governmental power. See Jon D. Michaels, Constitutional Coup: Privatization’s Threat to the American Republic 125–26 (2017) (“[I]n in the absence of clear, prohibitory language, courts have largely continued giving privatization a free pass. Specifically, courts have generally declined to treat contractors, deputies, and the like as the true recipients of delegated powers—and thus subject to the doctrinal bar on private delegations.”).

    [74] Or possibly the judiciary. See Mistretta v. United States, 488 U.S. 366 (1989).

     [76] Assoc. of Am. Rrs. v. U.S. Dep’t of Transportation, 721 F.3d 666, 670–71 (D.C. Circ. 2013) (“Even an intelligible principle cannot rescue a statute empowering private parties to wield regulatory authority.”), rev’d Dep’t of Trans. v. Assoc. of Am. Rrs., 135 S. Ct. 1225, 1231–33 (2015).

    [76] Assoc. of Am. Rrs., 721 F.3d at 675.

    [77] Id. (citing New York v. United States, 505 U.S. 144 (1992)).

    [78] See generally M. Elizabeth Magill, The Real Separation in the Separation of Powers Law, 86 Va. L. Rev. 1127 (2000) (describing this tension).

    [79] Dep’t of Trans., 135 S. Ct. at 1231–32 (noting that private actors are distinguished by a profit motive, in addition to federal control).

    [80] See generally Rachel E. Barkow, Insulating Agencies: Avoiding Capture Through Institutional Design, 89 Tex. L. Rev. 15 (2010). 

    [81] See, e.g., Stiglitz, supra note 6, at 1137–38 (applying such an empirical analysis to the legislative veto mechanism). There is not widespread agreement that this is the proper way to understand the Vesting Clause of Article II. See, e.g., Rao, supra note 10, at 1209–10 (“The recognition that ‘independence’ has an uncertain and unpredictable connection to presidential control and that administration depends largely on political factors makes it all the more important to revisit the constitutional framework of administration and to establish the boundaries for presidential control.”).

    [82] Adrian Vermuele, Law’s Abnegation: From Law’s Empire to the Administrative State 71 (2016).

    [83] Posner, supra note 12, at 1714. As an example, “[t]he reason we should care about constraints on the removal power is not that those constraints upset some balance between Congress and the President. The reason is that those constraints may improve or worsen the performance of the bureaucracy. To determine whether they do, one must consider the particular body in question and ask why the constraints might be useful or harmful.”

      [84] Much of Free Enterprise Fund’s dissent focused exactly on these question, looking to whether the for-cause removal provision presented any discernible limit on the President in practice. Free Enter. Fund v. PCAOB, 561 U.S. 477, 525–26 (2010) (Breyer, J. dissenting).

    [85] PHH Corp. v. CFPB, 881 F.3d 75, 84–91 (D.C. Cir. 2018) (en banc). In all, the case resulted in seven different opinions, addressing different constitutional and statutory questions.

    [86] Id. at 137, 146–48 (Henderson, J. dissenting) (discussing the removal from the annual appropriations process); id. at 188 (Kavanaugh, J. dissenting). Judge Kavanaugh focused on the removal question, consistent with the Supreme Court’s prior cases.

    [87] Id. at 124 (Griffith, J. concurring).

    [88] Id. at 122–23 (Wilkins, J. concurring).

    [89] Id. at 136 (Griffith, J. concurring).

    [90] Collins v. Mnuchin, 896 F.3d 640, 664 (5th Cir. 2018).

    [91] Barkow, supra note 80, at 49–63.

Click on a link below to access the full text of this article. These are third-party content providers and may require a separate subscription for access.

  Volume 104 / Issue 5  

Agency Design and the Zero-Sum Argument

By Daniel Richardson
104 Va. L. Rev. Online 136

A Tribute to Gordon Hylton

By Risa L. Goluboff
104 Va. L. Rev. 843

Subsidizing Segregation

By Joy Milligan
104 Va. L. Rev. 847

The Writ-Of-Erasure Fallacy

By Jonathan F. Mitchell
104 Va. L. Rev. 933