Regulatory Pluralism: An Empirical Analysis of Decision Making in Administrative Law

Epstein, Daniel

Introduction

Administrative procedures constrain regulatory investigations.¹ But such procedures are not self-executing.² Regulated parties cannot benefit from administrative procedures unless and until their counsel makes use of such provisions before an agency or in court. Yet even though lawyers may choose which procedures may benefit a client in response to an administrative inquiry, lawyers, particularly those who organize through large law firms to serve global corporations, face institutional constraints that bias against identifying—or present financial disincentives or relational risks vis-à-vis regulators that lead them to ignore—certain procedural safeguards in the law. This institutional constraint derives from the reality that those industry clients choose regulatory counsel who maintain relationships with a given regulator and who will advise them as to how a would-be regulator or a court will interpret the law. Whether a law is likely to be enforced as a policy matter is distinct from whether a given economic product or service is prohibited or permitted by law, particularly in an environment where regulators investigate to determine whether jurisdiction exists. Industry players, therefore, price this discretion by hiring lawyers who make conservative predictions about what a regulator might do or what a court might conclude.³ Entrepreneurs or small businesses in novel sectors unspecified by current regulatory jurisdiction often cannot afford regulatory counsel nor is compliance with the conservative prediction reasonable for companies facing barriers to market entry or growth. Moreover, regulatory counsel, whose fidelity will be to those clients providing repeat fees, will decline to represent those clients who are willing to evade compliance with their conservative predictions, for any such engagement threatens relationships with both incumbent clients and regulators. These realities present a federal regulatory state institutionally biased against small entrepreneurs.

In 2019, former President Trump sought to resolve the disproportionate harms administrative inquiries posed to small businesses and to those businesses that cannot afford large law firm counsel by restating extant public rights laws to identify available yet underutilized administrative procedures. Executive Order Number 13892 and Executive Order Number 13924 articulate statutory principles that, if subject to judicial review, would prohibit agencies from establishing policy through pre-enforcement investigations or adjudication without prior notice of the agency’s authority to do so.⁴ The big idea was to ensure nonpublic pre-enforcement inquiries by agencies complied with procedurally established due process protections even absent a petition by a well-resourced regulated party. Here, small businesses, notwithstanding their inability to afford expert counsel, would reap the benefits of administrative procedures. On January 20, 2021, and February 24, 2021, President Biden rescinded Executive Order Numbers 13892 and 13924, respectively.⁵

In this Article, I identify a number of procedural safeguards that are often overlooked in regulatory litigation against the bureaucracy. In Part I, I outline those “due process” procedures that constrain agency investigations yet are both underexamined by scholars and underutilized by practitioners. These include the Paperwork Reduction Act (PRA), which regulates general investigations as “information collections” and the original public rulemaking requirements of the 1946 Administrative Procedure Act (APA) now codified at 5 U.S.C. § 552(a)(1)(D). This presents a puzzle given the substantial acceptance by both political scientists and administrative law scholars of the theory that administrative procedures are the primary mechanism through which Congress ensures oversight over the bureaucracy.⁶ I suspect that for certain litigants, confronting the uncertainties and risks in judicial agenda setting concerning regulation (i.e., judicial discretion and deference) is necessary for their survival. But market incumbents can absorb the cost of pre-enforcement settlements and benefit when the absence of rules harms would-be competitors, disincentivizing business-side lawyers from bringing legal challenges to regulatory inquiries on behalf of individual entities.

In Part II, I contend that despite the clear remedies provided under the statutory rights discussed in Part I, regulated parties do not avail themselves of these remedies. I argue that this fact is suggestive of how we should think about “power” in the administrative state. I question why certain quasi-due process procedures often lie fallow, and I have developed an empirical model that contributes to the legal understanding of the federal regulatory state. The received legal view of executive power suggests that the president, through his appointees, chooses not to enforce these due process features. My first empirical model tested this received view and rejected it by showing that strategic litigants can shape legal outcomes affecting the executive branch through forum selection on administrative law claims. That successful regulatory challenges require strategic agenda setting by litigants is suggestive of the pluralistic, rather than unitary, nature of regulatory power. To say that administrative power is pluralistic is to say that administrative power is not concentrated in political officials but rather in the regulated parties who set those officials’ decision-making agenda. And yet regulated parties are not routinely advised by their counsel to secure quasi-due process rights, such as the requirement that agency jurisdictional statements be publicly noticed in advance, that all investigations must be for a rulemaking purpose, or that regulatory inquiries of industry members are information collections subject to Office of Management and Budget (OMB) review, approval, and notice and comment. The only rule of law justification for administrative procedures being dependent on regulated parties is to claim that the legal status of administrative investigations is excluded from the regulatory process. In other words, administrative inquiries are auxiliaries to law enforcement. Alternatively, it may be the case that administrative investigations are subject to regulatory procedures, but those procedures are not enforced by the legal representatives of regulated parties and thus must depend upon Congress or the courts for their enforcement.

In Part III, I hypothesize that the underutilization of these procedural tools by counsel is based upon a legal norm that bureaucratic investigations are adjuncts to the executive law enforcement power. As such, rather than inform bases for affirmative litigation challenges, administrative subpoenas inform counsel of the need to represent their client in a defensive posture. I show how Supreme Court jurisprudence prior to the APA made clear that pre-enforcement agency investigations are “legislative” in nature, not acts of law enforcement, but argue that the post-Nixon⁷ Supreme Court’s skepticism as to the continuing validity of Humphrey’s Executor v. United States⁸ has meant the reordering of administrative subpoenas as “executive” not “legislative” in nature. I briefly summarize empirical evidence showing that Congress has delegated its investigative powers over time to the bureaucracy. I then investigate what regulatory pluralism, as a model of the administrative state, means for pre-enforcement regulatory inquiries. If pluralism is a workable model for the administrative state, then it should have strong explanatory power for administrative activities that are alleged not to constitute rulemaking or adjudication, as is suggested with administrative subpoenas.

In my second empirical model, I expand upon my research showing that Congress delegates its investigative powers to the bureaucracy in order to maximize the electoral benefits of direct congressional oversight.⁹ My prior work showed that although Congress prefers ex ante methods of oversight, viz., private rights of action under statutes like the APA in order to constrain the executive branch, it still conducts ex post hearings and does so to maximize its electoral rewards. Because Congress may delegate administrative subpoena authorities to the executive branch and then rationally maximize electoral reward when those powers are used for law enforcement activities, I add executive branch law enforcement activities as an instrumental variable that I theorize mediates the relationship between, on the one hand, legislative delegation of investigative powers in the form of administrative subpoenas and, on the other, congressional oversight. I find, however, that increases in executive branch law enforcement activities do not advance Congress’s electoral goals. If, as scholars suggest,¹⁰ Congress’s preferred method of oversight over delegated power is through indirect monitoring via establishing administrative procedures, then we should be skeptical of the idea that administrative subpoenas are delegations of legislative power or otherwise skeptical of the use of administrative subpoenas as adjuncts to law enforcement, rather than agency rulemaking or adjudication. The results of this model suggest that doctrinal distinctions between executive versus legislative power within the bureaucracy are important within the context of separation of powers and democratic accountability. While the law distinguishes between “administration,” “implementation,” or “enforcement” of legal authority and formal “law enforcement,” virtually everything the executive branch does is law enforcement. In this sense, the federal regulatory state—an enforcement state—is representative of a strong unified executive. Yet in the unified enforcement executive, power is not centralized in the president with deference by the federal courts but is dispersed toward, or delegated to, the specially interested litigants who set the administrative agenda.

In Part IV, I conclude by outlining the implications of the arguments and empirical evidence marshaled herein for administrative law. The theory of the administrative state as both a unitary enforcement state and one governed by a pluralistic theory of power means that improving regulatory due process depends, in part, upon the ability for lawyers to inform the judicial decision-making agenda on administrative law. But the pluralistic enforcement state is a cautionary tale, for the diffusion of power to those without formal accountability raises legitimacy concerns, particularly given the unique problems of agency capture. And alternatives to the pluralistic enforcement state, for example, the nondelegation doctrine, may be impractical as the political evolution of administrative politics has ossified over time. It appears we may have achieved a unitary executive branch in form but pluralism in substance. I make the case for rethinking pre-enforcement agency investigations as legislative rulemaking, versus executive enforcement, as a solution to an executive defined by unified enforcement yet pluralistic power.

Underutilized Due Process Protections against Regulatory Inquiries

Informal investigations represent the primary manner in which federal agencies make regulatory policy.¹¹ In one year, the federal bureaucracy issued 3,367 rules compared to 397,834 adjudications.¹² Executive branch agencies and departments with law enforcement powers use “voluntary information requests,” “requests for information,” or “administrative investigations” or “inquiries” to obtain information concerning individuals’ and businesses’ noncompliance with statutes. Such voluntary information requests are often sent to a class of industry members if the agency or department is concerned that certain business sectors may not be complying with a statute. These voluntary information requests can lead to information that leads to further compulsory requests and subsequent enforcement activities.

Legal practitioners defend against these investigations as if they are indistinguishable from compulsory process in the form of congressional, grand jury, or administrative subpoenas. Yet unique rules apply to regulatory inquiries that do not apply to congressional or traditional law enforcement inquiries. Two crucial sets of rules derive from the statutes codified under Title 5 and Title 44 of the United States Code (U.S.C.), which set forth the metes and bounds of federal information law. Two notable laws include the PRA, which regulates general investigations as “information collections,” and the original public rulemaking requirements of the 1946 APA now codified at 5 U.S.C. 552(a)(1)(D) (the Freedom of Information Act or FOIA).¹³

PRA Protections

All federal agency investigations collect information. The PRA regulates federal agency¹⁴ collections of information from third parties or the public. Under the Act, information collections cover the obtaining or soliciting facts or opinions to, or identical reporting or record-keeping requirements imposed on 10 or more persons.¹⁵ OMB has interpreted the Act’s information-collection requirements concerning “ten or more persons” to mean that “[a]ny recordkeeping, reporting, or disclosure requirement contained in a rule of general applicability is deemed to involve ten or more persons. Any collection of information addressed to all or a substantial majority of an industry is presumed to involve ten or more persons.”¹⁶

The PRA requires that an OMB control number be displayed on every “information collection request” that is subject to OMB review.¹⁷ In particular, the PRA requires that an information collection is “inventoried, displays a control number and, if appropriate, an expiration date; . . . informs the person receiving the collection of information of . . . the reasons the information is being collected; the way such information is to be used; an estimate, to the extent practicable, of the burden of the collection; whether responses to the collection of information are voluntary, required to obtain a benefit, or mandatory; and the fact that an agency may not conduct or sponsor, and a person is not required to respond to, a collection of information unless it displays a valid control number[.]”¹⁸ Only criminal, civil, or administrative law enforcement investigations and proceedings against specific individuals or entities are exempt from OMB review under the PRA.¹⁹ An “administrative action or investigation involving an agency against specific individuals or entities” is distinguished under the PRA from “the collection of information during the conduct of general investigations . . . undertaken with reference to a category of individuals or entities such as a class of licensees or an entire industry.”²⁰

Beyond these procedures, agency information collections are subject to public notice and comment requirements.²¹ The PRA states, “An agency shall not conduct or sponsor the collection of information unless in advance of the adoption or revision of the collection of information . . . the agency has conducted the review established under section 3506(c)(1), evaluated the public comments received under section 3506(c)(2),” published a notice of the information collection in the Federal Register, and received OMB approval of the information collection.²² Finally, the PRA specifically requires that agencies not only meet the Small Business Act (SBA) mandate²³ to reduce information collection burdens on small businesses, but that they “make efforts to further reduce the information collection burden for small business concerns with fewer than 25 employees.”²⁴

Federal tax returns are perhaps the most visible example of OMB-approved forms the bureaucracy uses to collect information. Yet even information collections contained in a rule of general applicability²⁵ or requested from a substantial majority of an industry are subject to PRA requirements. As a technical matter, the Federal Trade Commission’s (FTC) requests for information (RFIs) from peer-to-peer file-sharing companies or the Consumer Financial Protection Bureau’s (CFPB) RFIs to payment card data processors arguably should have been subject to the PRA requirements of having a control number approved by OMB and subject to advance notice and comment. In what is arguably a form of PRA structuring, agencies like the CFPB and FTC will send RFIs to nine recipients on the grounds that the PRA does not apply. Yet while such requests avoid the magic “ten or more persons” standard in the statute, OMB’s own interpretation requires procedural compliance even for requests to an industry of one.

The federal courts have recognized the necessity for agency information collections to undergo OMB review and approval.²⁶ Courts have also recognized that the PRA states that “no person shall be subject to any penalty for failing to maintain or provide information to any agency if the information collection request . . . does not display a current control number assigned by the Director[.]”²⁷ The difficulty, however, is determining when an agency investigation rises to the level of an information collection. Complicating this difficulty is the reality that courts have indicated that statutory requirements for information collections are outside the scope of judicial review under the PRA.²⁸

Agencies and departments do not currently classify their voluntary information requests or subpoenas enforcing such requests as “information collections” required to be reviewed by the OMB or containing OMB approval numbers under the PRA. Nor does the federal bureaucracy state in their voluntary information requests that such requests are exempt from the PRA.

Small Business Regulatory Enforcement Fairness Act Protections

The preamble to the Small Business Regulatory Enforcement Fairness Act (SBREFA) states its legislative purpose is “to provide relief from excessive and arbitrary regulatory enforcement actions against small entities.”²⁹ Section 213 of SBREFA permits small entities to make inquiries “concerning information on and advice about compliance with” the statutes and regulations an agency enforces, requiring the agency to interpret and apply “the law to specific sets of facts supplied by the small entity.”³⁰ Moreover, should an agency pursue a civil or administrative action against the small entity, “guidance given by an agency applying the law to facts provided by the small entity may be considered as evidence of the reasonableness or appropriateness of any proposed fines, penalties or damages sought against such small entity.”³¹ Section 223 of SBREFA requires “[e]ach agency regulating the activities of small entities shall establish a policy or program within one year of enactment of this section to provide for the reduction, and under appropriate circumstances for the waiver, of civil penalties for violations of a statutory or regulatory requirement by a small entity.”³²

SBREFA also created an Ombudsman who shall “work with each agency with regulatory authority over small businesses to ensure that small business concerns that receive or are subject to an audit, on-site inspection, compliance assistance effort, or other enforcement related communication or contact by agency personnel are provided with a means to comment on the enforcement activity conducted by such personnel.”³³ The statute itself defines “means to comment” as “to refer comments to the Inspector General of the affected agency in the appropriate circumstances.”³⁴ In other words, enforcement abuse against small businesses is within the investigative responsibility of the Inspectors General. Section 231 of SBREFA states that “[i]f, in an adversary adjudication arising from an agency action to enforce a party’s compliance with a statutory or regulatory requirement, the demand by the agency is substantially in excess of the decision of the adjudicative officer and is unreasonable when compared with such decision, under the facts and circumstances of the case, the adjudicative officer shall award to the party the fees and other expenses related to defending against the excessive demand, unless the party has committed a willful violation of law or otherwise acted in bad faith, or special circumstances make an award unjust.”³⁵

The Supreme Court has never had occasion to analyze the small entity compliance provisions of SBREFA and few federal circuit courts have analyzed the law.³⁶

APA Protections

The public disclosure provisions of the APA require that each agency “separately state and currently publish in the Federal Register for the guidance of the public . . . substantive rules of general applicability adopted as authorized by law, and statements of general policy or interpretations of general applicability formulated and adopted by the agency; and each amendment, revision, or repeal of the foregoing.”³⁷ Under the APA, if a person lacks “actual and timely notice of the terms thereof” then “a person may not in any manner be required to resort to, or be adversely affected by, a matter required to be published in the Federal Register and not so published.”³⁸ While federal courts have examined these provisions in the context of informal rulemaking under section 553 of the APA,³⁹ they have had less occasion to examine due process-style arguments in this vein in the context of agency enforcement.⁴⁰

The APA’s public disclosure provisions clearly inform the information collection requirements of the PRA. A “rule of general applicability” – for instance an agency’s determination that it has jurisdiction over a particular act or practice – must be published in the Federal Register in advance of any collection of information from a party based upon such a determination. Further, such a collection of information must itself comply with the procedures established under the PRA.

Notwithstanding the general absence of particularized challenges to administrative inquiries on PRA grounds, the weight of jurisprudential evidence creates the presumption that such challenges would fail to be ripe. The PRA does not create a separate cause of action which means that challenges to administrative inquiries must be brought under the APA. The APA permits review of only “final agency action” where a “preliminary” or “procedural” agency action is subject to review only upon the final action of the agency⁴¹ – in the case of administrative inquiries that often means an enforcement complaint or, at minimum, a subpoena.

Administrative Neglect of Procedures

A number of agencies enforce statutes by issuing resolutions interpreting that agency’s authority over some form of conduct. These resolutions represent jurisdictional rules or rules of “general applicability.” However, these resolutions of jurisdiction are often not made accessible to the public through publication in the Federal Register. Further, agencies may use consent decrees or other settlement documents as a basis for establishing its statutory jurisdiction over public activities without publishing those documents or the standards promulgated therefrom. When regulators win enforcement cases before their commissions, they create “regulatory common law” that is given deference by reviewing courts. The Third Circuit, in Wyndham v. FTC, acknowledged the validity of the “common law” of consent decrees.

Issues related to such “secret law” were litigated before the FTC and the Eleventh Circuit by LabMD – a pathology business that qualified as a small entity under the size standards of the Small Business Act.⁴² The facts of LabMD have been described by other scholars.⁴³ But the case is of interest for purposes of this article because all of the procedural requirements of the PRA, the APA and SBREFA were violated yet LabMD did not argue any of these specific provisions. Cases like LabMD informed Executive Order 13892 and a subsequent order at 13924.⁴⁴

Executive Order 13892

E.O. 13892 was titled, “Promoting the Rule of Law Through Transparency and Fairness in Civil Administrative Enforcement and Adjudication.” It framed the procedural requirements established by the PRA, SBREFA, and the APA in terms of “fairness” and “transparency.” That framing is interesting in its own right considering the constitutional conservatives who shaped policy in the Trump Administration are presumed to subscribe to unitary executive principles and largely reject the validity of Humphrey’s Executor.⁴⁵ “Fairness” and “transparency” are terms often used by good government groups and congressional overseers to describe the importance of the legislative power of regulatory oversight. Such oversight is framed in terms of constitutional checks over delegated legislative power.⁴⁶ The framing of E.O. 13892 in terms of fairness and transparency over administrative investigations echoes this sense that the APA is a constitutional check on delegated legislative investigative power. In this sense, the typology of regulatory fairness presumes the validity of delegation.

E.O. 13892 identifies the procedural requirements of the APA (and points out that the 1946 APA’s public disclosure requirements are now codified as the Freedom of Information Act) and specifically finds that “departments and agencies (agencies) in the executive branch have not always complied with these requirements. In addition, some agency practices with respect to enforcement actions and adjudications undermine the APA’s goals of promoting accountability and ensuring fairness.”⁴⁷ Further, E.O. 13892 specifically directs that “[a]gencies shall afford regulated parties the safeguards described in this order, above and beyond those that the courts have interpreted the Due Process Clause of the Fifth Amendment to the Constitution to impose.”⁴⁸

E.O. 13892 also anticipated arguments that challenges to regulatory inquiries were not ripe under the APA. The E.O. defines “legal consequence” to mean “the result of an action that directly or indirectly affects substantive legal rights or obligations. . . . and includes, for example, agency orders specifying which commodities are subject to or exempt from regulation under a statute . . . as well as agency letters or orders establishing greater liability for regulated parties in a subsequent enforcement action. . . [i]n particular, ‘legal consequence’ includes subjecting a regulated party to potential liability.”⁴⁹ E.O. 13892 clarifies pre-enforcement, informal (i.e., no subpoena or compulsory process), and voluntary inquiries as “final” for purposes of the APA due to the legal consequences that attach in virtue of the public disclosure and notice requirements under FOIA and the PRA.⁵⁰ Both such investigations as well as the use of investigations and adjudication to establish “jurisdictional determinations” must be predicated upon “standards of conduct that have been publicly stated in a manner that would not cause unfair surprise[.]”⁵¹

While not explicit, E.O. 13892 makes clear how it understands the legal status of bureaucratic requests for information. The E.O. structures agency inquiries as cabined by rules. If an agency seeks to establish its statutory jurisdiction over some conduct, product, or service it must first publish its jurisdictional determination in the Federal Register. The APA makes clear that “rulemaking” is the process of formulating a rule⁵² and because “statements of policy” rules must structure inquiries and inquiries are auxiliaries to prospective rules, agency inquiries constitute rulemaking.

The concept of investigations as auxiliaries to standards of policy is central to the legislative inquiry power. When Congress investigates the private sphere for a “legislative purpose” it is revealing such an investigation as “legislative” in nature as opposed to an investigation whose purpose is to establish not merely legal consequences but legal penalties. E.O. 13892 thus makes the explicit distinction between agency investigations that are legislative in nature – and are therefore cabined by the APA and PRA – and agency investigations that enforce the law. In the simplest light, legislative investigations can be conceived as independent from the political oversight of the president whereas law enforcement inquiries are crucially subject to the chief executive’s purview.

Moreover E.O. 13892 adopts the standard of the Federal Rules of Civil Procedure that any claim for an act or omission must clearly state the standards for which compliance was required⁵³ and applies to how agencies proceed with enforcement notably independent of whether the agency proceeds before an Article III court or an in-house administrative law judge. As such, E.O. 13892 foreshadowed such cases like Jarkesy.⁵⁴

What makes the Biden Administration’s rescinding of E.O. 13892 interesting is the extent to which E.O. 13892 simply stated the law versus established or ordered agencies to act. To be sure, the Order did seek agencies to publish rules of procedure governing agency inspections,⁵⁵ but that requirement is arguably already established in law.⁵⁶ Indeed one obvious aspect of E.O. 13892 as simply a restatement of law is that it seeks to avoid “PRA structuring” where agencies like the FTC and CFPB send voluntary RFIs to nine persons in order to avoid the ten person magic number that kicks in the PRA’s procedural requirements by clarifying that “any collection of information during the conduct of an investigation” must comply with the PRA.⁵⁷

Executive Order 13924

The coronavirus pandemic occurred subsequent to the issuance of Executive Order 13892, prompting the issuance of an order to continue regulatory fairness in the form of Executive Order 13924 titled, “Regulatory Relief To Support Economic Recovery.” Like E.O. 13892, E.O. 13924 directed agencies to commit “to fairness in administrative enforcement and adjudication.”⁵⁸ Consistent with E.O. 13892, “administrative enforcement” is defined to include “investigations, assertions of statutory or regulatory investigations, and adjudications” thus, again, echoing a more ancient notion distinguishing between legislative and executive branch enforcement.⁵⁹ E.O. 13924, like its predecessor, was rescinded by the Biden Administration. E.O. 13924 is more explicit about SBREFA’s procedural requirements concerning pre-enforcement guidance on compliance noted in section 6(a) of E.O. 13892 yet applies SBREFA’s benefit to small businesses to be able to obtain such guidance “without regard to the requirements of section 6(a) of Executive Order 13892” that is, independent of whether the entity is a small business. Despite E.O. 13924’s brief existence, it is notable for establishing a regulatory “Bill of Rights” in section 6, entitled “Fairness in Administrative Enforcement and Adjudication:”

The Government should bear the burden of proving an alleged violation of law; the subject of enforcement should not bear the burden of proving compliance.
Administrative enforcement should be prompt and fair.
Administrative adjudicators should be independent of enforcement staff.
Consistent with any executive branch confidentiality interests, the Government should provide favorable relevant evidence in possession of the agency to the subject of an administrative enforcement action.
All rules of evidence and procedure should be public, clear, and effective.
Penalties should be proportionate, transparent, and imposed in adherence to consistent standards and only as authorized by law.
Administrative enforcement should be free of improper Government coercion.
Liability should be imposed only for violations of statutes or duly issued regulations, after notice and an opportunity to respond.
Administrative enforcement should be free of unfair surprise.
Agencies must be accountable for their administrative enforcement decisions.⁶⁰

As stated earlier, several of these principles are already codified within the federal public law. That an Order purportedly restating these extant legal principles has been rescinded – without legal challenge – is telling of a reality that administrative due process rights are intrinsic yet fallow. Part II seeks to explore why that is the case.

The Nature of Regulatory Power

In Part I, I examined statutes that restrain executive branch decision-making and noted their dormant nature in the sense that neither the president nor the executive branch agencies enforce these procedural requirements established by law. The two executive orders which sought to enforce these relevant statutes were both rescinded. What does it say about our administrative law that laws on the books may mean nothing if the sovereign – here, the chief executive, i.e., president – does not implement them? The discussion in Part I of this Article is suggestive of our administrative law on questions of political discretion. My objective in this Part, therefore, is to say something about our administrative law by examining the nature of administrative power. I am interested in whether nonenforcement of statutory procedures is a decision of the president or a decision of regulated parties. The relationship between public law and executive power is a central concern within the legal thought of German jurist Carl Schmitt and, on the American scene, Adrian Vemeule’s interpretation of Schmitt’s influence on administrative law. Vermeule’s use of Schmitt’s thought concerning the use of federal emergency powers presents an opportunity to analyze the degree to which administrative power is centralized within a single chief executive.

Vermeule’s thinking about presidential power during emergencies is relevant to discerning the degree to which our administrative law is dependent upon the political discretion of the president. Vermeule hypothesizes that post 9/11 lower courts, in administrative law matters, will be more deferential to the administration thus showing presidents can more lawfully exercise discretion during times of emergency. This would suppose that presidents have substantial control over the administrative law agenda. Using novel data combined with quantitative methods, I find no evidence to support Vermeule’s hypothesis. I then explore the implications of this finding for Vermeule’s broader theories about the American system of administrative law. I draw on the empirical political science literature concerning political power, interest groups, delegation and oversight to argue that our administrative law achieves legal formalism without succumbing to indeterminacy in the ways predicted by Schmitt.

This Part proceeds as follows: first, I analyze what Vermeule means in describing U.S. administrative law as “Schmittian;” second, I empirically test a specific hypothesis suggested by Vermeule predicting that lower court (sub-Supreme Court) judges will be more deferential to the administration on national security matters after 9/11 on emergency grounds. I contribute to the legal literature by collecting a novel set of data limited to judicial review of national security claims by the executive branch and use a logistic treatment effects model to test whether the intervention of the 9/11 emergency causally affected judicial deference in national security cases. Based on the empirical model, 9/11 did not have a statistically significant effect on deference. This indicates the plausibility that, contra Vermeule, our administrative law is not arbitrarily applied in the case of emergency exceptions. I then discuss the implications of my failure to reject the null hypothesis that 9/11 did not significantly change judicial deference, coloring the empirical findings by drawing on quantitative social science literature and recent case studies and data. The inferences drawn by rejecting Vermeule’s theory confirms a number of political features of American administrative law suspected in Part I, including the overdispersion, yet underenforcement, of rules that place interest groups as a central lever of political power in administrative decision-making. The claim is that American administrative law achieves both legality and legitimacy in being overdetermined by interest group politics.

Vermeule’s Schmittian Theory of Executive Power and Discretion

Administrative law scholars have increasingly incorporated German jurist Carl Schmitt’s constitutional critique of liberalism as a lens for understanding administrative law phenomena. For Schmitt, legal liberalism, to the extent it subscribes to a concept of law whereby political decisions are authorized in publicly available rules of law, fails as a theory due to the fact that its executives during states of emergency will make decisions with no formal basis in law and yet no liberal theory can successfully justify these exceptions.⁶¹ On a granular level, recent Americanists have described agency adjudication as “ruled by a norm of exceptionalism.”⁶²

To say that American administrative law is “Schmittian” is to say that the image of law as rule-bound fails in cases where the executive branch exercises discretion to permissibly violate legal rules during national emergencies. That liberal democracy tolerates or permits rule infractions during emergencies is suggested as evidence that the rule of law, and constitutional liberalism, fails to hold as a governing theory. Adrian Vermeule argues that any “aspiration to extend legality everywhere, so as to eliminate the Schmittian elements of our administrative law, is hopelessly utopian.”⁶³ For Vermeule, the failure of legality is evident when courts rely on emergencies to “increase deference to administrative agencies.”⁶⁴ Such deference is possible because of “open-ended standards” in administrative law that aspire to direct courts to constrain executive action but are substantively ineffective. These feckless legal standards result from “grey holes” in the law.⁶⁵ As applied to administrative law, grey holes, like the “arbitrary and capricious” standard for judicial review under the APA, “represent adjustable parameters that courts can and do use to dial up or dial down the intensity of judicial review” in emergency circumstances of war or threats to security.⁶⁶ Judicial review becomes “more apparent than real.”⁶⁷

The relevance of German state thinker Carl Schmitt arises because of the concern that political circumstances, not the legal code itself, best governs how the law is applied.⁶⁸ Schmitt’s implication, according to Vermeule, is that “[t]he legal systems of liberal democracies cannot hope to specify either the substantive conditions that will count as an emergency, because emergencies are by their nature unanticipated, or even the procedures that will be used to trigger and allocate emergency powers, because those procedures will themselves be vulnerable to being discarded when an emergency so requires.”⁶⁹ While not stated explicitly by Vermeule, Schmitt’s view is that when the chief executive (or administrative state) has the discretion to determine both the existence of an emergency and how to address it, the chief executive remains the only legitimate sovereign.⁷⁰

In the context of the public law, Schmitt presents two problems: first, lawmakers cannot craft rules that are sufficient for governing in emergency situations; second, as such, legislators therefore anticipate the need for executive branch discretion (during emergencies or otherwise) by creating “vague standards and escape hatches . . . in the code of legal procedure[.]”⁷¹ For Vermeule, statutes governing administrative action can, at most, specify which official is authorized to act during an emergency but cannot foretell those sets of facts that justify an exception from the general rule.⁷² This is why Vermeule concludes that “exceptions” to the general rules that delegate discretion to judges or administrative officials are built into the fabric of administrative law.⁷³ In practice, Vermeule shows how in a host of judicial decisions interpreting the APA, courts have simply excluded certain agency conduct from the scope of the APA without even analyzing whether the conduct was excepted or excluded under the act, reflecting the existence of “black holes.”⁷⁴ And for those administrative law decisions where agency conduct is subject to the APA, Vermeule argues that otherwise stringent standards of review and exceptions are relaxed in the face of emergency, thus reflecting “grey holes.”⁷⁵

Vermeule’s reasoning also indirectly responds to Jurgen Habermas, the legal and political theorist who has aggressively defended legal liberalism against Schmitt. Habermas argues that liberal democratic law is both formalistic and substantive by involving a distinction between principles and rules.⁷⁶ Further, judges resolving public law disputes can avoid merely deferring to the executive when rules are underspecified because they rely on liberal democratic background principles in interpreting and applying statutes.⁷⁷ Vermeule suggests that the idea that “judges would draw upon thick background principles of legality [e.g.,] principles of procedural regularity and fairness” is “a hopeless fantasy.”⁷⁸ Vermeule’s view is most strongly stated in the following terms: “it is an inescapable fact that judges applying the adjustable parameters of our administrative law have upheld executive or administrative action on such deferential terms as to make legality a pretense. In such cases, judicial review is itself a kind of legal fiction and the outcome of judicial review is a foregone conclusion – not something that is compatible, even in theory, with the banal liberal-legalist observations that administrative law contains standards and permits deference.”⁷⁹

The Schmittian theory of law during emergencies is that governments sidestep blackletter rules in order to exercise necessary political discretion. This theory would directly explain the phenomena identified in Part I, above, where rules on the books are ignored, and executive orders that largely restate the law are rescinded. Adrian Vermeule theorizes that in the United States, lower federal courts permit such sidestepping.⁸⁰

Vermeule proposes a hypothesis that is testable as an empirical model: “lower courts after 9/11 have applied the adjustable parameters of the APA – ‘arbitrariness,’ ‘reasonableness,’ and so on – in quite deferential ways, creating grey holes in which judicial review of agency action is more apparent than real.”⁸¹ Vermeule made the prediction that after 9/11, lower federal courts, that is sub-Supreme Court, would be more deferential to the federal government in “emergency” cases, particularly ones raising national security concerns. Vermeule argues that “[i]t is logically possible that judges might exercise vigorous review during perceived emergencies, but it is institutionally impossible for them to do so.”⁸² My goal is to test this hypothesis presented by Vermeule, particularly his claim that “as judicial perception of a threat increases, deference to agencies increases.”⁸³

Empirically Assessing Our “Schmittian” Administrative Law: Data and Methods

To assess Vermeule’s claims, I construct a novel set of data on all merits decisions involving exemption 1 of the Freedom of Information Act (FOIA), which permits the government to withhold information classified “under criteria established by an Executive order to be kept secret in the interest of national defense or foreign policy” and is “in fact properly classified pursuant to such an Executive order.”⁸⁴ Exemption 1 cases are the most common national security cases subject to review under the APA, which FOIA amended. That FOIA contains exemptions, exclusions and provides a private right of action for unlawful withholdings by the government creates a useful observational scheme for the sorts of issues Vermeule finds relevant in identifying grey holes. Further, the government’s interest in national security secrecy, particularly after 9/11, would well-fit the central expectations of Vermeule’s hypothesis as above-articulated.

A couple of comments on the methodology. First, in coding “deference” I would count any partial summary judgment to the government or partial reversals on appeal that gave appellant some relief as government losses (“0”), else they were government wins (“1”). Given Vermeule’s expectations, it makes little sense to expect that judicial threat perception would be addressed by only partially deferring to the government. In other words, partial deference is a loss for the government during an emergency. Second, if national security was at issue but the case was decided against the government on threshold questions like whether in camera review was necessary or affidavits were sufficient, I counted those decisions as a government loss. In other words, I coded ‘deference’ in such a way as to create a presumption against the government because I’m trying to avoid any error or bias that measures something other than what Vermeule predicted: federal judges voting deferentially toward the government after an exogenous shock in the form of a terror attack. Furthermore, threshold issues present fertile grounds for deference-leaning judges to craft grey holes particularly if case law on a merits question would restrain more engaged interpretations.

While a substantial portion of the exemption 1 FOIA cases are appealed, a coding scheme where not all district-level data has an appellate-level value could significantly bias any model. A number of features of the data explain how I carefully pared the data by coding only some appellate cases while dropping certain district court decisions to avoid any panel-level effects influencing my model: first, and perhaps unique to the FOIA context, the government did not appeal its district court losses so those cases in the sample never have corresponding appeals; second, because of this phenomenon, when a FOIA plaintiff appeals a district court loss and wins on appeal, the initial government win should be considered a loss (indeed as a matter of law it was in error) and therefore I drop these reversed district court deference decisions from the sample; third, in order to avoid homoskedasticity in the data, I drop appellate affirmances of district court decisions for the government. This coding choice effectively corrects what would otherwise be a problematic hierarchical model in which panel decisions were viewed as independent from district court decisions, which de novo review logically prevents.

I hand coded each case by date, whether the case occurred before or after September 11, 2001 (the emergency situation), whether the opinion deferred to the federal government on national security (the outcome variable, ‘y’ in my equation), and whether the decision was by a court within the D.C. Circuit to model any biases on deference that might occur due to the fact that most administrative law cases are brought within the D.C. Circuit. This allows me to test a potential sub-hypothesis: whether administrative law expertise leads to less deference in cases of emergency. In the 296 unique exemption 1 cases I studied from 1971 (when the first exemption 1 case appears) to the present, federal courts defer to the administration’s secrecy argument in 72.3% of all cases. And deference by the courts has increased after 9/11 to 74.10%. But that increase is not statistically significant when compared to pre 9/11 deference, for in the 157 exemption 1 cases decided by lower courts prior to 9/11, the courts deferred to the administration in 70.7% of cases. The dependent variable (deference) is binary and my model must aim to measure the effect of 9/11 on the likelihood of judicial deference to government secrecy claims. I also suspect that any effect on deference varies with whether the decision was by a judge or a panel of judges within the D.C. Circuit given their unique expertise in administrative law matters. I discuss my modeling choices and their interpretation in the next section.

Results and Discussion

Given deference is a binary (1 or 0 event) dependent variable, I start with a logistic regression model which predicts the likelihood ratio (log-odds) of the dependent variable occurring (in this case, deference) given each unit increase or occurrence in the independent variables. The results are shown in Table 1. Neither of the log odds of the coefficient estimates (increased likelihood of deference post 9/11 but decreased likelihood of deference by D.C. federal courts) reveal a statistically significant (p<.05) relationship with deference so we would fail to reject the null hypothesis that there is no association between the 9/11 attack and deference to national security secrecy by the executive branch.⁸⁵

Table 1. Logistic Regression Model: Effects of 9/11 on Judicial Deference to the Executive

*** p<0.01, ** p<0.05, * p<0.1

Vermeule’s claim that our legal system cannot specify the substantive conditions that will count as an emergency or the procedures for allocating emergency powers means that emergencies like 9/11 function as exogenous shocks to our legal institutions, thus providing a quasi-experimental condition for making causal inferences about the effects of emergencies on our legal institutions. I conceive of the event of the 9/11 attack as a treatment applied to the sample of judges deciding exemption 1 cases on 9/11 and thereafter. Because case assignments to district court judges or panel selection for appellate judges is random (or assumed random), the parameter error (standard deviation of the sample) for judges hearing exemption 1 disputes after 9/11 would be uncorrelated with the likelihood that a given judge defers to the government’s secrecy claims. Further because 9/11 was itself a random event (at least as it affected federal courts) the treatment is exogenous to the sample of judges deciding exemption 1 cases on or after 9/11. The functional form of a treatment effects model is:

and where the model examines the independent effects of the exogenous shock of 9/11 and administrative law expertise among judges, while also measuring the effects of interacting administrative law expertise and 9/11 on deference. Unlike the first probability model, a treatment effects model can measure the unique causal effects of the 9/11 attack on federal courts within the D.C. Circuit by regressing the difference in variation between the D.C. Circuit courts and all other federal courts as a result of exposure to the 9/11 attack. The results are shown in Table 2.

Table 2. Treatment Effects Model of 9/11 on D.C. Circuit Deference

*** p<0.01, ** p<0.05, * p<0.1

The effects of 9/11 when interacted with the effects of D.C. federal courts on deference are not statistically significant. In looking at the average treatment effect on the entire population of judicial decisions and the specific population of within-D.C. Circuit decisions, in order to measure the difference pre- versus post-treatment, we also cannot reject the null that 9/11 had no effect on deference. The results of these model specifications are displayed in Tables 3 and 4.

Table 3. Average Treatment Effect of 9/11 on all Judicial Decisions

Table 4. Average Treatment Effect of 9/11 on Decisions within the D.C. Circuit

Finally, as a robustness check and because the data contains both expert (within-D.C. Circuit) judges and non-expert judges and where members of (decisions within) each group are not exposed to the treatment (pre-9/11) as well as exposed (post-9/11), I use a causal inference technique called difference-in-differences to model effects of the difference between the treated and non-treated groups as they varied between D.C. federal courts and non-D.C. federal courts in order to model potential causal effects with an additional technique. The results are displayed in Table 5.

Table 5. Difference-in-Differences Model on Treatment Effects on Judges Before v. After 9/11

R-square: 0.01
* Means and Standard Errors are estimated by linear regression
** Inference: *** p<0.1

In the final causal model, we again cannot reject the null hypothesis of 9/11 having no effect on deference. The implication of these results is that lower court federal judges do not unmistakably defer to the executive even when presented with opportunities to fill black or grey holes and where constitutional deference to an administration’s secrecy needs during an emergency would easily outweigh a citizen’s right to access information.

These results indicate that when modeling causal effects of a terrorist attack on federal judges deciding executive branch arguments for an exception from disclosure on national security grounds, judges do not flex their discretion to rely on broad standards in the law as justifications to defer more substantially to the government. While the empirical results could be interpreted to support a number of theories, there are two clear implications of the results. First, that there were no statistically significant differences in deference before versus after the 9/11 emergency means that legal rules were applied consistently throughout. This means that even if judges ratchet up grey holes in order to achieve a preferred policy outcome, they do so independent of the presence of an emergency. Importantly, if motivated reasoning by public law judges is a feature of liberal legal institutions in the normal case and the exception, alike, then states of emergency would not predict jurisprudential change. Second, the data reflects that 9/11 did not alter the statistically observable differences in deference between courts within the D.C. Circuit and other federal courts. This finding supports a number of further claims, which I deduce from Figure 1, which visually models the data.

Figure 1. Total Disputes per Year Before and After 9/11 Attack

Figure 1 illustrates that prior to 9/11, the number of exemption 1 filings per year never exceeded around seven per year in either the D.C. federal courts or the non-D.C. federal courts. But after 9/11, the exemption 1 filings in D.C. federal courts increased (reaching as high as nine per year) while the number of exemption 1 filings outside of D.C. courts decreased (to a high of around five). Even without graphically representing deference, the likelihood that disputes were filed in the D.C. federal courts after 9/11 appears potentially significant. Indeed, there was a 29% relative decrease in non-D.C. Circuit exemption 1 filings after 9/11.

Hypothesizing a change in causal direction: Do strategic litigants shape deference outcomes?

While the statistical evidence supports failing to reject the hypothesis that 9/11 did not affect deference, the parameter measures may be relevant for rethinking causal direction. Rather than Vermeule’s hypothesis, viz., that courts ignore legal rules to increase deference after an emergency, the data suggests the possibility that as the number of judicial decisions (both deferring and not deferring) increases in both D.C.-based and non-D.C. based federal courts, petitioners (those suing the government) may view D.C. courts as more favorable and strategically decide that the likelihood of accessing otherwise secret documents after 9/11 is higher in those courts. In a regression-based means difference test we reject the hypothesis that there is no difference between filing in D.C.-based federal courts and other courts given likelihood of deference prior to 9/11.⁸⁶

I thus propose the following model (simply refactoring the variables from the Vermeule-informed model and changing the proposed causal direction):⁸⁷

In order to run a model with these parameters that are interactions of original model variables, I create new variables representing these products so that we can think of the parameters as influencing the likelihood of future petitions by plaintiffs. The outcome variable is the product of two prior variables: whether a decision was before or after 9/11 (treatment) by whether the decision was in a D.C.-based federal court. This product is labeled “Post 9/11 D.C. filings” as it tracks the likelihood of treatment. The key independent variables are first, a product of government deference and the binary variable indicating before or after 9/11, which given 65% of its observations are null, I’ve labeled “Pre 9/11 deference;” second, a count variable of exemption 1 disputes per year; and third, the product of deference and the variable identifying disputes in the D.C. federal courts, which I’ve labeled “D.C. courts’ deference.” Given the goal is a model that can predict the likelihood that cases will be pursued in the D.C. federal courts after 9/11 (i.e., the likelihood or odds the value of the binary variable is ‘1’), we can use the logistic regression model form from Table 1. The transformed model is shown in Table 4.

Table 6. Logistic Regression Model on Effects of Prior Judicial Deference on Likelihood of Filing in D.C. federal courts

*** p<0.01, ** p<0.05, * p<0.1

The results show that by transforming the original model by creating new variables out of the same data, the parameter estimates affect the outcome variable with a high degree of statistical significance. In transforming the estimates into odds ratios, we see that the likelihood of exemption 1 filings occurring in the D.C. courts is increased by all the independent variables at a statistically significant level. However, because the new variables may be biased by the increase of null observations (Pre-9/11 multiplied by a deferential opinion is ‘0’), I need a model design as a robustness check where I can remove the potential biasing effects of the treatment and the D.C. courts (now outcome variables) in the parameters. In other words, a potentially debiased model will examine the effects of deference as it varies over the number of disputes on the likelihood that a case is filed after 9/11 in a D.C.-based federal court. Because I predict that prior deference by courts is endogenous to the decision by a petitioner to choose the less deferential D.C.-based federal courts and because deference is a binary (indicator) variable, and further predict that the total number of prior exemption 1 cases will only affect forum choice after 9/11 through deference (and not simply as a result of time passing (e.g. ‘Year’)), I use an instrumental variable design that is appropriate for endogenous indicators. The results are displayed in Table 7, where I exclude from instrumentation the 9/11 effects and D.C. filings (the outcome variable) to ensure an unbiased relationship.⁸⁸

Table 7. Logistic Instrumental Variable Model on Effects of Prior Judicial Deference on Likelihood of Filing in D.C. federal courts

*** p<0.01, ** p<0.05, * p<0.1

The relevant values are the average marginal effects coefficients and since the number of exemption 1 disputes affects the likelihood of a post 9/11 D.C.-based federal court filing only through prior deference (endogenous and already assumed to be correlated with the outcome variable), the key is whether the exogenous regressor, “Year,” is statistically significant in order to evaluate the marginal effects of the increase in exemption 1 disputes as they influence deference. Here, we see a statistically significant relationship and positive marginal effects of exemption 1 disputes on the likelihood of a post 9/11 D.C.-based federal court filing. This confirms the statistical validity of the causal relationships observed in model 6 while factoring out the outcome factors to avoid potential bias concerns.

These empirical results present a strong counter-hypothesis to Vermeule’s. While further empirical investigation is justified by the results, the results of the models identified above provides strong support for the following predictions:

First, the theory of power underlying our administrative law is not focused on the chief executive as sovereign decider but instead the locus of power is found in interest groups who engage in strategic agenda-setting behavior. Administrative law disputes arise due to the legislative empowerment of regulated parties and their organized interests with judicial review rights which ripen through agency petitions. Those interests pursue their claims strategically, which is to say, they consider the past behavior of legislators and judges in making predictions for selecting which issues in which forums to pursue. Strategic choice by litigants depends upon a range of statutory options, for if rules or procedures regulating government action were limited in scope, we would fail to see the evidence of strategic litigation we observed in Model 6. Our administrative law, rather than maintaining holes or gaps in the rules, can be characterized by an overdispersion of legal rules and procedures. That is to say, no one regulatory dispute is resolved by reliance on the totality of germane rules and will always be rule-underinclusive.

Second, because of overdispersion in administrative rules and remedies (procedures), interest groups engage in agenda setting: selecting which procedures to present to courts, which limits the range of legal issues to be resolved by the courts. The role of interest groups, then, has a constraining effect on judicial discretion through agenda control. Thus, interest groups maintain substantial administrative-political power by determining the questions presented before and the relevant legal rules to be decided by judicial decision makers.

Third, because interest groups play a crucial role in setting the administrative law agenda, the fact that interest groups are strategic in seeking judicial relief (i.e., filing in district court within the D.C. Circuit versus a circuit with less regulatory expertise) in addition to pulling the fire-alarms that trigger congressional monitoring of the bureaucracy means that congressional oversight, in addition to judicial monitoring, is an avenue through which rules of law remediate administrative infractions. Importantly, the fact that the administrative law agenda is set by interest groups also means that those same rules on the books which are overdispersed are also underenforced, for interest groups may simply never raise certain procedural arguments.

Fourth, our administrative law anticipates and resolves the exception because it creates the agenda setting conditions whereby regulated parties can pull congressional oversight fire alarms whenever judicial remedies are unavailing (and vice-versa). Further, the competing monitoring of our administrative law from both the courts and Congress permits agency infractions to be evaluated in two senses of legal validity: formal, or technical, validity of rules and consequential, or justification-based, validity of rules. In our system, legality is maintained because overseers from Congress and the courts exercise discretion to enforce against rule infractions based upon public policy justifications. Hence the rule of law does not depend upon the ability to anticipate emergency situations and provide rules that cover the exception but instead thrives when legal decision makers have the flexibility to enforce rules strictly as well as on the basis of public policy and where legislative nullification attends to failing to foresee the consequences of a given rule infraction or providing a legal justification on policy grounds that drifts from the policy preferences of Congress or the president. As such, and because anticipating consequences is subsumed under the rules which structure our administrative law, rule consequentialism forces democratic accountability when agencies make predictions about which policy choices will lead to political oversight where congressional and presidential monitoring to punish infractions establishes governing standards over the bureaucracy. At the same time, agencies and the courts are bound by judicial precedents which are treated by agencies as legally valid statutory and regulatory interpretations.

In the next section, I provide context for these predictions with reference to the political science literature in addition to recent political matters. By analyzing key administrative law cases in the context of pluralist theories of power, I find theoretical support, in addition to the empirical support in Tables 6 and 7 and visible in Figure 1, for these predictions.

The Theory of Regulatory Pluralism

Despite the clear remedies provided under the statutory rights discussed in Part I, regulated parties do not avail themselves. And they must avail themselves because the president, as sovereign, does not exercise unilateral discretion to enforce procedural rules. The sovereign acts upon administrative rules only when subject to a petition by special interests. These findings are suggestive of how we should think about “power” in the administrative state. Administrative power is pluralistic. That is to say, administrative power is not concentrated in political officials but in the regulated parties who set those officials’ decision-making agenda. That regulated parties are not advised by their counsel to secure quasi-due process rights like the requirement that agency jurisdictional statements be publicly noticed in advance, that all investigations must be for a rulemaking purpose, and that all regulatory inquiries of industry members are information collections subject to OMB approval shapes official regulatory decision-making. To change outcomes requires a change in legal strategy.

Vermeule argues that legality fails in the context of the exception because rules cannot anticipate emergency situations. However, the empirical evidence suggests an alternative causal story. Rather than an insufficiency of rules to structure judicial decision-making, our administrative law has an oversupply of procedures; this oversupply of rules governing agency action permits interest groups to use legal challenges as an opportunity to select and apply which rules shape judicial review and test the salience of certain rules in providing a basis for striking down disliked agency decisions; third, that our administrative law is the result of strategic agenda-setting by publicly interested groups signifies how judicial discretion may be effectively cabined consistent with democratic norms. In this section, I highlight the political science literature that colors these inferences while being responsive to Schmitt and I highlight recent matters of bureaucratic infractions as illustrations of how these inferences operate in practice.

In the study of American politics, pluralism argues that political power is decentralized, where the government establishes conditions for interest groups to shape the policymaking process.⁸⁹ Nevertheless, even under pluralistic political theories, the question of political power was answered by who retained authority over decision-making.⁹⁰ In the last half-century, scholars rejected this concept of political power, identifying the political agenda itself as fundamental to political power and the ability to control what issues get placed on the decision-making agenda as more important politically than who has the ultimate authority to choose between alternatives.⁹¹ Thus interest groups become politically powerful to the extent they can shape decision-making agendas. Political scientists have argued that “[c]ourts, regulatory agencies, and congressional committees all require the presentation of policy proposals in specialized and arcane language, and all have complicated rules of formal agenda access. Hence, agenda entrance barriers will favor those able to master these rules or pay for specialists who do. Even with many venues, there remain substantial barriers to entry into the pluralist heaven.”⁹² Furthermore, scholars have observed interest group agenda setting as influential over judicial decisions.⁹³

For Schmitt, addressing interest group pluralism was crucial to his philosophical project. Schmitt argued that legal positivism, in discrediting the legal validity of state sovereignty,⁹⁴ legitimized the diffusion of political authority from a unitary authority to a “pluralist party-state” legally empowered to be “hostile” to the state.⁹⁵ Interest group scholars have argued that “[w]e may conceive of pluralist systems of governance as systems of institutionally-linked policy venues [which] give the opportunity for losers in one policy venue to search for more favorable venues elsewhere . . . [q]uestions of the distribution of political and economic power cannot therefore be considered without a discussion of the relative abilities of policy actors to manipulate image and venue.”⁹⁶ Not only was Schmitt aware of Americanist scholarship on pluralism but he foresaw the American administrative law system where ideological groups are granted authority to obtain judicial review of actions of the executive branch.⁹⁷

Schmitt’s response to liberal pluralism is the claim that in times of emergency “the exception” reveals “the subject of sovereignty” as a single executive decision maker.⁹⁸ For Schmitt, this sort of pluralism seeks to dissolve the “plural political unities” that represented the European nation-states, or, in the American context, the central governing role of the states, in favor of a pluralism defined as anti-state and with “universal and monistic concepts” concerning the unrestricted nature of participation in civil society.⁹⁹ Notably Schmitt did not reject the idea of civil society groups, for “[p]olitical unity can never be understood as absolutely monistic and destructive of all other social groups.”¹⁰⁰ The issue is whether such pluralist groups had governing legitimacy.¹⁰¹ Pluralists “aim not only to negate the state as the highest comprehensive unity, but above all to negate its ethical claim to be a different and higher sort of social relation than any of the many other associations in which people live.”¹⁰² Schmitt’s definition of politics (and therefore political power) is that it distinguishes between friend and enemy, where political friendship is found in the state’s exclusive ability to unify the differences celebrated by civil society and political enmity is found in “recognizing the opponent as a just enemy on an equal plane with oneself. This way one has the basis for a limitation of conflict.”¹⁰³

Interest group pluralism, which Schmitt readily conceded informed modern governance, limited the political power of a state by endorsing a politics of contestation rather than sovereignty. Schmitt argued that without an ability to distinguish legality from legitimacy, e.g., to have a governing theory about when a sovereign entity was empowered to suspend the legal order when essential to preserving it, the state was perpetually threatened.¹⁰⁴ Schmitt stated, “[t]he existence of the state is undoubted proof of its superiority over the validity of the legal norm.”¹⁰⁵ Thus Schmitt might argue that political power wielded in the form of interest group agenda setting, rather than sovereign decisiveness, is illegitimate.¹⁰⁶

Our administrative law, however, is not only not “Schmittian” – it maintains legality and legitimacy because it empowers interest groups with the power to enforce and change legal rules. The political science literature has well-anticipated Schmitt’s objections to pluralism by showing how Congress empowers interest groups by ensuring administrative procedures can be enforced to benefit the interest groups, thus reflecting the consequentialist nature of administrative legality. Scholars have found that “by controlling the details of procedures and participation, political actors stack the deck in favor of constituents who are the intended beneficiaries of the bargain struck by the coalition which created the agency.”¹⁰⁷ That rules are both procedural as well as interest-beneficial informs this deck-stacking behavior and the inference that rules are overdispersed.

Vermeule argues that “[b]lack holes arise because legislators and executive officials will never agree to subject all executive action to thick legal standards . . . they could not do so even if they tried[.]”¹⁰⁸ Yet public law scholars have found that “elected representatives can be expected to be unsure about the substantive details of their most desired policy, even though they are certain about who should benefit and how the costs should be shared.”¹⁰⁹ Because administrative procedure is not simply a formal requirement for, i.e., transparency, but a substantive benefit to an organized constituency, Congress can rely on the bureaucracy and courts to clarify the details of the law while using oversight to correct failures to enforce procedures consistent with the preferences of the intended beneficiary. Because administrative procedures “increase the efficacy of ex post sanctions,” scholars have identified that elected officials enable the content of legal rules to be determined through the remedies pursued by regulated parties in addition to using political oversight to prevent bureaucratic drift from policy goals.¹¹⁰ As Mathew McCubbins and his colleagues’ models have shown, “the organic statute can be vague in policy objectives, seemingly giving an agency great policy discretion, but the administrative process can be designed to assure that the outcomes will be responsive to the constituents that the policy is intended to favor.”¹¹¹ These scholars find that “[a]dministrative procedures have the advantage that their enforcement is left to constituents, who file suit for violations of prescribed procedure, and to the courts.”¹¹²

Political scientists have converged on the theory that Congress’s preferred form of political monitoring is authorizing interest groups with public rights. Because Congress cannot anticipate emergencies, it overdisperses public rights as benefits – that is, there are more procedures than there are resources available to enforce those procedures. Empowered by public rights, interest groups drive administrative law by setting the judicial and political oversight agenda. Once rules are understood as both procedural and consequentialist in enforcement, legality in our administrative law can be understood to proceed from both judicial review (which articulates rules in statute and crafts rules through precedent) and political monitoring (which determines the risks to the bureaucracy for certain rule violations).

Those scholars who view administrative law as arising from Congress’s overdispersion of constituent benefits (procedures) have not only explained judicial behavior in this context but also ex post congressional oversight as resulting from “a system of rules, procedures and informal practices that enable individual citizens and organized interest groups to examine administrative decisions . . . . to charge executive agencies with violating congressional goals, and to seek remedies from agencies, courts or Congress itself.”¹¹³ Scholars thus find that Congress provides procedural benefits to regulated constituents because of an expected electoral return, for those interest groups and individuals who set a given committee’s oversight agenda will also assist in providing electoral rewards for the members of that committee.¹¹⁴ Oversight, or political monitoring of the bureaucracy, derives from the same procedural enactments that inform administrative law agenda-setting before the judiciary, for “political leaders assign relative degrees of importance to the constituents whose interests are at stake in an administrative proceeding and thereby channel an agency’s decisions toward the substantive outcomes that are most favored by those who are intended to be benefited by the policy.”¹¹⁵ And because the rules themselves can be interpreted in terms of the benefits they assign, Congress has designed its oversight authority through procedures – what scholars call “a fire-alarm policy” where “potential supporters can in most cases bring to congressmen’s attention any violations that harm them and for which they have received no adequate remedy through the executive or judicial branch.”¹¹⁶

Although not directly addressed by the scholarly literature, implicit in “fire alarm” theories of congressional oversight and pluralistic theories of judicial agenda-setting is the fact that some legal procedures will be underenforced because interest groups will select other procedures to pursue that may be more promising in terms of an expected benefit or remedy. Because rules are both formal and consequential in value, our administrative law can maintain legality even when rules are formally violated because the political stakes of such violations may simply be too low for remediation by oversight or judicial redress. I illustrate this in the next section.

To further exemplify how pluralism, rather than centralized sovereignty, governs administrative power, I highlight two recent legal phenomena involving oversight of the administration in the context of the administration’s use of discretion in the context of national security and foreign relations – one from the Supreme Court and the other from Congress – to support the counter hypothesis to Vermeule e.g., that our administrative law supports legality and legitimacy while underenforcing certain rules. In these two examples, the limitations on discretion employed by Congress and the Supreme Court actually violate blackletter rules of our administrative law. In the congressional context, the counterfactual might simply mean that congressional oversight is a political, rather than legal, kind of activity. Rather than the filling of a “grey hole” with standards that are more apparent than real, the procedural monitors disregarded a clear statement of law, effectively recognizing a black hole in the law where one did not previously exist.

The rising action for the first impeachment of President Trump was a whistleblower’s disclosure to the Intelligence Community Inspector General (ICIG). The ICIG’s support among the congressional impeachment managers in the U.S. House of Representatives reflected a public mood supporting the procedures that enabled Congress and the public to be informed that the President engaged in a quid pro quo with the President of Ukraine for purposes of targeting a potential (and ultimate) political rival. In this context, there were administrative procedures empowering a detailee in the National Security Council within the Executive Office of the President to report information to the ICIG; there were administrative procedures authorizing the ICIG to document the information as a “whistleblower disclosure” and, there were administrative procedures directing the ICIG to report the disclosure to Congress. Here, the legality of our administrative law where procedures empower interested parties and secure political benefits to Congress appear to work well. This example illustrates a concept of sovereignty that is possible when administrative procedures have consequentialist values in terms of political accountability.

I identify this example of success in American proceduralism because in this same instance of procedures securing political benefits, clear statement rules were violated. The ICIG acted directly contrary to the publicly stated notices on the information collection form for which the whistleblower made his disclosure. The “Disclosure of Urgent Concern” form version in force during the Ukraine complainant’s August 12, 2019 disclosure was the May 24, 2018 form, which limited the ICIG to collecting only first-hand information for purposes of interpreting the Intelligence Community Whistleblower Protection Act.¹¹⁷ Under a statute known as the Paperwork Reduction Act, “collections of information” from the federal government must be conducted on an OMB-approved form and the publicly noticed purpose of the collection is subject to public comment, thus having the force and effect of law and limiting the scope of the agency’s power to collect information.¹¹⁸

The ICIG conceded that the information he received was secondhand. On September 30, 2019, the Intelligence Community Inspector General issued a press release addressing criticisms that the IG processed and reviewed a second-hand whistleblower complaint inconsistent with the ICIG’s public interpretation of the Intelligence Community Whistleblower Protection Act.¹¹⁹ At the time the whistleblower complaint was received, it was policy of the ICIG to interpret the IC Whistleblower Protection Act’s requirement that information be “credible” to mean that the information be “first-hand information.” “[S]econd-hand knowledge of wrongdoing” was insufficient under the policy.¹²⁰ The ICIG, recognizing the whistleblower complaint was based upon second-hand information, ignored his own policy and stated, “there is no such requirement set forth in the statute.”¹²¹

In effect, the ICIG determined that the prior form, which had legal validity under the Paperwork Reduction Act, was “not in accordance with law” under the APA.¹²² As is obvious, the role of setting aside agency action is with the courts, not an Inspector General. Further, amendments to the APA known as the Freedom of Information Act (FOIA) prevent the ICIG from changing the agency’s interpretation of the Intelligence Community Whistleblower Protection Act sua sponte, for “interpretations of general applicability formulated and adopted by the agency” must be published in the Federal Register.¹²³ And the statute is explicit: “[e]xcept to the extent that a person has actual and timely notice of the terms thereof, a person may not in any manner . . . be adversely affected by, a matter required to be published in the Federal Register and not so published.”¹²⁴

The point is that the ICIG ignored if not directly violated a procedural right without immediate sanction. From a formal perspective of our administrative law, we can say that the harmed party, i.e., the President, lacked standing to remedy the infraction. But from a consequentialist perspective, we must say the obvious: in this case, the procedural requirements under the Paperwork Reduction Act and FOIA are low value rights.¹²⁵ Even in the context of the procedural regulation of the federal government writ large, that is, the U.S. Constitution, a technical violation of the Constitution can be determined to be secondary to more consequentially valuable violations. – like those rising to high crimes or misdemeanors. For, as a technical matter, to authenticate second-hand information and authorize a complainant to disclose confidential presidential communications to Congress without informing the White House (which has legal equities in the information) is, in addition to being ultra vires and arbitrary and capricious, violative of the President’s core confidentiality interests under the Constitution, which will always supersede any statutory basis for disclosure.¹²⁶ This example not only illustrates the overdispersed and underenforced nature of our administrative law (where a given administrative action may be covered by multiple procedural requirements only some of which are followed or relied upon by a petitioner) but it also confirms the politically-dependent value of public rights. That our administrative law is conditioned on the electoral and partisan interests of Congress reflects its democratic legitimacy. The President ultimately exercised his administrative remedy: after he was acquitted from impeachment, he removed the ICIG from office.

In 2012, the Department of Homeland Security established an immigration program known as Deferred Action for Childhood Arrivals (DACA). In 2017, after a change in presidential administration, the Department of Homeland Security rescinded DACA on grounds that DACA was a policy unauthorized by law. In the June 2020 decision of Department of Homeland Security v. Regents of the University of California,¹²⁷ Chief Justice Roberts wrote on behalf of the Supreme Court majority that when a federal agency issues a public statement regarding a decision-making policy, any agency action to rescind that policy must include an adequate explanation of the change or is otherwise arbitrary and capricious, and therefore invalid.¹²⁸ The APA, section 552 of Title 5, subsection (a)(1)(D) (also known as the Freedom of Information Act) governs agency policy statements, like the Deferred Action against Childhood Arrivals (DACA) memorandum and its rescission, which are not interpretations of statutes requiring a rulemaking record. Chief Justice Roberts concurred with this view, concluding the DACA memorandum created a “program for conferring affirmative immigration relief.”¹²⁹ These rules can be described as “policy rules” or “public guidance” and are regulated as “substantive rules of general applicability” or “statements of general policy” requiring publication in the Federal Register yet which do not need to go through the notice and comment process contemplated under section 553 of the APA. As specifically enumerated at subsection (a)(1)(E), “revision” or “repeal” of a policy rule requires only publication in the Federal Register – and nothing else. Section 553 of the APA, subsection (b), refers to this aspect of FOIA, specifically excluding from notice and comment rulemaking “general statements of policy.”

This case is noteworthy because nowhere in Chief Justice Roberts’ opinion does he address the clearly germane procedural remedy specified in 5 U.S.C. § 552. Instead, Roberts spent substantial ink distinguishing the rescinding of the DACA memorandum from an exercise of enforcement discretion to invent a non-textual requirement of the APA mandating that “reasoned decision-making” accompany any rescinding of a policy statement.¹³⁰ But political power in our administrative system does not rest with the decision maker but in the interest group empowered to set the terms of the decision-making agenda before the courts. Here, petitioners the United States did not brief the specific remedy of 5 U.S.C. § 552(a)(1) in response to the respondents.¹³¹ As Chief Justice Roberts stated in footnote 4 of the opinion, “Justice Kavanaugh further argues that the contemporaneous explanation requirement applies only to agency adjudications, not rulemakings. . . . But he cites no authority limiting this basic principle—which the Court regularly articulates in the context of rulemakings—to adjudications. The Government does not even raise this unheralded argument.”¹³² This is a feature of our administrative law: if the interested parties do not search for, identify and raise a winning argument from the breadth of rule overdispersion (for whatever reason, perhaps there are only certain arguments strategic litigators seek to run), then the courts, lacking the political authority the interested party has on administrative matters, will not be bound to find the correct law. As a recent administrative law opinion stated, “we need not probe this undeveloped argument further, as ‘[m]entioning an argument in the most skeletal way, leaving the court to do counsel’s work, create the ossature for the argument, and put flesh on its bones’ is tantamount to failing to raise it.”¹³³

That agency rule changes must be announced publicly and in advance of a person’s being affected¹³⁴ is a procedural requirement designed to avoid subjecting individuals to unfair surprise or regulatory burdens for which they could not possibly have advanced notice. As stated earlier, violation of this principle of administrative law made the first impeachment of President Donald Trump possible.¹³⁵ The “Disclosure of Urgent Concern” policy changed by the ICIG was substantively similar to the Obama administration’s 2012 DACA memorandum such that rescinding or changing the policy required public notice¹³⁶ and would now, post-DHS v. Univ. of Cal., require a reasoned explanation in addition. In University of California v. DHS, we see, the courts fashioned a remedy to a technical violation that in the context of presidential impeachment was disregarded by administrative law in practice. Because legality is consequentialist, we can explain the divergent behavior beyond mere policy grounds, for the examples reflect that the ability for interest groups to obtain redress to infractions can be achieved even if irremediable under different circumstances.

Legal rule consequentialism¹³⁷ governs executive branch rule infraction and subsequent judicial deference during emergencies. As such, rule violations only threaten the rule of law if there is no political consequence for failing to monitor the violation of a rule that would otherwise advance greater constitutional interests when enforced than when violated.¹³⁸ The diffusion of administrative law disputes in the United States means rules and procedural requirements are saliently violated or ignored by the executive branch. And these rule infractions are not all vindicated by interest groups going to court or reporting to Congress. As a result of interest group disinterest in these rule infractions, elected officials do not face electoral risks for their permissibility towards this class of rule violations. This causal story does not appear to prevent a consequence of constitutional or democratic harm. But other features of our administrative law anticipate constitutional and democratic concerns.¹³⁹ Because Congress creates more rules and procedures than are actually enforced at a given time (if at all), it provides enormous political power to organized interest groups which serve public interests. Because of rule overdispersion, these interest groups reverse the presumption of informational dissymmetry between the government and regulated parties by being able to strategically select which procedural infractions will most likely lead to judicial or congressional remedies for their members. And because interest groups make a strategic choice to set the legal agenda before the courts or the oversight agenda before Congress and the president, or both, a system endures where judicial and political oversight of rules can yield permanent effects on agency behavior whereas judicial or political reticence does not prevent the opportunity for monitoring over a future interest group challenge to a rule infraction. Under this system, administrative rules are deontologically valid and their violation is justiciable or subject to political oversight in principle.¹⁴⁰ Yet whether monitoring or sanction is imposed is a consequence of whether an issue properly shapes the political or judicial decision-making agenda. This system thus allows the rule of law to function in the face of discretion and deference.

The Pluralist Model and the Law of Bureaucratic Investigations

The Constitution places the president as the nation’s chief law enforcement officer.¹⁴¹ As such, the president’s discretion in how to implement and enforce the law is not generally subject to judicial review or political oversight even as the Constitution regulates the acts of federal officials once an enforcement decision is made.¹⁴² But this picture becomes more complicated when considering the advancement of modern regulatory administration. In the abstract, our public law categorizes federal administrative agents as different in kind from federal enforcement officials.¹⁴³ And yet for the regulated person (and those who counsel her) investigations and orders from administrative agencies present as “enforcement” in type (one hires a “white collar” defense lawyer) or substance, where it bodes similar costs, consequences and penalties for a target. This blurred dividing line in practice exhorts justifications of a unitary executive. Unitary executive theorists have long held that the holding in Humphrey’s Executor, distinguishing administrative agencies from presidentially-supervised departments by deeming administration as non-executive and therefore non-law enforcement in nature, has been effectively reversed by the courts and governmental practice.¹⁴⁴ In this sense, all executive branch enforcement of statutes, administrative, civil, or criminal, is constitutional law enforcement and therefore subject to presidential supervision and control.¹⁴⁵ There is no daylight between what we mean by law enforcement and law administration or implementation.

This revised picture of the administrative state challenges assumptions by scholars about legislative delegation and political control by raising the specter that delegation is unconstitutional even with “intelligible principles” ascribed to it.¹⁴⁶ And what might appear to be delegation in fact is simply authorization, by law, within the law enforcement branch’s purview.¹⁴⁷ Such a state of affairs lets the administrative state have its cake and eat it too, for the regulatory enforcement agent is suddenly entitled to qualified immunity protection for his official enforcement acts while also exempt from constitutional procedures attendant to traditional law enforcement investigations.¹⁴⁸

These scenarios are not hypothetical. Consider that in Daugherty v. Sheer, the D.C. Circuit did not even question whether a Federal Trade Commission civil service attorney was an officer or official for purposes of qualified immunity.¹⁴⁹ Thus while the Supreme Court, in Harlow v. Fitzgerald, was careful to distinguish that only high-level executive branch officials were entitled to qualified immunity,¹⁵⁰ the D.C. Circuit has effectively rendered a civil servant conducting an investigation at an independent agency to have the same law enforcement discretion as a Cabinet member. And as the president’s inherent constitutional discretion over law enforcement expands in scope, the line demarcating Congress’s power to investigate political acts via impeachment, on one hand, from Congress’s power to monitor regulators implementing legislative powers via the oversight power, on the other, is suddenly blurred.¹⁵¹

But the expanded unitary nature of presidential administration may advance executive strength by appearance only, for congressional oversight responsively extends beyond the monitoring of delegated power (the ministerial) to target political officials for political purposes (the discretionary), diluting executive discretion as a result.¹⁵² No statute clarifies whether and when such line blurring is justified or otherwise distinguishes a law enforcement agency occupied by agents with badges and guns from a supposedly legislative one occupied by civil servants with advanced degrees. And yet our administrative law still hangs together. Rather than seek to narrow the breadth of governing principles that structure our constitutional jurisprudence in this area, my more modest interest is interrogating how our administrative law appears to function notwithstanding the lack of a singularly applicable law or decision maker governing the complex possible circumstances.

In this Part, I sketch the jurisprudential development behind the thinking of contemporary bureaucratic investigations within the valence of executive law enforcement power. I further interrogate this jurisprudential project under the lens of bureaucratic decision-making riding on a pluralistic theory of power.

Bureaucratic Investigations as Legislative Inquiries

Congress conducts oversight (monitoring the bureaucracy) but also investigates presidential administrations and the private sector. While contemporary oversight disputes present the legislative oversight power as justified whenever a congressional claim over the executive branch has a legislative purpose,¹⁵³ that understanding has evolved over history. The first century and a half of congressional investigations reflected that only congressional probes into the private sphere were considered relevant to a legislative purpose, as congressional inquiries into presidential administrations were always politically motivated.¹⁵⁴ Congress’s private sector probes waned as administrative agencies were created initially with investigative and adjudicative powers subject to congressional approval and only later were empowered to write legislative rules.¹⁵⁵ Modern oversight thus arose from the political need to monitor the legislative activities of the bureaucracy yet was justified under the legal guise of the private sector inquiries Congress undertook in order to create regulatory agencies in the first place. Today, Congress’s power to investigate the administration, the private sector and the bureaucracy itself has been assigned to Inspectors General, regulators, and litigants.¹⁵⁶ In this section I explore the nature of investigative powers known as administrative subpoenas, which the early 20^th century Supreme Court has identified as “legislative in character.”¹⁵⁷

In distinguishing between the executive departments and the independent agencies, the Supreme Court in the 1935 case Humphrey’s Executor v. United States clarified that independent agencies were, in part, simply administrative bodies whose task was to implement legislative policy in accordance with statutory standards.¹⁵⁸ Echoing McGrain v. Daugherty from eight years earlier, the Humphrey’s Executor Court stated, “[i]n making investigations and reports thereon for the information of Congress . . . in aid of the legislative power, it acts as a legislative agency.”¹⁵⁹ The notion that investigative requests by agencies like the Federal Trade Commission or Securities and Exchange Commission were not law enforcement activities but legislative in nature would be puzzling to most contemporary legal observers of the American regulatory scene. But Humphrey’s Executor recognizes that Congress has the power to delegate to agencies its own power to conduct investigations and issue reports. Interestingly, the “legislative purpose” limitation of McGrain v. Daugherty has never been translated as a rulemaking purpose requirement once an independent agency has obtained the discretion to conduct quasi-legislative investigations even though this is arguably what 5 U.S.C. § 552(a)(1) requires.

Humphrey’s Executor established that to the extent that an independent agency exercises any executive function “it does so in the discharge and effectuation of its quasi-legislative or quasi-judicial powers, or as an agency of the legislative or judicial departments of the government.” While Humphrey’s Executor was framed within the context of the president’s removal power, the case has implications for quasi-legislative agency investigations. In 1946, the Supreme Court in Oklahoma Press Publishing v. Walling further complicated the distinction, announced in Humphrey’s Executor, between agencies within the President’s Article II supervision and quasi-legislative or quasi-judicial agencies – which historically characterizes the independent agencies like the FTC or SEC. In Oklahoma Press Publishing, the Court was confronted with a challenge to the enforcement a subpoena issued by the U.S. Department of Labor’s Wage and Hour Administration pursuant to § 11(a) of the Fair Labor Standards Act (“FLSA”). FLSA, while implemented by a cabinet department (the Department of Labor) “incorporates the enforcement provisions of the Federal Trade Commission Act[.]”¹⁶⁰

Oklahoma Press Publishing dealt with two questions about the tension between individual rights and separation of powers that continue to inform legal challenges to the federal regulatory state. First is the question addressed by the Court in McGrain v. Daugherty almost twenty years before: must a legislative inquiry establish jurisdiction prior to a subpoena (compelling responses to government inquiries) being judicially enforced? This is what is meant by challenging the “coverage” or jurisdiction of a subpoena by claiming that the organic statute from which the subpoena is issued does not cover the respondent’s acts or practices. And such challenges to Congress’s legislative inquiries as being without authority were common. The second question is whether the procedural checks attendant to law enforcement subpoenas (warrant requirement; review and enforcement by a neutral, Article III magistrate; grand jury requirement for an indictment before issuing a complaint) apply to quasi-legislative administrative agency subpoenas.¹⁶¹

While Humphrey’s Executor held that the agency’s power to investigate was legislative in scope, Oklahoma Press Publishing can be traced as the source for converting quasi-legislative agencies into law enforcement agencies. The Oklahoma Press Publishing Court, in a footnote, recognized Congress’s ability to delegate both quasi-legislative powers of inquiry as well as direct law enforcement powers by crafting a distinction between general and specific investigations.¹⁶² Under a Humphrey’s Executor regime, all legislative investigations are general because they concern policymaking, which is always a matter of national, and thus general, interest. Under Oklahoma Press Publishing, the Supreme Court invented a legislative power of inquiry that is specific in nature – in other words it concerns specific persons or corporations, not classes of persons or industries (the scope of policy) – and, in so doing, it justified congressional delegation of a power that the Constitution does not afford to Congress: the power to enforce its own laws.

The Oklahoma Press Publishing Court explicitly confused the relevant constitutional principles. The Court, in analyzing the Labor Department’s subpoena under FLSA, stated, “[t]he very purpose of the subpoena and of the order, as of the authorized investigation, is to discover and procure evidence, not to prove a pending charge or complaint, but upon which to make one if, in the Administrator’s judgment, the facts thus discovered should justify doing so.”¹⁶³ Proving a case is the hallmark of adjudicatory proceedings, not law enforcement. The ability to seek judicial relief sets the zenith of the president’s law enforcement powers. Government prosecution of a matter is supervised by the courts, not the president. The process of discovering and procuring evidence to determine a violation of law versus a need for policy is precisely how the executive ensures that the laws are faithfully executed – it is a core trait of law enforcement, not a legislative function.

The Oklahoma Press Publishing Court, aware of its prior decision in the 1924 case of Federal Trade Commission v. American Tobacco Co.,¹⁶⁴ where Justice Holmes struck down an FTC investigation on Fourth Amendment grounds, permanently abated the effect of that holding by firmly exonerating agency investigations from warrant requirements, declaring, “[i]t is enough that the investigation be for a lawfully authorized purpose, within the power of Congress to command.”¹⁶⁵ And, yet, while the collection of evidence may be within Congress’s power or a hearing within the courts’ power, the act of a complaint with notice is neither quasi-legislative nor quasi-judicial, it is the epitome of law enforcement beyond Congress’s or the courts’ powers.¹⁶⁶ Oklahoma Press Publishing determined that whenever an investigation is “authorized by Congress, is for a purpose Congress can order, and the documents sought are relevant to the inquiry” then the requirement of “probable cause, supported by oath or affirmation” as required in a warrant, “is satisfied.”¹⁶⁷ After these pre-APA cases, the Court established that investigations to determine whether regulatable entities were subject to a legislative act and whether they violated the act were within Congress’s power to authorize and, by Congress’s decision to delegate such power, were impliedly within an agency’s power to implement.¹⁶⁸ Only in the Supreme Court’s 1946 decision in Oklahoma Press Publishing v. Walling, issued in the twilight before President Truman’s signature of the APA of 1946, can McGrain be understood as a presaged justification for presidentially insulated quasi-legislative, quasi-judicial agency investigations. When McGrain was decided, it was a year subsequent to the question of the President’s power to remove a postmaster official in Myers v. United States¹⁶⁹ and subsequent to a number of administrative law challenges filed in the Court of Claims, which heard claims arising under the Constitution or statute that entailed money damages (the constitutional challenge to removal in Humphrey’s Executor was likewise filed in the Court of Claims).

Oklahoma Press Publishing v. Walling addressed the question of a private target’s challenge to an administrative agency subpoena for records and information. For the first time, the Court had to evaluate the question of validity when Congress delegates its investigative powers to a non-law-enforcement agency for purposes of investigating conduct covered by statute. In its reference to McGrain,¹⁷⁰ the Court analogized an agency investigation as effectively a delegation of Congress’s own inquiries for a legislative purpose (“[i]t is enough that the investigation be for a lawfully authorized purpose, within the power of Congress to command”).¹⁷¹ In effect, if Congress could validly delegate its investigative powers to committees certainly it could delegate such powers to administrative agencies charged with implementing regulatory norms established by Congress.

If McGrain is read to govern legislative inquiries of private citizens and Oklahoma Press Publishing applies that principle to regulatory inquiries by agencies created by Congress, then the APA’s definition of “rule” as encompassing part of an agency statement designed to implement law,¹⁷² and its definition of “rulemaking” as governing the process for formulating such a statement, can be understood in a new light. Just as congressional investigations are bound by a rulemaking purpose, so too must agency investigations be considered a process for formulating a rule, i.e., “an agency statement of general or particular applicability and future effect designed to implement [] or prescribe law or policy[.]”¹⁷³

By misreading McGrain, American public law jurisprudence has not only mistakenly approved the adjudication of interbranch information disputes but simultaneously failed to treat regulatory (by agencies) investigations as antecedent to rulemaking (the clear holding of McGrain). And such inquiries, bound by a regulatory purpose, are ones the APA requires to be disclosed publicly in advance. But the concept of administrative subpoenas as legislative inquiries is not the received view of the law.¹⁷⁴ By extending McGrain to provide judicial review of congressional oversight of administration, our jurisprudence has ignored the extent to which Congress’s own delegation of its investigative functions are not subject to due process requirements. In narrowing McGrain to its proper holding, the courts may be better positioned to exercise review over the regulatory power of investigation which, while evolving from congressional oversight, has certainly evaded its contemporary attention.

While the jurisprudence on Congress’s investigative power being incidental to its general legislative powers may be clear, the distinction has become confused by subsequent judicial analysis. The Supreme Court held that Congress may delegate authority to the executive branch through a legislative act without impermissibly delegating its constitutional powers.¹⁷⁵ This doctrine upholds the congressional delegation of legislative authority as an implied power so long as Congress provides an “intelligible principle” to guide the executive branch in implementing a regulation.¹⁷⁶ Crucially, this doctrine is limited to the context of Congress’s delegation of regulatory authority (what is conceived by administrative law scholars as legislative rulemaking).¹⁷⁷ Missing from this judicial treatment is the prior doctrine that Congress’s power to investigate is incidental to its power to legislate. As such, any lawfully delegated executive power to regulate must imply a power to investigate to effectively craft rules. While not speaking directly to its “intelligible principle” doctrine, the Supreme Court has recognized the power of Congress to delegate legislative powers of inquiry to the executive as “comprehended in the ‘necessary and proper’ clause, as incidental to both [Congress’s] general legislative and its investigative powers.”¹⁷⁸ However, the federal courts have never required that the executive’s investigative powers, particularly those by independent agencies, be incidental and thus cabined by agency rulemaking.¹⁷⁹ Just as Congress is constitutionally restricted from exercising law enforcement or judicial powers, so too would any attempt to delegate such powers. Yet the delegation of investigative powers to the executive without a requirement that such powers be incidental to a rulemaking function allows the executive to exercise legislative-like investigations for law enforcement purposes without the doctrinal restrictions attendant to separation of powers concerns and more substantive threats to fair notice and due process. This central separation of powers concern is crucial to institutional motivations behind oversight yet is noticeably overlooked in the literature.¹⁸⁰

Bureaucratic Investigations as Law Enforcement

The president is the nation’s chief law enforcement officer. As such, the assumption is that executive branch activity is, above anything else, law enforcement activity. In this sense, investigations by agencies would be thought to advance executive branch law enforcement interests.¹⁸¹ While the early Supreme Court accepted the bureaucracy as a creature of legislative delegation and therefore understood regulatory investigations – subpoenas by the FTC or the Securities and Exchange Commission, for instance – to be legislative, not law enforcement,¹⁸² the contemporary Supreme Court readily classifies administrative subpoenas as part and parcel of law enforcement.¹⁸³ Administrative subpoenas have been described as the ‘backbone” of “regulatory enforcement.”¹⁸⁴

If the congressional authorization of administrative subpoena powers by the bureaucracy serves a law enforcement purpose, then it raises the question concerning why Congress would empower the bureaucracy with investigative powers given the theory that Congress establishes the bureaucracy and administrative procedures solely for purposes of controlling public policy. Oversight is theorized to be ex post delegation.

If Congress authorizes the bureaucracy to issue administrative subpoenas which permit greater law enforcement activities and this activity maximizes congressional oversight efficiency, then a number of inferences can be made. First, because granting administrative subpoena power to the bureaucracy permits greater law enforcement, administrative procedures here would actually decrease congressional control over policy by increasing bureaucratic discretion to enforce the law.¹⁸⁵ Second, congressional oversight would be theorized as less about bureaucratic control than, for instance, exposing executive branch activities that could threaten the president and his party electorally.¹⁸⁶ Third, and most important, to the extent oversight involves the management of delegated power, Congress would be largely abdicating that responsibility and relying on law enforcement subpoenas and Inspector General investigations for ideologically salient findings.¹⁸⁷ The United States Department of Justice has increasingly relied upon administrative subpoenas for administrative, civil, and criminal enforcement.¹⁸⁸ Any mechanism where Congress relies on fire-alarm alerts of procedural violations by the bureaucracy to ensure control would suddenly be irrelevant.

The Supreme Court, the same year that both the Legislative Reorganization Act and the APA became law, held that agency exercises of the “subpoena power for securing evidence” with “the aid of the district court in enforcing it” is an “authority . . . clearly to be comprehended in the ‘necessary and proper’ clause, as incidental to both its general legislative and its investigative powers.” Administrative investigations, as a historical matter, were considered legislative in nature. But the current Court views pre-enforcement subpoenas issued by administrative agencies as essential to the president’s “duties as the head of the Executive Branch” and “to ensure that [the president’s] subordinates serve the people effectively and in accordance with the policies that the people presumably elected the President to promote.”¹⁸⁹ The Supreme Court in City of Los Angeles v. Patel dealt with the question of what constitutional process applied to administrative subpoenas and it held that the law or regulations authorizing administrative inspections must permit pre-enforcement review by a neutral adjudicator or, if the target industry is “closely regulated” the inspecting officers’ discretion must be constrained by standards governing which industrial targets to inspect and under what circumstances. Implied within the Court’s reasoning is that administrative subpoenas are law enforcement activities, not legislative ones.

In this sense, Patel overturns a central holding in Oklahoma Press Publishing Co. v. Walling: that administrative subpoenas were “incidental” to Congress’s “legislative” powers. 327 U.S. 186, 214 (1946). Four months after the Court’s Oklahoma Press Publishing decision Congress passed and the president signed the APA, which stated, “[e]xcept to the extent that a person has actual and timely notice of the terms thereof, a person may not in any manner be required to resort to, or be adversely affected by, a matter required to be published in the Federal Register and not so published.” Because Oklahoma Press Publishing held that agency subpoenas were ancillary to legislation and the APA authorized congressional delegation of its legislative rulemaking power, the APA’s requirement that “interpretations of general applicability formulated and adopted by the agency” be “published in the Federal Register” applies to the FTC’s decisions to authorize compulsory process. In short, the “Resolution authorizing compulsory process in Section 5 cases concerning data privacy” must be published in the Federal Register for public inspection. Shivers ruled that the officers engaged in a discretionary function and that” no federal statute, regulation, or policy that specifically prescribes a course of action that the prison employees here failed to follow.” With FTC “officers” like Alain Sheer, however, there is federal law that limits their discretion. Alain Sheer cannot independently subpoena a company – he must seek the full Commission’s approval and must have a previously approved resolution authorizing compulsory process for the target industry. Thus, the discretionary function exception does not apply to FTC employees as they have no independent enforcement power. While this all gets messy – in the context of Bivens, the D.C. Circuit in Sheer says FTC employees are law enforcement officers entitled to immunity from suit because there is no clearly established right against investigation due to a retaliatory animus – the consistent thread in Patel, Sheer, Oklahoma Press is that 1) FTC employees do not have discretion to use compulsory process only the full Commission does, 2) clearly established law requires that Commission resolutions of compulsory process authority are subject to public notice, 3) there is a clearly established right to pre-enforcement review of an administrative subpoenas and that review must be meaningful in the sense that a reviewing court must ensure that applicable requirements (APA public notice, Fourth Amendment “clarity and regularity” requirements, or ex parte warrant requirements) were met. Even if the FTC Commissioners had the discretion analyzed by the Supreme Court in United States v. Gaubert, 499 U.S. 315, 324 (1991), the policy or internal guidelines (i.e., the Resolution) relied upon are required to be published in the Federal Register or otherwise noticed to an investigatory target.

Delegation of the Legislative Power of Inquiry

In prior work, I provided empirical support to the idea that congressional oversight means the legislative monitoring of delegated power.¹⁹⁰ Further, I showed, consistent with other scholars, that Congress prefers to attach those administrative procedures to delegated authority that enables its public monitoring (through hearings) to most efficiently advance electoral interests of members.¹⁹¹ That scholars of congressional oversight have theorized that Congress principally exercises oversight through administrative procedures which govern delegation¹⁹² is well-established yet is nevertheless relevant to the context of Part I, infra, which reveals the extent to which administrative procedures are often unenforced. The mid-20th century Supreme Court considered both agency investigations aimed to determine whether a private party’s conduct was subject to a particular statute and whether that party violated the statute as part of the “legislative power.”¹⁹³ This is distinct from law enforcement activity carried out under the supervision of the president, viz., “to prove a pending charge or complaint” and seek legal consequences for that party’s failure to meet its obligations.¹⁹⁴

Administrative subpoenas provide a fertile area for examining delegation as well as measuring the utility of administrative procedures as instruments of political control. Subsequent to the Court’s Oklahoma Press Publishing decision, Congress enacted the APA, which McNollgast highlight as exemplar of their theory that administrative procedures are Congress’s preferred tool for controlling the bureaucracy.¹⁹⁵ As applied to agency subpoenas as legislative delegations, the APA requires advanced public statements by the agency addressing both the question of whether a statute applies to certain conduct as well as the standards governing a determination that the law was violated. And yet the Court has never determined that the APA’s rulemaking requirements apply in advance of administrative subpoenas nor have regulated parties argued that rulemaking applies to agency determinations that a law governs some conduct, even when such a principle strikes at the core of the power to prescribe the duties by which citizens are regulated.¹⁹⁶

In the context of the issues raised in this Article, I am interested in determining whether the electoral benefits Congress receives from delegation of its investigative powers is affected when those investigative powers are used for law enforcement purposes. Unlike bureaucratic exercises of ministerial responsibilities, executive branch law enforcement is generally thought to be exempt or excluded from congressional oversight.¹⁹⁷ In the context of the political economy literature on delegation, this raises the possibility that congressional delegation empowers the bureaucracy with law enforcement powers yet does not rely upon the sorts of administrative procedures that allow Congress to control bureaucratic policy implementation.¹⁹⁸ In this sense, the executive branch control over delegated legislative power creates a problem for legislative accountability.¹⁹⁹

I theorize that if Congress’s revealed electoral preferences can be advanced through authorizing greater law enforcement power to the executive, then Congress has maximized its opportunities to publicize political infractions by the bureaucracy even as it has ceded actual control over legislative authority. In such a context, the executive may be actively policing statutory violations by regulated parties and the bureaucracy itself and transmitting reports to Congress that enhance Congress’s ability to ensure public policy is consistent with its ideological preferences even as that executive branch decision-making is fully excluded from congressional purview.

Modeling the Pluralist Theory in the Context of Administrative Investigations

Data and Methods²⁰⁰

Administrative subpoenas, also known as civil investigative demands, from Offices of Inspectors General (OIGs) and agencies like the Federal Trade Commission, Securities and Exchange Commission, National Labor Relations Board, Commodities Futures Trading Commission, Food and Drug Administration, Federal Communications Commission and Consumer Product Safety Commission are commonly used to form the basis for criminal enforcement actions.²⁰¹ Congress is more likely to respond to decisions to file criminal cases underlying subpoena-obtained information than it would to decisions to bring administrative or civil cases. First, while the executive branch can enforce the laws administratively and civilly, administrative enforcement is often only publicized through a final decision and order when an administrative complaint proves successful. The public filing of criminal cases is independent from the outcomes of those enforcement matters. Second, the political leadership of the executive branch, typically the U.S. Department of Justice, does not have oversight over administrative enforcement by boards and commissions but does exercise prosecutorial discretion as to law enforcement referrals from agencies and Inspectors General. Third, because congressional enactment of civil enforcement authority evolved over the time period examined (1946-2020), we cannot assume that each filing in the sample is independent from and equally distributed as any other filing and thus civil filing data is likely skewed.²⁰² Finally, criminal enforcement filings are often based on referrals where prosecutorial discretion is independent of the decision to proceed administratively and civilly, yet solely pursuing civil enforcement often means that a decision was made not to pursue a matter criminally, thus further leading to bias in estimates based on civil filings.

Congress’s inserting language in statutes that authorize bureaucrats to issue administrative subpoenas is the key manner in which Congress delegates its investigative powers to governmental agents. Administrative subpoena authorizing statutes are uniquely suited to empirical work on congressional oversight and administrative law. Administrative subpoena statutes authorize agencies to subpoena regulated parties as part of administrative investigations to determine whether a statute is being complied with. As the Supreme Court recognized in Oklahoma Press Publishing Company v. Walling, administrative subpoenas are delegations by Congress of its compulsory power to obtain information necessary to regulate. Contemporary scholars have employed tools of measuring statute length or conducting text analysis as proxies for legislative delegations.²⁰³ Studying administrative subpoena statutes presents a more direct way for regulatory scholars to measure delegation for at least two reasons. First, when a statute says “the Secretary shall have the power to issue subpoenas” the statute is much more readily interpreted by the executive agency as delegating a specific power through a specific process versus, e.g., the Energy Policy Act’s statement that “the Secretary shall make guarantees under this or any other Act for projects on such terms and conditions as the Secretary determines” which is where agencies exercise rulemaking or adjudicatory discretion.

Second, subpoena grants are much clearer evidence of intentional delegation by Congress than “as the Secretary determines” grants (typical of rulemaking). In other words, whether Congress delegates authority to an agency to issue a regulation or conduct a hearing, clear language is not necessary for an agency to exercise discretion, as the agency receives substantial deference to an interpretation needed to implement a policy command. However, in the case of subpoena grants, agencies are empowered with quasi-enforcement tools – a power to compel – and both Congress and the executive branch have incentives to ensure no uncertainty exists as to the interpretation of the delegation.²⁰⁴

As such, these statutes represent a pure form of delegation of legislative authority because there is no ambiguity as to how much discretion or power is delegated – once the agency has subpoena authority under some statute, it is authorized to compel information productions and enforce that compulsory process in court via the Department of Justice or some other organic act that permits independent litigating authority for noncompliance. There is no need for interpretive discretion by the agency.

Key Variables

Political scientists conceive of oversight over delegated power to mean political control. The literature well-establishes that congressional oversight investigations and hearings most effectively control the bureaucracy when members can utilize oversight powers to advance their electoral goals. Because delegation is causally necessary for oversight, my prior work tests whether a power is legislatively delegated to the bureaucracy in terms of its likelihood of more effectively advancing the electoral interests of members through oversight hearings. Through a series of empirical models, I showed that Congress is able to more efficiently advance the electoral goals of its members by delegating the power to monitor political officials to “fire chiefs” (IGs, the Special Counsel).²⁰⁵ Such delegation allows Congress to rely on its fire chiefs to supply members on oversight committees with information necessary to benefit their electoral prospects through credit-claiming or otherwise exposing alleged misconduct by a politically-adverse administration.

In examining congressional hearing data, I developed a dependent variable indicating resource expenditures (in terms of time, in days, spent per hearing) for congressional oversight from 1946 to 2020. As discussed above, Congress delegates to reduce legislative resource burdens and it delegates oversight to reduce the resource costs involved in monitoring delegated authority. Reducing the workload of oversight enables Congress to use oversight to serve members’ electoral goals and so Congress will be expected to prefer to delegate in ways that ensure such a payoff. In prior work, I find that Congress has a preference to conduct oversight in a manner that reduces the amount of time Congress has to spend on a given hearing while increasing the total number of hearings held by committees with oversight jurisdiction.

The key independent variable is administrative subpoena activity. Entities that issue administrative subpoenas – independent agencies, GAO, or Inspectors General – are statutorily required to report their findings to Congress, independent of whether they had to use a subpoena to get the information.²⁰⁶ The IG Act created permanent audit and investigative officers in every federal department and, by far, the IGs issue the most administrative subpoenas by the bureaucracy. As I have previously found, delegation of congressional investigative power to Inspectors General allows Congress to delegate oversight to federal officials in order to increase to increase its political payoff via information at the lowest cost in terms of internal resources.²⁰⁷

Because Congress delegates subpoena authority each time an agency or a component within an agency is created, every enacted subpoena power is part of a larger statute establishing one or more regulatory norms. The reporting requirement in §7(a) of the Presidential Threat Protection Act of 2000 required the U.S. Department of Justice to report to Congress on every legislative authorization of administrative subpoenas for use by the executive branch.²⁰⁸ The Department of Justice data covered statutes on the books from the period of the early 1900’s to 2001, which, in prior work, I updated to the present.²⁰⁹ The number of statutory provisions delegating subpoena power have been fairly consistent over time.²¹⁰

One difficulty with measuring the effects of administrative subpoena authorizations is that a single statutory authorization may yield dozens of subpoenas from a single agency in a year. Relatedly, given the large numbers of subpoenas issued by the bureaucracy each year, congressional oversight committees are unlikely to be directly responsive to such subpoena activity even when subject to administrative procedures. Here, legal doctrine informs the empirical modeling approach. If administrative subpoena powers are legislative in nature and delegated by Congress to the bureaucracy, then they must be bound by a legislative purpose and would only affect Congress if instrumented through agency rulemaking. But if administrative subpoenas are adjuncts to law enforcement, then there can be no corresponding legislative review.²¹¹ If subpoenas influence Congress via law enforcement filings, then it tells us that Congress can delegate authority and reap electoral rewards without any need to monitor bureaucratic law enforcement activity.

In order to test the hypothesis that the electoral payoff of efficient oversight hearings should not be affected by congressional authorization of more investigative discretion for law enforcement activity through administrative subpoena authorizations, I obtain criminal filings from the Department of Justice for all available years in which information is made available.

Because I am interested in the effects of investigative delegations on oversight as mediated by criminal filings, I create an interaction term between those two variables. To control for any effects being caused by regulatory or other factors, I introduce a discrete measure of all federal rulemaking per year, a continuous variable of total rules on the books, and a control based on whether the Inspector General Act had become law.

Results and Discussion

Because the leading indicator looks at how current independent variables affect continuous oversight activity a year in advance, I use negative binomial regression to examine the effects of the subpoena and criminal filing interaction and the rulemaking control on oversight hearing counts. The results of this model are displayed at Table 8, below. The interaction term has no statistically significant effect leading me to fail to reject the null hypothesis that Congress does not have incentives to abdicate legislative authority in order to magnify executive law enforcement power. In the next model shown in Table 8, which models based on every hearing versus clustering per Congress, I use a non-leading measure of resources dedicated to oversight as well as a count of total rules per year as a control and do not observe a statistically significant effect of the interaction term. The rulemaking control is negative and significant indicating that the expected effect – subpoenas affect oversight resources through rules not enforcement – holds true.

Table 8. Effects of Law Enforcement on Congressional Oversight Costs (1946–2020)

Standard errors in parentheses.
*** p<0.01, ** p<0.05, * p<0.1

The legislative power to investigate the private sector is how modern legislatures revise or reform business regulations to support the general welfare. If neither Congress nor the bureaucracy uses that power of regulatory investigations to inform legislative rules, an essential aspect of modern policymaking is disregarded. The results in Table 8 suggest that Congress maximizes its oversight preferences by authorizing the executive branch to utilize investigative powers for policy, not law enforcement, purposes.²¹² Certainly, delegation implies that Congress has the power to engage in the delegated act itself and Congress lacks any law enforcement power.

Concerning legislative inquiries over the private sector, this is an area where Congress has traditionally stood on strong legal grounds to rely on the courts to enforce subpoenas for information and documents. Congressional delegation of this power to the bureaucracy would be possible if subpoenas were subject to the legislative requirement that the investigation be for a legitimate lawmaking purpose. In the context of the bureaucracy, this requirement takes the form of administrative procedures to ensure that administrative subpoenas are for a legitimate rulemaking purpose.

The results counteract the theory that administrative subpoena authorities are understood not as legislative delegations subject to legislative control but as authorizations subject to executive branch law enforcement discretion. Thus while the U.S. Department of Justice reports, “Inspector General [subpoena] authority is mainly used in criminal investigations”²¹³ the results indicate that it is not the criminal filings that incentivize Congress but the policy goals advanced through such inquiries which advance Congress’s interests in monitoring such delegation.

Preserving Executive Unity in the Face of Pluralism

The previous sections have shown that the bureaucracy does not enforce procedures simply because they exist on the books. What is more, such non-enforcement is not solely a question of executive branch enforcement discretion. It turns out that the picture of administrative power as organized around presidential decision-making suffers empirical flaws. Instead, the evidence suggests that procedures are enforced or not enforced depending upon whether regulated parties bring them before the agencies and reviewing courts by setting the regulatory oversight agenda. That procedural enforcement may depend upon parties, not a central decision-maker, appears to challenge the image of the executive branch as unitary. At the same time, however, the modern bureaucracy mixes quasi-legislative enforcement powers with traditional law enforcement powers. The picture that emerges is a bureaucracy where law enforcement discretion tends to be unitary and indivisible while procedurally required restraint is pluralistic. This picture presents immediate concerns for constitutional liberty. The final empirical model shows, however, that the electoral gains associated with congressional oversight – the supposed constitutional capacity to check bureaucratic policymaking – is not affected through delegation of investigative powers used for law enforcement compared to explicit policymaking by the bureaucracy. This finding suggests that whether the executive branch is more or less aggressive in prosecuting defendants, Congress does not effectively use those outcomes for electoral (credit-claiming, position taking) purposes. This finding supports a reality of a bureaucracy in tension: efficiency dictates that a unitary executive merge administrative, civil and criminal enforcement yet compliance with statutory oversight depends upon political agenda-setting, rendering the unitary executive inefficient in ensuring due process during those enforcement activities.

The argument suggested by the empirical work in this Article defends pre-enforcement regulatory inquiries as rulemaking. In this Part, I outline two key confusions in jurisprudence that have led to the rejection of the notion that the APA’s rulemaking requirements apply to pre-enforcement inquiries. First, the law misunderstands the nature and power of legislative inquiries. Second and related, law mistakes the scope and extent of the delegation of the legislative power of inquiry.

The Metes and Bounds of Oversight

Congress exercised a rulemaking power, e.g., to investigate the private sphere in order to write bills, for over a century before its relatively recent decision to use its investigative authority to conduct oversight over the bureaucracy, which is justified by political scientists on the grounds that oversight is the principal’s power to oversee the agents to whom it has delegated authority. This reality is the result of Congress’s decision to delegate its power to investigate the private sphere to the bureaucracy, which the courts have historically examined under the same framework it did for direct congressional investigations.²¹⁴ The centrality of administrative procedures to oversight of the bureaucracy signals the manner in which the practice of administrative law is simply one face of Janus: the other face is congressional oversight; but while each utilizes different procedures the goal of regulatory oversight is the same.

The “Necessary and Proper” Clause justified Congress’s power to investigate the private sphere as well as its authority to delegate that power to the bureaucracy. And it would logically justify the monitoring of such delegation. But not for our current legal institutions. The legal doctrine supporting oversight derives from the begrudging acceptance of bureaucratic policymaking best reflected in Justice White’s dissent in Bowsher v. Synar where he states, “with the advent and triumph of the administrative state . . . the Court has been virtually compelled to recognize that Congress may reasonably deem it ‘necessary and proper’ to vest some among the broad new array of governmental functions in officers who are free from the partisanship that may be expected of agents wholly dependent upon the President.”²¹⁵ The Department of Justice’s Office of Legal Counsel, whose opinions bind the president and the bureaucracy, has challenged “the continuing validity of Humphrey’s Executor”²¹⁶ to deny that bureaucratic policymaking exists. Instead, all bureaucratic activity, including rulemaking and adjudication, are supposed as law enforcement activities subject to the supervision of the president, as chief executive.

Both the executive branch and Congress have interpreted the investigative authority Congress has over the private sphere as justifying “oversight” – Congress’s ability to control bureaucratic policymaking.²¹⁷ To illustrate this move, the Office of Legal Counsel at the U.S. Department of Justice conflates Congress’s “authority to make official inquiries into and to conduct investigations of executive branch programs and activities” with “the implicit authority of each house of Congress to gather information in aid of its legislative function.”²¹⁸

Just like the Office of Legal Counsel (OLC) has rejected the validity of presidentially-insulated agencies exercising executive powers, OLC has also rejected the executive branch’s position, from presidents Washington through Nixon, that responding to congressional oversight requests is within the executive branch’s discretion as to whether compliance would advance the public interest. Once Congress prevailed in the federal courts concerning its requests for information from the Nixon White House,²¹⁹ and the Supreme Court concluded that executive privilege was not absolute,²²⁰ the executive branch moved to concede the legality of oversight in Congress’s investigative powers writ large while arguing that the activities and communications of the president, his advisers and high-level executive branch officials are entitled to a presumption of confidentiality.

Thus rather than rely on the noticeable absence of constitutional text, the federal government and the courts have determined that “oversight” is a power “so far incidental to the legislative function as to be implied.”²²¹ Because of the courts’ refusal to recognize the constitutional legitimacy of congressional delegation, Congress, the executive branch, and the federal courts have justified congressional investigations of the bureaucracy under the holdings established through disputes between Congress and individuals, where oversight is justified whenever it is for a legitimate legislative purpose.²²² This compromise has been counterbalanced by the OLC’s position that the Constitution creates a judicially rebuttable presumption of the confidentiality of executive branch communications.²²³

Rethinking Delegation of Legislative Power

Since the Founding, Congress has exercised the following distinct powers underexamined by scholars beyond the power to write publicly binding rules: (1) the power to investigate, and legally compel, private persons and businesses for regulatory purposes; (2) the ability to investigate the president and his copartisans for political purposes, with no power of legal compulsion; and (3) the power to adjudicate public rights granted to individuals in the form of private laws.

Congress’s ability to investigate the private sphere as well as its ability to expose the president are distinct powers with differing historical and legal developments yet scholars and jurists describe both as part of “congressional oversight.” And yet each one of these aforementioned powers, (1)-(3), has been delegated to administrative agencies. The legislative power of inquiry for a regulatory purpose was first delegated to committees of Congress then to the Interstate Commerce Commission and exists today in what is described as enforcement-related inquiries by agencies.

The power to adjudicate public rights for purposes of writing private laws was initially delegated to the joint Committee on Claims and then to the Court of Claims and has since been delegated to individuals in the form of statutory private rights of action. The ability to engage in political inquiries for electoral purposes (what is often meant by “congressional oversight”) has been delegated to the administrative state in the form of post-Nixon Title 5 civil service entities (e.g., the several Inspectors General, the Office of Special Counsel) as well as some Title 5 statutes (FOIA, Privacy Act, the Ethics in Government Act).

For as long as there has been delegation of legislative power, there has been oversight over that delegation. Congress has a political preference not simply to delegate its legislative powers (what I call “first wave delegation”) but to also delegate its oversight over that power (“second wave delegation”). Scholars have identified post-APA statutes that grant regulated parties the right to obtain judicial review of agency policymaking as the primary method for which Congress has delegated oversight over regulatory policymaking. The scholarship on delegated oversight has focused on Congress delegating its rulemaking authority and rather than overseeing such delegation directly, it establishes administrative procedures (where regulated parties can pull “fire alarms”) as its preferred means for controlling agency rulemaking.²²⁴ While it is true that when “Congress legislates, it delegates”²²⁵ the target entity is not necessarily an executive function notwithstanding the fact that what constitutes an executive branch versus, say, a legislative branch, agency is a question of law.²²⁶

Pre-enforcement Inquiries as Rulemaking

Before Congress ever delegated its policy-writing powers to the bureaucracy, it delegated its investigative powers over the private sphere. “Oversight” as the term is used arises from the political need to monitor delegated power. Throughout the 19th century, as Congress delegated investigative powers to committees and then eventually to agencies, Congress exercised oversight by retaining the power to make rules of law even as agencies conducted investigations of the private sector. When, in the early 20th century, Congress delegated rulemaking authority to the bureaucracy, oversight subsequently became more formal – typified by the Legislative Reorganization Act, which empowered standing committees to exercise bureaucratic oversight, and the APA, which codified a series of norms governing agency law creation. As these statutes were being debated in Congress, the Supreme Court, in Oklahoma Press Publishing v. Walling, affirmed the historical picture of an administrative state formed through delegation and oversight in holding that agency investigations of the private sector, backed by subpoenas, were a “delegated power” by Congress “incidental to both [Congress’s] general legislative and its investigative powers.”²²⁷ Thus, precisely four months before the APA became law, the Supreme Court held that administrative investigations were governed by the same requirements that applied to congressional investigations of the private sphere: where the power to compel documents and testimony is “an appropriate adjunct to the power of legislation” when cabined by a “legislative purpose” articulated in the authorizing resolution.²²⁸ In the context of the APA, then, it should be a rather obvious principle that agency investigations must be cabined in advance by a publicly noticed rulemaking purpose.

The history of the federal administrative state is punctuated by three great waves of delegation.²²⁹ First, Congress delegated its power to investigate private businesses and individuals to the executive branch through the creation of administrative agencies. Here, Congress reserved its rulemaking authority. In this sense, “rulemaking” was how Congress exercised oversight over the investigative bureaucracy. In the second wave, Congress delegated its rulemaking power to the bureaucracy. At this point, Congress created committees to engage in direct oversight of this more legislative bureaucracy. In the third wave, the present one, Congress delegates its oversight powers to the bureaucracy, relieving its committees of direct oversight responsibility and enabling them to focus on election-oriented activities. These wave reflect congressional attempts to increase the role of the federal government in promoting social welfare while maintaining congressional capacity to check federal power. Within these waves, the reservation of rulemaking while other legislative powers were delegated was understood as essential to preventing the abdication of legislative power. But rulemaking as an intelligible principle gave way to the institution of oversight which has given way to electoral gains – all supposed, within a given historical incrustation, as essential to legislative activity.

Delegation is the subject célèbre among public law scholars. A popular debate involves the claim that Congress delegated legislative power at the founding, thus casting doubt on the claim that the original understanding of Article I of the Constitution prevented delegation. The non-delegation doctrine deems any ceding of legislative power to the executive branch to be impermissible unless constrained by intelligible policy principles. The logic is that Congress must be in control of policy – not unelected bureaucrats. But such logic also betrays that those originalists who cast aspersions on delegation are not Article I apologists. The intelligible principle doctrine has a real world analogue familiar to political scientists and legislative actors alike: oversight. Congressional oversight has evolved to accommodate the collapsing of the historical and legal distinction between executive law enforcement and bureaucratic regulation as articulated in Part III. The political development of the administrative state began as an investigative state, evolved to a rulemaking state and is now retiring (or ossifying) into an enforcement state. Each wave of delegation of legislative power has transformed the institutional design of congressional oversight.

The social scientific theory of delegation as a means for Congress to achieve efficiencies is relevant to legal theories of delegation especially given the role administrative procedures play in constraining grants of legislative power. Our current model of American administrative law should conceive of the administrative state as following a form of punctuated delegation where the regulatory state was preceded by an investigative state and is followed by the current punctuation as an enforcement state. The congressional choice to delegate monitoring to regulated parties through fire alarm powers or to fire chiefs reflects the transformed nature of regulatory power and the utility of political pluralism as a model of contemporary regulatory power. Those delegatees of legislative power shape the regulatory agenda and the evolution of the regulatory state into an enforcement state is the result of political preferences of regulated parties and their counsel. In a sense, the sovereign is the lawyer whose argument on behalf of a regulated party becomes the rule of decision.

In this Article, I defended administrative pluralism by countenancing the strong critique against pluralism presented by Adrian Vermeule. The inferences drawn by rejecting Vermeule’s theory reveals a number of political features of American administrative law, including the overdispersion, yet underenforcement, of rules that place interest groups as a central lever of political power in administrative decision-making. American administrative law achieves both legality and legitimacy in being overdetermined by interest group politics. The political features of American administrative law, beyond the APA and its judicial interpretation, ensure both legality – where legal rules can accommodate exceptions to norms – and legitimacy – legal rules are not employed arbitrarily. What we observe is that our administrative law involves a robust cross-institutional dynamic: (1) congressionally-originated statutes empower interest groups to petition for federal judicial review of administrative decisions, (2) federal judicial decisions are constrained by the issues presented by the interest groups, (3) because precedent serves to codify statutory and regulatory interpretations, judicial decisions function with statutory force, and (4) both presidents and Congress can respond to judicial opinions through oversight of agencies and programs.²³⁰ These political features, where both political officials and courts can provide normative content to rules through enactment, adjudication, and oversight, ultimately reveal how our federal administrative law, inclusive of discretion, is rule-constrained, for administrative rules function as both formal procedures and consequentialist norms directed toward policy outcomes.

Since 1978 the administrative state has further evolved where those powers delegated to commissions, boards and independent agencies have now been delegated to executive agencies and cabinet departments, thus blurring the brightline between those agencies supervised by the president and independent agencies as well as the brightline between ministerial and discretionary executive power.²³¹ As such, those post-1946 powers delegated to committees have, beginning in 1978, been delegated to the executive branch. To use the Trump administration as an example, a substantial amount of the activities of the presidentially-appointed Inspectors General were incidental to congressional oversight and impeachment. While the law of delegation and oversight led to the creation of an investigative state, it was supplanted by a regulatory state where political circumstances motivate rethinking assumed legal principles. Congress’s compulsory powers have been subsumed by executive branch law enforcement yet such a state does not affect political incentives for oversight. Congress is thought to only delegate its political oversight powers to advance electoral interests. But if administrative investigations are auxiliary to law enforcement, the law should require the same constitutional procedures that apply to formal enforcement inquiries, rather than deferring to such activities as quasi-legislative.

Pre-enforcement agency investigations are adjuncts of rulemaking, not adjudication, and are not law enforcement activities otherwise excluded from the APA. As such, when agencies investigate prior to enforcement, those inquiries must be limited by a legislative purpose. Under the APA, rulemaking is simply the agency process for formulating a rule and pre-enforcement inquiries are not inconsistent with that definition. Chenery II states that an agency can choose between rulemaking and adjudication in how it makes policy. But contemporary administrative policymaking reveals that adjudication often involves enforcement of rules while processes antecedent to the formation of rules involve adjudication. An agency can set policy through orders by directly enforcing rules on the books before hearing examiners via adjudication or, if it seeks to investigate in order to determine whether rules apply to some corporate conduct, it must precede through rulemaking – literally the process of forming a rule.²³²

No agency may act without relying upon the plain rule established by statute or engaging in a public interpretation of a statute in order to clarify the rule or standard governing a proposed agency action. Whenever an agency other than the Federal Bureau of Investigation or an office subject to presidential control conducts an investigation of a business prior to the filing of a complaint, it is engaged in 552-rulemaking (as opposed to 553-rulemaking) and the jurisdictional basis for an investigation and the conduct prohibited or required in a complaint are rules which must be disclosed to a party in advance. Similarly, the power of the Comptroller General, the several Inspectors General and the public rights statutes codified in Title 5 (the APA, the Privacy Act, and the Freedom of Information Act) paint the picture of delegation not simply of rulemaking power but of the oversight power itself and thus to deem the panoply of powers held by the investigative agencies like the FTC, SEC, CFTC or CFPB as executive raises severe separation of powers concerns. Rethinking pre-enforcement inquiries as legislative, versus executive, is a solution to problems of nondelegation and due process that apply to such inquiries when considered within the executive power.

By identifying underenforced procedural constraints on regulatory conduct this Article contributes to the literature on the underenforcement of constitutional norms.²³³ Despite a host of statutes containing explicit standards for the judicial management of litigation, I find these rights remain fallow. The executive branch does not enforce these constraints leaving these rights’ viability conditional on private enforcement. Yet parties, through their representative interests (lawyers), also fail to challenge government infractions of these rights in court. Further research in this area will further examine why regulated entities do not become motivated litigants on behalf of the procedural rights identified in this Article. Additional research will also consider whether agencies’ compliance with these procedural norms of constraint is strategic. In other words, social scientists predict that if courts do not vitiate these rights, then affected parties will seek remedies with Congress through its oversight power. This suggests that agencies can choose to enforce such constraints depending on the ideological medians of Congress and the revealed oversight preferences of the key committees.

Regulatory Pluralism: An Empirical Analysis of Decision Making in Administrative Law

Executive Summary

Resources

Focuses

Introduction