An open access publication of the American Academy of Arts & Sciences
Summer 2021

The Uncertain Future of Administrative Law

Jeremy Kessler and Charles Sabel
View PDF

A volatile series of presidential transitions has only intensified the century-long conflict between progressive defenders and conservative critics of the administrative state. Yet neither side has adequately confronted the fact that the growth of uncertainty and the corresponding spread of guidance–a kind of provisional “rule” that invites its own revision–mark a break in the development of the administrative state as significant as the rise of notice-and-comment rulemaking in the 1960s and 1970s. Whereas rulemaking corrected social shortsightedness by enlisting science in the service of lawful administration, guidance acknowledges that both science and law are in need of continual correction. Administrative law has the resources to ensure that the provisionality of guidance does not lead to the abuses that conservatives fear. But to deploy those resources–and to carry through the reforms of administrative organization that are their natural complement–progressives must rethink their commitments to judicial deference to administrative authority and administrative deference to presidential authority, commitments on which the progressive defense of the administrative state currently depends.

Jeremy Kessler is Professor of Law at Columbia University. His forthcoming book is Fortress of Liberty: Conscription and Constitutional Change in Twentieth-Century America.

Charles Sabel is the Maurice T. Moore Professor of Law at Columbia University. He is the author of many articles in law reviews and academic journals. His forthcoming book is Fixing the Climate: Strategies for an Uncertain World (with David Victor, 2021).

In an interregnum, as the old order subsides and the new order struggles for definition and recognition, debate is often Janus-faced: now looking backward in an anguished attempt to save the receding world or put it in its best light; now looking forward to make the most of the possibilities for renewal that the emergent world affords. Such is the situation in American administrative law today. Under pressure for decades, the vulnerabilities of the administrative state were cast in harsh relief by the electoral victory of Donald Trump. This victory helped reveal and accelerate the erosion of public confidence in traditional models of growth and social mobility, the institutions of government, and the governing elite. The return of the White House to Democratic control will do little to abate this erosion. Within the legal academy and the federal judiciary, the long-smoldering debate between progressives and libertarian-minded conservatives over the New Deal legacy has burst into flames. Progressives defend current arrangements–and the internal operations of the administrative state in particular–as harmonizing professional expertise, energetic presidential direction, and the rule of law. Their opponents see an autonomous administrative state, largely playing by the rules that administrators have created for themselves, as a growing threat to the balance of judicial authority, legislative sovereignty, and presidential democracy on which constitutional freedoms depend. This debate can be so absorbing that its participants–losing sight of the mixture of achievements and shortcomings that make the actual administrative state a vexation, but perhaps a tolerable one, for the public–come to suspect that but for the machinations of the other side, there would be no real problem at all.

But alongside this first debate is a second, focused resolutely, yet without fanfare, on what administrative law can or should become. Its point of departure is the growing importance of uncertainty–the inability to anticipate future states of the world with enough confidence to assign them probabilities–to administrative decision-making. As is often the case in the early stages of epochal change, there is little attempt to give precise definition to the shift in progress, measure its extent, or explain its causes. Instead we are confronted with arresting descriptions of new developments, counterintuitive to the world we thought we knew and inexplicable but for the postulated break with the past.1

For one leading commentator, Adrian Vermeule, the rise of uncertainty turns many substantive decisions into “coin flips,” foreclosing the reasoned ranking of alternatives or “first order” reason-giving that makes administrative decisions orderly and intelligible to reviewing courts. Outcomes will depend on “second order” considerations having nothing to do with the law, such as the determination that, right now, any decision is better than none, or that one policy is more easily administrable than equally imperfect alternatives.2 Judges, recognizing this limit, should–and, more often than not, do–defer broadly to administrative discretion rather than futilely interrogate administrative actions that cannot be justified on conventional grounds. For Vermeule, a strong administrative state, freed by the judges themselves of all but the most minimal judicial check, is indispensable for realizing the popular will in otherwise uncertain times. That such a state does not conform to liberal democratic norms is no loss for Vermeule; that it invites domination by strong, popular leaders is if anything a gain.

For administrative law scholar Nicholas Parrillo, by contrast, the effect of uncertainty on the administrative state has been the “oceanic” spread of guidance: a provisional form of regulation that tentatively advises private parties and public officials about how an agency intends to exercise its discretion or interpret its legal authorities.3 In issuing guidance, an agency can give regulated entities an understanding of what compliance means right now, without committing itself to what will be required of regulated entities in the near future, should conditions change or new facts come to light. The public and private practitioners whose views inform Parrillo’s account regard guidance as “essential to their missions.” A former senior Food and Drug Administration official “cannot imagine a world without guidance”; according to one current Environmental Protection Agency official, guidance is “the bread and butter of agency practice.”4

But just as a firm, having achieved economies of scale for efficiency reasons, may use its size to compete unfairly, so Parrillo finds that the use of guidance as a legitimate response to uncertainty can open the way to abuse. To advance their own institutional interests or win the favor of particular constituents, administrators may use guidance too flexibly, avoiding the procedural burdens that normally must be assumed to change policy, and frustrating the consistency required by the rule of law. Or administrators may use guidance too inflexibly, knowing that, once issued, guidance creates de facto legal obligations and permissions, and thus as a practical matter can regulate with the force of law while again evading the procedures normally required to do so–and often judicial scrutiny as well.

For Parrillo, judicial efforts to winnow the good uses of guidance from the bad have reached their limit. If guidance is to serve as an adaptive response to uncertainty while limiting the risks of abuse, the administrative state should itself be reformed to further develop the institutional capacity for continuous but lawful adjustment to change and variety. Parrillo calls this capacity “principled flexibility”: a method of decision-making in which requests for the revision of guidance are easily made (reducing the risk of expedient compliance by habit) and exposed to appropriate discussion within the agency and its public (reducing the risks of clientelism). Principled flexibility implies an innovation in administration, already visible in some agencies, that connects front-line decision-makers with their superiors, on the one side, and regulatory parties and beneficiaries, on the other, to ensure an integration of rule-application (or enforcement), rule-making, and rule-revision that conforms to the public interest and the rule of law, not simply to the self-interest of an agency as an organization or its lobbies.

Neither Vermeule’s nor Parrillo’s views provide anything approaching a comprehensive vision of a future administrative state, nor were they intended to do so. In important ways, they are in error. But they do clear a path to exploring how, in response to deep changes in the very circumstances of decision-making, administration has begun to purposefully adapt, and might well emerge from the interregnum better equipped to meet, effectively and legitimately, the demands of a volatile world. That is the path that we follow and try to extend in this essay.

We begin with the premise that the growing reliance on guidance–a kind of provisional “rule” that invites its own revision or qualification–marks a break in the development of the administrative state on the order of the transition from regulation by case-by-case adjudication to regulation by notice-and-comment rulemaking in the 1960s and 1970s. Where rulemaking aimed to correct social shortsightedness by applying science in the service of lawful administration, guidance–sprung from uncertainty–enables administration in the public interest when both science and law are recognized as themselves in need of continual correction. Guidance is the kind of law that uncertainty admits.

The centrality of guidance to contemporary administration, in turn, calls into question politically progressive defenses of the administrative state. These defenses ground the legality and legitimacy of administrative decision-making in the technical authority of professional administrators and the democratic authority of their political superiors, two forms of authority that are harmonized and largely self-disciplined by internally generated hierarchical norms, which also serve as a shield against intrusive judicial review. Today, such deference to authority and hierarchy seems incautious, even reckless. Even when understood as a strategic bargain, necessary to protect the autonomy of the administrative state from a judiciary suspected, often rightly, of congenital animosity to administration itself, the progressive defense resembles a pact with the devil.

The emerging law of guidance, and the reality of uncertainty to which it responds, points toward a different, and more defensible, conception of the administrative state, one that is aware of its own fallibility, that routinely invites challenges to its technical and political authority, and that continually responds to these challenges with reasons that are legible to the courts and to the public at large.

This is an essay in reinterpretation, and exploratory reinterpretation at that. Few if any of the observations or references to empirical developments are original to us. What is different is the perspective. Most commentators, and especially the ones from whom we have learned the most, strive to fit guidance within an overarching structure of mutually reinforcing doctrinal and organizational elements: the traditional administrative state, put in its best light. When the fit is awkward, the commentators, as the excellent lawyers they are, explain the discrepancy while preserving the structure. We see the same incongruities and ask instead if they might not prefigure a new frame and, if so, how elements of a reconfigured administrative state might fit into it. We go just far enough in that direction, we hope, to suggest the plausibility of switching figure and ground.

Though we want to show that the increase in uncertainty puts tectonic pressure on the administrative state, we begin by arguing that legal reasoning, of the kind familiar to appellate courts, does not run out under uncertainty. In suggesting otherwise, Vermeule adopts the vantage point of the lone judge or administrator, deciding cases on the basis of records whose quality gradually degrades as new and puzzling circumstances arise, beyond the range of current understanding. When familiar sources of evidence become less reliable or irrelevant, the only way to decide is by tossing a coin or appealing to second-order  reasons.

But uncertainty does not come to agencies in isolated instances or as a gathering nebulosity that defeats understanding. It arrives as a trickle, then a stream, and then a flood of indications that new approaches are needed. Often it is the agency itself that makes the limits of knowledge manifest and actionable, by proposing to set a standard at or beyond the frontier of technology and, in doing so, calling attention to the uncertainty that must be overcome to make compliance eventually possible. Either in response to puzzling novelty or as part of a deliberate effort to press beyond the known, the agency actively seeks the cooperation of all the actors in a position to learn a way through uncertainty. Such cooperation is likely to be forthcoming as regulators, regulated entities, and the beneficiaries of regulation, jointly ignorant of the risks they may face, share a vulnerability that can motivate joint action.5

There should be nothing surprising in such a reaction. We expect the rational actor, facing uncertainty, to inquire after further information the better to direct action. Nor should it be surprising, however, that organizing such inquiry is easier said than done. Ideally, when decisions are reversible and information is easily available, action and inquiry can proceed together, so that initial decisions are corrected without loss in light of later knowledge, even if, that too, remains provisional. In fact, information is typically costly to acquire–not least because at the outset of inquiry it is not clear what needs to be known–and decisions can turn urgent and irreversible. Furthermore, inquiry can and from time to time does fail, though failure does not discredit the enterprise as a whole, any more than science as a whole is discredited by its reverses. In practice, what is called for in the face of uncertainty is “measured action,” in which each step is intended to inform the next.6 We can think of measured action as straddling the boundary between firstand second-order reasons, or, rather, devising by second-order considerations the means for enabling first-order choosing. Reducing administration to a choice between firstand second-order decision-making ignores the core of what administrators do.

A prerequisite of measured action in regulatory settings is the formation of regimes that signal when and where further inquiry is opportune or urgent, and when additional measures must be taken to further defer final decision or to act decisively.7 Regimes directed to this purpose arise, for example, when regulators: induce monitoring of ecosystems for the protection of particular species or the environment generally; mandate extensive reporting of the failures or side effects of products authorized for sale to detect latent hazards; or organize information-pooling to set or reset (technology-forcing) limits on permissible levels of pollution as knowledge of what counts as dangerous exposure to the pollutant, the technologies of pollution control, and the costs of the latter change rapidly and unpredictably.8 Often resource constraints hamper formation of these regimes; sometimes their operation is frustrated by raw conflicts of interest or the very rigidities of administrative law that we will discuss in relation to guidance.9 In the best cases, such regimes allow the agency to cooperate with outside actors–firms, NGOs, property owners, other public bodies–to learn what can be known; together they find a way through uncertainty that none could have found alone.

Are the reasons given to explain and justify the choice of a particular variant of this kind of rulemaking process, or to evaluate its performance with regard to a given outcome, unintelligible to lawyers generally, and especially to courts? It is hard to see why. Evaluation of first-order reason-giving will focus on the validity of the process by which reasons are produced, because it is easier for outsiders to judge the suitability of the decision-making process than the substance of an arcane decision. Such outsiders can be expected to ask: Has there been an effort to canvas all the available evidence? Have competing views been considered and, as needed, reconsidered? Were channels left open for the easy registration of dissent? Evaluation of the suitability of an information-gathering regime under uncertainty will focus on process in the same way.

In administrative law, this kind of interrogation of the reasoned quality of decision-making has since the 1970s been associated with “hard-look” judicial review or, more helpfully, process review, to distinguish it from judicial review of agency conformity to required procedures, or of the overall adequacy of some regulatory output.10 As its name suggests, hard-look review requires an agency to assure that it has given searching consideration to relevant alternatives. From the first, this check on the conformity of the chain of administrative decision-making to a minimum standard of rationality has been understood to include the agency’s response to uncertainty. Perhaps the best statement of the requirements of process review, the D.C. Circuit’s 1974 opinion in Industrial Union Department, AFL-CIO v. Hodgson, held that when an administrator is “obliged to make policy judgments where no factual certainties exist or where facts alone do not provide the answer, he should so state and go on to identify the considerations he found persuasive.”11 Courts are thus equipped to interrogate agency management of monitoring regimes. Indeed, many of the cases that courts hear in which first-order reasons are insufficient, leaving the agency to decide by coin toss or appeal to second-order reasons, arise from the failure of a poorly designed or carelessly operated information-gathering system, with the judge called on to evaluate the legal and practical merits of competing remedies. The question before the court (including in many of the cases Vermeule cites to illustrate his view of reason-giving under uncertainty) is not only what the agency should do given that it does not have the information needed to decide, but rather how it got into this situation in the first place, and what to do about that.12 We are not suggesting that lawyers or courts, by asking such questions, should take the lead in designing the agency process of decision-making under uncertainty any more than they currently do. Our point is only that there is nothing alien to the legal mind in judging the sufficiency of an agency process against criteria no more or less vague than those used to evaluate the adequacy of administrative decision-making under more familiar conditions.

As public and private administrators come to appreciate that they are working under conditions of real uncertainty, they will seek whenever possible to make provisional decisions, learn quickly from them, and revise accordingly. They will use guidance, a form of administrative action defined by its provisionality. Guidance is an extraordinarily heterogeneous category, comprising all those written documents that administrators issue to inform their own staff and the public at large about how they currently understand and intend to enforce their legal authorities, whether derived from congressional statute or prior administrative regulation. As Parrillo notes, guidance comes in countless forms–“advisories, circulars, bulletins, memos, interpretive letters, enforcement manuals, fact sheets, FAQs, highlights, you name it”–and is nowhere systematically collected. As a rough estimate, the number of pages of guidance that agencies produce “dwarf[s] that of actual regulations by a factor of twenty, forty, or even two hundred.”13

Formally, the signal feature of guidance is its provisionality. Agency rules are generally issued and amended by means of a costly and time-consuming process, called “notice-and-comment” or “legislative” rulemaking, in which the agency must elaborately explain its purposes, expose its evidence-gathering and deliberation to public scrutiny, and explain its reactions to criticism. Guidance, in contrast, can be issued and amended quickly, with little if any formal process.14 Because of this informality, guidance lacks “the force of law” and thus “the power to bind” private parties characteristic of agency rules. Guidance is, formally, “only a suggestion–a mere tentative announcement of the agency’s current thinking about what to do in individual adjudicatory or enforcement proceedings, not something the agency will follow in an automatic, ironclad manner as it would a legislative rule.”15 As such, guidance not only permits but demands flexibility: “If a particular individual or firm wants to do something (or wants the agency to do something) that is different than what the guidance suggests, the agency is supposed to give fair consideration to that alternative approach.”16 Similarly, while an agency may choose to depart from its guidance without formal process, it should in principle give a reasoned explanation for such departures.17 Understood in this way, guidance is a tool for measured action.

Yet government-by-guidance has provoked significant legal controversy, for at least three related reasons.18 First, as notice-and-comment rulemaking became accepted as the paradigmatic mode of administrative rulemaking, the less procedurally onerous issuance of guidance began to strike some scholars, litigants, and judges as a potential cheat. As Justice Elena Kagan put it during an oral argument in 2015, this is the recurrent concern that “agencies more and more are using interpretive rules and are using guidance documents to make law and that . . . it is essentially an end run around the notice and comment provisions.”19 A second, related fear is that because the provisionality of guidance documents makes them difficult to challenge in court, agencies can use guidance to evade not only the pre-issuance notice-and-comment process but also post-issuance judicial review as well.20 A final concern relates to the proliferation of doctrinal deference to agencies’ interpretation of their statutory mandates and prior regulations. Critics warn that such deference perversely shelters agency interpretations announced in guidance documents from judicial scrutiny, even though they do not reflect the deliberation and evidence-gathering required by the notice-and-comment process or by formal agency adjudication.21

Underlying these technical legal objections are deeper normative concerns about the relationship of regulation as “current thinking” to conventional forms of legal authority–legislative, executive, and judicial–that help to explain why guidance continues to bedevil American courts and legal commentators. Guidance, unlike notice-and-comment rules, cannot be seen as analogous to and directly descended from legislation as a natural outgrowth of the constitutional order. But neither does guidance have the finality that marks the culmination of lawful executive or judicial action. Unlike prosecutorial indictments and administrative enforcement actions, it does not purport to represent the executive branch’s determination that a particular private party has violated the law; unlike administrative orders and judicial decisions, it is not an assessment of the guilt or liability of an accused party. Guidance is always ripening into a conclusive decision, but it is never ripe; for this reason, unlike administrative rules and orders, it is not reviewable by the courts as a matter of course.

In the paradigmatic forms of legal decision, the ultimate decision-maker is the agent of a constitutional principal–the legislature, the executive, the judiciary– or one of the constitutional principals themselves. Each of the latter is in turn ultimately the agent of the popular sovereign. In general, a principal sets the framework within which decisions are made and reviews an agent’s actions for conformity to its intent. With guidance, and under uncertainty generally, this principal-agent relation breaks down. The framework for decision-making becomes indistinct as the field of possible actions and the criteria for evaluating them become less and less determinate, especially as seen, by the principal, from afar. The diligent administrative “agent,” in dialogue with external actors (regulated parties and regulatory beneficiaries) as much as or more than with internal superiors, is as likely to make decisions presuming new frameworks as to apply old ones. Accordingly, the same circumstances that make guidance inherently provisional make it inherently refractory to the forms of control that traditionally confer legal validity and democratic legitimacy. To the extent that guidance regimes increasingly reflect ongoing collaboration among a hierarchically diverse group of public and private actors, the content and legal effect of guidance at any given moment will have an increasingly attenuated (or at least increasingly uncertain) relationship to legislative mandates, presidential directives, or judicial orders. However scholastic or opportunistic debates about the legality and legitimacy of guidance can be, they articulate a very real problem: whether liberal democratic societies are capable of managing uncertainty without reinterpreting what it means to make law, and the values that underpin lawmaking, when law itself must often be provisional.

Well-established groups of practitioners in and around the agencies, anticipating our own approach, have over the years proposed working solutions to this problem, devising new standards by which guidance regimes can be developed with principled flexibility in mind, while remaining legible to courts and the public at large. The practitioners’ understandings are reflected in a variety of “institutional pronouncements”–by the American Bar Association, the Administrative Conference of the United States, and guidance about guidance produced by the agencies themselves–which describe the sorts of reasons that agencies should be prepared to give to regulated parties, regulatory beneficiaries, and the courts whenever they depart from or adhere to a practice or norm established by prior guidance documents.22 Like the commentators who refer approvingly to their views, the practitioners have no broad program of reform. But it is worth noting that, since the 1990s, successive cohorts have been at the forefront of such standard-setting, not resistant to it–as conservative fears about the widespread abuse of guidance might lead one to conclude.

Many liberal and left-leaning scholars, meanwhile, are so intent on defending the administrative state–conceived along midcentury lines–from conservative rollback that they are insufficiently attentive to the accumulating innovations in practice that might provide resources for responding to real deficiencies in the legality and legitimacy of administrative decision-making. Preoccupied with preservation, the progressive defense of the administrative state has given rise to an organizational and doctrinal synthesis of the governance mechanisms–agency expertise, presidential oversight, judicial deference–that, taken together, arguably legitimate in a novel and compelling way the traditional, and implicitly unchanging, forms of administrative governance. Below, we present this interpretation of administrative law and indicate its shortcomings as a framework for measured action under uncertainty.

The progressive synthesis is our name for a constellation of arguments that ground administrative legality and legitimacy in external political control–presidential control in particular–and internal professionalism. This synthesis is the product of historical evolution, in which elements with different origins co-evolve, in the manner of the components of what would become the human eye, as they are enlisted into the service of a common function. Like all evolutionary stories, it has a “just so” aspect. We present it as a fully integrated whole to highlight its basic institutional and doctrinal commitments.23

Here, in outline, are the essentials: Externally, or at the outer boundary of the administrative state, Congress delegates authority to administrative agencies and the president directs how agencies wield it. Internally, this authority triggers processes of deliberation and decision conforming to the professional habits and norms of the scientific and legal experts who work within individual administrative agencies and supervisory institutions, such as the Office of Management and Budget (OMB). Thanks to this layer of professional mediation, the ultimate expression of congressionally delegated and presidentially directed authority– administrative decisions–can be expected, most of the time, to be both scientifically rational and legally valid. Energetic and continuing presidential direction, meanwhile, ensures that administration is not frustrated by the bureaucratic torpor and rivalry common to large organizations, and that decisions have the necessary democratic pedigree, even when they are far removed, in time and effect, from what Congress might have contemplated when first delegating authority to a particular agency.24

The synthesis of presidentialism and professionalism is effected by and embodied in internal administrative law: a body of norms, growing out of the practice of administration itself, that harmonizes the demands of political will, legal regularity, and scientific rationality. In the progressive account, internal administrative law is hierarchical: within agencies, it subordinates lower-level decisionmaking to higher-level review, with the head of the agency as the ultimate authority; within the executive branch, it subordinates agencies to the direction of the president. Institutionally, the routine operation of these nested hierarchies is policed by superor meta-agencies–such as the OMB and, within it, the Office of Information and Regulatory Affairs–housed within the Executive Office of the President.25

Because presidentialism and professionalism continually imbue administrative decisions with democratic legitimacy, legal validity, and scientific rationality, courts lose their centrality as the guardians of the constitutional conformity of the administrative state. Partisans of the progressive synthesis rarely disclaim the virtues of judicial review altogether, but they focus on its vices: the tendency of courts to subordinate the rationality of the administrative process to their own professional and political preferences. These limits warrant a general policy of judicial deference to administrative decision-making. More often than not, courts should avoid reviewing administrative decisions at all; when review is called for, courts should defer to administrators’ expert views of the meaning of the laws their decisions implement, the procedures that were necessary to get the job done, and the evidentiary and policy rationales for deciding one way or another.

Whatever its merits as a response to anti-administrative attacks, the progressive synthesis is deficient in important ways. First, it is conceptually incomplete, if not incoherent. As has been manifest since the New Deal, decision-making by political directive and decision-making by expertise proceed by different methods. They may reach the same or compatible solutions, but they may not. The synthesis provides no process or principles for systematically reconciling them. Instead, following the promptings of doctrine, it assumes that both can be exchanged into the common currency of hierarchy and made interchangeable. But as the countless conflicts between the previous administration and the professional staffs of the agencies attest, the reality is otherwise. In fact, courts are routinely called on to decide whether, in a particular case, deference is owed to hierarchical authority of one kind or another, political or professional. Familiar tensions are recast, not overcome.26

These tensions point to a second and politically salient defect of the synthesis: its limited resources for responding to presidential overreach. For the progressive synthesis, the chief threat to constitutional democracy continues to be, as it has been since the New Deal, a conservative judiciary and, more generally, the frustration of decisive and synoptic leadership, whatever its origin. Expansive defense of doctrines of judicial deference to administrative hierarchy responds to the first danger, while defense of the authority of the Executive Office of the President to supervise, coordinate, and direct decision-making across the administrative state responds to the second. But when the corrective of an empowered White House itself becomes a threat–perhaps the threat–to democratic, lawful, and effective administration, these responses become worse than useless.

A third deficiency goes to the inadequacy of the progressive synthesis as a response to uncertainty. The synthesis validates authority in its various forms while seeking to coordinate their exercise. Uncertainty calls authority in all its forms into question. It does not, to be sure, simply devalue expertise or political leadership. But it emphasizes the role of the expert and the political leader as organizers of open-ended inquiry, rather than as repositories of tried-and-true solutions that must merely be adapted to new contexts.

If not by the progressive synthesis, how then can administrative law help make professional expertise and political control usefully commensurate, in ways that improve the capacity of the administrative state to respond to uncertainty, conform it to minimal liberal democratic norms, and limit the dangers of presidential overreach? We offer a provisional answer to this question in the form of an alternative doctrinal program and a reminder of the fundamental need for organizational reform.

Our programmatic alternative treats uncertainty not simply as a challenge to be met, but as a guide. Uncertainty is the great leveler. In revealing the limits of current knowledge, it limits the claims of authority–all forms of secular authority at least–and presses authorities of all sorts into a more or less open inquiry that continually updates the nature of a given problem and its provisional solutions.

Such inquiry, moreover, regularly brings to light the interdependence of disparate types of authority, as political action is found to have technical prerequisites and vice versa. Forms of administrative law and organization that improve the quality of reason-giving and exchange between the political and technical authorities, and enable courts and the public to better judge these authorities’ respective claims, make the administrative state more effective in responding to uncertainty and more likely to respect core liberal democratic norms in doing so.

The alternative program emphasizes that judicial review, despite its susceptibility to ideological capture and prolonged history of anti-administrativism, remains an indispensable tool for resolving conflicts between presidentialism and professionalism, sociopolitical and sociotechnical reason-giving.27 Current doctrine invites litigants and judges to attack guidance as practically binding regulations that should have gone through the notice-and-comment process. We agree with the progressive view that this doctrine should be reformed, not least to avoid formalistic maneuvering on all sides that leaves administrative action less susceptible to post-issuance input, learning, and revision on the fly. But we part company with the progressive synthesis when it comes to the availability and scope of judicial review of guidance.

The traditional argument against the reviewability of guidance begins in prudence but ends in formalism: guidance is said to lack the finality and ripeness necessary for effective and lawful judicial review under the Administrative Procedure Act (APA), the Constitution, and the common law. The objection to this argument is that guidance is so well-suited to regulating in an uncertain world precisely because it encourages participants in the administrative process to act now: to experiment with new technologies and methods of organization, to share information, and to give reasons why other participants should act differently in light of these experiments and information exchanges. If a proposed guidance regime is legally or practically flawed, and that flaw is presently known, it should be presently addressed. Otherwise, the resulting regime risks discouraging the exploration and experimentation it is meant to foster.28

With respect to the scope of judicial review, the progressive synthesis endorses and would extend a body of doctrine that focuses on whether guidance documents are permissible interpretations of preexisting statutory and regulatory provisions. Instead of asking how well-considered, responsive to objections, and consistent with prior explanations agency reason-giving is, these doctrines encourage judges to determine only whether administrators are offering minimally reasonable interpretations of the statutory and regulatory provisions that authorize them to regulate at all.29 In recent years, progressives have argued that this type of reasonable-interpretation inquiry should, in most circumstances, be as narrow and formalistic as possible; a semantically plausible interpretation–an excuse that can be understood as excusing–is good enough. This formalism may well limit the extent to which judges interpose their own views for those of administrators, but it does so at the price of obscuring and thus tolerating or even encouraging incompetent or self-serving decision-making.

Our alternative program would refocus courts and agencies on what some commentators have called process review: review of the chain of reasoning that has led an agency to adopt and maintain a particular guidance regime.30 Such review asks whether the agency’s decisions in the form of guidance are well-considered, responsive to objections, and consistent with prior explanations of agency behavior and previously established regulatory presumptions. When an agency departs from prior explanations or presumptions, or chooses to maintain them despite objections, process review asks whether the agency has acknowledged its departures (or refusals to change course) and whether it has given plausible reasons for its choices. So described, process review is distinct both from procedural review, which asks whether a guidance document is so impactful that it should have been issued by more onerous procedures (such as notice-and-comment rulemaking), and from outcome review, which asks whether a guidance document is substantively reasonable in light of the evidence the agency had at its disposal. The deferential review of an agency’s interpretive choices preferred by progressives is one kind of outcome review, while the suspicion that agencies use guidance to avoid notice-and-comment rulemaking frequently leads conservatives and more classical liberals to advocate an exacting kind of procedural review.31 Process review steers clear of these extremes and, more importantly, encourages agencies to acknowledge uncertainty and manage it in a reasonable manner.32

Happily, there are resources in current law that permit judicial review of agency action under uncertainty to take the form of process review and, in doing so, improve the quality of administrative reason-giving. One such resource is hardlook arbitrary and capricious review, the standard that judges apply when reviewing an agency’s exercise of its policy judgment rather than its interpretive acumen. While hard-look review can resemble either process or outcome review, depending on the kinds of questions that a court asks when assessing an agency’s policy judgment, the canonical hard-look case, Motor Vehicle Manufacturers Association of the United States v. State Farm Mutual Automobile Insurance (1983), nicely exemplifies the sort of process inquiry we favor.

Another doctrinal resource that already encourages process review is the Skidmore framework.33 Skidmore instructs judges to give weight to an agency’s legal interpretation according to “the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade.”34 Subsequent doctrinal developments significantly narrowed the reach of Skidmore in favor of more hierarchically minded, formalistic, and deferential standards, such as Chevron (which directs judges to accept facially reasonable interpretations of an agency’s ambiguous statutory mandates) and Auer (which directs judges to accept facially reasonable interpretations of an agency’s own ambiguous regulations). These latter standards are today most vociferously defended by progressives. Recently, however, judges and scholars from across the political spectrum have begun to explore the possibilities of Skidmore anew.35

In the 2018 Kisor decision, Justice Elena Kagan, whose defense of Clintonian presidential administration has become a keystone of progressive thinking, wrote an opinion narrowing the scope of Auer deference, much to the disappointment of progressive scholars. Kagan’s opinion directed lower court judges to ask a set of threshold questions before applying Auer when reviewing agency interpretations of their prior regulations. These threshold questions, somewhat analogous to ones that already confine the application of Chevron in statutory interpretation, effectively move away from the formalism of Auer and back toward the more holistic Skidmore endeavor of calibrating the degree of deference to the persuasive quality of agency decision-making.

Our alternative program would accelerate this movement, urging, for example, that guidance documents always be evaluated under Skidmore, whether they are construed as interpreting an agency’s prior regulations or its statutory mandates. Going further, the alternative program would extend Skidmore to notice-and-comment rules, displacing the Chevron framework altogether and thus limiting agency incentives to regulate in a more inflexible manner when guidance would do.

Of course, the pursuit of Chevron deference is not the primary reason that agencies regulate by notice-and-comment rulemaking. Statutory requirements and judicial expectations, as well as genuine desire for public input, all drive agencies toward the notice-and-comment process, at least some of the time. To the extent that many so-called statutory requirements are themselves products of judicial interpretation (whether of the APA, an agency’s organic statute, or both), this is another area where our alternative program would recommend greater, rather than less, judicial deference. Unless an agency’s organic statute explicitly requires notice-and-comment rulemaking, judges should permit the agency to proceed by guidance.36

Finally, the alternative program would recommend that hard-look review and the Skidmore inquiry, both understood as process rather than outcome tests, be used more or less interchangeably. In this way, judicial review of the agency’s reasoning process would converge on a single standard in nearly all cases, mitigating the incentives agencies have to choose one form of rulemaking (more rather than less formal) or reason-giving (interpretive rather than policy-based) over another in order to garner greater judicial deference.37

Together these recommendations recast judicial review, at least with respect to guidance and thus action under uncertainty, as asking not whether agency decisions possess a sufficiently hierarchical pedigree to merit deference, but whether they give reasoned explanations for action (or inaction).

Our alternative program is not, however, simply a call to recenter the courts. We recognize that internal administrative law and organizational reform are vital complements to process review as a means of improving agency reason-giving. The progressive synthesis, however, is drawn to a peculiarly hierarchical conception of internal administrative law. By embedding administration in hierarchy, the synthesis seeks to reconcile presidentialism and professionalism and, in doing so, justify a more limited role for the courts. But under uncertainty, this approach imputes a burden of certitude on technical and political authorities that neither can bear.

Administrative law scholar Jerry Mashaw’s historical reconstruction of internal administrative law, informed and in some measure inspired by his investigations of decision-making in federal agencies around 1980, suggests an alternative interpretation, one that is less hierarchical and more capable of managing uncertainty.38 Mashaw understands internal administrative law as a system of quality control by which an agency corrects flaws in its process of decision-making to better serve its public purpose, where that purpose notably requires the process itself to be perceived as fair. Internal administrative law as quality control is always relative to the kind of decisions whose quality is in question. When, as was the case when Mashaw did his empirical research, the public interest was served by the hierarchical application of fixed rules to decide the validity of individual claims, internal administrative law aimed to improve the quality of the information available to the hierarchy and the reliability of its treatment of that information. But when, as is increasingly the case, the public interest under uncertainty is served by full and fair ventilation of reasons for action in changing contexts that resist hierarchical control, internal administrative law turns to improving the quality of those processes. Thus, one of the most thorough recent case studies of internal administrative law documents such a shift, showing how the Environmental Protection Agency replaced an unworkable standard-setting procedure–which advanced from regulatory goal-setting to the choice of regulatory means in the familiar hierarchical or principal-agent sequence–with a nonhierarchical process that invites criticism of emergent decisions at each stage of their development.39 Enlarged to include, as it plainly does, nonhierarchical routines, internal administrative law can indeed be an instrument for harmonizing different forms of expertise. But like judicial review of process, it does so by requiring the clarification of differences, not by assuming that different ways of knowing and deciding can be ranked according to the degree they are authoritative.

Both process review and internal administrative law, properly understood, can improve the incentives for sound reason-giving. But neither separately nor together can they substitute for the demanding organizational reform of agencies needed to make principled flexibility a practical routine. For Parrillo, this rewiring of the circuits of decision-making is paramount: “Mitigating the binding power of guidance documents entails interventions that are essentially structural and managerial.”40 His research reminds us again and again of the betwixt-and-between character of our agencies, with front-line departments rigidly enforcing rules and a separate, superior department making and revising them, even as questions of private-sector “compliance” increasingly implicate questions about the advisability of rule changes. Often, we expect, reorganization aimed at overcoming these barriers to the fluid internal collaboration required for principled flexibility will trigger corresponding changes in internal administrative law; sometimes revisions of internal law will trigger further institutional reform. In either case, change requires both reformed practices and new legal norms validating them.

In the end, going beyond the progressive synthesis means allowing administrative innovation in the face of uncertainty to run its course. Like the progressive defense of the administrative state, our alternative program depends on administrators acting in good faith, most of the time. But the alternative program breaks with the progressives’ reliance on the formalism of principal-agent relationships and hierarchically ordered authorities to legitimate administrative action.41 That reliance is out of character, given progressivism’s origins in a revolt against formalism. But above all, it is self-defeating. Against the conservatives’ formalistic objections to administrative governance, progressives have taken refuge in a formalism of their own, one that fetters administration with an outmoded model of decision-making too self-confident for an uncertain world.

© 2021 by the American Academy of Arts & Sciences. Published under a CC BY-NC 4.0 license.


  • 1For an influential study in this genre that links the failure of traditional forms of product development and marketing, and the need for alternative approaches, to the rise of uncertainty, see Clayton M. Christensen, The Innovator’s Dilemma (New York: Harper Business, 2013).
  • 2Adrian Vermeule, Law’s Abnegation: From the Law’s Empire to the Administrative State (Cambridge, Mass.: Harvard University Press, 2016), 125–154.
  • 3Nicholas R. Parrillo, “Federal Agency Guidance and the Power to Bind: An Empirical Study of Agencies and Industries,” Yale Journal on Regulation 36 (1) (2019): 165.
  • 4Nicholas Parrillo, Federal Agency Guidance: An Institutional Perspective (Washington, D.C.: Administrative Conference of the United States, 2017).
  • 5Under stable conditions, cooperation is typically impeded by an information asymmetry: because the regulator is less knowledgeable about risks and the costs of abating them than the regulated entity, the latter has an incentive to conceal information in order to reduce the eventual costs of compliance. Under uncertainty, ignorance is symmetric.
  • 6Michel Callon, Pierre Lascoumes, and Yannick Barthe, Acting in an Uncertain World: An Essay on Technical Democracy (Cambridge, Mass.: The MIT Press, 2009).
  • 7On such regimes generally, see Charles F. Sabel and William H. Simon, “Minimalism and Experimentalism in the Administrative State,” The Georgetown Law Journal 100 (2011): 53.
  • 8See, for example, Michelle Ouellette and Charles Landry, “The Western Riverside County Multiple Species Habitat Conservation Plan: Looking Forward After Ten Years,” Natural Resources and Environment 29 (3) (2014): 40; and Robert Ball, Melissa Robb, Steven A. Anderson, and G. Dal Pan, “The FDA’s Sentinel Initiative–A Comprehensive Approach to Medical Product Surveillance,” Clinical Pharmacology & Therapeutics 99 (3) (2016): 265–268. See the detailed explanation of the process by which the Environmental Protection Agency periodically resets the Nation Ambient Air Quality Standards in Sidney Shapiro, Elizabeth Fisher, and Wendy Wagner, “The Enlightenment of Administrative Law: Looking Inside the Agency for Legitimacy,” Wake Forest Law Review 47 (2012): 463.
  • 9J. B. Ruhl, “Regulation by Adaptive Management–Is It Possible?” The Minnesota Journal of Law, Science & Technology 7 (1) (2005): 21. In a subsequent article, Ruhl and Robin Craig argue that the continuing adjustments of policy required for ecosystem governance fit awkwardly within the procedures of conventional administrative law, and that judicial oversight of such regimes would be better exercised by creating a procedural “track” that takes account of their distinct features. See Robin Kundis Craig and J. B. Ruhl, “Designing Administrative Law for Adaptive Management,” Vanderbilt Law Review 67 (1) (2014): 1. Daniel Tarullo argues on the same grounds for a “more tailored” application of administrative law regarding bank supervision. Daniel K. Tarullo, “Bank Supervision and Administrative Law,” paper presented to the Yale Journal on Regulation 2019 Administrative Law Conference, Washington, D.C., 2019. The accumulation of such weighty exceptions with common features points to the opportunity for a new rule.
  • 10For a discussion of the distinguishing features of process review and its relation to the hard-look doctrine, see Gary Lawson, “Outcome, Procedure and Process: Agency Duties of Explanation for Legal Conclusions,” Rutgers Law Review 48 (2) (1996): 313, 318.
  • 11Cited in ibid., 324.
  • 12Tucson Herpetological Society v. Salazar, 566 F.3d 870, 874–875 (9th Cir.); Greater Yellowstone Coalition, Inc. v. Servheen, 665 F.3d 1015, 1019–1021 (9th Cir. 2011); Sierra Club v. Environmental Protection Agency, 873 F.3d 946, 951–952 (D.C. Cir. 2017); and Robertson v. Methow Valley Citizens Council, 490 U.S. 332, 333–334, 109 S. Ct. 1835, 1837–1838, 104 L. Ed. 2d 351 (1989).
  • 13Parrillo, “Federal Agency Guidance and the Power to Bind,” 167–168.
  • 14See Mark Seidenfeld, “Substituting Substantive for Procedural Review of Guidance Documents,” Texas Law Review 90 (2011): 331, 340–341.
  • 15Parrillo, “Federal Agency Guidance and the Power to Bind,” 168–169.
  • 16Ibid., 169.
  • 17See, for example, ABA Recommendation 120C, 118-2, A.B.A. Annual Report 57 (1993).
  • 18For a useful overview, see Peter Strauss, “Domesticating Guidance,” Environmental Law 49 (3) (2019): 765.
  • 19Perez v. Mortgage Bankers Association, 135 S. Ct. 1199 (2015) (Nos. 13-1041, 13-1052), Oral Argument at 11:27.
  • 20See, generally, Steven J. Lindsay, “Timing Judicial Review of Agency Interpretations in Chevron’s Shadow,” The Yale Law Journal 127 (8) (2018): 2448.
  • 21See, generally, Christopher J. Walker, “Attacking Auer and Chevron Deference: A Literature Review,” Georgetown Journal of Law & Public Policy 16 (1) (2018): 103.
  • 22For discussion of these institutional pronouncements, see Ronald M. Levin, “Rulemaking and the Guidance Exemption,” Administrative Law Review 70 (2) (2018): 263, 278–287.
  • 23Exemplary works include: Jerry L. Mashaw, Creating the Administrative Constitution: The Lost One Hundred Years of American Administrative Law (New Haven, Conn.: Yale University Press, 2012); Blake Emerson, The Public’s Law: Origins and Architecture of Progressive Democracy (New York: Oxford University Press, 2019); Daniel R. Ernst, Tocqueville’s Nightmare: The Administrative Emerges in America, 1900–1940 (New York: Oxford University Press, 2014); Nicholas Bagley, “The Procedure Fetish,” Michigan Law Review 118 (3) (2019): 345; Jeremy K. Kessler, “The Administrative Origins of Modern Civil Liberties Law,” Columbia Law Review 114 (2014): 1083; Kathryn E. Kovacs, “Progressive Textualism in Administrative Law,” Michigan Law Review Online 118 (2020): 134; Elizabeth Magill and Adrian Vermeule, “Allocating Power within Agencies,” The Yale Law Journal 120 (5) (2011): 1032; Gillian E. Metzger, “1930s Redux: The Administrative State Under Siege,” Harvard Law Review 131 (1) (2017): 1; Gillian E. Metzger and Kevin M. Stack, “Internal Administrative Law,” Michigan Law Review 115 (8) (2017): 1239; Jon D. Michaels, “Of Constitutional Custodians and Regulatory Rivals: An Account of the Old and New Separation of Powers,” New York University Law Review 91 (2) (2016): 227; and Cass Sunstein and Adrian Vermeule, “Libertarian Administrative Law,” The University of Chicago Law Review 82 (1) (2015): 393.
  • 24Two other sources of legitimacy that sometimes have been invoked by progressive scholars are public participation in agency decision-making (paradigmatically, through notice-and-comment rulemaking) and transparency about such decision-making. See, for example, Maggie McKinley, “Petitioning and the Making of the Administrative State,” The Yale Law Journal 127 (2018): 1538, 1600–1605 (discussing progressive arguments for participatory administration); and David E. Pozen, “Transparency’s Ideological Drift,” The Yale Law Journal 128 (2018): 100, 107–123 (discussing progressive arguments for transparent administration). Yet the leading progressive defenses of the administrative state treat these potential sources of legitimacy as either derivative of or subordinate to presidentialism and professionalism. While public participation can supplement the democratic pedigree provided by presidentialism and improve the information that the professionals have at their disposal, the availability and quality of public participation is dependent on, and rightfully subject to revision by, professionals. Indeed, one of the functions of administrative legitimacy, from the progressive point of view, is to ensure this professional control of participation. It is professionals who should determine when participation is most necessary and appropriate, who should channel that participation in effective ways, and who should ensure an equality of opportunity in participation, even going so far as to correct barriers to equal participation that stem from social and economic inequalities. A similar story can be told about transparency. Transparency requirements can help keep presidents and professionals in line, encouraging them to make decisions in a manner that will be broadly acceptable to the bar, the bench, the media, and the public at large. But the tendency of contemporary, progressive scholars is to second-guess transparency’s benefits and to insist that the authority to weigh the costs and benefits of transparency be left, in most circumstances, to presidents and professionals. In general, the most passionate advocates of participation and transparency make proposals that are inconsistent with or antithetical to the visions of reform offered by the leading progressive defenders of the administrative state’s legality and legitimacy. With respect to participation, see, for example, David J. Arkush, “Direct Republicanism in the Administrative Process,” The George Washington Law Review 81 (2013): 1458; Jessica Mantel, “Procedural Safeguards for Agency Guidance: A Source of Legitimacy for the Administrative State,” Administrative Law Review 61 (2009): 343; and Lisa Schultz Bressman, “Beyond Accountability: Arbitrariness and Legitimacy in the Administrative State,” New York University Law Review 78 (2) (2003): 461. With respect to transparency, see Pozen, “Transparency’s Ideological Drift,” 123–146 (discussing “transparency’s rightward drift”).
  • 25See Metzger, “1930s Redux,” 1256.
  • 26Some progressive scholars have recognized the utility of the courts in navigating these tensions. See, for example, Gillian E. Metzger, “Ordinary Administrative Law as Constitutional Common Law,” Columbia Law Review 110 (2010): 479, 492. “Hard look review prioritizes expertise and technocratic decisionmaking within the agency, in the process downplaying more raw political considerations. At the same time, requiring that agencies explain and justify their actions also arguably reinforces political controls by helping to ensure that Congress and the President are aware of what agencies are doing.” As this description suggests, however, the ideal judicial intervention, from a progressive perspective, will maintain a balance of power between technical and political authorities, leaving the reconciliation of real differences to a process beyond the purview of the courts.
  • 27For a related argument, see Daniel Farber and Anne Joseph O’Connell, “Agencies as Adversaries,” California Law Review 105 (2017): 1375.
  • 28For a related argument, see Seidenfeld, “Substituting Substantive for Procedural Review of Guidance Documents.”
  • 29Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984) (with respect to statutes); and Auer v. Robbins, 519 U.S. 452 (1997) (with respect to regulations).
  • 30See Lawson, “Outcome, Procedure and Process,” 323–324.
  • 31Interpretive review counts as a form of outcome review because it asks whether the outcome of the agency’s interpretation as expressed in a guidance document is substantively correct in light of the meaning of preexisting statute or regulation that is being interpreted. For further discussion, see ibid., 325. Outcome review can be very deferential or very demanding, depending on the quality and quantity of evidence that courts demand.
  • 32Our colleague Thomas Merrill has recently proposed a hybrid of outcome, process, and procedural review, which invites courts to defer to agency interpretations of prior statutes and regulations when those interpretations are promulgated in a manner that resembles (but need not replicate precisely) the procedures used in notice-and-comment rulemaking. See Thomas W. Merrill, To Say What the Law Is: The Supreme Court’s Chevron Doctrine and the Future of the Administrative State (forthcoming). While Merrill’s proposal is very much in the same spirit as our own, we think it is both too demanding and too lax for an uncertain world: agencies should not have to use procedures that resemble notice-and-comment to merit judicial respect, nor should judges assume that the use of notice-and-comment-like procedures is an adequate proxy for reasoned decision-making.
  • 33See Skidmore v. Swift & Co., 323 U.S. 134 (1944).
  • 34Ibid., 140.
  • 35See, for example, Kisor v. Wilkie, 139 S. Ct. 2400 (2019); Blake Emerson, “The Claims of Official Reason,” The Yale Law Journal 128 (8) (2019): 2122; Daniel A. Farber and Anne Joseph O’Connell, “The Lost World of Administrative Law,” Texas Law Review 92 (2014): 1137, 1186; and Daniel E. Walters, “Litigation-Fostered Bureaucratic Autonomy: Administrative Law Against Political Control,” Journal of Law and Politics 28 (2013): 129, 175–178.
  • 36We thank Gillian Metzger for pressing us on this point.
  • 37For a related discussion of the virtues of a single standard of review, see David Zaring, “Rule by Reasonableness,” Administrative Law Review 63 (2011): 575.
  • 38Jerry L. Mashaw, Bureaucratic Justice: Managing Social Security Disability Claims (New Haven, Conn.: Yale University Press, 1983), 161. For discussion, see Charles F. Sabel and William H. Simon, “The Management Side of Due Process in the Service-Based Welfare State,” in Administrative Law from the Inside Out: Essays on Themes in the Work of Jerry L. Mashaw, ed. Nicholas R. Parrillo (Cambridge: Cambridge University Press, 2017).
  • 39See Shapiro et al., “The Enlightenment of Administrative Law”; and Lisa P. Jackson, “Process for Reviewing National Ambient Air Quality Standards,” memorandum (Washington, D.C.: U.S. Environmental Protection Agency, 2009).
  • 40Parrillo, “Federal Agency Guidance and the Power to Bind,” 183.
  • 41In this regard, our alternative program is consistent with Nicholas Bagley’s recent critique of “the procedure fetish,” although that critique styles itself as a defense of progressive administrative governance. See, generally, Bagley, “The Procedure Fetish.”