By: David Arkush*

This Essay examines the three ideals that underlie most models of administrative legitimacy—the rule of law, sound public policy, and democracy—as well as their associated models of administration, and it argues that administrative legitimacy efforts are best focused on the democracy ideal.  Reforms guided by the rule of law and public policy ideals have far less potential to contribute to administrative legitimacy for two reasons: there is little evidence that the ideals are underserved in present administration, and each ideal suffers from deep conceptual problems that inherently limit its contributions.

“[T]here is plenty of reason to believe that administrative legitimacy is currently undermined by a democracy deficit . . . .”

Reforms driven principally by the democracy ideal also have fallen short.  Indeed, unlike the rule of law and public purposes ideals, there is evidence that the democracy ideal is underserved by present administration, which suggests that progress in realizing the ideal could enhance legitimacy.  In addition, unlike the other ideals, the most prominent challenges for realizing the democracy ideal are matters of practical design, not flaws in the ideal’s very conception.  This analysis suggests that it may be possible to make administration more democratic and that doing so should be the most fruitful path to improving administrative legitimacy.


Administrative law is said to have been in crisis since Congress first began to establish modern regulatory agencies.  The central concern is the “legitimacy” of the administrative process.[1]  Agency officials write laws of general applicability but lack the political accountability of elected legislators.  They decide individual matters with binding authority but lack the independence of Article III judges.  At the same time, the administrative process is often inaccessible to the public, despite many features designed to make it transparent and open to participation, and the public lacks tools to assess adequately the quality of regulatory policies and outcomes.[2]

Observers have not always made clear what is meant by the term legitimacy,[3] but the ordinary sense of the term often suffices, with its evocation of a set of characteristics related to public perceptions of legality, propriety, and efficacy.  The principal reason for concern over the legitimacy of the administrative process is that it often involves the exercise of “substantial public power by unelected agency officials.”[4]  The lack of public accountability, as well as agencies’ poor fit within the constitutional scheme that separates legislative, executive, and judicial powers, means that agency decisions run a higher risk than other government actions of being viewed as unlawful, unsound, or undemocratic.

This triad of values or ideals—the rule of law, sound public policy, and democracy[5]—captures much of what drives legitimacy concerns as well as the models of administration proposed in response.  Each ideal is associated closely with a particular model.  The rule of law ideal is linked to a formalist model of administration, in which the law binds administrators tightly, leaving them with little to no discretion.  The public purposes ideal is linked to models of technocratic administration, in which agency discretion is legally broad but constrained and channeled by sound science.  The democracy ideal is linked to models of enhanced citizen participation, the most prominent being participation by interest groups.[6]

One might question why we should discuss the formalist and technocratic models at all if it is true that, as Gerald Frug noted years ago, “no one believes in them anymore.”[7]  To be sure, this Essay will not dwell for long on the rule of law ideal or the associated formalist model, as they have little expression in contemporary reform debates and proposals.  But the public purposes ideal and associated technocratic models remain prominent in administrative law.  They have been a powerful force in administrative reform efforts in recent decades—not just in academic debate but also in political and legal reform efforts, some of which have been successful.  At present, expertise-based models animated by the public purposes ideal are ascendant in Congress, the current presidential administration, and the courts.

This Essay argues that models based on the rule of law and public purposes ideals can be expected to make only limited improvements to legitimacy for two reasons.  First, there is little evidence that the contemporary administrative process fails to satisfy the models sufficiently for legitimacy purposes.  Second, the very conception of each ideal embeds a significant limitation on its potential contribution: each ideal reflects an aspiration to constrain administrative discretion through means which, by their nature, cannot accomplish the task.

In contrast, the democracy ideal should command more attention and energy.  Unlike the rule of law and public purposes ideals, there is evidence that present administration underserves the democracy ideal and administrative legitimacy suffers as a result.  Moreover, the limitations of the democracy ideal are less intractable than those of the other ideals.  The principal challenges in realizing the democracy ideal are practical in nature, rather than inherent flaws in the concept of a more democratic form of administration.  These problems are not easy to solve, to be sure, but there is no reason to believe they are intractable.  As a result, the democracy ideal may offer the best path to strengthening administrative legitimacy.

I.  The Rule of Law Ideal and the Formalist Model of Legitimacy

The rule of law ideal responds to the problem of discretion by holding that the law constrains and directs agency action.  Administrative action should “adhere to the dictates of public laws laid down in advance by the sovereign legislature.”[8]  The core concepts are borrowed from adjudication.  Foremost, the rule of law ideal delineates political and legal decision making.  Political decision making, conducted by the legislature, involves weighing policy and enacting it into law, while judicial decision making, the realm of the courts, involves applying the law to particular facts.[9]  Substantive lawmaking power is located exclusively within the legislature because that is the bargain struck in the Lockean social contract.[10]  The ideal conceives of administrative agencies as similar to courts, making legal and not political decisions.

The earliest legitimizing model of administrative law reflected the rule of law ideal closely.[11]  Termed the “formalist”[12] or “traditional”[13] model, it views the administrative agency as a discretionless “transmission belt”[14] or “machine”[15] for implementing legislative directives.  The model holds that agency mandates are clear, having been set out by the legislature, and the administrator’s sole task is to apply preexisting law or policy, generated outside the agency, to particular facts.[16]

The ideal’s divide between law and politics immediately encounters basic problems.  Unlike courts, administrative agencies have rulemaking powers that are closely analogous to the legislative function.  One response to this problem, embodied by the “nondelegation doctrine,” is to restrict the law-giving functions of agencies.  In theory, the doctrine limits the extent to which Congress can delegate its lawmaking functions to administrative bodies.  But in practice, the doctrine authorizes virtually all delegations of legislative power rather than forbids them.  The cases hold that Congress may delegate rulemaking authority to administrative agencies so long as it provides an “intelligible principle” to guide agency action.[17]  In turn, the “intelligible principle” standard requires only that Congress “clearly delineate[] the general policy, the public agency which is to apply it, and the boundaries of this delegated authority.”[18]

Another response is to view agency rule writing as a form of rule applying—to say that even when it writes prospective rules of general application, the agency is directed by legislation, and law rather than politics has guided its decisions.[19]  Imagine courts that, in addition to deciding individual cases, also write regulations clarifying, augmenting, or modifying statutory text, and the content of these regulations is dictated by statutory language, purpose, and context.  The rule of law ideal conceptualizes agencies as performing this function.  As a result, the ideal is subject to the objections that apply to legal formalist claims that courts are limited to law, not politics.  Whatever their merit in the context of judicial decision making, observers have found formalist models unconvincing in the context of administrative law.  The common view is that it is impossible to grant authority to administrators without also granting them discretion over policy.[20]

In addition, a more fundamental problem renders most of this discussion academic: Congress often gives administrative agencies broad discretion to make policy, and it has been doing so for the last century.  Even if we accept that agency rulemaking functions could be restricted to a type that we could fairly view as the mere application of law, Congress plainly gives agencies much broader authority.

In light of the nondelegation doctrine’s anemia and Congress’s routine grant of broad authority to agencies, the principal expressions of the rule of law ideal in administrative law are procedural—the requirement of procedural regularity and the aspiration to proper procedures.[21]  These goals are modest and largely undisputed.  Parties interested in particular administrative actions might dispute the precise procedures required and whether they have been satisfied, but there is little disagreement over the importance of procedural regularity as a general matter.  This retreat to procedural formality leaves behind much of the rule of law ideal’s thrust.  The ideal seeks not only to delineate how law is made but by whom.[22]  As a result, to the extent that the rule of law ideal’s core aspiration is that “law, rather than politics, ultimately has governed the administrative action,”[23] it plays a limited role in current administrative law discourse.

II.  The Public Purposes Ideal and Technocratic Models of Legitimacy

The public purposes ideal holds that the administrative process should generate sound public policy.  It has been expressed in two principal ways.  The first is associated with the New Deal Era and James Landis.[24]  In a period of pessimism about the health of an economy unguided by competent administration, agency officials’ expertise became an important justification for granting them broad policymaking discretion.[25]  In this view, administrative action was best constrained not by the courts or Congress but by the science of public health or sound management in a given area of the economy.[26]  Given the necessary time and freedom of action, administrators eventually would converge on the right answers to regulatory problems.[27]

The New Deal model of expert administration was short lived as a persuasive justification for regulatory discretion, but the public purposes ideal and related claims regarding agency expertise retained a strong place in administrative law.  In the 1970s and 1980s, the ideal underwent a profound resurgence,[28] finding expression in a model of “comprehensive rationality”[29] of administration that continues to play a prominent role in the legal literature and in practical reform efforts.  Comprehensive rationality stems from the notion that administrative agencies do a poor job of setting regulatory priorities or enact policies that produce greater social costs than benefits.[30]  In brief, agency outcomes are often irrational, and the goal, then, is to rationalize them.  The most prominent tool for this purpose is cost-benefit analysis, in which a policy maker weighs the social costs and benefits of a proposal, or all potential proposals, before enacting it.

The models animated by the public purposes ideal cannot provide an adequate response to legitimacy concerns.  Foremost, they cannot constrain administrative discretion adequately.  Technical expertise and science can resolve questions of fact, and those facts, in turn, can either inform policy decisions that must be made or can combine with policies already set to compel a certain course of action.  But the facts alone cannot make a decision.  The agency expert must make policy judgments.[31]  In addition, one cannot make a judgment impartially; without a stake in a matter, there is no reason to decide one way or another.[32]  Broad understanding of these points is likely the reason why the Landis model was so short lived,[33] even though its themes have continued to resonate in legal and political discourse.

One answer to the inadequacy of expertise is to posit a more expansive, sophisticated, politically informed, and generalist expertise.[34]  One can suggest that an important public health policy will be set not by agency scientists but by a politically appointed agency head who combines scientific expertise with a deep sense of public values, political acceptability, and a commitment to the public interest.  But this response assumes away the problem by presupposing the existence of a public official whom we trust to make good decisions that will be viewed as legitimate.  To redefine the word “expert” as someone who possesses faculties of judgment that alleviate legitimacy concerns is to dodge the central issue.

Despite the broad understanding that scientific expertise cannot resolve policy questions, and therefore cannot constrain administrative discretion, cost-benefit analysis has undergone a dramatic ascendance.  In recent decades, the methodology has been enacted and reenacted in numerous executive orders and statutes,[35]and the courts have increasingly reviewed agencies’ economic analyses with demanding rigor.[36]  Like other empirical inquiries, cost-benefit analysis, at best, can only supply the answers to factual questions, and cannot constrain the value judgments of decision makers.  An economic analysis might purport to demonstrate that a policy or set of policies is illogical by certain standards or is a product of bounded rationality, but to cite the data is to argue about what is good policy, not to win the argument.

Further, cost-benefit analysis is subject to a more fundamental critique: it lacks an objective basis.  As has been discussed extensively in the legal literature, one cannot conduct a cost-benefit analysis without assigning values to the objects of the analysis.  It is true that these inputs determine the result of the analysis, but there is no empirical or even agreed-upon basis for the values assigned.  Instead, they must be based on the very contested policy judgments that a neutral analysis is intended to help resolve.[37]  In turn, this indeterminacy gives rise to another critique of cost-benefit analysis: it risks diminishing the transparency and democratic accountability of agency decisions by masking important value judgments behind a veil of inaccessible analysis that appears to be scientific, but is not.

The inability of expertise models to resolve policy questions is related to a fundamental problem inherent in the public purposes ideal: it begs the question at issue.  Embedded in the ideal is an assumption that people are likely to view administrative agencies as legitimate if the agencies make good policy decisions.  This assumption may or may not be merited.  But even if we accept it, a greater problem remains: there is no objective answer to whether an agency’s policy decisions are “good.”  To answer by stating that experts can discern what is “good” policy would render the ideal circular: we are concerned that agencies have too much discretion and might fail to serve the public.  The ideal proposes that agencies can assuage these concerns by making good policy decisions.  But who is the judge of policy?  The agency.

Because we are concerned with public perceptions of legitimacy, a better answer is that the notions of policy soundness in which the public purposes ideal is grounded should come from the public.  If agencies can enhance their legitimacy with the public by producing sound policy, it is likely that they must do so by producing policy that the public views as sound.  In this sense, the public purposes ideal arguably collapses into the democracy ideal.

To make this discussion more concrete, let us assume that an agency—say, the Environmental Protection Agency (“EPA”)—acts with great efficiency to enact a regime that is the wonder of public policy schools, economists, and environmentalists nationwide.  If the public does not like the result, then nothing has been accomplished to improve legitimacy.  In addition, supporters of the policy might lose through the political process what they won through purportedly sound administration.

What remains of the public purposes ideal is the assertion that good science and rigorous analysis are important and useful tools for sound administration.  Although their presence cannot ensure legitimacy, their absence can doom it.  This is a retreat into proceduralism that, like that of the rule of law, is unobjectionable, adds little to legitimizing efforts, and abandons much of the ideal’s original, substantive content.

Indeed, the rule of law and public purposes ideals share a common structure.  Each responds to the problem of discretion by arguing it away: agencies have little discretion because their decisions are compelled by something exogenous to the administrative process, either the law or the facts.  Each fails to quell legitimacy concerns because neither can direct the decisions that administrators must make.  After careful scrutiny, all that remains of each is its proceduralist shadow.

There is a final point to be made regarding prominent, contemporary expertise models.  At present, they are, in some sense, a solution in search of a problem.  Sound science and analytical rigor are important values, but there is scant evidence that they are missing from the regulatory process (except when deliberate political interference frustrates them) or that the public believes they are missing (subject to the same caveat).  Indeed, much of the evidence that underlies the drive toward “comprehensive rationality” has been criticized as deeply flawed.  Comprehensive rationality seeks to remedy misguided regulation—for example, the promulgation of regulations with costs that exceed their benefits or the poor prioritization of regulatory initiatives. Without disputing that these phenomena exist, critics argue, they are vastly overstated.  Some studies purporting to demonstrate regulatory irrationality have been shown to suffer from errors; other studies are persuasive only if one agrees with contestable value judgments made by the authors.[38]  In contrast to these studies, a growing volume of evidence demonstrates that the benefits of regulations usually outweigh the costs, often by a wide margin.[39]

III.  The Democracy Ideal and Models of Increased Participation

In Sargentich’s telling, the democracy ideal envisions a high degree of citizen participation in the administrative process, or at least strong democratic accountability for agency officials regarding whether they actively consider public views.[40]  To his description, this Essay adds a substantive component: rather than merely serving as factors for consideration, public values should be reflected in or, to the extent possible, embodied by agency outcomes.[41]

The democracy ideal differs from the rule of law and public purposes ideals in multiple ways.  First, former ideals respond to the problem of discretion by attempting to suppress it out of the administrative process, claiming that it is obviated by exogenous sources of authority—laws or facts.  In contrast, the democracy ideal squarely admits that discretion exists in administration and attempts to import a basic source of legitimacy—citizen preferences—into the process.[42]  Second, there is reason to believe that progress toward realizing the democracy ideal might enhance administrative legitimacy.  This is because there is evidence that the ideal is deeply underserved at present and because progress toward more democratic administration would almost inherently combat the source of that failure.  Third, the democracy ideal’s conceptual problems are less severe than those affecting the rule of law and public purposes ideals.  The principal challenges facing the democracy ideal are matters of practical design, not limitations or contradictions embedded in the ideal itself.  As a result, it should be possible to make progress toward realizing the democracy ideal in administration.

A.     Evidence That the Democracy Ideal Is Underserved

We would care little for an administrative process that perfectly realized the rule of law and public purposes ideals if it did not also enjoy democratic support.  Indeed, as discussed above,[43] it is not clear what it would mean to satisfy the public purposes ideal without reference to democratic support for administrative policies.  Conversely, an administrative process can enjoy strong public support even if certain experts believe it generates poor policy outcomes according to their metrics.  This is certainly not to say that the democracy ideal alone could legitimize administration.  Satisfying the rule of law ideal and incorporating expertise properly are likely necessary as well.  And the democracy ideal also has significant gaps.  Most prominent is that it neither prevents nor remedies a significant source of illegitimacy and injustice—majoritarian tyranny.[44]  The point here is only that the democracy ideal appears to have more affirmative legitimizing potential than the others.  Unlike the rule of law and public purposes ideals, there is strong evidence that the democracy ideal is deeply underserved by current administrative law and practice.

One important phenomenon in this regard is the persistent perception that regulated interests and their perspectives dominate the regulatory process.  For decades, observers across the political spectrum have agreed that agencies are too often “captured” by industry.[45]  The term “capture” is a metaphor for a range of ways in which an agency comes to reflect the values or viewpoints of the industry it regulates.[46]  For example, an agency may be inclined toward compromise rather than conflict because it needs industry cooperation to accomplish certain tasks[47] or because industry has power with its overseers in Congress and the White House.[48]  Agency officials may have a history of employment in the regulated industry or may hope for future employment there.[49]

Empirical evidence confirms that industry participates in the administrative process much more than citizens or public interest groups, creating what Sid Shapiro has termed “representational capture.”[50]

A 1977 Senate committee report examined the previous ten major rules written by seven regulatory agencies, finding that “in agency after agency, participation by the regulated industry predominates—often overwhelmingly.”[51]  For example, 75% of Federal Power Commission rulemakings involved no public interest representatives even though the matters had a “clear consumer and public impact.”[52]  The remaining rulemakings had industry-to-public-interest participation ratios ranging from 4:1 to 12:1.[53]  In addition, public-interest representatives had participated in just 10% of the Federal Communication Commission’s (“FCC”) last thirty adjudications.[54]

Cary Coglianese studied twenty-five significant EPA rules written under the Resource Conservation and Recovery Act[55] between 1989 and 1991, finding that businesses participated in 96% of rulemakings, trade associations in 80%, and environmental and citizen groups (combined) in just 12%.[56]  Of the groups that participated, 59% represented regulated entities, while citizen and environmental groups comprised only 4%.[57]

Scott Furlong examined reports of registered lobbyists who lobbied both Congress and the executive branch to influence policy on environmental and natural resource issues in 1996.  Ninety-four percent of lobbyists worked for businesses or trade associations, while only 3% were from public-interest groups.[58]  An examination of the clients of lobbying firms revealed that 73% were businesses or trade associations and 6% were public-interest groups.[59]

Melissa Golden studied comments on eleven proposed rules at the EPA, the National Highway Traffic Safety Administration (“NHTSA”), and the Department of Housing and Urban Development.  In eight rules proposed by the NHTSA or the EPA, business interests filed between 66.7% and 100% of the comments.  For five of eight rules, there were no public-interest comments.[60]

Jason Webb Yackee and Susan Webb Yackee studied thirty rulemakings by four agencies from 1994 to 2001 and found that business interests filed 57% of comments, compared to 19% by government interests and 22% by nongovernmental interests, which included 6% by public-interest groups.[61]

The principal cause of these differential participation rates is thought to be resource disparities between regulated industry and public interest groups,[62] as participation in the administrative process is expensive.[63]  Indeed, merely monitoring administrative activity well enough to identify actions in which one might wish to participate is costly.[64]  Overall, the imbalance in expenditures is stark.[65]

Wendy Wagner recently identified another variant of capture that stems from resource disparities not just between interest groups but between regulated entities and the agencies themselves: “information capture,” meaning “the excessive use of information and related information costs as a means of gaining control over regulatory decisionmaking in informal rulemakings.”[66]  The law prohibits an agency from “shield[ing] itself” from a “flood of information” and “developing its own expert conception” of a matter.[67]  To the contrary, “the agency is required to ‘consider’ all input that it receives.”[68]  The flood of information can cripple an agency as well as hamper the participation of less well-funded interest groups.[69]

A phenomenon apparently related to the perception of capture is the recent hyperpoliticization of the administrative process, in what has been termed administrative law as “blood sport.”[70]  Tom McGarity, borrowing a phrase from former Securities and Exchange Commission Chair Arthur Levitt, suggests that some high-stakes rulemakings now operate under a “blood sport” paradigm that differs vastly from the conventional model of a deliberative, lawyer-dominated process.[71]  These rulemakings have entered the realm of “politics as usual”[72]—meaning they are characterized by a rare, if not unprecedented, degree of political warfare.  The typical blood sport rulemaking may involve the flooding of an agency with information; unusually intense lobbying, including lobbying members of Congress and the White House;[73] public relations campaigns, coordination with think tanks, media pundits, and bloggers;[74] and intense congressional oversight, including stridently adversarial hearings, lengthy confirmation battles, invocation of the Congressional Review Act, which provides fast-track procedures for Congress to overturn a final agency rule within sixty days, and attempts to strip an agency’s funding or authority.[75]

Despite all of the findings reviewed above, evidence of a causal relationship between representation and influence is scarce and, at best, mixed.[76]  But the ambiguity of evidence on capture has done little to diminish concerns among students of administrative law and the broader public.  Improving the expression of the democracy ideal in administration should help reduce capture, whether real or merely perceived, in turn enhancing administrative legitimacy substantially.

B.     Models of Administration Associated with the Democracy Ideal

The dominant model associated with the democracy ideal is interest representation, in which interest groups that represent relevant segments of the public are afforded the opportunity to participate more extensively in the administrative process.  There is broad agreement that the interest-representation model has proved inadequate.  Two of the most common critiques are that the model fails to assure that the right interests will be represented, or represented properly,[77] and that resource disparities give industry groups an overwhelming advantage over public interest groups.[78]  Increased opportunities for participation are used disproportionately by organized interests, compounding the perception, if not the reality, of inadequate popular representation and pervasive industry capture.[79] Still, it is easy to overstate the shortcomings of interest representation, and the shortcomings have received far more attention than the achievements.  Sid Shapiro has begun to argue persuasively that, when measured in pragmatic terms by its actual accomplishments rather than its ability to meet a theoretical ideal, the interest-representation model “has been reasonably successful in narrowing administrative discretion in meaningful ways, adding to the accountability and legitimacy of the administrative state.”[80]

There have been other attempts to enhance citizen participation or representation but none nearly as broad or linked to actual legal and political reforms as the interest-representation model.  The most noteworthy is a handful of experiments and proposals related to deliberative democracy.  In these models, an agency convenes a discussion between affected groups, and sometimes randomly selected individuals—with the goal that they will reach a consensus or at least find some common ground on a contentious regulatory issue—then make recommendations to the agency.[81]  These proposals have merit but suffer from several flaws.[82] One is that their actual impact on agency decisions is uncertain at best, as they do not require the agency to follow a deliberative group’s recommendations.[83] More important is that they are highly resource intensive for a host of reasons, making them unlikely candidates for regular use throughout a vast bureaucracy.[84]

Like the rule of law and public purposes ideals, the democracy ideal has found expression in models that abandon significant aspects of the ideal.  The most dominant of these secondary models is accountability of administrators to the political branches.[85]  These models rely on Congress or the President to hold agencies accountable and, in turn, rely on the overseers’ accountability to the American public.[86]  These models suffer from serious shortcomings, foremost that the political branches are incapable or unwilling to oversee the administrative process adequately and that their own democratic responsiveness leaves something to be desired.[87]

C.     The Challenge of Democracy in Administration

The accountability and interest-representation models may abandon more of the democracy ideal than is necessary.  A critical point regarding the democracy-based models, including attempts at more robust citizen participation, is that some of their most important problems are practical matters, not limitations inherent in the democracy ideal.

A few inherent limitations exist, to be sure.  One is that decisions regarding how better to achieve democracy necessarily predate their own implementation, meaning the decisions themselves cannot be adequately democratic.[88]  Another is that, to some extent, each of the models employs the basic arrangement that gives rise to the problem of discretion in the first place—the reliance on representation rather than direct democracy.

As fundamental problems go, however, these are not so bad.  They do not cast doubt on the possibility that advancing the democracy ideal would enhance legitimacy.  They suggest only that models based on the ideal cannot completely resolve legitimacy problems.  This is in stark contrast to the gravest problem inherent in the rule of law and public-purposes ideals: they assume administrative discretion away more than respond to it.  The predominant critique of the democracy ideal is only that it is “difficult to achieve.”[89]  Through better design, we may make significant progress toward achieving it.

Two of the most significant design needs are a means of insulating agency officials from undue interference without sequestering them from the public as well and, conversely, a means of increasing citizen participation that does not also increase the influence of factional or private interests.  Special interests can take advantage of most of the opportunities for administrative oversight that are available to ordinary citizens, as well as others that most of the public lacks.  Recall that a common concern regarding the interest-representation model is that efforts to provide more opportunities for public participation or oversight often empower regulated entities as much as, if not more than, public-interest groups or citizens.[90]

Another challenge is how best to blend an agency’s technical expertise with citizen preferences.  The perceived tension between expertise and democratic values has been a subject of persistent debate in administrative law.[91]  We want agency officials to embody neutral expertise rather than mistaken public preferences or partisan or private goals, and therefore we partially insulate them from the political process.  At the same time, regulation requires value choices, not just the discovery and application of facts.[92]  For this reason, we also demand that agency decisions incorporate public values.  The challenge is how to place more discretion in citizens’ hands without sacrificing agency expertise, or place more in agency officials’ hands without sacrificing democratic values.

It bears emphasizing that these do not appear to be intractable problems rooted in irreconcilable conflict between democracy and expertise but matters of institutional design on which progress should be possible.  In theory, one can envision an arrangement that sacrifices little of either: a proceeding in which experts provide the relevant facts and law (to the extent is it clear)—and perhaps even set out the policy choice to be made—and then citizens or political representatives supply the actual decision.[93]  The principal form of expertise we seek in administration is the ability to find facts and to discern the limitations of our knowledge.  To a lesser extent, we seek legal expertise regarding an agency’s organic statute.[94]  We also seek the agency’s wisdom regarding policy decisions.  But the decisions themselves are the core exercise of discretion that gives rise to legitimacy concerns.  Perhaps they can be made by elected officials or—better yet for the democracy ideal—citizens.[95]


Many contemporary regulatory reform efforts promote technocratic models of administration.  Certainly, those models predominate in legislative proposals and executive orders.  This Essay has attempted to redirect some of that energy toward more productive ground for strengthening administrative legitimacy: citizen-participation models animated by the democracy ideal.

The technocratic models have not substantially improved legitimacy to date, and there is little reason to believe they will achieve more in the near future.  There is little evidence that technical deficits in administration are in fact diminishing legitimacy, and the models are based on an incoherent ideal of administration that confuses empirical analysis with policymaking.  In contrast, there is plenty of reason to believe that administrative legitimacy is currently undermined by a democracy deficit, and some of the most significant challenges for democracy-based models are practical matters rather than fundamental limitations of the ideal of democracy in administration.  Through careful institutional design, we may be able to make substantial progress on models of democratic accountability and citizen participation, thereby strengthening administrative legitimacy.

        *   J.D., Harvard Law School, 2003.

        [1].   See James Freedman, Crisis and Legitimacy (1978); see also Gerald E. Frug, The Ideology of Bureaucracy in American Law, 97 Harv. L. Rev. 1276, 1279–81 (1984); Mark Seidenfeld, A Civic Republican Justification for the Bureaucratic State, 105 Harv. L. Rev. 1511, 1513 (1992); Richard B. Stewart, The Reformation of American Administrative Law, 88 Harv. L. Rev. 1669, 1676, 1679–80 (1975).

        [2].    Cf. Freedman, supra note 1, at 6–7.

        [3].   See, e.g., Paul H. Brietzke, James O. Freedman, Crisis and Legitimacy: The Administrative Process and American Government, 14 Val. U. L. Rev. 361, 362–65 (1980) (book review).

        [4].   Thomas O. Sargentich, The Reform of the Administrative Process: The Contemporary Debate, 1984 Wis. L. Rev. 385, 393 (1984).

        [5].   This typology derives from, and largely follows, one expounded by Thomas Sargentich.  See generally Sargentich, supra note 4.  However, this Essay intends neither to follow Sargentich’s approach closely nor to comment on all points of departure.

        [6].   The associations between the ideals and models are not exclusive, to be sure.  Each ideal could be expressed in any of the models, and any model could be justified by reference to any of the ideals.  For example, the democracy ideal is most naturally associated with models of increased citizen or interest group participation, and formalist models are typically based in the rule of law ideal.  But democracy ideals could be expressed through a formalist model, and a formalist model based on democracy.  In the typical formalist model, administrators are directed by legislative policy choices set into laws.  But the core of formalism requires only the constraint of administrative discretion, not its constraint by any particular means.  A formal model could be built on plebiscites, with administrators mechanically implementing direct citizen choices.  For ease of discussion, this Essay discusses each ideal with the model with which it is most commonly associated.

        [7].   Frug, supra note 1, at 1297.

        [8].   Sargentich, supra note 4, at 397.

        [9].   Id. at 399.

      [10].   See, e.g., id. at 397–98.

      [11].   See Frug, supra note 1, at 1282; David Arkush, Direct Republicanism in the Administrative Process, 81 Geo. Wash. L. Rev. (forthcoming 2013) (manuscript at 8), available at

      [12].   Frug, supra note 1, at 1297.

      [13].   Stewart, supra note 1, at 1669.

      [14].   Id. at 1675.

      [15].   Frug, supra note 1, at 1297–98.

      [16].   Id. at 1298–99; Sargentich, supra note 4, at 398–99; Stewart, supra note 1, at 1675.

      [17].   Hampton v. United States, 276 U.S. 394, 409 (1928).

      [18].   Mistretta v. United States, 488 U.S. 361, 372–73 (1989) (quoting Am. Power & Light Co. v. SEC, 329 U.S. 90, 105 (1946)).

      [19].   See Sargentich, supra note 4, at 399.

      [20].   See, e.g., Frug, supra note 1, at 1312–14.

      [21].   See Sargentich, supra note 4, at 404–06.

      [22].   See supra text accompanying note 6.

      [23].   Sargentich, supra note 4, at 399.

      [24].   Id. at 411.

      [25].   See Stewart, supra note 1, at 1677–78.

      [26].   Id. at 1702.

      [27].   Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2260–61 (2001); Stewart, supra note 1, at 1677–78.

      [28].   See, e.g., Sargentich, supra note 4, at 411–13.

      [29].   See Richard B. Stewart, Administrative Law in the Twenty-First Century, 78 N.Y.U. L. Rev. 437, 443–45 (2003).

      [30].   See, e.g., Stephen Breyer, Breaking the Vicious Circle: Toward Effective Risk Regulation 10, 21 (1993); Cass R. Sunstein, After the Rights Revolution: Reconceiving the Regulatory State 74 (1990); Sargentich, supra note 4, at 412–13; Sidney Shapiro, Pragmatic Administrative Law, in Issues in Legal Scholarship 2005, at 10–12 (2005); Stewart, supra note 29, at 443.

      [31].   See, e.g., Stewart, supra note 1, at 1684.

      [32].   See, e.g., David. J. Arkush, Situating Emotion: A Critical Realist View of Emotion and Nonconscious Cognitive Processes for Law and Legal Theory, 2008 BYU L. Rev. 1275, 1354–55 (2008); see also Frug, supra note 1, at 1330 (“[G]enuine absence of personal involvement in an issue precludes the making of a judgment: ‘the participation of the speaker . . . is part of any sincere statement of fact.’  In short, impersonal judgment is a contradiction in terms, and every attempt to split the difference—to allow some personal involvement but not too much—simply creates a structure for manipulation.”).

      [33].   Kagan, supra note 27, at 2261–62.

      [34].   Freedman, supra note 1, at 51–55; Frug, supra note 1, at 1319.

      [35].   See Paperwork Reduction Act, 44 U.S.C. §§ 3501-20 (enacted 1980); Regulatory Flexibility Act (RFA), 5 U.S.C. §§ 601-12 (enacted 1980, amended 1996); Unfunded Mandates Reform Act, 2 U.S.C. §§ 1532-38 (enacted in 1995); Small Business Regulatory Enforcement Fairness Act, 5 U.S.C. § 601 (enacted 1996, in part amending RFA); Exec. Order No. 12,044, 43 Fed. Reg. 12,661 (Mar. 23, 1978) (requiring analysis examining cost-effectiveness of alternatives to proposed major rules); Exec. Order No. 12,291, 46 Fed. Reg. 13,193 (Feb. 17, 1981) (directing agencies to refrain from rulemaking unless net benefits of a rule outweigh net costs, and to maximize benefits and minimize costs); Exec. Order No. 12,866, 58 Fed. Reg. 51,735 (Sept. 30, 1993) (replacing Exec. Order No. 12,291 but maintaining similar cost-benefit analysis requirements); Exec. Order No. 13,563, 76 Fed. Reg. 3821 (Jan. 18,2011) (reiterating requirements of Exec. Order No. 12,866); Exec. Order No. 13,579, 76 Fed. Reg. 41,587 (July 11, 2011) (encouraging independent agencies to comply with parts of Exec. Order No. 13,563); OMB Circular A–4, Regulatory Analysis (1996, amended in 2003) (describing “best practices” for agencies to comply with Exec. Order No. 12,866).  On the rationales behind Executive Order 12,291, see James F. Blumstein, Regulatory Review by the Executive Office of the President: An Overview and Policy Analysis of Current Issues, 51 Duke L. J. 851, 858-59 (2001); Richard H. Pildes & Cass R. Sunstein, Reinventing the Regulatory State, 62 U. Chi. L. Rev. 1, 3 (1995).

      [36].   See, e.g., Bus. Roundtable v. SEC, 647 F.3d 1144, 1146 (D.C. Cir. 2011).

      [37].   See, e.g., Frank Ackerman & Lisa Heinzerling, Priceless: On Knowing the Price of Everything and the Value of Nothing (2004); Arkush, supra note 32, at 1338; Lisa Heinzerling, Regulatory Costs of Mythic Proportions, 107 Yale L.J. 1981, 1985–86 (1998).

          [38].   See, e.g., Heinzerling, supra note 37.  For brief reviews of this debate, see Sidney A. Shapiro, Administrative Law After the Counter-Reformation: Restoring Faith in Pragmatic Government, 48 U. Kan. L. Rev. 689, 723–25 (2000); Shapiro, supra note 30, at 11–13.

      [39].   The leading evidence is the OMB’s annual report to Congress analyzing the costs and benefits of the major regulations of the previous ten years.  Since their inception under the administration of George W. Bush, the analyses have been overwhelmingly positive.  For example, OMB’s 2011 report found that regulations issued between October 1, 2000 and September 30, 2010 resulted in benefits ranging from $132 billion to $655 billion, compared to costs ranging from $44 billion to $62 billion.  Office of Mgmt. & Budget, 2011 Report to Congress on the Benefits and Costs of Federal Regulations and Unfunded Mandates on State, Local, and Tribal Entities 13–14 (2011), available at; see also Shapiro,supra note 30, at 12 (“[R]egulatory critics tend to adopt assumptions that make government look irrational, and the adoption of other equally plausible assumptions make government regulation look entirely reasonable in terms of its costs and benefits.  Moreover . . . a substantial literature contests the claim that traditional regulatory policies have failed.”).

      [40].   See Sargentich, supra note 4, at 425.  For a stronger statement of the ideal, see Frug, supra note 1, at 1296 (“[O]ther advocates of reinvigorating the notion of democracy . . . understand the term ‘democracy’ to refer to the process by which people create for themselves the form of organized existence within which they live.  Only by creating these forms together can people confront the intersubjective nature of social life.  Moreover, unless people do so themselves, the artificial structures through which they operate will threaten to function beyond their control.”).

      [41].   The addition of a substantive element is why this Essay adopts the name “democracy ideal” rather than Sargentich’s “democratic process ideal.”  Strong objections have been raised to including under the rubric of the democracy ideal the proceduralist versions that rely only on representation, and particularly interest-group competition.  See Frug, supra note 1, at 1374, 1376.

      [42].   Cf. Sargentich, supra note 4, at 425–26 (“The most direct expression of the democratic process ideal in the contemporary debate is the commitment generally to expand public participation in administration.”).

      [43].   See supra text accompanying notes 23–39.

      [44].   A reconciliation of the democracy ideal in administrative law with the problems of majoritarian rule is beyond the scope of this Essay, but two provisional points are worth suggesting.  First, some matters of regulatory policy, and in particular many issues of health, safety, or consumer protection, might prove less controversial than those that animate the fiercest debates regarding the protection of minority rights or viewpoints.  A question such as the permissible level of a toxin in the workplace might pose fewer problems for democratic theory than one like the permissibility of same-sex marriage.  Second, it is possible that the protection of minority rights from the will of the majority should be left to constitutional law, not administrative law proper.

      [45].   See, e.g., Steven P. Croley, Theories of Regulation: Incorporating the Administrative Process, 98 Colum. L. Rev. 1, 3–4 (1998); Stewart, supra note 1, at 1684–85 (“Critics have repeatedly asserted . . . that . . . agencies unduly favor organized interests, especially the interests of regulated or client business firms and other organized groups at the expense of diffuse, comparatively unorganized interests such as consumers, environmentalists, and the poor.”); id. at 1721–23.

      [46].   See Stewart, supra note 1, at 1685–86; Cass R. Sunstein, Factions, Self-Interest and the APA: Four Lessons Since 1946, 72 Va. L. Rev. 271, 286 (1986) (“[T]he notion of mechanical-reaction-to-pressure must sometimes be understood as a metaphor for a complex process in which administrators come to share the values of particular affected parties and their approaches to regulatory issues.”).

      [47].   Stewart, supra note 1, at 1686.

      [48].   Id. at 1685.

      [49].   See, e.g., Paul J. Quirk, Industry Influence in Federal Regulatory Agencies 19–20 (1981).

      [50].   Id. at 2, 4.

      [51].   Staff of Senate Comm. on Governmental Affairs, 95th Cong., Study on Federal Regulation: Public Participation in Regulatory Agency Proceedings 12 (Comm. Print 1977) [hereinafter Study on Federal Regulation]; see also id. at 16 (“Organized public interest representation accounts for a very small percentage of participation before Federal regulatory agencies.  In more than half of the proceedings, there is no such participation whatsoever.  In those proceedings where participation by public groups does take place, typically, it is a small fraction of the participation by the regulated industry.”).

      [52].   Id. at 13.

      [53].   See id.

      [54].   See id. at 16.

      [55].   42 U.S.C. § 6925 (2006).

      [56].   Croley, supra note 45, at 129.

      [57].   Id. (citing Cary Coglianese, Challenging the Rules: Litigation and Bargaining in the Administrative Process 46–47 tbl.2-x (Dec. 23, 1994) (unpublished Ph.D. dissertation, University of Michigan)).

      [58].   See Scott R. Furlong, Businesses and the Environment: Influencing Agency Policymaking, in Businesses and Environmental Policy: Corporate Interests in the American Political System 155, 174 (Michael E. Kraft & Sheldon Kamieniecki eds., 2007).

      [59].   Id. at 175.

      [60].   Marissa Martino Golden, Interest Groups in the Rulemaking Process: Who Participates? Whose Voices Get Heard?, 8 J. Pub. Admin. Res. & Theory 245, 250, 252 (1998).

      [61].   Jason Webb Yackee & Susan Webb Yackee, A Bias Towards Business? Assessing Interest Group Influence on the U.S. Bureaucracy, 68 J. Pol. 128, 133 (2006).

      [62].   See, e.g., Sidney A. Shapiro & Rena Steinzor, Capture, Accountability, and Regulatory Metrics, 86 Tex. L. Rev. 1741, 1754 (2008); Stewart, supra note 1, at 1764 (citing Benjamin W. Heineman, Jr., In Pursuit of the Public Interest, 84 Yale L.J. 182, 188 (1974)).

      [63].   See, e.g., Croley, supra note 45, at 120–25.

      [64].   See id. at 124–25.

      [65].   The Senate committee report, for example, stated that:

The regulated industry consistently outspends public participants by a wide margin in regulatory agency proceedings.  In every case or agency reviewed, industry spent many times more on regulatory participation than their public interest counterparts.  In some instances, industry committed as much as 50 to 100 times the resources budgeted by the public interest participants.

Study on Federal Regulation, supra note 5, at vii.

      [66].   Wendy E. Wagner, Administrative Law, Filter Failure, and Information Capture, 59 Duke L.J. 1321, 1325 (2010).

      [67].   Id.

      [68].   Id.

      [69].   See id.

      [70].   See generally Thomas O. McGarity, Administrative Law as Blood Sport: Policy Erosion in a Highly Partisan Age, 61 Duke L.J. 1671 (2012).

      [71].   Id. at 1680.

      [72].   Id.

      [73].   Id. at 1703–07.

      [74].   Id. at 1708–10.

      [75].   Id. at 1711, 1714–16, 1718.

      [76].   See, e.g., Shapiro & Steinzor, supra note 62, at 1754–55.

      [77].   Arkush, supra note 11 (manuscript at 32).

      [78].   Id.

      [79].   Id. (manuscript at 41).

      [80].   See Shapiro, supra note 30, at 5; see also id. at 5–10.

      [81].   See Arkush, supra note 11 (manuscript at 23–25).

      [82].   See id. (manuscript at 26).

      [83].   See id.

      [84].   See generally id.

      [85].   See Sargentich, supra note 4, at 431.

      [86].   See Arkush, supra note 11 (manuscript at 22–26).

      [87].   See id. at 24.

      [88].   See, e.g., Sargentich, supra note 4, at 431.

      [89].   Id. at 429.

      [90].   See supra text accompanying notes 77–80.

      [91].   See Susan Rose-Ackerman, American Administrative Law Under Siege: Is Germany a Model? 107 Harv. L. Rev. 1279, 1279 (1994) (“Modern democracies need to strike a balance between popular control and expertise.”); Sunstein, supra note 46, at 281 (“The debate over the respective roles of ‘expertise’ and ‘politics’ in agency decisionmaking has proved to be one of the most persistent in administrative law.”).

      [92].   Peter L. Strauss & Cass R. Sunstein, The Role of the President and OMB in Informal Rulemaking, 38 Admin. L. Rev. 181, 183 (1986).

      [93].   For further discussion of such a proceeding, as well as the theory underlying it, see Arkush, supra note 11, at 42.

      [94].   Chevron, U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 865 (1984).

      [95].   Frug, supra note 1, at 1298.