Administration as Democratic Trustee Representation

Abstract

The “folk” theory of democracy that typically justifies the administrative state cannot help but lead to a discourse of constraint. If agency action is only legitimate when it mechanically applies the will of the voters as transposed by Congress through statutes, then the norms guiding that action will inevitably restrain agency discretion. As a result, attempts to establish the democratic credentials of the administrative state ironically obstruct the application of collective power. But this “folk” theory of democracy is bad theory. It is empirically incredible and, alarmingly, facilitates dangerous populist politics. Political theory instead suggests that a theory of democratic representation not only better explains legislative outcomes, but also deprives demagogues and deregulatory partisans of the fictions that prop up their agendas. After a brief survey of representative theory, this article will demonstrate that a model of trustee representation adequately describes administration, reinforces its democratic credentials and constitutes a space for politics in shaping the regulations that govern us all.

I. Introduction

Scholars and jurists struggle to justify the policymaking power of the administrative state. Their theories most often rest upon a transmission belt² conception of democratic legitimacy. Simple and superficially appealing, this conception holds that democracy is vindicated by aggregating voter preferences through elections. The popular will is next transposed by members of Congress into statutes. Finally the popular will, now codified, is applied mechanically by administrative agencies who should, in the words of Justice Kavanaugh, “fill in the details” using their technical expertise.³ Administration is to be a “machine without a mind”⁴ that makes no value judgments. Thus, the people's will travels on the back of the franchise from voter to administrator with, it is hoped, little disfiguration. Woodrow Wilson, for example, imagined administration operating according to “nomothetic” law that would carry out the orders of democracy's “corporate, popular will.”⁵ The nondelegation doctrine, which holds that federal lawmaking power is held exclusively by Congress, also makes use of the transmission belt model. If Congress delegates policymaking authority to bureaucratic agents, it interrupts the seamless transit of the popular will from voter to representative, from representative to statute, and, finally, from statute to administrative mechanism.

Many political theorists have likewise failed, as Foucault observed, to cut off the king's head. Like conservative jurists, they remain skeptical of any policymaking power that is not tied directly to the sovereign voters and their elected representatives. “There is a persistent tendency,” notes political philosopher Joseph Heath, “to imagine the state as a single vertically integrated hierarchy, in which all major decisions are taken at the apex of power.”⁶ This idea “leads to an overwhelming emphasis on legislatures as the primary locus of public decision-making.”⁷ Agencies are supposed to be the sovereign's Hobbesian “nerves and tendons,” not a mind capable of judgment.⁸ Indeed, many follow Weber and identify democracy with “an imagined domain of rational deliberation, direct participation, or public reason that stands apart from and steers machine-like administrative structures.”⁹

The transmission belt model leads to the conclusion that administration will lack democratic legitimacy unless some very specific conditions obtain: that there is a discernable popular will on a policy issue; that Congress accurately transcribes this will in sufficient detail; and that agencies implement it mechanically and straightforwardly. One upshot is that theoretical and doctrinal work addressing administrative legitimacy is often oriented around constraint: it asks how one might force the bureaucratic behemoth into its proper placement on the transmission belt. Some scholarship suggests methods that restrain the human beings tending the administrative apparatus when they might otherwise be tempted to use its powers for their own purposes.¹⁰ Others address how best to get administrators to stick to their statutory mandates and how those mandates ought to be interpreted—and by whom.¹¹ Some scholarship even abandons the democratic justification of administrative power altogether. It concentrates instead on how to best reign in this inevitable, but normatively problematic, feature of the state. Conceding that administration will invariably make value-laden policy choices outside those contemplated by Congress, they will talk about how to temper administrative discretion by applying liberal¹² or fiduciary¹³ norms that courts might enforce. These scholars resign themselves to the expertocracy of administration, coldly comforted with an assurance that it might, at the very least, avoid violating liberal or fiduciary values.

Yet, when we see administration as a democratic embarrassment, we kneecap the project of collective governance. When the administrators tasked with protecting the environment and safeguarding public welfare dare reach beyond the clear, unambiguous meaning of an implementing statute, when they make judgments that cannot be derived directly from some tiny piece of codified text, they find themselves apologizing for their existence. Boxed in by the transmission belt theory, they must concede that their actions are undemocratic, but are nevertheless justified by necessity or pragmatism.

To be sure, if administrative decision-making power is not subject to any kind of democratic constraint, we will be shaken by the bureaucratic nightmares envisioned by theorists from Weber to Foucault to Habermas.¹⁴ Our anxieties about “technocracy, the relentless conquest of instrumental rationality, and roads to future serfdom and despotism”¹⁵ are not unfounded. Fortunately, the transmission belt model is not the only way to incorporate popular participation into administration.

In the following sections, I will demonstrate that the transmission belt model fails to provide a solid foundation for democratic legitimacy, let alone the legitimacy of the administrative state. Namely, it relies upon a conception of popular sovereignty that takes literally several fictions: first, that the people have an identifiable “will”; and second, that this will can be applied regularly and without controversy by administrators as they interpret and execute the law. As a result, critics of administration often ascribe to congressional lawmaking more democratic credentials than it deserves. But this failure does not mean that administration is doomed to democratic exile. Instead, like many other contemporary lawmaking institutions, it benefits from the democratic legitimacy derived from political representation. Accordingly, after briefly surveying a few models of democratic representation, I will make the case that the trustee model of democratic political representation provides a convincing standard by which to measure the democratic credentials of the administrative state. The article concludes by showing how some superficially undemocratic moves by agencies might indeed claim democratic legitimacy.

II. Disassembling the Transmission Belt

The transmission belt model relies upon a problematic organ-body conception of popular sovereignty: that there is a “people” with a “will” that can be captured in the statutes granting authority to agencies. The model posits that elected political representatives’ statutory control over agencies lends agencies democratic legitimacy because “agency policy decisions … will presumably reflect the will of the people and achieve the consent of the governed.”¹⁶ This hierarchical conception of legitimate authority is an upshot of democratic legal positivism. If natural law or some other transcendental norm is not to govern, then polities are left only with the commands of their sovereign.¹⁷ In a democracy, where kratos (power) is vested in the dêmos, that sovereign is the people itself.

The idea that the popular will is “transmitted” from dêmos to agency relies on several cascading fictions. First, that “the people” has a will; second, that Congress will reproduce this will; third, that Congress accurately transcribes it into statutes; and fourth, that agency officials straightforwardly apply it in their rulemaking and adjudications. Holding to these fictions requires acceptance of various implausible premises. For example, Congress—a collective body of diverse human beings—has no “intent,” at least as it may be understood as some kind of psychological state unique to human individuals.¹⁸ As a result, courts reach to a variety of interpretive and substantive canons¹⁹ to lend it anthropomorphized motivations. The major questions doctrine is one example: it ascribes to Congress an intention not to delegate policy-making power over questions of major social and economic importance.²⁰ To be sure, there may be good reason to act as if Congress has such an intention, but that reason, under the transmission belt model, has to be because it somehow enables Congress to replicate the will of “the people.” Otherwise, it will be some outside rationale, not popular will, that shapes the law.

Legal scholars are familiar with the fictions used to attribute an intent to Congress when it passes laws. They are also familiar with the fiction of agencies’ role as statutory transcribers: administrators routinely make value-laden policy choices that cannot be directly identified within statutory text.²¹ The Legal Realist insight that statutes are indeterminate (or underdeterminate)²² is well received. Legal scholars are less familiar, however, with the fiction used to ascribe a will and an intent to “the people” itself. They take it as a given, as A.V. Dicey did, that a legislature would “produce coincidence between the wishes of the sovereign and the wishes of the subjects.”²³ For them, it is a matter of common sense: legislative primacy derives from legislators’ unique connection to the popular sovereign.²⁴ The problem is that it comes with some intractable empirical and normative challenges.

Empirically, it is absurd to believe that a country of hundreds of millions could agree to anything, and certainly not so much that they can act as a unitary macro-subject.²⁵ As political science research confirms time and again, citizens have no clear, identifiable policy preferences that are capable of aggregation into a coherent, collective purpose. Hannah Pitkin, a seminal theorist of democratic representation, noted decades ago that “a constituency is not a single unit with a ready-made will or opinion on every topic; a representative cannot simply reflect what is not there to be reflected.”²⁶ Often, voters only come to have opinions and beliefs about policy because representatives, activists, and other political entrepreneurs have won them over. Many cast ballots based on party identification alone. Further, citizens remain necessarily ignorant of most policy proposals and their consequences.²⁷ They instead rely entirely upon the judgment of lawmakers. Nowhere is this clearer than when Congress considers policy that lobbyists and staff scribe in backroom offices.²⁸ The vote, moreover, is a coarse tool for conveying information about voter preferences; survey answers change when questions are reworded.²⁹ Both the President and Congress claim electoral mandates. Yet both often disagree on policy. Whose views should take precedence is not obvious.³⁰

It is not clear, therefore, that lawmaking reflects the popular will and not the will of, for example, elites and manipulators. Especially given the sophisticated surveillance, targeted messaging, and misinformation applied by political actors, lawmaking outcomes beg more questions than they answer.³¹ Yet, even if we concede that the political process forms or informs voters’ policy preferences, the problem remains. The populace is too diverse to achieve collectivized reason and consensus—and certainly not in fine legislative detail.³² Everyone might agree, for example, that there should be freedom of speech. But because of the “inherently formal and indeterminate nature” of principles like freedom, there will be many reasonable but inconsistent elaborations.³³

Consequently, the principal–agent transmission belt model of popular sovereignty is, as political theorists David Runciman and Mónica Brito Vieira³⁴ describe it, plainly inadequate to describe what is going on. Voters do not behave in a way that makes them a plausible principal.³⁵ Yet pretending that they do yields disquieting implications. If anyone claims to have identified an intention that is shared by all, or even just a majority, one is not terribly likely to conclude that we have achieved social harmony. Rather, something more nefarious may be afoot. “The abstract idea of a sovereign people,” notes political theorist Bryan Garsten, “[tends] to become concrete in the form of demagogues claiming to rule in the name of the people.”³⁶ Attempts to establish mimetic identity between representative and represented is not the realization of democracy. Instead, as F.R. Ankersmit argues, it is an invitation to tyranny.³⁷ Closing the conceptual gap between ruler and ruled forecloses public contestation, effaces social difference, and gives politicians an excuse to engage in a “relentless and unchecked pursuit of their particular vision of the good.”³⁸ The fiction of organ-body popular sovereignty, if taken literally, therefore puts us “in the business of electing dictators who can rule by decree while hiding behind ‘the will of the people.’”³⁹ The idea that the people can be accurately represented is simply a lie. “[L]ike all such political lies, it [can] be maintained not only at the cost of truth but of blood as well.”⁴⁰

As a result, whatever it is that Congress is doing, it does not, and should not pretend to, literally translate the “will of the people” into law. The Schmittian temptation to equate the leader with the people applies not only to chief executives, but also to legislators. Much political theory accordingly assigns to legislative bodies a different function. While the function assigned varies widely, each deploys some model of democratic representation that preserves a space for legislators’ independent thinking and decision-making. For some, Congress should help resolve political conflict under decision-making procedures that give all citizens equal respect and concern.⁴¹ For others, it should serve as a deliberative, and deliberately counter-majoritarian, body capable of identifying truths about what is in the general interest,⁴² not uncommonly in a manner favorable to counterrevolutionary interests.⁴³

While many theorists would insist that the notion of popular sovereignty is a fiction, some suggest it might be a useful one. If we act as if the sovereign was popular, we might be encouraged to include those excluded from the political process and to attend to the interests of all.⁴⁴ If popular means everyone, then everyone counts. The dream of popular sovereignty is one where citizens all share a role in shaping the laws that bind them by, for example, expanding the franchise and equalizing political resources.⁴⁵ Further, the obvious fictionality of the idea is itself useful. It emphasizes the gap between the imaginary sovereign “people” and its real-life representatives. As a result, a government grounded in the idea popular sovereignty stirs up perpetual and interminable debates about who the people are, where their interests lie, and what it is they actually want.⁴⁶ Such debates don't just motor democratic discourse; they also ensure that any answer provided by any politician in power is always contingent. The obvious fictionality of popular sovereignty therefore helps preserve for future electorates access to collective decision-making. No one generation can credibly claim to reach the final answer to a polity's collective problems. Relatedly, if the place of sovereign power is conceptually empty—because its occupant does not, in fact, exist—then any claim to occupy it permanently is an undemocratic usurpation of power.⁴⁷

In light of these normative and empirical problems, the transmission belt model fails to legitimize administration. No matter how much the agency's discretion is constrained, no matter how “intelligible” the principle imparted might be, citizens will still experience regulation not as their own will, but instead as a rule imposed upon them. Claiming otherwise is itself an invitation to tyranny because it forecloses contestation: voters cannot complain because the agency is diligently filling in the mundane details of their own orders. The democratic justification of the administrative state, if there is any to be had, must lie elsewhere.

III. Democratic Political Representation

Fortunately, it can lie within a theory of democratic representation. Such a theory explains not only why any collective governance, including statutory lawmaking, can be democratic; it also demonstrates why the administrative state carries its own democratic credentials. I suggest that the trustee model of political representation provides a home for administration within representative democracy. The next section, after roughly defining democratic political representation and several models within it, offers a few justifications for this move.

A. Democratic Political Representation: A Summary

The “folk” understanding of representation often mimics the unfortunate transmission belt model.⁴⁸ The people, as principals, direct their agents, their elected representatives, to do their bidding. Representatives transmit exogenous citizen preferences, the “bedrock for social choice,” in a linear, bottom-up process that translates their preferences into law.⁴⁹ As a result, it is the task of political institutions to minimize the “agency costs” incurred when representatives disobey their orders—for example, through electoral sanctions.⁵⁰ When working properly, the individual human representative disappears, drowned out by the loud transmission of public opinion into legal authority. Although a significant amount of political science scholarship relies on this model,⁵¹ it, as explained above, has both empirical and normative drawbacks. It does not—and cannot—describe what is actually going on. If democracy requires perfect congruence between voter preference and law, no actually existing democracy can claim to be democratic.⁵²

It is time to jettison the folktale. As political theorist Bryan Gartsen explains, “The impossibility of fully and completely representing the people's will is integral to the concept of representation itself.”⁵³ Although it provides a framework “for realizing the democratic ideal of giving kratos to the dêmos, power to the people,”⁵⁴ representation is not an institutional shorthand used only because direct, unmediated democracy is impracticable or otherwise undesirable.⁵⁵ Unmediated rule, which perfectly transcribes the “will of the people” into law, would indicate presence, not representation.⁵⁶ Instead, political representation is best understood as rule explicitly undertaken by and through mediators.

Although it does not promise that the will of the popular sovereign will govern, this scheme of institutional mediation can promise that decisions will be made by representatives acting in the name of all⁵⁷ rather than by someone imposing their personal judgments on everyone else. To accomplish this “omnilateral” orientation,⁵⁸ models of democratic political representation identify how citizens can participate in decision-making in a way that helps to (1) maintain the normative priority of the ruled over their rulers and (2) ensure that the ruled have some equally distributed role to play in their own governance. Thus, representative systems are often assessed according to how well they incorporate equal involvement in political decision-making—a fundamental norm that Urbinati and Warren call “democratic autonomy.”⁵⁹ Representation vindicates democratic autonomy by allowing the otherwise absent citizens to be present during decision-making in some sense—that is, to have their interests considered evenhandedly, to have their authorization solicited with equal vigor, to enjoy an equal chance to have their ex-post judgments registered, and so on. For example, Runciman and Brito Vieira explain that citizens can be present in lawmaking if they are able to object to what is being done by their rulers.⁶⁰ Representatives will decide in light of these possible objections, perhaps changing their choices and their behavior as a result. Citizens thus attend legislative sessions as specters to haunt the legislators.

There are many ways to accomplish citizen presence within the lawmaking process. Representation accommodates substantial institutional creativity.⁶¹ The menu of techniques that can encourage citizen involvement and motivate lawmakers to take their citizens’ interests seriously is capacious. As a result, political theorists use a variety of models to describe democratic political representation. A successful representative democracy will exhibit many of them, often in combination and with overlap.⁶² Shrewd institutional design will mix, match, and commingle different models to maximize democratic autonomy.

First, there is the “congruence,” “mandate,” “compliance,” or “principal–agent” model, which reproduces the transmission-belt theory addressed above. Sometimes, the best way to incorporate some citizens’ presence is to treat them and their representatives as principals and agents, their agency costs policed via electoral sanction. Yet it involves the issues addressed above.

Most other models, in contrast, make good use of the autonomous judgment deployed by representatives. Rather than deny representatives any independent causal role in governance, these models accept it. They prove particularly useful in circumstances where direct links between citizens and representatives, like regular voting, are difficult to operationalize. None of them, further, relies upon any pre-existing social consensus or homogeneous “popular will” to lend legitimacy to political decision-making. They also explain why the mere diversification of leadership positions—whether on the bench, in the boardroom, in the presidential cabinet, or in Congress—can be counted as a democratic improvement, all else being equal.

Descriptive representation holds, roughly, that “someone like me will pursue my interests.”⁶³ Selected because of the characteristics they share with citizens (e.g., gender identity, religious beliefs, class, race, sexual orientation), descriptive representatives are motivated by personal reasons, not by the possibility of electoral sanction or constituent instruction. These representatives are left to decide how their identities will express themselves in the lawmaking process. This model, however, has some drawbacks. It risks essentialism⁶⁴ and obscures the fact that interests can be endogenous to the political process rather than derived from exogenous empirical characteristics such as racial identity. Finally, any decision-making body that perfectly resembled the people would simply reproduce the people itself. It would not, therefore, be representation at all. Accordingly, some judgment must be exercised as to which people, which characteristics, and which interests merit descriptive representation. Those judgments are far from obvious and noncontroversial.

The gyroscopic model posits that citizens authorize representatives who can be trusted, without oversight or sanction, to pursue goals that the voters find attractive.⁶⁵ Gyroscopic representatives, who reliably “spin on their own axes,” are understood as honestly motivated, skilled advocates who are therefore given discretion to deploy their judgment as they see fit. To illustrate, Ohio voters concerned about commercial exploitation might find representation in Massachusetts Senator Elizabeth Warren, a banking and financial law expert with credible commitments to working-class and consumer interests. The risks include (1) short-shrifting commercial interests and (2) pleasing only a radical base of supporters.⁶⁶ Gyroscopic representation works best, therefore, in a system that draws representatives from many different viewpoints.

The trustee model of political representation, unlike the dyadic models described above, has a tripartite structure. It is, as a result, a highly mediated form of rule. A principal (party one) authorizes a trustee (party two) to act for a beneficiary (party three). For example, a voter may authorize a representative to use their best judgment while acting in the whole country's interest. This model, unlike the others, enables the representation of persons (both natural and fictitious) who cannot speak and act for themselves.⁶⁷ As a result, a trustee can, for example, be said to act for “the people”—“the people” who cannot, normatively and empirically, act for themselves. Arguably, all members of Congress are trustee representatives. They are elected (authorized) by a particular geographic constituency yet make decisions that hold for the entire country and do so in the form of laws that will hold for generations to come.⁶⁸

The absent beneficiary is made present (represented) through an obligation on the part of the trustee representative to act in its “best interests,” even when they conflict with the wishes of the authorizer.⁶⁹ Using some logical deduction, we might tentatively hold that to act in someone's “best interests” means to act carefully and with the subjective belief that the action is in the beneficiary's interest and is not, at the very least, self-serving.⁷⁰ Meanwhile, although the authorizing party can set the terms and scope of the trustee's authority, it retains no day-to-day control over them. As a result, the trustee enjoys an ominous amount of autonomy⁷¹ when determining the beneficiary's best interests.

Admittedly, many associate trustee representation with the paternalism of Edmund Burke⁷² and James Madison.⁷³ After all, it concedes to the trustee the authority to make decisions on behalf of other human beings who are capable of deciding for thesmelves. But we should not be so quick to jettison the model. The autonomy granted a trustee representative can enhance democratic autonomy in a few ways. First, trustee representatives can provide presence to citizens who cannot speak and act for themselves. They can represent those not yet born, the disenfranchised, children, the incapacitated, and so on. Even if such voices cannot, as a matter of physics or legal right, be present, they can nonetheless play a role in collective decision-making if someone speaks and acts for them. Trustee representatives can also speak and act for those wrongfully shut out of the political process because of economic and social inequalities. When political resources are stacked in a lopsided manner, when the marginalized are so overburdened by the demands of ordinary life that they cannot speak for themselves, only a trustee can give the marginalized any (admittedly ventriloquized) voice at all.⁷⁴ Similarly, when matters are too complex for ordinary citizens to understand—let alone form a viewpoint about—only a trustee representative can act on their behalf.⁷⁵ Of course, many would rather citizens speak and act for themselves. But, in an imperfect and unequal world, the trustee model is more democratically appealing than a model that prioritizes the voices of those with outsized power, wealth, and influence. During collective decision-making, a trustee can prevent the effacement of ordinary citizens and their interests. Further, acts that a trustee representative might undertake in the interest of disempowered beneficiaries might enable them to participate on their own behalves in the future.⁷⁶

Moreover, the trustee model can, unlike other models of representation, provide an answer to what is, according to Thomas Hobbes and Bernard Williams,⁷⁷ the first question of politics. The trustee's duty to act in its beneficiary's “best interests” imposes on it a duty to establish some amount of security and safety. Although, as explained below, a “best interests” determination is always contestable, it assuredly includes this minimum. Yet it is not obvious that citizens always know which actions might successfully stifle the outbreak of a deadly virus or defend against violent invasion. Moreover, presuming that citizens do have such knowledge, it is not obvious that citizens will possess sufficient collective capacity to act on it. Other models of representation nevertheless suggest that leaders follow voters’ myopic preferences, even if they are epistemically unsound or prioritize conflicting policy goals. In contrast, a trustee model lets experts be experts. This is not to argue that experts should remain insulated from critique and contestation. Expert trustees’ claims to act in the public interest, because both interpretive in nature and reliant upon imperfect knowledge, should face and respond to the crucible of citizen objection.⁷⁸ Facts will always present with an Arendtian slipperiness;⁷⁹ truth claims can always be challenged. But a trustee does not need to apologize for “following the science.” Instead, it is probably a trustee representative's duty to do so.⁸⁰

Recently, representation has taken a constructivist turn.⁸¹ Important scholars have now abandoned the idea that voters come to politics with ready-made preferences and interests. Instead, voters’ opinions are the result of successful appeals made by representatives. Representation is not a spatial concept, “re-presenting” voters located in geographical districts through appointed agents located in a governing body. Instead, representation is better understood as a communicative process whereby representatives call the represented into being by making claims that are accepted as representative by an audience. Emphasizing the tutelary aspect of democratic discourse,⁸² a representative's claim to represent a constituency is considered successful if it is taken up by them by signaling their approval with a vote, a campaign contribution, and so on. Laudably, constructivists do not rely on organ-body sovereignty or the existence of a people with exogenous preferences or objective interests that a democracy should translate into law. Moreover, all theories of representation must involve some amount of interest construction as representatives predict how voters might respond to policies and craft their public appeals to secure citizens’ support. In particular, the trustee model, which counts on trustees to “interpret the group's interest, and in doing so put forth a claim to be representing it,”⁸³ has constructivist traits.

Unfortunately, though, constructivist theories lead one to a normative dead-end.⁸⁴ Either they are too relativist or they are too rationalist.⁸⁵ They provide no way to distinguish successful manipulation from ordinary representation without also relying on standards of judgment that are perfectionist—viz, whether a voter's decision to take up the representative's claim is rationally correct or made only after fulfilling certain deliberative desiderata.

B. Representation and Lawmaking

If successfully implemented, models of democratic representation lend legitimacy to the outcome of representatives’ decision-making. These outcomes, including legislation, will be authoritative because they are the result of processes that vindicate democratic autonomy: their creation involves citizens in some way and in a manner that prioritizes them (including their preferences and interests), not the lawmakers. Such outcomes deserve our respect to the extent that they are the expression of democratic autonomy. As a result, there is good democratic reason to treat congressional lawmaking as authoritative. Regular elections, a near-universal franchise right, protections for free speech, and an increasingly diverse legislature lend it democratic credentials.

But congressional lawmaking is not democratically legitimate merely because it is congressional; it is legitimate because the procedures that yield legislation reflect democratic autonomy. Accordingly, federal statutes may be more or less justified depending on how well their procedures perform. If representative processes are skewed, legislation is accordingly less democratic and less justified. For example, legislation carries less democratic weight if large portions of the dêmos are excluded from participation due to onerous voting requirements and extensive gerrymandering. Legislation will carry less democratic weight if corruption or supermajority rules persuade lawmakers to regularly prioritize a favored elite at the expense of others’ equal claims for attention.

Moreover, since Congress holds no monopoly on representative institutions, the outcomes of other decision-making procedures likewise carry democratic weight. Indeed, representative systems explicitly discard the “formalist vestiges of prerogative, sovereignty, and even lordship … framed primarily around juristic conceptions of delegation, the separation of powers, the rule of law, and constitutional limitations.”⁸⁶ Rather than locating power in a specific body, democratic representation permits the proliferation of rulemaking settings. State and local legislatures certainly enjoy democratic legitimacy. The processes used to select policy-setting executive leaders and judges may also respect democratic autonomy. If the standard is democratic legitimacy rather than constitutional formalism, Congress deserves primacy only if no other institution better instantiates democratic autonomy.

Moreover, if other representative institutions contribute to or elaborate upon its decisions, Congress's democratic credentials might grow even more authoritative. Congressional lawmaking in isolation does not maximize the amount of democratic autonomy that citizens might enjoy. Congress is but one (important) part of a representative system that includes both formal and informal representative processes occurring within and between multiple institutions.⁸⁷ Bicameralism and presentment, for example, weave together three different representative institutions in the creation of statutes. The result, one hopes, is a greater level of democratic autonomy than each would yield on its own. These statutes prompt further representative processes—for example, implementation at the state level through local representatives; appointment of reviewing judges; interpretation by chief executives; and, as explained below, elaboration within agencies. Meanwhile, many rest upon both formal and informal representative processes involved in the selection of intra-party candidates and policies. Each step in the life of a statute, from drafting to enforcement, provides an opportunity to enhance its democratic credentials. Thus, to the extent that there is any kind of popular macro-agent, it is not an organic popular sovereign, an identifiable group person with a will of its own. It is instead the ongoing and immanent production of decisions made through representative processes that help vindicate the norm of democratic autonomy.⁸⁸

Of course, those holding to a rigorous (idolatrous?)⁸⁹ separation of powers doctrine and a strict interpretation of the Vesting Clause⁹⁰ will reject this intersection and accumulation of representative practices. But they can do so only at great cost. They must select from a set of undesirable conclusions. First, they could concede that Congress, although the only legitimate lawmaking power, may not be as democratic as one might like. Second, they could insist that Congress is maximally democratic, but only by adopting the unfortunate transmission belt model. This is the only model that attributes to Congress the power to speak full-throated with the “voice of the people.” Finally, they might prioritize the text of the Constitution (as they understand it) over democratic autonomy. This means that they must either (1) adopt a vulgar democratic legal positivism that prioritizes decisions made hundreds of years ago over those made in the present or (2) value the rights protected by constitutional constraints (like the separation of powers) over democracy itself.

C. Administration as a Representative Institution

Accordingly, representation can lend democratic legitimacy to all kinds of public decision-making, not just the laws passed by Congress—including actions taken by administrative agencies. There are a few reasons why this move makes sense. Administrators are part of a state conceived as separate from citizens, and the work they do is, at least notionally, on behalf of those citizens. As public policy scholar Anthony Bertelli puts it, “this is representation.”⁹¹ Political representation makes space for administration's division of labor, a mediation that accommodates citizens’ expectation that an “intelligent, civic-minded state official [will] take care of problems.”⁹² Those officials will invariably address questions of political import. Their decisions involve both “facts and value commitments, both ends and means”—all of which are “inextricably intertwined in political life.”⁹³ These are precisely the types of questions with which representation is designed to deal. There is a morphological fit, in other words, between models of political representation and the role played by administration in governance.

Theories of representation can also open new avenues for research and reform. One common critique of the administrative state is that it has a “democracy problem”: it makes value-laden policy choices without the popular consent implied by elections. Democratic representation preempts this complaint because it does not require—indeed, often conspicuously excludes— citizens’ knowing understanding and prior approval of everything an official does in their name. Instead, it uses indicia of democratic autonomy to assess the democratic quality of an institution's output. As shown above, the franchise is far from the only mechanism that can be used to accomplish this. Administration's so-called “democracy problem” might be mooted as a result of the considered use of, for example, descriptive, gyroscopic, and trustee representation in the design of administrative institutions. The next section will demonstrate the fruitfulness of this approach by applying a trustee model.

More specifically, many models of democratic representation avoid the transmission-belt theory's organ-body sovereignty conundrum. As a result, they can suggest something meaningful about the democratic legitimacy of administrative action beyond whether or not it carries out the orders of a Congressional principal who itself is thought to follow the orders of “the people.” A theory of representation therefore allows administrative law scholars to transcend interminable theoretical debates that focus narrowly on the proper constraints to place on administrators as they interpret and apply implementing statutes. It allows for the possibility that administrative action is the product of democratic representation and therefore legitimate even if there is no crystal-clear statutory directive to implement. As a result, it suggests that there are ways to democratize administration beyond, as the U.S. Supreme Court has recently implied,⁹⁴ holding Congress's feet to the fire and forcing it to codify policy in ever more exhaustive detail.

IV. Why the Trustee Model is Appropriate

The remainder of this article will suggest that administrative agencies can serve as democratically successful trustee representatives. This is not to argue that other models of representation are irrelevant or inapt. A central bank's democratic credentials, for example, might be improved if its significant offices are held by descriptive and gyroscopic representatives rather than by the designees of financial concerns. I select the trustee model, however, because of its tight fit to extant administrative practice. As a result, it can be used to assess administration's democratic quality as it currently functions. Second, its theoretical resonance with pathbreaking scholarship on fiduciary theories of government means that scholars applying it will be well served by an existing body of rich literature. Finally, the trustee model fills a normative vacuum left by other models of representation and boasts several features congenial to democratic autonomy.

A. Conceptual Fit

First, the trustee model's tripartite framework reasonably describes administration's legal infrastructure: the model suggests that an authorizer (e.g., Congress, through statutes) authorizes a trustee (the agency, understood as a collective body) to take care of the interests of a beneficiary regarding some delimited subject (e.g., the people's health and welfare, consumers, racial minorities, and so on).

Further, the model expressly contemplates that the beneficiary is unable, for one reason or another, to act for itself. When it comes to the regulation of complex financial schemes and scientifically intensive public health interventions, this is precisely the public's circumstances. When an agency looks out for the interests of political minorities and the economically disadvantaged, it is looking out for those citizens whose political voices are silenced. Many administrative agencies pursue the interests of those shut out of the political process. For example, the Equal Employment Opportunity Commission (“EEOC”), the Department of Justice Civil Rights Division, the Federal Trade Commission (“FTC”), and the Consumer Financial Protection Bureau (“CFPB”) advocate on behalf of those subject to discrimination and without the resources necessary to mobilize on their own behalf. Relatedly, agencies can act for an entire country even though a country cannot, as a practical matter, act for itself. The model accommodates the constructivist and Hobbesian notion that there is no concept of “people” prior to politics.⁹⁵ An agency dedicated to defense and security, for example, may carry out tasks in the (loosely defined) “national interest.” There may be no “nation” that can speak for itself and its interests, but the Department of Defense can do so in its stead.

Moreover, baked into the trustee model is the inevitable discretion wielded by administrators as they interpret and apply their statutory mandates. There are a few reasons why administration's decision-making autonomy is inevitable. First, the inherent ambiguity of agencies’ organizing statutes will require interpretation and elaboration. No amount of rule-tightening or definitional glossaries will ever remove all ambiguity from the law.⁹⁶ Legal interpretations are commonly controversial and, even when the law's meaning is clear, how it ought to apply in unanticipated circumstances may not be obvious.⁹⁷ Appealing to legislative intent will not eradicate the ambiguity. Ascertaining legislative intentions, which requires ascribing intentions to a collective agent, is also a matter of interpretation—one that involves linguistic ambiguities, problems of systemic coherence, and so on.⁹⁸ And when law is indeterminate or underdeterminate,⁹⁹ agencies must exercise interpretive discretion that cannot help but prioritize some values at the expense of others.¹⁰⁰ A statutory instruction to evaluate drugs for safety and effectiveness doesn't tell administrators a lot about how much risk should be traded off for marginal increases in efficacy.

Second, given their expertise and experience, administrators will have greater understanding of the issues in question, and therefore enjoy an independent source of political influence over lawmakers.¹⁰¹ Greater popular participation cannot expunge this source of autonomy. Citizens rarely know what regulations cover and consequently cannot evaluate them responsibly.¹⁰² But their elected representatives may fare no better. Legislators’ capacity and incentive for oversight is limited, constrained to episodic formal investigations and debates focused on scandal rather than policymaking.¹⁰³ Politically appointed agency officers may enjoy only a limited understanding of their duties.¹⁰⁴ In contrast, agency officials’ expertise, connections to well-organized stakeholders, and reputation for effectiveness give them authority. They leverage this authority to shape the legislative process itself as they propose bills and amendments.¹⁰⁵ As political philospoher Joseph Heath observes, the civil service “is often the source of highly effective policy entrepreneurship.”¹⁰⁶

Finally, the model does not suppose that the beneficiary (e.g., the public) must vote for the trustee. The trustee's authority to decide on its behalf and in its interests is not necessarily electoral. Rather, it comes from a distinct party, one with authority to appoint the trustee. Administration's “unelected bureaucrats” can therefore find not exile from, but rather a home within, the trustee model.

B. Resonance with Fiduciary Theories of Government

In addition to the snug conceptual fit between the trustee model and actual administrative practice, extant literature applying fiduciary concepts to administration already exists.¹⁰⁷ The theoretical overlap between these fiduciary theories and political trusteeship not only gives scholars interested in trustee representation a head start, but also demonstrates the plausibility of the project. Indeed, some of this pathbreaking scholarship is cited below. However, the trustee model of democratic political representation—at least as it is deployed here—differs from such fiduciary theories. Most importantly, whether addressing administration, the U.S. Constitution,¹⁰⁸ the judiciary,¹⁰⁹ the executive,¹¹⁰ or otherwise,¹¹¹ fiduciary theory does not attend to the justification of power as such.¹¹² Inspired by Locke's liberalism, many focus instead on how that power ought to be limited by the (judiciary's)¹¹³ application of fiduciary duties. It is a fungible theory that, as its advocates put it, applies to any relation of authority,¹¹⁴ regardless of how it comes about. As a result, it shares the same weaknesses of many liberal and republican¹¹⁵ theories of government mentioned in the introduction of this article: it leaves unexamined the constitution of collective power. This neglect implies that such power is, although perhaps inevitable, au fond unjustifiable—a beast to be captured. In contrast, the trustee model of political representation can be part of a political framework that expresses democratic political autonomy. As illustrated in Habermas's analogy of democracy with the two-headed Janus,¹¹⁶ it is part of the left-facing constitution of collective decision-making. Fiduciary theories, like many republican and liberal theories, are part of the right-facing constraints on that decision-making. Many of its proponents invoke it precisely because they, equating democracy with consent, are skeptical of democracy itself.¹¹⁷ As explained above, however, democratic representation requires no such consent. At the extreme, a democratic theory requiring universal consent is an invitation to populism—or worse.

C. A Stronger Model of Representation

To the extent that any model of political representation requires that representatives exercise their own judgment when constructing and interpreting constituent interests, a model that recognizes and addresses the autonomy implied by such construction is more empirically accurate.¹¹⁸ Because it is more accurate, it can help scholars to examine the normative implications of this autonomy. The trustee model may, therefore, be a better model of representative democracy than the others. This is the third reason for the selection of the trustee model.

To explain, recall the tutelary character of constructivist models. They describe how representatives mobilize voters’ political interests into existence. Such mobilization requires a representative to think and act independently of the electorate because the electorate is not as yet aware of these interests. Descriptive models of representation may also require representatives to autonomously construct constituent interests as they decide which aspects of their group merit attention and which ones do not. Even mandate models must build in room for a representative's independent judgment. Representatives face votes only every few years, and that vote contains very little information about voters’ preferences on particular policy issues—even if a constituency could form a coherent unified preference.¹¹⁹ Yet each of these models has very little to say about the inevitable discretion wielded by representatives. Ironically, they therefore have very little to say about how representatives ought to go about their jobs and how citizens ought to evaluate them.¹²⁰ The trustee model, however, is not so hobbled.

In fact, the obligation to act in the beneficiary's interest lays upon the representative a normative yardstick to measure how well representatives exercise their inevitable judgments. Recall that constructivist models’ desiderata fail to screen out successful voter manipulation. So long as voters seem to take up (or acquiesce to) the claims made, representation is judged a success. Similarly, a successful mandate representative will be judged a success upon election, even if they take up the most ill-advised and ill-informed voter preferences. A trustee model, on the other hand, might suggest that manipulation and blind obedience is a violation of the representative's duties to serve their beneficiaries’ interests. In failing to serve the public's interests, a representative is failing to represent the public. Whatever the representative is doing, it is not representation. The representative is instead perhaps acting in their own interests. Further, to assess the success of descriptive representatives beyond their descriptive characteristics, one cannot help but reach for the fiduciary standards suggested by trustee representation: Did they decide after proper research and investigation? Did they have good reasons? Did they try in good faith to articulate the interests of those like them, or did they act for self-interested or reckless reasons? Simply, the other-regarding duties laid on the representative are what makes them representative of others.

As a result, the trustee model furnishes a normative standard missing from other models of democratic representation. That standard should prove attractive to advocates of democracy. To illustrate its appeal, consider a trustee representative authorized to act on behalf of public health. Of course, the public interest, including the interest in public health, is subject to deep contestation. The notion of a public is fictional and socially constructed—but it is not entirely open-ended. Namely, it suggests that policymaking should not reflect the partial preferences of the few at the expense of the many.¹²¹ Decisions have weaker claim to be in the public interest, and therefore to represent the public, if they reflect concern and respect only for a small portion of the population. In other words, the notion of the public interest is a principle of inclusion rather than exclusion.¹²² As William Novak argues, the growth of government institutions oriented to the public welfare arose contemporaneously with the idea of universal citizenship ascribing equal rights to all.¹²³ Consequently, the trustee serving the public makes present a greater swath of the population than a mandate representative who openly serves partial interests. At the very least, a commitment to the public interest forbids highly resourced private parties from hijacking the collective decision-making process to serve their own ends.¹²⁴

The autonomy enjoyed by trustee representatives to decide and act in the beneficiary's interests brings additional democratic benefits. First, it respects the lack of consensus among the represented. It does not matter that citizens, both present and future, cannot form preferences regarding (or agree about) the actions that the representative should take. To act legitimately, it only matters that the trustee pursues their best interests as authorized (leaving aside, for the moment, how “best interests” and “authorization” are defined).¹²⁵ Especially in situations where it is not obvious that government inaction is any more democratic than government action, the trustee representative's autonomy unshackles public officials from dangerous gridlock or inertia.

Furthermore, the conspicuous tripartite distinction it draws between the authorizers, the trustee, and the beneficiary is appealing because it prevents trustee representatives from convincingly claiming to speak with the “voice of the people.” Unlike other models, which might condone any action taken by lawmakers so long as they garner popular support, a trustee model prevents a ruler from claiming an identity¹²⁶ with the ruled. As J.S. Mill¹²⁷ and de Tocqueville¹²⁸ once pointed out, majority sentiment can quieten public debate. On the contrary, the trustee model emphasizes the non-identity between the ruling trustee and the governed beneficiary. It therefore invites constructive popular contestation where other models might smother it. Ironically, the thick institutional mediation separating the trustee representative and citizen beneficiary may encourage democratic feedback rather than suppress it. Of all the models of democratic representation discussed in this article, it is least susceptible to populist politics.

Representative independence from the electorate can also make them better advocates.¹²⁹ They can therefore lend citizens priority and presence in decision-making that they would not otherwise enjoy. An attorney presenting a case on behalf of her client, for example, is expected to use both their expertise and their experience to secure outcomes that are superior to those that clients might achieve on their own. As Dewey once observed, although the man who wears the shoe knows best that it pinches and where it pinches, it is the shoemaker who can best judge how the trouble is to be remedied.¹³⁰ Likewise, a trustee advocate within a legislative body, exercising independent judgment and expertise, can press the interests of a constituency more effectively than it can itself.

Finally, a trustee's obligation to pursue the public interest should gratify scholars with deliberative commitments. The idea of the public, a multitude of human beings existing coherently together, attends to the ways individual experience affects, and is affected by, broader social and economic conditions. It evokes the fact of human interdependence and, as a result, suggests that public norms should reflect the same kind of reflexivity and generality championed by social contract philosophers from Kant to Habermas. In any event, many participatory and deliberative models of democracy tacitly rely upon a version of the trustee model. Theories that promote the use of deliberative mini-publics, for instance, ascribe to them the kind of autonomy, informed advocacy, and other-regarding cognitive orientation attributed to trustees.¹³¹ These bodies, acting for the population at large, deliberate and decide on which actions will be taken on its behalf. Members are instructed not to parrot popular views or pursue naked self-interest, but instead to bend their views according to the unforced force of the better argument.¹³² They must imagine what the interests of others might be and how a proposed rule might impact them. And members are encouraged to endorse decisions that, thanks to deliberation's generality and reflexivity desiderata, articulate some notion of the public good. Some theories of legal interpretation would likewise find friendship with a trustee model of representation. Those that lean on deliberative norms rather than popular participation require decision-making officers to take everyone into account and give meaningful weight to their different points of view,¹³³ or give “reasons calling on some understandable vision of the public welfare or purpose” and reasoned responses to public objections.¹³⁴

Unlike deliberative theories of democracy, however, the trustee model incorporates the idea of the public interest without making any commitment either to its precise content or the acceptable ways in which contributors might participate in public reasoning. First, the trustee model is less sanguine about the possibility that trustees land on some rationally correct or unanimous answer to the question of the common good. A trustee's claim to authoritatively speak in the beneficiary's best interests is always provisional. Indeed, as explained in the next section, the idea of “the public interest” in trustee representation is a political, not a juridical, question. It accepts that citizens may object to, and should have the means to object to, the trustee's interpretation. A trustee is no Judge Hercules.¹³⁵ Second, the idea of the public interest accepts reasons that are both “communicative” and “instrumental,” both value-based and technical. It therefore subjects both science and norms to political judgment. Further, the trustee model does not require that that the only reasons acceptable in the justification of collective decision-making are those that each and every citizen might or should accept. As a result, the trustee model may appeal to those skeptical of the legitimizing power of public reason.¹³⁶

V. Unpacking Some Definitions

Accordingly, the trustee model of political representation should provide some traction into the democratic legitimacy of administrative agencies. Before applying the model, however, it is worth unpacking a few of its elements and addressing some potential objections.

A. The Bounded Autonomy of the Trustee Representative

As mentioned above, a trustee representative enjoys autonomy in its decision-making. The trustee does not, however, operate without any constraints whatsoever. On the contrary, it is bound to act in the interests of its beneficiary.¹³⁷ Only then can the trustee claim to give presence to the beneficiary in its decision-making. Some minimal, but meaningful, substantive conditions can be deduced from this basic obligation. First, as fiduciary theorists point out, the trustee cannot act in its own self-interest.¹³⁸ If trustee representation consists of acting on behalf of another, trustees cannot act only for themselves. For example, the Social Security Administration, as a trustee, cannot steal resources set aside for entitlements in order to build lavish offices. Second, a trustee cannot act in such a way that no one could ever believe that it is acting in the beneficiary's best interests. The Center for Disease Control (CDC), as a trustee of the public health, cannot pour deadly poison into the public water supply. Third, the trustee must in good faith (actually) believe that the actions it is taking are in the beneficiary's interests. The Open Markets Committee of the U.S. Federal Reserve cannot unthinkingly and substantially increase interest rates without any reason to believe that its intervention will either stabilize prices or increase employment rates. Whatever good faith means, it is certainly not intentional or reckless disregard of the beneficiary's interests.

These sorts of obligations, under private fiduciary law, are typically enforced by courts in lawsuits. The Delaware Court of Chancery, for example, specializes in holding fiduciaries to their duties of loyalty (including good faith) and care.¹³⁹ While courts may therefore be competent to police the political trustee's similar duties, there is reason for some skepticism. First, institutions more permeable to democratic participation might also enforce them—for example, electoral sanctions, legislative investigations, administrative ombudsmen, and so on.¹⁴⁰ To the extent that a strict interpretation and application of these duties import a judge's personal values in a way that shapes outcomes, using these mechanisms may be more democratically desirable. There is no reason to believe, moreover, that courts are better placed than an expertly staffed review panel to assess the quality of care taken by a trustee making policy under conditions of economic, social, or scientific complexity.¹⁴¹ Accordingly, if judicial review is appropriate, it probably should be deferential. The Delaware Courts themselves recognize their own lack of business acumen, for example, by implementing a highly deferential “business judgment rule.”¹⁴² They refuse to intervene with corporate decision-making unless it is plainly reckless or self-serving. Furthermore, prophylactic measures might be taken to prevent the breach of such duties. “Good types” or “gyroscopes,” unlikely to breach, might be authorized as trustees through the use civil service tests, professional vetting, and so on.

B. Politicizing the Public and Its Interest

Notwithstanding these minimal constraints, the duty to act in the beneficiary's interest is left open. Indeed, the parameters are so vague that many might reject the trustee model at the outset because it does little more than encourage paternalistic policymaking by electorally unaccountable officials. As hinted at above, the notion of the “public” and its “interest” are indeed fictions, and the business of defining them might be monopolized by powerful actors. Further, both representatives’ and citizens’ honest beliefs about what serves the public's best interest routinely conflict. The open-endedness of the “best interests” calculation even drives some scholars to spurn fiduciary theories of government because it lends too much power to the institution granted the authority to define it.¹⁴³ There is good reason, according to critics, for colonizers to have justified their colonization using fiduciary excuses.¹⁴⁴ They also complain that it provides little traction into the question of what public officials ought to do.¹⁴⁵ A procedural question about how much deference to lend the trustee is often really a substantive question about how the trustee ought to act.¹⁴⁶

I suggest that this open-endedness is a feature, not a bug. This is because what is in a beneficiary's best interests is subject to construction and contestation; it is always permeable to democratic politics. Determinations about what's in the “public interest” is a constructed product of politics, not a precondition that establishes the need for politics.¹⁴⁷ Contests about who ought to be included in the dêmos, where its interests lie, and how best to achieve them are the fuel that motor political discourse. Especially in conjunction with the highly mediated tripartite trustee model, citizens’ democratic instincts will accordingly drive them to critique trustees’ best interest determinations. To illustrate, consider an agency tasked to promote public education. Debates will arise over who, exactly, is included in this public and what, exactly, its best interests are.¹⁴⁸ Is the public interest best served by targeting the education of minor children or adults? Is it best served by prioritizing elementary education or higher education? The maths and sciences or the humanities? The determination of “interest” is not the only thing politics needs to construct. It also needs to construct what it means by “public.” Indeed, “the functioning of representative democracy depends upon politicians being able to offer competing visions of the people to the people.”¹⁴⁹ Are we a technologically advanced society dedicated to industrial innovation, or a society of letters and literature? Are we an agricultural heartleand or a cosmopolitan center of commerce? Often, our competing answers to these questions map on to our partisan cleavages. For example, are immigrants included in the dêmos? Is it in the interest of the dêmos to ensure a minimum standard for all students, or should we prioritize choice and competitive merit? Will we best achieve these goals through privatization or by funding public schools? Finally, because interests cantilever between objectivity and subjectivity, they require construction by political actors who can and should be challenged within democratic politics.¹⁵⁰ The agency on public education might prioritize high-level reading skills because, objectively, reading is a prerequisite to satisfying and remunerative employment. But what about those who prefer working with their hands?

Stated more directly, the bests interests calculation is a political question, not a legal one. It should not be surprising, therefore, that the democratic politics surrounding administrative policymaking erupts during presidential elections. Selecting a president becomes a proxy for selecting, for example, the best interests of the public when it comes to immigration policy, climate policy, competition policy, and so on. So long as the trustee representative obeys its minimum substantive duties, courts should avoid interfering. Meanwhile, like most political questions, there is room for democratizing the answers. Additional institutions can lend citizens and their interests more presence within a trustee representative's decision-making. For example, trustees can accept public feedback and respond to public objections. Their decisions can be made transparent and publicly available. Further, if the trustee is not an individual, but a body, that body might be staffed by elected, gyroscopic or descriptive representatives. Using carefully designed representative systems, unlike direct participation mechanisms, can help ensure that proceedings remain efficient and elude capture by well-resourced interests.¹⁵¹

C. Trustee Authorization

Finally, it is important to make clear what authorization purports to do—and what it does not. From Hobbes,¹⁵² we know that authority is conferred by an author, the person who has the right to act. Within the context of trustee representation, this authorizer is recognized as having legitimate power to appoint a trustee to care for the interests of a beneficiary. It is also recognized as having the legitimate power to define the scope of a trustee's duties: to determine what is ultra vires and thus not an act of representation.¹⁵³ Sometimes this legitimacy arises from a pre-existing normative relationship between the authorizer and beneficiary.¹⁵⁴ For example, a parent might authorize a fiduciary to invest funds on behalf of a minor child. Unlike in a principal–agent model, this relationship is not identity: there is reason to treat authorizers as distinct moral agents whose interests may diverge from those of the beneficiary.¹⁵⁵ Despite the relationship between parent and child, the trustee is not the parent's agent. In fact, to fulfill its duties, a trustee might find itself acting against the parent's express wishes. The parent might want the trustee to contribute trust funds to her pyramid scheme. The trustee, duty bound to act in the child's best interests, should refuse.¹⁵⁶

Within the context of democratic politics, the capacity to authorize a trustee representative is usually attributed to a sovereign “people,” sometimes through a founding constitutional document.¹⁵⁷ Of course, as Hobbes pointed out, “the people” does not exist prior to its representation. There is only the multitude.¹⁵⁸ As a result, his analytical scheme of political legitimacy required the concurrent appointment of a representative and the creation of the people through that representative.¹⁵⁹ A more democratically appealing option is for individual voters, each of whom has a claim to be part of “the people,” to authorize trustees by participating in elections.¹⁶⁰ Specific individual voters jointly authorize members of Congress and the President, as trustee representatives, to decide collectively on the laws that best serve the people-as-beneficiary. Moreover, the model recognizes that the-people-as-beneficiary is a moral agent distinct from the-people-as-authorizers. As a result, these representatives might provide for the interests of those who cannot vote, those who voted for someone else, those who are not yet born, and so on. This gap also opens up space for various trustees’ rival claims to represent the interests of the people-as-beneficiaries. It thus is congenial to a constitution that separates powers and facilitates political opposition. For example, particular subsets of the population elect members of Congress. Another distinct subset elects the President. Both Congress and the President have authority to represent the people-as-beneficiaries, whose interests cannot be reduced to those of their (distinct) authorizers.

Of course, the boundaries of a trustee's authority are not always clear. Congress may be authorized by “the people,” acting through the Constitution, to regulate interstate commerce,¹⁶¹ but hesitate over whether its authority includes firearm regulations that protect victims of domestic violence. At this point, one observation should be clear. Given the autonomy ascribed to the trustee, it should not simply take the authorizer's interpretation at face value. Its duty is owed to the beneficiary, not the authorizer. After all, the trustee representative is not representing the authorizer. It is representing the beneficiary. And the authorizer may have interests that conflict with those of the beneficiary. Because the trustee is duty bound to wield its authority in the beneficiary's best interests, it should reject any instructions that do not serve those interests. Accordingly, from the trustee's viewpoint, the question of authority often resolves, like any best interests determination, into a political question—not a legal one.

VI. Agencies are Trustee Representatives

With the parameters of the trustee model defined, it can now be applied to administration. Here, I argue that administration can be understood as a form of democratic trustee representation. Presuming that the model lends legitimacy to political decision-making by helping to establish democratic autonomy, there is consequently good reason to believe that administrative agencies enjoy a source of democratic legitimacy—even if they fail to follow the precise marching orders of a Congressional principal. Furthermore, applying the model to administration explains why certain superficially undemocratic agency actions can nonetheless claim democratic credentials.

A. Coherence with Administrative Practice

The idea that an administrative agency is a trustee representative acting in the public interest fits comfortably with actual administrative practice. As Joseph Heath observes, “there is little disagreement that the civil servant should be committed to serving ‘the public interest,’ or ‘the general good’.”¹⁶² The Interstate Commerce Commission, established in 1887, was charged by statute with “protecting and promoting the public interest” by establishing “reasonable and just” and non-discriminatory railway rates.¹⁶³ Historically, courts have afforded autonomy to civil servants because they worked within institutions designed to evade capture by private, special, or pecuniary interests.¹⁶⁴ Unlike courts, agencies self-consciously engage in general, all-things-considered policymaking that impacts a broad group of individuals rather than opine, as courts do, on the rights of particular parties.¹⁶⁵ Furthermore, the development of public administration within the United States developed concurrently with social sciences that articulated a concept of public welfare and the public good.¹⁶⁶

The case that administrative agencies are successful democratic trustee representatives is elaborated in more detail below.

1. Agencies Act in the Public Interest

As noted above, a trustee representative has a duty to act in the public interest. The notion of the public and its interest promotes in trustee representatives an inclusive cognitive orientation that is congenial to democratic autonomy. Historically, agencies have fulfilled this duty. In fact, it is this inclusive cognitive orientation that made independent administrative agencies so attractive to early twentieth century progressives. Believing that political decision-making too often followed the preferences of big business “plutocrats,” they suggested the use of independent commissions charged with acting in the “public interest” precisely because they could give voice to the voiceless.¹⁶⁷ For example, the Interstate Commerce Commission assessed the impact of railroad monopoly rates on disempowered Midwest farmers. To help justify the government setting of prices, Congress characterized railroads as public utilities. Moreover, formally enslaved people, having been shut out of electoral politics, found presence in the Reconstruction era Freedman's Bureau.¹⁶⁸

Also consistent with the model, there is no reason to characterize administrators as Platonic guardians of a justiciable public good. As Klein,¹⁶⁹ Emerson,¹⁷⁰ and Rahman and Gilman¹⁷¹ argue, administration opens up a space for democratic evaluation, debate and judgment. Happily, there are many points of entry for citizens, or their representatives, to intervene in agency decisions about the nature of the public and its interest. One such entry point is the politics surrounding the presidency. As remarked earlier, presidential elections often center upon agency policy: how much pollution the Environmental Protection Agency (“EPA”) should eliminate; how much commerce the FTC should regulate, and so on. Hoping to secure electoral support, presidential candidates will mobilize voters by deploying competing policy visions. Once in office, more democratic doorways open as cabinet officials are nominated, vetted by both Congress and the press, and appointed. Meanwhile, appointees themselves serve as successful descriptive or gyroscopic representatives.¹⁷² Each step involved in the election and operation of a presidential administration thus provides opportunities not only for citizens to participate, but also for an administration to prioritize their interests. Presidential influence in agency policymaking is one way whereby elections “mean something,” giving citizens a cognizable role to play in shaping the regulations that bind them by, for example, voting, surveilling, and objecting. Popular demands for civil rights protections and pollution control percolate into executive orders commanding agencies to assess the costs of discrimination and climate change. It also vindicates democratic autonomy when it is suffocated elsewhere—for example, if Congressional action is gridlocked or disfigured by gerrymanders, filibusters, and other antimajoritarian obstacles.

Of course, the President cannot speak with the “voice of the people” any more than Congress can. Presidential influence always flirts with Schmittian decisionism. Moreover, given the complexity of policy, voters are vulnerable to manipulation by an enterprising chief executive who convinces them, for example, that “building a wall” will solve immigration issues. Regardless, extant law tempers the possibility of presidential diktat by forcing agencies to address and incorporate other voices while basing their decisions on research and evidence. In other words, it forces agencies to act like trustees, not like hirelings and toadies. The Administrative Procedure Act (APA)¹⁷³ not only requires agencies to solicit public comment for formal rulemaking, but also to provide reasoned responses and cogent explanations for its actions.¹⁷⁴ They compel agencies to expose their reasoning process to the public as early as possible to facilitate dialogue and objection.¹⁷⁵ Concomitantly, it forbids ex parte communications between the President and the agency during formal rulemaking,¹⁷⁶ and the justifications provided by agencies must be public-regarding, based on claims that incorporate some understandable vision of the public welfare or public purposes.¹⁷⁷ Meanwhile, Congress retains some¹⁷⁸ power to insulate agency decision-making from executive control. The Pendleton Civil Service Act¹⁷⁹ and the Civil Service Reform Act,¹⁸⁰for example, require that positions within administration should be distributed based on merit, not through political patronage. Congress might improve upon these populist prophylactics by, for example, demanding that agency decision-making bodies be staffed by gyroscopic and descriptive representatives that are likely to keep faith with their statutory mandates. Finally, courts ensure that agencies do not, at presidential behest, interpret their statutory authority beyond reason.

2. Agency Authorization is Trustee Authorization

The trustee model coheres with administrative practice for another reason. Under the model, a trustee representative must be authorized by a party with the right to act in relation to the beneficiary. In current practice, the agency's authority to act has statutory provenance.¹⁸¹ As explained above, congressional statutes command authority because, and to the extent that they, result from processes that vindicate the democratic autonomy enjoyed by the populace. Congress, using the procedures of bicameralism and presentment, entrusts administrative agencies to care for the interests of beneficiaries identified within organizing statutes that establish and define the duties of, for instance, the FTC¹⁸² and the Securities Exchange Commission (SEC).¹⁸³ The “intelligible principle”¹⁸⁴ provided by statutory language, presuming it is so provided, lays out the scope of the agency's authority. This authorization procedure respects the idea that authority derives from by the people by incorporating democratic presence indirectly, via the election of lawmakers, and directly, by virtue of any public debates that shape the statute's content.

Although the statutory authorization of administrative agencies is not an explicit component of the trustee model of representation, the two share many characteristics. First, the original trustee model likewise contemplates the use of a binding text: the Constitution. Elected lawmakers, like trustee administrators, must also point to codified authority. Second, agencies’ implementing statutes are not precise marching orders provided by a principal to an obedient agent, as a transmission belt theory might imply. Rather, they are consistent with the autonomy afforded a trustee representative. Despite their “intelligible principles,” agencies’ organizing laws are often open-ended and vague, resembling more closely a deed of trust than a system of command-and-control. Indeed, many organizing statutes explicitly contemplate that agencies will promulgate binding rules as they undertake their duties. For example, under the Federal Trade Commission Act, the FTC may “prescribe (A) interpretive rules and general statements of policy with respect to unfair or deceptive acts or practices … and (b) rules which define with specificity acts or practices which are unfair or deceptive.”¹⁸⁵ Although a trustee may not act beyond the scope of its authorization, the parameters of its authorization are often underdeterminate or indeterminate. Which practices are “unfair,” and thus subject to FTC rulemaking, is far from obvious. Questions about the scope of an agency's authority thus become questions of statutory interpretation. So, according to the model, how should a statute be interpreted and who should interpret it?

First, an agency need not wait upon Congress for further instruction that may never issue forth. If an agency is a trustee, it is not an agent of Congress. As trustees, agencies are bound instead to act on behalf of and in the interest of the beneficiaries identified in their authorizing statutes. Consequently, questions about what Congress intended, or would have intended, had it considered the specific issue at hand are largely irrelevant from the trustee's point of view. Instead, the agency trustee must interpret and implement an indeterminate statute in a manner that best serves its beneficiary's interest—subject to the minimum constraints set forth above (diligence, good faith, care, and so on). Of course, this conclusion resonates with the deferential Chevron doctrine,¹⁸⁶ whereby courts would yield to agencies’ interpretations of their own authority so long as they are not “arbitrary, capricious, or manifestly contrary to the statute.” These features provide a third reason why statutory authorization comports with the trustee model.

Before moving on, it is useful to note that empowering agencies to interpret their own authority need not be seen as antidemocratic. Recall that democratic congressional authority is not sui generis; it is legitimate only to the extent that it exhibits democratic autonomy. Recall also that other representative processes can likewise express democratic autonomy. As a result, there is no reason to think that only Congress, through bicameralism and presentment, can make democratically legitimate policy—including the policymaking inherent to statutory interpretation. Further, as argued above, administrative agencies are commonly obliged to act on behalf of beneficiaries that are politically disempowered vis-à-vis other citizens. If these agencies interpret their own authorizing statutes, they can therefore ameliorate political inequalities within other decision-making venues. To illustrate, citizens suffering from socio-economic hardships might find democratic representation in the EEOC's interpretation of its own statute, especially when judicial intervention or congressional influence might prove hostile to their interests. Agency interpretation is also helpful when a mechanical application of statutes works against the public interest. Even street-level bureaucrats will better serve the public if they bend a few rules.¹⁸⁷ Nor is there a compelling reason to wait for Congress to elaborate upon the meaning of an ambiguous statute. Agency inaction may be the least democratic action available because it preserves a status quo that many citizens might find untenable.

This approach has, at least, more democratic appeal than delegating to courts the duty to provide determinative interpretations of indeterminate authorizing statutes. When a court, for example, forces an agency to wait for further congressional orders, it swaps the agency's interpretation for its own by holding that what is clear to the agency is not, in fact, clear to the court.¹⁸⁸ Given agencies’ expertise and experience, they are better placed to interpret in a manner that serves the public interest in a way that recognizes the interests of many. Courts, of course, may only recognize the interests of litigants sub judice and can deploy no special expertise in determining their decision's social and political knock-on effects.¹⁸⁹

3. The Autonomy of Agencies is the Autonomy of a Trustee

Administrative agencies, as explained at the outset of this article, invariably enjoy decision-making autonomy. Unlike the transmission belt model, the trustee model of representation can accommodate it without logical gymnastics and a naïve belief in fantastic fictions. Instead, it is part and parcel of successful trustee representation: a trustee independent of both authorizer and beneficiary can represent persons and interests overlooked by other representative institutions.

4. Agency Legal Constraints are Trustee Constraints

As alluded to earlier in this section, the trustee model and administrative law also correspond in the constraints they each impose on officials’ decision-making. First, agencies, like trustees, must be authorized to act. In each context, reviewing courts can play a role in guarding against actions taken ultra vires. Meanwhile, the scope of agency authorization is often indeterminate, involving political choices best left to political institutions. The intelligible principle and Chevron doctrines¹⁹⁰ capture this notion neatly.

Furthermore, to act in the beneficiary's best interests, and therefore provide representation to that beneficiary, a trustee must assume several derivative duties: care, loyalty, and good faith. Fortunately for the success of the trustee model, these normative benchmarks parallel extant legal rules governing administrative decision-making. Administrative officers, for example, cannot be reckless in their conclusions nor act through blatant self-interest. Under the APA, courts may invalidate actions that are arbitrary, capricious, or an abuse of discretion.¹⁹¹ A CDC official cannot, in good faith, claim how to address contagious disease without any supporting research or investigation. They should base decisions on facts, not fairy tales.¹⁹² The APA's formal rulemaking requirements,¹⁹³ which require a judicially enforced, trial-type evidentiary hearing, enforce a duty of care and diligence.¹⁹⁴ Like private fiduciary duties, judicial review of agency decisions can be more or less deferential. In fact, courts seem to be moving in an invasive direction, demanding “that agencies provide increasingly detailed and persuasive justifications for their discretionary policy decisions.”¹⁹⁵ This movement is uncongenial to the politics-friendly trustee model discussed here.

5. A Tripartite Model for a Tripartite Practice

Finally, the model's separation of administrative trustee from both authorizer and beneficiary mirrors common understandings of how administration ought to operate.¹⁹⁶ Namely, administrators are not employees of their authorizers. They are servants of the public. The U.S. Federal Reserve System does not take marching orders from elected representatives who hope that a little inflation might improve their chances at the voting booth. Instead, an independent board must attempt to “maintain long run growth of the monetary and credit aggregates commensurate with the economy's long run potential to increase production, so as to promote effectively the goals of maximum employment, stable prices, and moderate long-term interest rates.”¹⁹⁷ The CDC's job is not care for the health of voters living in 1946, when it was established, but care for the populace as it exists today and will exist tomorrow.

B. The Benefits of the Trustee Model

Thus, the trustee model of political representation reasonably describes and therefore lends democratic credentials to administrative agencies. Citizens have “presence” within agencies not only because agencies respect their duty to act in citizen beneficiaries’ interests, but because extant law and culture encourage them to do so. Further, agencies’ best interests determinations are appropriately politicized, permeable to the input of citizens and their representatives. Moreover, citizens enjoy presence in the lawmaking process that results in agencies’ statutory authorization. One upshot of this conclusion is that it can explain why some agency actions that appear to be undemocratic may not be undemocratic at all.

Agency officials occasionally find themselves confronting elected lawmakers hostile to their mandates. The EPA, authorized to protect the public from pollution, might face representatives who would rather scrap its regulations because of the onerous burden they place upon businesses and landowners. Industry-aligned presidential political appointees might mobilize to erase EPA policy, and congressional hearings might commence with the aim to embarrass and delegitimize the EPA before the public. Driven by a democratic instinct, one might conclude that the EPA should capitulate. If this is the direction that popular winds blow, who is an unelected agency official to object? The trustee model explains why it should not—and why its refusal is democratic. The EPA, as a trustee representative, is no presidential mouthpiece. Nor is it the appointed agent of any particular congressional representative—or their wealthy constituents. The EPA is instead a trustee representative duty bound to serve the public interest identified in its statutory mandate. One reason that this kind of representation is so appealing is precisely because it stands firm in the face of both changeable popular opinion and the pressure applied by a highly resourced, self-interested minority. It provides a bulwark against populist inclinations. It provides representation for beneficiaries whose voices are not adequately represented by congressional and presidential politics. Thus, to radically undermine EPA policy (or burn it to the ground), lawmakers have to pass new legislation.

Another benefit of the model is that it suggests agencies are at least sometimes permitted to make policy that lacks clear statutory provenance. To illustrate, the Department of Education might implement a universal student loan forgiveness plan under a statute meant to address national economic emergencies. It does so knowing that a powerful industry lobby, using its plentiful cash reserves, routinely convinces lawmakers to block any and all legislation related to forgiving the debt of ordinary consumers. Rather than conclude that the Department's plan is egregious executive overreach, we might also consider that the agency attended to interests of those illegitimately excluded from a corrupt legislative process. As Emerson notes, proponents of New Deal agencies promoted them because they could protect people who, as a result of the complexity and inequality intrinsic in the modern economy, could not initiate action to protect themselves.¹⁹⁸ “[G]iven the constant inequality generating pressures of the capitalist economy,” notes political theorist Kevin Olson, administration may be “required to maintain the conditions under which the sovereignty of citizens can be fairly and accurately translated into law.”¹⁹⁹

Finally, administrative agencies can help to maintain constitutional representative democracy itself. Consider a scenario where a single party captures the executive and the legislature. That party then seeks to change election rules in its favor, securing power for years to come. An agency committed to civil rights might step in to protect the franchise, even if it seems to do so against democratic currents and even if such questions are deemed nonjusticiable “political” questions.²⁰⁰ The trustee model can tell us why agency intervention is nevertheless democratic: it protects democratic autonomy, securing representation for those excluded from the political process. Consider also a U.S. Attorney General, a political appointee, countermanding the orders of a president to weaponize the Department of Justice in a bid to overturn the results of an election. Despite the President's appointment power, the U.S. Attorney General is not his personal lawyer. The Attorney General, rather, is authorized by the Judiciary Act of 1789 to represent “the people.” The U.S. Attorney General, if abiding by his duty to serve the people and its interest, can refuse and still deserve to be called democratic.

VI. Conclusion

Much of modern administration, given the democratic credentials lent it by the trustee model of representation, is likely legitimate. It facilitates democratic autonomy by giving citizens presence in decisions that, by their nature, elide direct participation. It is, of course, unlikely that courts hostile to administrative policymaking will take seriously the public trust held by administrators. But for political and legal theorists, thinking of agencies as trustee representatives invites us to consider that agencies, far from being embarrassing leviathans requiring restraint, might be part and parcel of a system of collective, democratic power.