A Modest Proposal for Reforming the Administrative State

The problem of administrative government in the United States is well known to the readers of Law and Liberty. Congress delegates broad legislative power to agencies, sacrificing something of our republican character. An agency thus exercises not only executive power as part of that department of government, but also legislative power delegated to it by the Congress. Independently, agencies often exercise judicial power as well, adjudicating not only public rights cases but also cases involving common-law private rights traditionally reserved for Article III courts. We have sacrificed not only some part of our republicanism, then, but also a much greater part of constitutionalism itself: the distribution of the powers of government in a manner calculated to prevent a centralization of powers inimical to liberty under any form of government.

What is less known, or at least explicitly addressed far less often, is that our constitutional doctrine today pretends that neither the delegation of power, nor the violation of separation of powers, is occurring.

The Supreme Court, under the so-called “non-delegation” doctrine, pretends that Congress is not delegating legislative power: When agencies make rules, they are supposedly just “implementing” the congressional will (no matter how broadly that will may have been expressed). It may be quasi-legislative, but it is still at bottom executive power. This is so because under the Constitution, only Congress can exercise legislative power, and the administration can only execute the law. Hence the term “non”-delegation. The problem, however, is that Congress routinely delegates broad power to agencies to make law but the Supreme Court has only twice in our history struck down these delegations of power as legislative. Notwithstanding two dramatic departures from our original constitutional order, our law today thus pretends that neither departure has occurred.

Constitutional conservatives have offered only variations on one theme: restore a proper non-delegation doctrine, thereby reserving to the Congress the legislative power and ensuring that the administration exercises only executive, and not also legislative (or judicial), power. What this solution amounts to is a call to do away with the entire administrative state. As much as we may wish for this to happen in a first-best world, we live in no such world. Politics should always be, and can only be, aimed at the best practical regime.

Let us propose something different. To accept the reality of the administrative state, we must accept the reality of legislative delegation. Does it follow that we must also accept the serious blow modern administration deals to the separation of powers? I do not believe it does. Indeed, by accepting delegation, I contend that we not only will recover a large measure of separation of powers, but we can also mitigate the harms to republicanism from delegation itself.

Let us imagine what would happen were we to accept as a matter of doctrine what is already true as a matter of reality. Let us accept that Congress delegates power to agencies. What follows from that acceptance? By accepting delegation, we can recognize that the administration exercises not only executive, but also legislative and judicial, power. That recognition then allows us to reframe the separation of powers question from who exercises power in the first instance—the constitutional branches of government, or the agencies?—to who controls the exercise of that power by the agencies.

If we accept delegation and thus that the administration exercises all three functions of government, each constitutional branch of government can control the function of administration corresponding to its own constitutional function. Congress can control the legislative functions of administration (rulemakings, for example); the President can control the executive functions (enforcement activities); and the courts can control the judicial functions (a subset of the adjudications that agencies conduct). Modern doctrine precludes this approach because, to repeat, it pretends that the administration is only executing the law, and so it is impossible to differentiate the functions in this way.

The implications of this approach are numerous. To give only the most prominent, under this model of administration, which I call “constitutional administration,” a properly conceived legislative veto is constitutional. That is to say, a legislative veto of legislative acts—of administrative rulemakings, for example—is constitutional. The administrative state can operate almost entirely unimpeded—its rules by default will still become law—but if Congress wanted to veto a specific rulemaking, Congress would have the power to do so.

This essay is intended to summarize, in an introductory way, the implications of accepting the delegation of legislative power. I will sketch the implications for legislative power, and briefly touch upon executive and judicial power. My proposal for accepting delegation—to be more fully developed in future writings—will put forward a simple, constitutionally significant, and politically practicable reform in each area. There are no obvious political hurdles to enacting them. Although they would work a profound constitutional reform of the administrative state, none would require a tectonic shift in administrative practice.

The Legislative Veto

Consider why, under modern doctrine, Immigration and Naturalization Service v. Chadha (1983) was rightly decided. In Chadha, the U.S. House of Representatives vetoed the decision of the Attorney General of the United States to withhold the removal of six illegal aliens. The veto was permitted by the statute enacted by Congress and the President granting the Attorney General the discretion to withhold deportation; that discretion was granted with the understanding that Congress would be able to supervise—and sometimes countermand—that discretion. The Supreme Court held that because the veto was a legislative act, it could only have the force of law if enacted by both houses of Congress and signed by the President. A legislative veto was thus pronounced unconstitutional.

That decision must be right under modern doctrine because if Congress does not like a particular implementation (or execution) of the law, it has no choice but to change the law—to clarify it, amend it, or perhaps repeal it. That requires the approval of both Houses of Congress and the President’s signature. Hence the Congressional Review Act of 1996 requires the coordination of all three. But what if our doctrine recognized that Congress is, in fact, delegating legislative power to the administration in at least some circumstances? Take what is at least superficially an easy case—legislative rulemaking.

Once we recognize that the administration is not exercising executive power, but rather legislative power delegated to it by the Congress, a legislative veto of the rulemaking would be constitutional because Congress would not have to change an existing law. If the rulemaking has not yet become law, Congress can step in and terminate the process. It would not need the assent of the President. To be sure, this process would not work for genuinely executive functions; the result in Chadha may still be correct because the decision to withhold removal is assumed to be an executive power. But Congress could step in and terminate any legislative act, such as a regulatory rulemaking, before it is consummated.

Put a different way, if Congress could delegate its legislative power, it could delegate it with conditions. It could choose, for example, to delegate only the power to propose laws. Congress would then have to enact every administrative rule before it became law. That is essentially the Regulations from the Executive in Need of Scrutiny Act of 2015 without a monetary floor. (Versions of the REINS Act have been introduced in Congress going back many years.)

But that does not do much work. After all, Congress wants to delegate power to agencies so it can pass the buck on tough policy questions. That is why the REINS Act is politically impracticable. Congress could do something else, however: It could delegate its power to enact law under the condition that Congress must have, say, seven months to consider each new rule. Only if Congress did not act within those seven months would the rule become law. But if Congress does take some action, such as enact a veto, then that rule cannot become law.

Notice a few things about this proposal, which generalizes the Rules Enabling Act of 1934 for federal court rulemaking with its seven-month wait period to all agency rulemakings, and adds a legislative veto provision.

First, much of the administrative state would operate as usual. Most of the time, for most rules, Congress and the President would assent by taking no action. Or perhaps Congress would take no action and the President would enthusiastically take ownership of the new rulemakings. Either way, that is what happens routinely in the modern administrative state. This model is consistent with much modern practice. It accepts that Congress can consent by taking no action. It accepts the constitutionality of delegation.

Under this proposal, the agency would not have authority to promulgate a rule without affording Congress an opportunity to object. Congress would not be exceeding its constitutional authority by vetoing the proposed rule; rather, the agency would be exceeding its statutory authority if it exercised power without meeting the conditions of its delegation.

A one-House veto could also be constitutional under this approach, because the Constitution also requires bicameralism. But we need not necessarily advocate a one-House veto. Recall that when Congress and the President delegate their power, they can decide how much of it to reserve. One could imagine a reservation of power in which only the House and Senate together could veto a rulemaking. For this same reason, the proposed statute could also reserve a veto power to the President—but it does not need to do so. It would be up to Congress and the President to decide on the threshold for the exercise of the administrative veto power.

Consider the proposal’s many administrative virtues: Congress delegates rulemaking responsibility to experts. In the Rules Enabling Act, it delegates to the Supreme Court and the Judicial Conference of the United States. The experts use their expertise to come up with a proposal. Then whether and how to use that expertise, which is invariably a political question, becomes a matter for Congress and the President to determine. And practically speaking, only truly important rulemakings would register on the radar of the political branches. This model thus fuses technocratic and administrative values with constitutional procedure and political accountability.

Rulemaking as Lawmaking

Some will criticize this approach on the ground that it would become impossible to distinguish among exercises of legislative, executive, and judicial power. It is surely right that at least some exercises of power will be hard to distinguish. What I claim, however, is that there are important classes of administrative functions that can be properly classified as either legislative or executive (or judicial). Giving Congress control over some of those classes of cases will be no small achievement.

Spelling out everything that may be an administrative exercise of legislative power must await further elaboration of this proposal. We are in search of workable rules, and we can adopt a simple one: Any action requiring rulemaking on the part of an agency would be considered legislative action, because such rulemakings usually demarcate for the first time permissible and impermissible conduct.

It may be that some rulemakings are truly administrative only; perhaps they are made only for convenience. But just as surely, our workable rule would also omit other kinds of legislative acts. When criteria are so vague as to provide no true guide to behavior—for example, when licenses shall issue if they are “in the public interest”—then an act determining what is or is not permissible even in a particular case may be legislative in nature. If the Attorney General’s immigration-related discretion is extraordinarily broad, he or she may be exercising legislative and not executive power. But we are in search, again, of workable rules, and stopping at rulemakings would be a substantial advance over modern practice. Although a fuller explication must, as I said, await another day, agency rules made through adjudications—such as those regularly created by the National Labor Relations Board—ought also to be subject to the legislative veto power.

Presidential Administration and a Modified Unitary Executive

Conservatives find themselves in a quandary when it comes to executive power. Many believe that a unitary executive is constitutionally required. Thus, the President should be responsible not only for the rulemakings and other activities of executive branch agencies, but also for the activities of independent commissions over which he currently has far less control. On the other hand, they lament the tremendous growth of the chief executive’s power with the advent of the administrative state. The combination of an unconstitutional state of affairs—the toothless non-delegation doctrine—with a constitutional unitary executive would seem a frightening prospect indeed. If Congress is to delegate great authority, would it not be better to divide up that power rather than have it accumulate in one unitary executive?

Under “constitutional administration,” in which we accepted the delegation of legislative power, the President would have unitary authority over all of the administrative state—including independent commissions as well as executive branch agencies—but over that state’s  executive functions only. In other words, Congress, if it wished, could delegate its legislative power to independent bodies whose commissioners would be insulated from presidential removal authority. Indeed, perhaps Congress could appoint and remove these commissioners without the involvement of the President at all. But what Congress would not be able to do is insulate the enforcement functions of any administrative agency, whether independent or in the executive branch.

The solution is easy, and would not run afoul of any modern doctrine: Most independent commissions already have directors of enforcement. All that is required is that these positions be made constitutional appointments—Senate-confirmable and removable by the President. A simple statutory enactment of a few short paragraphs, taking the statute creating U.S. Attorneys as a model, would do the trick.

Judicial Review of Adjudications

Judicial review of judicial acts is the one component of the proposal that does not follow from our accepting the delegation of legislative power. That is because Congress does not have any judicial power to delegate. Thus, if agencies are exercising judicial power, it is not because Congress granted them that power. It is because the courts have abdicated their own judicial responsibilities.

Article III vests the “judicial power” of the United States in federal courts whose judges enjoy constitutional protections against political influence, including salary protections and lifetime tenure during good behavior. Thus, individuals cannot be deprived of life or liberty—and historically could not be deprived of the fruits of their labor—without an adjudication in an Article III court presided over by judges enjoying these protections. And yet, today, administrative agencies often adjudicate facts and law relevant to such rights without de novo review by Article III courts.

There is a rather simple solution to this degradation of Article III in the administrative sphere. It is well known that in the bankruptcy context, when private rights are at stake, bankruptcy judges can only make reports and recommendations (of both law and fact) and Article III district judges must review those reports and recommendations de novo. The same solution could be applied to all determinations of private rights in administrative adjudications. Congress could borrow from the existing Bankruptcy Code and the Federal Magistrates Act of 1968, which provide the textual precedent for requiring reports and recommendations subject to de novo Article III review.

Judicial Review of Other Powers: A Brief Chevron Detour

When considering the judicial power under a theory that accepts delegation, we must separate judicial review of legislative, executive, and judicial functions; after all, we now recognize that the administration exercises all three kinds of power. I leave the reader with some final thoughts on judicial review of legislative functions and the implications for Chevron deference.

When we understand that agencies are not merely interpreting governing statutes but are rather making law themselves, that might give us an entirely different intuition as to which interpretive approach to adopt. If agencies are making law, ought not courts engage in interpretation as if they were interpreting a congressional enactment? It has been observed before that “the relationship of the Constitution to Congress parallels the relationship of governing statutes to agencies.” Constitutional administration would make this parallel all the stronger because it recognizes that agencies are in fact exercising the same power as Congress.

Courts ought to be limited to making sure the agencies are acting within their delegated discretion, just as the courts make sure Congress is acting within its own delegated discretion—within the legislative power granted to it by the Constitution—when reviewing acts of Congress. So long as it is within the bounds of the delegation, the agency has discretion to make policy choices—just as Congress freely makes policy choices within its legislative power. Indeed, Chevron deference has a specific parallel in the interpretation of congressional statutes under questions of constitutional delegation: the presumption of constitutionality. In cases where the constitutional question is not clear, the courts “defer” to Congress’ interpretation of the Constitution.

But the presumption of constitutionality is not so easily deployed. Judges do not give up as easily on constitutional interpretation as they seem to give up on statutory interpretation in the context of administrative delegations. No justice of the Supreme Court, as far as I can discover, has ever argued that merely leaving the constitutional text ambiguous on its face is sufficient to confer interpretive authority upon Congress. Rather, the courts must look at text, context, intent, purpose, historical background, general background principles of law, conventions, and so on. Usually, only when “traditional tools of statutory construction” run out, do courts deploy the presumption of constitutionality. Because agencies are making law and not merely interpreting existing law, the same tools ought to be used in deciding whether a congressional statute genuinely sought to grant agencies the lawmaking power in question. The exact same constitutional values are served by adopting the same method of interpretation in both contexts.


I confess that, when reading modern administrative law cases, I often feel torn. On the one hand, the formalism of modern doctrine tends to ensure rule-of-law values and certainty, and is more faithful to the constitutional text on the surface than is functionalism. On the other hand, there can be no doubt that had the Framers conceived of the modern administrative state, they would be aghast at the power of the executive branch. Thus, even those who have recognized that certain functionalist tools might be unconstitutional as originally understood argue that such tools ought to be permitted to balance the accretion of power in the executive. There is something incredibly compelling about Justice Byron White’s functionalist vision for administrative law in his Chadha dissent. It is not quite what the constitutional text says, but it looks a lot more like what the text may have been intended to create than the majority’s formalist vision.

Constitutional administration advances this debate between formalists and functionalists in the context of separation of powers and the administrative state. Formalism only requires an accretion of tremendous power in the executive branch if we accept the fiction that Congress does not delegate legislative power and thus that its agents in the executive are always exercising executive power. Once we recognize and permit delegation, we can apply formalist reasoning to achieve what were originally functionalist results in many separation-of-powers cases. A formalist, for example, would permit a legislative veto of agency rulemaking, thereby reserving significantly more power to Congress than it currently enjoys.

If modern doctrine reflected these insights, many constitutional conservatives would breathe more easily when contemplating the administrative state. Functionalists who seek more accommodation between Congress and the President than current doctrine allows also ought to breathe more easily. The administrative state, if it looks unconstitutional at all, would suddenly look a lot less unconstitutional. The activities and powers of each branch of government would be closer to their intended original operation. Congress would have more power over legislative matters, the President over executive matters, and the courts over judicial matters.

Progress can be made. We need only accept a de facto precedent—the actual delegation of legislative power to agencies—that we have refused to acknowledge for several decades. We need only reorient our thinking on delegation.