The very idea of separate administrative courts—and even the term “administrative law”—came to be regarded in England (and to a lesser extent in the United States) as the denial of the rule of law. Thus, by his attempt to vindicate the rule of law as he saw it, [A.V.] Dicey in effect blocked the development which would have offered the best chance of preserving it. He could not stop the growth in the Anglo-Saxon world of an administrative apparatus similar to that which existed on the Continent. But he did contribute much to prevent or delay the growth of institutions which could subject the new bureaucratic machinery to effective control.
—F.A. Hayek, The Constitution of Liberty (1960)
The administrative law debate has for decades been preoccupied (especially on the conservative side) with Chevron’s metaphysics and judicial review of rulemaking proceedings. Lately, however, agency adjudication has reemerged as a subject of judicial and scholarly attention, and even of public concern. Our “hidden judiciary”—the 12,000-plus administrative law judges and administrative judges who handle the overwhelming portion of administrative cases—has proven sufficiently arbitrary and capricious of late to grab headlines.
What has brought us to this pass are actions like the imposition of civil fines by bureaucratic edict; sudden changes of agency policy, accomplished by means of adjudication and without fair warning to the parties; the opportunistic shifting of enforcement proceedings from Article III courts to agency tribunals; and the administrative invalidation of invention patents, en masse and including patents that cannot be canceled in any court of the United States. This pattern of conduct is of a piece with bureaucratic practices that deprive citizen of any kind of adjudication—“non-final” enforcement actions that effectively thwart private citizens’ business operations or use of their land, or the withholding of permits or licenses unless and until applicants meet extortionate demands.
Such irregularities and abuses aren’t lapses. They are the natural and, for the most part, fully intended consequences of the “appellate review” model that has been the bedrock of American administrative law for well over a century. Under that model, agency adjudication comes first, followed by highly deferential, on-the-record judicial review. It is encapsulated in the Supreme Court’s foundational decision in Crowell v. Benson (1932), and it is embodied in the Administrative Procedure Act (APA).
A Time to Be Bold(er)
The appellate review model, I believe, is beyond incremental reform or repair. I propose to decimate and to replace it, over a wide range of governmental action, with a full-scale system of independent administrative courts. Those courts should be endowed with the incentives and the institutional capacity to provide a meaningful check on the regulatory state, and with the capacity to develop, over time, doctrines of administrative law that actually merit that honorific.
Legal systems like that exist in many and in fact most civilized countries (common law, and civil law). The prototype I envision is what F.A. Hayek had in mind: Germany’s administrative courts, which Hayek proffered as a model in a chapter of The Constitution of Liberty tellingly entitled, “Liberalism and Administration: The Rechtsstaat.” (The epigraph to this essay concludes that chapter.) Those courts are strictly confined to administrative disputes. But they operate outside of administrative agencies, and they exercise the full powers of independent courts.
Congress contemplated such a regime in the 1930s, but the proposal failed to gain traction. Instead, the debates of that period gave us the APA’s appellate review model and its rule-of-law-ish pretensions. It may be too late in the day to revisit that decision. But if there were ever a good time, the current ferment in the AdLaw field provides it.
Administrative Justice on Trial
Virtually all administrative adjudication in the United States occurs in agencies, not courts. Agency adjudicators preside over challenges to agency decisions; enforce orders and impose fines; issue or withhold permits or licenses; and decide disputes among private parties. Some 2,000 of them are administrative law judges (ALJ’s), the great majority (85 percent) of whom serve in the Social Security Administration. Other administrative judges (AJ’s) bear various titles (that of immigration judge, for example). There are more than five times as many AJ’s as there are ALJ’s.
Both groups of officials are attached to a particular agency and are appointed by that agency. Most enjoy a measure of functional independence, such as salary protections and protections against removal. However, the decisions of ALJ’s and AJ’s are virtually always subject to review and reversal by agency heads. This prospect of executive reversal distinguishes administrative judges from actual courts and judges.
The latter-day contentions over this century-old system have been kindled by the abuses and irregularities noted earlier. Underneath that surface, however, lurk graver doubts—some arising from seemingly technical but nonetheless vital separation-of-powers concerns, others having to do with elementary notions of due process and constitutional government.
In Lucia v. S.E.C. (2018), the U.S. Supreme Court held that the Securities and Exchange Commission’s ALJ’s are “officers of the United States.” They must therefore be appointed in conformity with the Constitution’s Article II, either by the President or by the agency heads (as the commission’s ALJ’s were not). That may not look like a terribly big deal. In earlier decisions, though, the Court has held that appointed officers must also be removable at will, by the President or an agency head. In that light, the ALJ’s and AJ’s functional independence looks increasingly tenuous; and while the Court sidestepped the removal question in Lucia, most legal scholars believe that this particular shoe will soon drop. Justice Breyer’s concurring opinion in Lucia pointedly warned against a separation-of-powers jurisprudence that risks “unraveling, step-by-step, the foundations of the Federal Government’s administrative adjudication system as it has existed for decades.”
A more fundamental attack has come from scholars who argue that “administrative justice” is an oxymoron and unconstitutional over a wide range. The principal rule-of-law problem is usually conceptualized as a “due process” concern: We cannot have the same set of officials sit as prosecutors and judges.
In some settings, that intuition is still universally shared. For an obvious example, we would be horrified if the U.S. Department of Justice could adjudicate criminal convictions inside the agency, subject to direction and reversal by the Attorney General and with near-conclusive effect (judicial deference to the executive branch on fact and law) in federal court. Nor would we change our minds on that score if the department’s internal arrangements conformed to all the niceties of “process.” Ultimately, due process is not just about the “how” but also, and mostly, about the “who”: The agent who prosecutes you cannot convict you. That is why we have courts that are independent from the executive; that is how Article III hangs together with the Due Process Clause.
Why do these basic intuitions give way to a ready acceptance of administrative adjudication pretty much across the rest of the board? Well, short of criminal convictions or constitutional violations, we want agencies to make policy by means of adjudication, the better and more flexibly to employ their expertise. To that end, the adjudicators must be inside the agencies and reversible by them. A court undertaking review must grant deference, lest that court rather than the agency end up as policymaker-in-chief.
It might seem unfair to subject regulated parties to adjudication by decision-makers who are appointed by the regulating agency, imbued with its mission and institutional perspective, and reversible by that agency. However, the APA—the administrative law profession assures us—makes ALJ’s and AJ’s functionally independent, and it contains elaborate formal adjudication procedures. Organic agency statutes often provide for comparable protections.
Those assurances, though, have begun to wear thin. Widespread discontent with the operation of the adjudicatory system has generated recommendations for more formalized agency procedures; more functional independence for ALJ’s and AJ’s (a proposal that runs headlong into Lucia); “fair notice” requirements before an agency changes course by means of adjudication; and more probing judicial review, at least on questions of law.
None of these proposals, however, goes to the root of the problem: the agency’s dual role as policymaker and adjudicator. Agencies might provide procedures approaching those of ordinary trial courts, but at the end of the day, the adjudicator is meant to bring a certain bias to the proceeding and to exercise discretion and expertise in line with the agency’s mission and agenda. Thus, the most one can expect from administrative adjudication is an appearance of impartiality.
Similarly, within the framework of the appellate review model, the case for judicial deference to the executive branch is overwhelming. When judges review policy decisions, they have no comparative advantage over agency administrators. The obvious question—why make the same decision twice?—naturally draws courts back towards deference.
Finally, the reform programs—more judicialized agency procedures, more intense judicial review—pull in opposite directions. The more “process” to which regulated parties have been subject within an agency, the less inclined an Article III court will be to second-guess a decision that has already been made.
The only escape from the conundrum—the only way to restore lawful and regular administration—is a judicial system that subjects government action to comprehensive, genuinely legal and judicial control. The following section contains my proposed solution: a new system of administrative courts.
The prototype of a general administrative judiciary, and the rule-of-law alternative to the appellate review model, is Germany’s system of Verwaltungsgerichte—the Verwaltungsgerichtsbarkeit, if you want to be Teutonic about it. Hayek identified it as the institutional core of the Rechtsstaat; he was and continues to be right about that. Obviously, not every element of that system can be transported across the pond. However, it would be perfectly possible to mimic the institutional logic and to replicate the basic design features.
The rock-bottom premise of the Rechtsstaat is that every administrative act is lawful until it is set aside by an independent court of competent jurisdiction (a Verwaltungsgericht or, in tax disputes, a Finanzgericht). Conversely, however, every such act must be challengeable by citizens who are affected in their rights. That cannot be done by a handful of general-purpose judges. It demands an entire judicial machinery, devoted to no other purpose than to keep administration within legal bounds and off the citizens’ backs. Germany built that machinery, and it is working as intended. Granted, Germans do complain about red tape and excessive judicialization; at the same time, our concerns over rampant lawlessness and abuse are nearly incomprehensible to Germans, jurists and deplorables alike.
Four design features are essential attributes of an effective Administrative Judiciary:
- Administrative courts must be independent of and from the executive. They must not be removable by the executive, and their decisions must not be revisable by an administrative body.
- The administrative courts hear only administrative cases, but substantially all administrative cases.
- The administrative courts must be open to every citizen who claims to have been violated in his private rights by an administrative act; but they are and must remain closed to anyone else. (For example, a person’s claim that an agency has failed to regulate some other person isn’t cognizable under German law, for lack of what we call “standing.”)
- While the executive branch agencies may use hearings and other adjudicatory devices in the interest of fairness or for purposes of obtaining information, any decision eventually reached by an agency is not owed judicial deference (a term unknown in German administrative law). Every judicial determination is and must be de novo, on all questions of law or fact.
Germany’s administrative judiciary exhibits these four core features, and so should America’s. I’ll describe the rough contours of such a judiciary: its establishment and appointment, its proceedings, and its jurisdiction.
Establishment and Appointment
Congress should create a system of administrative courts that are independent of administrative agencies—say, 100 courts with 1,000 or so judges, spread across the country. A Federal Administrative Court (a Bundesverwaltungsgericht) would sit at the system’s apex and there would perhaps be appellate courts in between. As principal officers of the United States, the Federal Administrative Court’s justices would be appointed by the President, subject to Senate advice and consent. Lower-level judges could be appointed in the same manner or, alternatively, by the justices of the Federal Administrative Court. All members of the Administrative Judiciary would serve fixed terms (say, 10 to 15 years), with removal only allowable “for cause” and at salaries not to be diminished during their tenure.
Congress should establish the Administrative Judiciary pursuant to its powers under Article I of the Constitution. It has done so many times for more specialized purposes, as with territorial courts or military tribunals. The U.S. Tax Court, which by all accounts is working very well, is a close cousin of, and in some respects an institutional model for, an administrative Article I judiciary.
Separation-of-powers fundamentalists may insist that judges appointed pursuant to Article I are executive officers (they can’t be members of the legislative or judicial branches) and that, therefore, the justices are “principal” officers to be removable at will by the President. The argument is not implausible, but I doubt that the Supreme Court will go there and declare the U.S. Tax Court unconstitutional. Regardless, the risk is worth taking. Any attempt to create a veritable army of life-term Article III courts and judges would produce heated partisan conflict and insurmountable opposition from the existing Article III judiciary. (Federal judges’ prestige depends on there not being too many of them.) An Article I Administrative Judiciary, renewable as administrations come and go and separate from our Solons, would be far less likely to stir comparable passions.
Wholly apart from political considerations, Article III adjudication would often require jury trials, which is an excessively costly proposition in this context. Most consequentially, an expanded Article III judiciary—even and especially if the new judges were limited to administrative adjudication—would likely revert to the Supreme Court’s AdLaw thinking, so-called. In that event, we would have gained next to nothing. I’ll come back to that point below.
Cases brought in the administrative courts would emphatically not be appellate actions for “review.” Rather, they would be original actions. The standard of review would be de novo. Courts would give zero deference to the administrator, on questions of law or of fact.
The proceedings would be conducted in accordance with the Federal Rules of Evidence and the Federal Rules of Civil Procedure. They would not be limited to the agency record. Private plaintiffs or, for that matter, the administrator could introduce external evidence. The cases would be litigated without a jury. Judges would make all determinations of fact as well as law.
Unlike in an appellate review proceeding, the usual form of relief would not be a remand to the agency that would require further adjudicatory proceedings; it would be an affirmative, non-reviewable order to the agency to give the plaintiff the relief to which he is entitled (excepting claims for monetary relief in the nature of damages). Successful claimants would be entitled to reasonable attorneys’ fees.
The most consequential question regarding an Administrative Judiciary is the scope of its jurisdiction. The four design features above suggest the answer. The point of the German model is not to proceduralize administration or to ensure administrative regularity or strict legality across the board. If it were, you’d want the tribunals inside the bureaucracy. The point of independent courts is to block coercive government interferences with private citizens’ legitimate expectations to go about their ordinary business, within the limits of the law. In short, the machine to be reined in is not the administrative state but the regulatory state.
To that end, the administrative courts’ jurisdiction must extend across the entire range of administrative agencies and disputes. One can and should exempt tax disputes, as well as private law-like claims that sound in contract or tort. But no subject-matter distinctions must be permitted. Agencies may and should be specialized and “expert.” The judiciary must be generalist and rule-oriented. Yield an inch at that front, and the entire enterprise collapses back into “policy” and judicial deference to the executive branch.
It is equally important to limit the administrative courts’ jurisdiction so as to focus them on their principal mission of providing systematic, readily available protection against coercive and unlawful interference in the affairs of private citizens. They should not entertain questions regarding the internal operations of the government (especially its employment decisions) or benefit determinations, such as disability or veterans’ benefits. I am no great fan of the way those systems currently operate; but because they merely distribute public benefits and do not interfere with anyone’s private affairs, their adjudication by administrative tribunals rather than courts poses no constitutional or rule-of-law problem. By the same token, their adjudication by a new system of administrative courts would threaten to transform those courts into small-claims tribunals, on matters far removed from their core purpose.
Two further jurisdictional matters, peculiar to the American context, warrant attention.
First, our administrative law (unlike Germany’s) permits judicial pre-enforcement review of administrative rulemaking proceedings. That form of review is structurally incompatible with an Administrative Judiciary, whose mission is to protect this private party from that unlawful executive imposition. Pre-enforcement rulemaking review is policy-laden, and so it naturally pushes toward judicial deference. It addresses the overall lawfulness and effects of a general rule, not its application in a particular case. It puts the interests of “regulatory beneficiaries” on a par with those of regulated parties, and it draws a vast array of affected parties into the proceedings. In short, it is an interest-group sport. The job of an Administrative Judiciary would be to correct that mode of adjudication, not replicate it. Thus, challenges to administrative rules should remain in Article III courts.
Second, the Supreme Court recognizes a category of “private rights” that cannot be committed to non-Article III tribunals. That category covers criminal proceedings and constitutional claims. Somewhat perplexingly, it also includes certain claims made under state laws among private parties in bankruptcy proceedings. It may also include common law claims between private parties in disputes under a handful of administrative schemes, and perhaps certain trademark (but not patent) claims.
Leaving these matters in the hands of Article III courts poses no impediment to establishing an Administrative Judiciary. (German administrative courts do not entertain criminal or constitutional cases, either.) In fact, the Supreme Court’s private rights jurisprudence strengthens the case for an Administrative Judiciary. The scandal of our appellate review system is that outside the smallish “private rights” category, just about all statutory claims or defenses against the government (as well as many claims and defenses that arise among private parties under administrative schemes) are matters of “public right”—which Congress may commit to administrative, non-Article III tribunals. Even so, Congress need not house those tribunals inside administrative agencies; join prosecutors and adjudicators; or leave regulated parties to the tender mercies of deferential review by Article III courts. Congress can and should establish an Administrative Judiciary.
Rethinking the Foundations
Newly created institutions, including judicial institutions, have a natural incentive to establish credibility and build constituencies, whether among legislators, members of the bar, litigants, or members of the public at large. Courts in particular can build institutional power and prestige in case-by-case decisions over time. An Administrative Judiciary would, as the years passed, generate a credible body of administrative common law.
Very likely, that corpus juris would differ materially from the administrative “law” with which we are familiar. That is not a problem: it is the point of the enterprise. The appellate review model and the legal doctrines on which it rests fail to satisfy minimal rule-of-law requirements. No amount, as I said, of institutional tinkering or doctrinal improvisation can fix the design flaw at the core of the system: the conjunction of adjudicator and enforcer; of adjudication and policymaking. Most assuredly, the courts we have created under the Constitution’s Article III cannot be expected to subject administrators to meaningful, systematic legal control. They lack the institutional capacity and do not want to be bothered with run-of-the mill administrative decisions in the first place. (One central reason for the judiciary’s ready embrace of the appellate review regime is that it clears the federal courts’ dockets.)
In sharp contrast, an independent Administrative Judiciary would have both the capacity and the constituency-building incentives to develop doctrines that are conducive to its mission of serving as a serious check on the regulatory state.
That, to be sure, would be a long-term, all-encompassing enterprise. But let me flag three areas where we might reasonably expect improvement.
Judicial Deference. For contemporary critics of the administrative state, judicial deference—Crowell v. Benson, Chevron, and associated canons—is the great white whale. Harpoon it, which is to say have Article III courts decide questions of law de novo, and behold: the rule of law. That will not work, because it cannot work. As Professor Adrian Vermeule has rightly observed, Crowell’s appellate review framework systematically pushes toward deference. The executive branch agency gets to adjudicate first because it is the “expert,” and courts will let the agency adjudicator’s factual findings stand because otherwise, what’s the point of having an administrative tribunal?
Alas, once administrators find “the facts” with near-preclusive effect, the distinction between facts and law becomes far too thin to prevent a migration of deference, first to “mixed” questions of law and fact, and then to the law proper. In fact, one has to ask what an Article III judge can really contribute to the policymaking exercise by making the same decision again. The answer to that “marginalist” (Vermeule) question is, not a whole lot.
An independent Administrative Judiciary would counteract these baneful tendencies. Unlike agency-appointed ALJ’s, institutionally independent judges would have no incentive to tailor their decisions to an agency’s mission or agenda, for they would not belong to, nor could they be reversed by, that or any other agency. At the same time, these new judges would preside over individual cases, not rulemaking proceedings. Thus would the legal question (“What did they do to you?”) predominate over the policy questions that currently prompt judicial deference to the executive branch in rulemaking proceedings among contending interest groups.
Worried about the administrative judges’ lack of expertise? I’m not. The parties (and if need be, special masters) would supply whatever technical knowledge might be required. In any event, “lack of expertise” is the point. The “expertise” that spooks through our AdLaw canons is a polite word for tunnel vision and agenda-driven decisionmaking. The point of having the same decision made twice (first by an expert and then by an actual court) is to have the initial decision examined from a different perspective by someone who is a generalist and precisely not an expert; someone who looks to rules, not utility. The appellate review system deliberately slights the difference; an independent administrative court system would accentuate it. It is the answer to Professor Vermeule’s marginalist question.
Adjudication versus Rulemaking. Among the elementary doctrines of administrative law is the so-called Chenery II doctrine: Agencies may make policy by rulemaking or by adjudication. The choice is left to the agencies’ well-nigh unreviewable discretion, as their “expertise” or convenience might commend. (This is one of the reasons why ALJ’s and AJ’s are placed inside the agencies, and why their decisions are reviewable by agency heads.) There is no way that Article III courts would challenge this doctrine. (It is hard to think of a plausible theory that would tell us when and why an agency must proceed in one way rather than the other.)
An administrative court system, in contrast, would render the Chenery doctrine a virtual nullity almost by default. Adjudication would be conducted outside of the agencies and without regard to their mission and agenda. Review would be de novo, and the verdicts would be unreviewable by the executive. If agencies wanted to make policy and buy themselves deference, they would have to write rules (provided they had the statutory authority), reviewable by Article III courts in the ordinary fashion. No judicial theory is needed: the choice is the agencies’.
Private versus Public Right. Any system of administrative law that is worth having needs a reasonably coherent, robust set of rights that run against the government. Under German administrative law, these are called “subjective public rights.” (You learn what they are in your first semester of Verwaltungsrecht.) The basic idea is that the rights correspond to, and serve to fend off, coercive government interference with private conduct—the right to hire and fire; to sell your product at a price that willing customers will pay; to make such use of your land as you see fit without harming others; to go about your business without fear of extralegal impositions.
A rather Anglo-Saxon notion, come to think of it. Except it’s enshrined in German law—since 1949, constitutionally enshrined. Meanwhile it is explicitly repudiated under our law. German AdLaw is rights-focused. Our system, in contrast, is rights-neutral. How much (if any) judicial review you get under the APA depends on the form of administrative action (adjudication, rulemaking, formal, informal), not on what’s at stake for us-the-administered.
“Your business is our business,” the APA tells us, “but first let’s have some procedures and now are you happy?” And while the Supreme Court periodically stamps its constitutionalist foot and insists that the adjudication of private rights is the “irreducible core” of the judicial power and thus must not be committed to non-Article III tribunals, any legal entitlement that flows from or is part of some governmental scheme and runs against the executive is “public” not private—and may therefore be disposed of administratively. That includes not just public benefits (say, disability payments) but also licenses, franchises, operating permits, and much else besides. It even includes invention patents, on the authority of a recent case.
The Supreme Court is exceedingly unlikely to revise this doctrine except at the outer margins. (It is much more likely to ramp up the standard of appellate review in the vain hope that doing so will somehow restore constitutional order.) Fine, then. Commit the adjudication of “public” rights, within the scope of the jurisdiction described earlier, to administrative tribunals—just not agency tribunals but actual courts. Over time, an independent Administrative Judiciary is bound to generate a tolerably coherent system of nominally public but substantively common law-like rights. Its institutional prestige would hang on such a development.
We adopted the appellate review model, and rejected the German model, over 100 years ago and then enshrined it in the APA in 1946. The confluence of the Supreme Court’s “dismantling” of ALJ independence, the ongoing, raucous scholarly debate over the foundations of administrative law, and the renewed attention to the defects of administrative adjudication provides a splendid opportunity to confront the appellate review regime with a promising alternative: a system of independent administrative courts.
Advancing that proposal would require far more than a few scholars’ creative thinking. It would require a serious and sustained academic, public, and legislative debate of the kind that preceded the APA. It would require careful thought about how exactly the Administrative Judiciary would work and how to integrate it into the existing legal system. Not only that, the APA and a vast number of organic agency statutes would have to be rewritten.
All that, you say, in this poisonous political environment and with this Congress? Of course, I am skeptical. But I also despair of the prospect of making meaningful progress in the direction of lawful government within the confines of the appellate review model. Sure, we could overrule Chevron. For the reasons mentioned above, any gains on that front are likely to prove illusory. Sure, we could make administrative law judges more judge-y and their procedures more due-process-ish. That, though, is the game of latter-day New Dealers who insist that an internal division of administrative functions is a fine substitute for the actual separation of powers. I don’t buy it.
What we need is a very different legal system to stem systemic official misconduct and protect the rule of law. I’m with F.A. Hayek. We need something like Germany’s administrative law and its judiciary. You can talk me out of the idea. But if not this, my rule-of-law friends, then what?
 For the purposes of this Liberty Forum essay, I’ll have to keep the description very brief. A splendid overview of German administrative law is Florian Becker, “The Development of German Administrative Law,” 24 George Mason Law Review 453 (2017).
 Of course, an administrative judiciary is a form of specialization. However, that division within the judiciary has nothing to do with administrative expertise or judicial deference. It simply recognizes that a case between a private citizen and the state is never quite the same as a lawsuit among private parties: the state can make law, and a private defendant can’t. Our law used to accommodate the difficulty by permitting citizens to sue individual officers as, or as if they were, private parties. But those “officer lawsuits” are long dead and gone. German administrative law rests on the recognition that the private plaintiff against the state confronts not an equal but, well, the state; and it features a raft of doctrines that block the government’s Flucht ins Privatrecht—that is, governmental attempts to circumvent the strictures of public law.
 The excellent Mike Rappaport, who upon information and belief will be one of the respondents to this essay, presented a strikingly similar proposal at a recent conference sponsored by the Antonin Scalia Law School’s Center for the Study of the Administrative State. We really did think of this independently. Great minds, etc.
Michael Greve’s solution sounds pretty good but let’s get a better definition of the problem first.
Could executive branch agencies be turned into helpful adjuncts to federal courts without usurping judicial power? Yes they could.
Michael Greve is right, it’s time to address the serious infirmities that afflict agency adjudication — but my reform proposal is better.
I am willing to gamble that a new Administrative Judiciary would, over time, develop a sensible, rights-centered body of law.