Since procedural decisions should be made to serve the substantive task, it follows that expertness in matters of substance [is] relevant to the exercise of procedural discretion.
— Louis L. Jaffe, Judicial Control of Administrative Action 567 (1965).
Using modern due process analysis, the Agency, in the second step of its Chevron analysis, carefully weighed the risks and benefits of informal hearing procedures . . . determining that these procedures would not violate the Due Process Clause of the Constitution.
— Amendments to Streamline the National Pollutant Discharge Elimination System Program Regulations: Round Two, 65 Fed. Reg. 30,886, 30,896 (May 15, 2000) (codified as amended in scattered parts of 40 C.F.R.), appeal dismissed, Dominion Energy Brayton Point, LLC v. Johnson, 443 F.3d 12 (1st Cir. 2006).
In the textbooks, procedural due process is a strictly judicial enterprise. As the story runs, the Court in Mathews v. Eldridge settled on a balancing test for determining what process is due, while in Cleveland Board of Education v. Loudermill the Court finally decided that although the political branches may determine substantive entitlements, it is for courts to decide independently what process the Constitution requires. The notion that procedural due process might be committed primarily to the discretion of the agencies themselves is almost entirely absent from the literature.
The facts on the ground are very different. Thanks to converging strands of caselaw after Mathews — partly involving due process, partly involving post–Chevron U.S.A. Inc. v. Natural Resource Defense Council, Inc. cases on agency interpretation of procedural provisions, and partly involving the long shadow of Vermont Yankee Nuclear Power Corp. v. Natural Resource Defense Council, Inc. and its latest incarnation, Perez v. Mortgage Bankers Ass’n — agencies themselves are now the primary frontline expositors and appliers of the Mathews test. Administrative constitutionalism has already come to the Due Process Clause.
In the second epigraph, the Environmental Protection Agency, referring to itself in the third person, solemnly determined that it had correctly applied the Mathews factors to decide how much procedure to afford in a class of permit decisions. (The epigraph also shows how closely the due process issues are entangled with questions of Chevron deference, when agencies are interpreting procedural provisions in organic statutes. I will disentangle all that later.) Similar examples are legion, arising from all sectors of the administrative state. The Nuclear Regulatory Commission applied Mathews to formulate rules of procedure for the licensing of plant operators. The Bureau of Alcohol, Tobacco, and Firearms applied Mathews to decide whether it should use formal hearings in the denial or revocation of “certificates of label approval” for alcohol. The Treasury Department applied Mathews to determine how much procedure to afford when deciding whether to revoke certificates of surety upon complaint from agencies. The Centers for Disease Control and Prevention even applied Mathews in a proposed rule to determine procedures for quarantine of travelers suspected of carrying a communicable disease, although the rule was never made final. There are examples from the Department of Labor, the immigration agencies, the National Labor Relations Board, the Federal Election Commission, the Department of Housing and Urban Development, and on and on — anywhere and everywhere. Descriptively, agencies are the first to apply the marginalist cost-benefit approach to due process.
The courts, for their part, have no settled approach to such decisions, despite the confident outlines of procedural due process law found in the hornbooks. Some agency applications of Mathews are apparently never reviewed in court at all, leaving the agency as both the first and last body to apply the test. When cases do reach court, the law in action is inconsistent. Nominally speaking, the conventional wisdom is that courts are supposed to apply Mathews independently, and some cases do just that. Others emphasize language in Mathews that requires deference to legislative or administrative judgments about what process is due in a given domain — a strand whose relationship to the standard framework of procedural due process is undertheorized, to say the least. Another line of cases slices the same pie differently, saying that where deference is appropriate, Mathews does not apply in the first place — thereby preserving the nominal independence of the test, but only by contracting its domain. Finally, there are many cases that apply Mathews without any express or acknowledged deference, but in a fashion that can only be described as cursory, implicitly crediting agency assertions unless they are patently unreasonable. Overall, whatever form deference may take, the case-law reviews agency determinations about due process with a light hand.
That judicial posture of deference is exactly right, or so I will argue. Rather than decide for themselves “what process is due,” courts should ask only whether the agency offered a rational justification for providing whatever process it chose to provide. Although courts should continue to apply the reigning Mathews calculus, according to which the process that is due is a function of the (marginal) risks and costs of error of the given procedure and the (marginal) gains of additional procedural safeguards, courts should not independently assess what the Mathews calculus requires in the circumstances. Rather they should defer to reasonable agency decisions about the design of procedural arrangements, reviewing the agency’s choices for arbitrariness, but not correctness, just as courts conduct arbitrariness review in ordinary administrative law. Although the Mathews calculus will supply the governing legal norm, courts will relegate themselves to the institutional margins, reviewing agencies’ execution of the Mathews calculus rather than performing it themselves. This approach extends to all procedural due process contexts, a view that Justice Kennedy has recently advocated in the setting of immigration: so long as the agency offers a “facially legitimate and bona fide reason” for its procedural choices, “‘courts will neither look behind the exercise of that discretion, nor test it by balancing its justification against’ the constitutional interests of citizens.”
Viewed from one angle, the theory is radical; from another, it attempts to make coherent sense of a number of converging strands of recent caselaw, and is in that sense conventional. The theory is radical insofar as it implies a greatly reduced role for courts in overseeing agency procedural choices under the rubric of procedural due process. In a Dworkinian spirit, however, the theory attempts to combine justification, the best account of the principles underlying the precedents, with fit, a coherentist account of the law’s path in recent decades. Dworkinism tends to be associated with the expansion of Law’s Empire, but that is a strictly contingent association, stemming in part from Dworkin’s own proclivities. Fit-and-justification may just as well counsel abnegation of authority by law in favor of administrative bodies; I will suggest that it does exactly that in this setting.
On the dimension of fit, surrounding developments in the law since Mathews was decided in 1976 support a reduced due process role for courts. Those developments include: (1) the “very basic tenet of administrative law that agencies should be free to fashion their own rules of procedure,” announced by the Supreme Court in Vermont Yankee, two years after Mathews, and recently reaffirmed in Perez v. Mortgage Bankers; (2) the watershed of Chevron, decided in 1984, and its key premises that, on grounds of both expertise and accountability, agencies are better positioned than courts to interpret governing statutes; (3) the growing body of caselaw that affords agencies Chevron deference even on procedural provisions in organic statutes — a body of caselaw that has witnessed agencies taking charge of the Mathews calculus, and that squarely rejects due process counterarguments; (4) the Court’s recent emphatic pronouncement that Chevron may actually grant agencies the power to determine the scope of their own jurisdiction; and last, but certainly not least, (5) the number of post-Mathews due process cases that explicitly or implicitly expand upon Mathews’s own cryptic suggestion that “[i]n assessing what process is due . . . substantial weight must be given to the good-faith judgments of the individuals charged by Congress” with administering the relevant statutes. Although in principle courts do not defer to agencies on the application of the Mathews calculus, this body of caselaw shows that sometimes deference of that sort does occur — and I believe for good reason.
On the dimension of justification, I suggest that these converging developments after Mathews rest on a master principle, the one identified by Louis Jaffe in the first epigraph: for many of the same reasons that agencies are better positioned than courts to interpret the procedural provisions contained in their organic statutes, agencies are also better positioned than courts to assess the marginal costs and benefits of additional increments of procedure for program beneficiaries and regulated actors, as Mathews requires. The traditional lawyer’s instinct is to say that procedural questions are for courts, especially when the relevant procedure is in some sense constitutionally mandated. But these instinctive commitments no longer reflect the governing premises on which the law operates. Rather the law now takes into account the interdependence of procedure and substance, and understands that agency choice of procedures is an exercise in system design, which must allocate risks of error and determine the marginal benefits and costs of decisionmaking in light of administrative goals. Generalist judges, who observe the system only episodically, should play a subsidiary reviewing role, asking only whether the agency’s procedural choices are rationally defensible.
* John H. Watson Professor of Law, Harvard Law School. Thanks to Adam Cox, Paul Daly, and Cass Sunstein for helpful comments, and to Evelyn Blacklock for outstanding research assistance.