The Yale Law Journal


A Debate Between Peter Strauss and Cass Sunstein

26 Sep 2006

In Beyond Marbury: The Executive’s Power To Say What the Law Is, 115 Yale L.J. 2580 (2006), Professor Cass Sunstein argues that Chevron is the Marbury v. Madison of our age, and that it is now the province of the executive branch to "say what the law is." Professor Peter Strauss responds that Chevron deference must remain "within" Marbury, and that the duty of the courts to set limits on executive claims of authority is as vital now as it has ever been. Check back through the coming week for installments of this debate.

Within Marbury: The Importance of Judicial Limits on the Executive’s Power To Say What the Law Is

Anyone reading the newspapers today must be aware of our President’s repeated insistence that he is the constitutional “decider”—that in many contexts (such as national security) he needs to be the one in charge of determining what the law is. Even in the more mundane context of domestic legislation and regulation, distinguished scholars have argued that a dominant presidential role in determining legal outcomes is appropriate as a matter of policy if not commanded by the Constitution. That these claims remain controversial is suggested by this summer’s considerable flap over the President’s use of signing statements to limit the reach of legislation he was formally approving—a flap that was highlighted by the ABA’s condemnation of the practice and that extended as well to blogs like Balkinization and The Georgetown Law Faculty Blog. Professor Cass Sunstein’s recent essay in The Yale Law Journal, discussing the Supreme Court’s sensible decision in Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., finds in that decision a basis for preferring executive over judicial interpretations, even when the former are made informally and without the benefit of public procedures. While he does not discuss the recent presidential claims of authority to determine legal issues, elements of his analysis appear to endorse them, in ways that in my judgment threaten the very foundations of our culture of legality.

The Virtues of Simplicity
I. Agreements and Concerns

Simplicity has vices as well as virtues. If the law consists of simple rules, it may badly misfire as compared with a more flexible, less rule-bound approach. But in many areas, simple rules are best. When courts are setting out doctrines to govern scope of review of executive action, they usually do well to favor simplicity. Complexity can have unfortunate systemic effects, and those effects cannot easily be justified by the effort to ensure greater accuracy. A clear formula, informing courts and litigants about the proper approach, reduces the risk of interminable debates over threshold issues. Sophisticated multifactor tests might well disserve the legal system, simply because they create undue complexity.

One disagrees with a scholar as respected as Cass Sunstein at risk; and he is right that there is much we agree on. In the brief format that remains (our readers will be glad to know that we now are to observe a word limit), I focus on checks-and-balances considerations that in my judgment he mistakenly puts aside.

Costing Mead
To know whether Professor Strauss’s arguments are right, we need to know exactly what is lost and what is gained by the lower level of judicial deference required by United States v. Mead Corp. Let me explain why more might be lost than gained, even or perhaps especially when the agency is responding to the will of the President.

The Virtue of Checks
In another of the provocative essays in the Symposium featured in this Journal’s current issue, Professor Neal Kumar Kaytal reminds us that, with Chevron in place, Congress is severely handicapped should it disagree with an executive judgment to which courts will defer—any rejoinder must survive not only the internal veto-gates of Congress, but also the veto itself. Thus, congressional checks on executive action are sharply limited. That leaves the courts and—his contribution to the dialogue—the professionalized civil service who within the executive branch can provide an expert counterweight to presidential politics. Eliminating the holding of United States v. Mead Corp. would prejudice both these remaining checks on presidential will.

Chevron articulated a two-step analysis of statutory interpretation: the first (judicial) step is to determine the possible meanings of the statute, and the second (agency) step is to choose among those possibilities. Chevron held that courts must accept agencies’ “reasonable” interpretations of statutes if those interpretations fall within the possible range of a statute’s terms. The opinion’s term of reference, however, was a considered agency judgment made following established public procedures in relation to the agency’s particular mandate, and not presidential instructions issued from the privacy of the White House, in the absence of public procedures, and affecting the whole field of governmental action. In eliding these differences, I suggest, Professor Sunstein’s analysis overextends the executive’s authority to interpret statutes and would lead courts to abandon their fundamental job of defining the scope of agency authority and supervising the exercise of that authority. A powerful and expert executive does need room for action, but to have such an executive we must also have a judiciary willing to define and police the bounds of the executive’s authority.

Entitling his essay “Beyond Marbury,” Professor Sunstein suggests that Chevron is properly understood as a kind of counter-Marbury for the administrative state.” Marbury v. Madison famously asserted that it is “emphatically the province and duty of the judicial department to say what the law is.” Chevron would be a counter-Marbury if it had held that statutory interpretation is, rather, the responsibility of the executive branch—and therefore, in the eyes of those who celebrate a strong, unitary executive, the responsibility of the President. But in my judgment, the Supreme Court’s sensible decision in Chevron can and should be read in a manner entirely consistent with Marbury.

With Chevron, the Court gave up the illusion that each question of statutory meaning has one sole determinate answer. Particularly in the administrative context, it is often impossible for courts to say more than what a statute could mean—to do more than to describe the boundaries of a set of possible meanings. But it is for the courts, always and irreducibly, to say what those possibilities are—to police the range of outcomes the statute permits. That is the Marbury function at a time when we acknowledge that statutes are indeterminate; that is the first step of the two-step Chevron analysis; and that question of range is indeed one to be “settled by the inclinations and predispositions of federal judges.” Within that range, policed by the courts, administrators can plausibly claim authority to determine the best policy outcome for today—at least, so long as they do so “reasonably,” following appropriate procedures and exercising what a court can find to be rational judgment. And while the range of possible meanings of a statute may not change with the years, particular resolutions of matters falling within that range well may.

But Professor Sunstein is alarmed at the political nature of courts’ judgments and seems to intend a lesser role for them. He argues instead that courts ought to cede a greater role to the executive. In so doing, Professor Sunstein rejects the Supreme Court’s signal in United States v. Mead Corp. that Chevron’s analysis applies only to procedurally intensive or regular agency behaviors, and not to more casual agency decisions. More dangerous still, Professor Sunstein’s approach would aid the recent inclination of several Justices to reduce further the legal constraints on executive action.

Let me illustrate the point in relation to the Supreme Court’s recent decision in Gonzales v. Oregon, which Professor Sunstein mentions glancingly (and in the briefing of which, the reader is entitled to know, I participated as an amicus curiae). In Gonzales, the Court considered the Attorney General’s claim that his office had authority to interpret the Federal Controlled Substances Act of 1970 (CSA)—a law that makes it a crime to possess or distribute addictive or psychotropic drugs—to preclude a doctor’s lawful prescription of morphine to assist another’s suicide in accordance with Oregon’s Death with Dignity Act. Without public procedure or consultation of any kind, the Attorney General had displaced Oregon’s law regulating the medical profession and had defined a new crime. Not surprisingly, a majority of the Court, with Justice Kennedy writing, rejected the Attorney General’s claim; perhaps the only real surprise was that Justice Scalia, with the new Chief Justice and Justice Thomas concurring, would have accepted it, relying on Chevron. The division was roughly the same as would occur in last Term’s final and more portentous discussion of presidential authority, Hamdan v. Rumsfeld.

The Attorney General’s claim was an entirely predictable, and professionally appropriate, claim by the government’s lawyer. Of course one argues for judicial deference to one’s client’s view of the law, if it might be entitled to it in a litigation in which its interests are at stake. Whether that view is entitled to influence, however,—and, if so, to what kind of influence it is entitled—are questions for others, the judiciary, to address. Merely saying it is so does not make it so. When the judiciary does address these questions, one should expect careful attention to the risks and rewards of answering them in one way or another. And these equations may differ with one’s assessment of the kinds of powers Congress has conferred on a given executive actor, the extent of that actor’s engagement with the public in the formulation of its view, the actor’s particular expertise, and so on.

The circumstances of Gonzales made it extremely unlikely that Congress had conferred such a power. In that case, the Attorney General had acted outside the structures of expert consultation that Congress provided in the CSA—indeed, had acted without any public procedure at all. On his own, the Attorney General had both displaced a considered state law regulating the medical profession and purported to define a new crime. It is difficult to imagine a more concerted assault on our ordinary principles of legality.

That the three dissenting Justices were the members of the Court as then constituted most strongly committed to a strong executive suggests the difficulty with Professor Sunstein’s otherwise brilliant analysis. Attentive in some respects to the problems of foxes in chicken coops, it pays insufficient care to the virtues of having someone other than the executive tend the fence around executive authority and take care that that authority is exercised in a manner subject to public participation and control. One can accept the bulk of Professor Sunstein’s analysis and nonetheless maintain that it is “emphatically the province and duty of the judicial department to say what the law is,” by framing the executive’s power to pinpoint the law within the bounds framed by judicial exercise of that duty, not outside of it. Doing so will lead one to just about the same place as Professor Sunstein reaches in respect of Chevron itself, but not to most of the specific conclusions that he reaches in Part II of his analysis, “Marbury’s Revenge?” The paragraphs following trace this analysis.

As early as 1940, the Supreme Court unselfconsciously revealed the tension between judicial and agency interpretation in an opinion long (and erroneously) thought to have sent “plain meaning” statutory interpretation to its grave. The case involved interpretation of a statute governing the Interstate Commerce Commission’s (ICC) regulatory authority, which the ICC had consistently interpreted (with support from other relevant executive agencies) to deny it the authority in question. At an early point in its analysis the Court reiterated the Marbury proposition that “[t]he interpretation of the meaning of statutes, as applied to justiciable controversies, is exclusively a judicial function.” Yet just a few pages later, after reciting the ICC and other interpretations, the Court remarked that

  [i]n any case such [executive branch] interpretations are entitled to great weight. This is peculiarly true here where the interpretations involve “contemporaneous construction of a statute by the men charged with the responsibility of setting its machinery in motion, of making the parts work efficiently and smoothly while they are yet untried and new.” Furthermore, the Commission’s interpretation gains much persuasiveness from the fact that it was the Commission which suggested the provisions’ enactment to Congress.  

Here resolution of the tension between court and agency prerogative is not so hard. One sees that the Court is deciding the statutory question for itself, (one might say, Chevron step one) but in doing so is according any respect merited to the views of responsible others. The same approach—unmistakable judicial decision of the issue, but with attention in doing so to the agency’s view—characterizes its later iconic decision in Skidmore v. Swift & Co.

Chevron’s addition of a second step to the analysis makes it different, as was another iconic case of Skidmore’s times, NLRB v. Hearst Publications, Inc. In both these cases, the Court’s initial analysis of statutory meaning—its performance of the Marbury function—led it to conclude that within the bounds of statutory possibility, Congress would have wanted subsidiary questions, which one might frame as interpretation, to be decided by a responsible agency rather than by the courts, subject only to the usual judicial review of administrative action for “reasonableness.” In Hearst the relevant question was who was an “employee”; in Chevron it was whether a number of pollution sources on one industrial site could be treated as a single “stationary source.”

Professor Sunstein properly argues that these two cases are not equivalent—that there was an actual delegation in Hearst and merely a fictive one in Chevron. The thing to note, however, is that in both settings the Court attended to what it constructed, on its own, to have been Congress’s wishes. Operating within the province that is “emphatically” its own, it found for itself a sufficient reason in the statute to believe a task had been assigned primarily to another; the constitutionality of such an assignment, under appropriate judicial safeguards for its regularity, was and is not open to doubt. Once a court has found such an assignment, it is merely following its own nose when it concludes that its task is not to assign meaning to contested terms de novo but rather to assure itself that an agency’s interpretation of those terms has the indicia of administrative legality—for example, that the agency’s action falls within its legal authority, was made following any requisite procedure, is supported to whatever extent is called for by the known facts, and reflects reasonable judgment.

I doubt that Professor Sunstein and I would disagree about any of the foregoing. The difficulties arise when one moves from administrative decisions that exercise delegated authority, following public procedures and subject to judicial supervision, to less formal executive behavior—to behavior that does not profit from the same level of public engagement and may, indeed, escape the effective oversight even of politically responsible agency heads. Sunstein relies on cases like Hearst and Chevron—cases in which agencies followed formal, public procedures in carrying out delegated authority—for the much broader claim that courts must also defer to less formal executive behavior—to choices made away from public scrutiny and without full administrative oversight.

It is striking that the “most important” case Professor Sunstein invokes as exemplifying the Court’s tendency to sometimes give “relatively little deference to agencies,” Citizens To Preserve Overton Park v. Volpe, was the latter sort of case, and not the product of APA rulemaking or adjudication. The practice of intensive, “hard look” review of the reasonableness of agency action that this case foreshadowed, with its insistence on reason-giving and regularity, does not embody distrust of agency discretion in the sense of denying primary agency responsibility for the matter being reviewed. Instead, “hard look” review accepts the agency’s responsibility, but asserts the need for careful review given the consequences of agency action—and does so out of the same congeries of factors that Professor Sunstein evokes, including a healthy respect for the fox-henhouse problem. If we put aside the ordinary procedures for public engagement in executive action, including the possibility of reviewing it on the basis of some kind of administrative record, haven’t we simply set the fox free? There is a “high risk of unreliable or biased interpretations” where agency interpretations are not the product of public procedures; and that risk increases when the interpreter has some assurance that the very fact of its making an interpretation may entitle it to prevail.

Professor Sunstein most strongly urges the persuasiveness of Chevron analysis for these informal kinds of decision in Part II of his essay, entitled “Marbury’s Revenge?” As he has in previous essays, he expresses doubt about the soundness of the Court’s decision in United States v. Mead Corp., in which the Supreme Court refused to apply Chevron to a Customs Bureau interpretive letter that had not been the product of APA rulemaking or other public procedures. In my judgment, however, Mead and the decisions that have followed it—Gonzales most recently—get matters about right. Within Marbury, the courts have the duty to determine agencies’ authority to act—that is, the duty to determine whether Congress intended the agency to have the primary responsibility to interpret the law, and if so following what procedures, subject only to the usual judicial review of administrative action; or, rather, whether that primary responsibility lies with the courts. That duty is “emphatically the province and duty of the judicial department,” and is “exclusively a judicial function,” the irreducible minimum of the judicial task. And this is the duty the Court performed in Mead. This is not a restriction of Chevron’s domain unless we are to assume from the outset that Chevron marked an unprincipled retreat from courts’ duties. If it did mark such a retreat, we should be relieved that in Mead the judiciary has now discovered that it, and not the fox, is responsible for identifying the relevant fences.

The problem with the proposition that “pure questions of law” are for the courts, and mixed questions of law and fact are for agencies, is about the same. Administrative Law teachers for decades enjoyed torturing their students with the contrast between Hearst, where the Court said it had been given to the NLRB to say whether newsboys were “employees” within the meaning of the statute, and Packard Motor Car Co. v. NLRB, which seemed totally to disregard the NLRB’s view of whether foremen could come within the statute. But if we regard the fence—the difference between the question what a statute could mean and the