Originally posted at Notice & Comment, a blog of the Yale Journal on Regulation and the American Bar Association Section of Administrative Law & Regulatory Practice, as part of an online symposium entitled Reflections on Seminole Rock: The Past, Present, and Future of Deference to Agency Regulatory Interpretations. Reprinted with permission.
The Separation of Powers Restoration Act, or more easily known as SOPRA, is not a complicated bill. If enacted, it would amend the Administrative Procedure Act to require courts to decide de novo all questions of law, whether constitutional, statutory, or regulatory. As the House Report makes abundantly clear, the intent is to overrule statutorily both Chevron, USA, Inc. v. NRDC and Auer v. Robbins (and its forebear Bowles v. Seminole Rock & Sand Co.), but not Skidmore v. Swift & Co.
The idea is not new. Indeed, beginning in 1975, well before Chevron began its journey through the courts, Senator Dale Bumpers introduced bills (or amendments to bills) that were intended to have the same effect. Although versions of his proposal were passed by the Senate on at least two occasions, the so-called Bumpers Amendment never became law because cooler heads prevailed. Undoubtedly the same fate will meet SOPRA. Why?
First, as several law review articles have tried to demonstrate, Chevron and Auer rarely decide a case differently than the case would have been decided in the absence of the doctrines. One can probably count the number of cases on one hand in which the court said it would reach a different conclusion but for applying Chevron or Auer deference. Articles like David Feder's blog here that purport to show that Auer matters only show that judges cite Auer. They do not show how those judges would have voted in the absence of Auer. Even in David Feder's blog, the disagreements between the judges was over whether the regulations were or were not ambiguous, and Auer does not factor in making that decision. Indeed, in one of the cases, the court that invoked Auer said the agency's interpretation was not only reasonable but also the best one, clearly demonstrating that it would have ruled the same way in the absence of Auer.
Second, much of the motivation for SOPRA, as reflected by the floor speeches of its proponents, is to reduce government regulation, viewed as improper extensions of the laws passed by Congress. But one must have a short memory to support SOPRA on that basis. Chevron itself involved a Reagan administration attempt to lessen regulation, and it was the D.C. Circuit applying a SOPRA-type approach that set aside that more lenient regulation; it was the Supreme Court applying Chevron that upheld the more lenient regulation.
Third, and more academically, the theoretical justification offered to support overruling Chevron is faulty. If one reads the House Report on SOPRA, there is one set of explanations given for overruling Chevron and a different one for overruling Auer. The explanations for overruling Chevron are that Chevron is at odds with the text of the APA, because the APA says that the court is to "decide all relevant questions of law," not let agencies decide them; it is at odds with Marbury v. Madison, because the courts have given to the executive the power to say what the law is; and it is at odds with the Separation of Powers, because the courts have impermissibly delegated to agencies the judicial power to say what the law is. The problem with these explanations is that they all incorrectly characterize Chevron, as the late-Justice Scalia would have told them.
In step one of Chevron a court says what the law is and is not. Often a court will find that the law is discernible and uphold or set aside the agency action on that basis. However, if a statute is truly ambiguous – or as the Court's opinion in Chevron described it, if after using all the ordinary tools of statutory construction, a court ascertains that Congress did not have an intention on the precise question at issue – then there is no "law" for the court to find. In short, Congress in the statute simply did not make "law" regarding this particular application. Thus, the court has decided de novo the question of law – the statute does not decide the case. Because the statute as written and interpreted does not provide the law necessary to decide the case, the court in Chevron recognized that it had two options. Either the court itself could fill the void and make the law, not interpret it, or it could interpret the vacuum left in the statute as Congress leaving to the agency the authority to make the law, so long as the agency's action is permissible under the statute as it can best be interpreted.
However one wishes to characterize it, once a court has faithfully applied step one and found that it cannot discern a congressional intent reflected in the statute that would decide the case, someone needs to provide the "law" to decide the case. In Chevron, the Court assumed that Congress would prefer the agency to make that "law," because Congress had given that agency the responsibility for implementing the law. Presumably, Congress could overrule Chevron and tell courts that it prefers the courts to make the law, but why would it prefer courts to do that? Even if the administration is in the hands of a different party than Congress, Congress has more influence on agencies' decisions than it does on courts' decisions.
The theoretical justification for overruling Auer is different. Here there is extended citation to the late-Justice Scalia's concurring and dissenting opinions in which he outlined the policy reasons why he believe Auer should be overruled. Professors Sunstein and Vermeule have in their blog here and in their Chicago Law Review article provided their policy reasons for why they believe Auer should be retained. I will not add to that debate except to second, on the basis of my experience as a government lawyer writing regulations, their belief that Auer does not in fact result in agencies writing vaguer regulations than they otherwise would.
Missing from both the House Report and Sunstein and Vermeule's article, however, is a consideration of a different incentive created by Auer deference. That incentive is to clarify rules by means of an interpretive rule or policy statement, rather than by issuing a new, clarified rule after notice and comment. The former is faster, cheaper, and easier in every respect, so there is already a great incentive to avoid notice-and-comment rulemaking. If Auer deference – which is equivalent to Chevron deference – will be accorded to the interpretive rule or policy statement, one of the incentives of using notice-and-comment rulemaking – to qualify for Chevron deference – disappears, and the balance between the two options is greatly tilted. And it is the misuse of interpretive rules and statements of policy that have exercised both the bench and the academy. Unlike the incentive to write ambiguous regulations in order to retain flexibility for later interpretation, for which there is no empirical support for agencies acting on that basis, the incentive to avoid notice-and-comment rulemaking is strong, and there is a wealth of empirical support for the fact that agencies indeed try to cut corners, especially given the number of cases challenging agency interpretive rules as improperly adopted legislative rules. This may suggest that Auer deference should not be at one with Chevron deference but instead should be equated with Skidmore deference – a lighter deference, one with the power to persuade if not to bind. Agencies then could make a choice between taking the longer route with resulting greater deference, or the shortcut and less deference. The more obviously correct (or inconsequential) the interpretation is, the stronger the case for the shortcut. The more the interpretation raises important policy issues that may be highly contested, the stronger the case might be for notice-and-comment rulemaking. And shouldn't that be the correct direction for the incentives to work? Thus, the substitution of Skidmore respect for the great deference of Auer might well be worthwhile. But is it worth a statute? Perhaps, but not SOPRA with its effect on Chevron.