|Improving Deference: Chevron as a Voting Rule|
|Jacob E. Gersen and Adrian Vermeule, Wednesday, 24 January 2007 [View as PDF]|
Of central importance to administrative law and theory is the question whether, and when, courts will defer to agency interpretations of law. In Chevron v. Natural Resources Defense Council, the Supreme Court replaced earlier answers to that question with a new framework: courts should defer to an agency interpretation unless the relevant statute is clear or the agency interpretation is unreasonable. In the past two decades, however, the Chevron framework has come under increasing strain. Doctrinally, there are many ambiguities and uncertainties about the nature of the inquiry at the first and second steps of Chevron, including questions about the admissibility and weight of various legal sources. In practice, recent evidence suggests that Chevron’s effect varies markedly with the ideological and political preferences of the judges who apply it.
These problems arise, in part, from a dubious premise of the Chevron enterprise, one that should be rethought. The dubious premise is that the legal system should adopt a doctrinal solution—the Chevron rule-for what is, after all, an institutional problem: the allocation of interpretive authority between agencies and courts when congressional instructions are silent or ambiguous. In our article in this Journal we explore an alternative, which is to cast Chevron as a voting rule, thereby institutionalizing deference to administrative agencies. Imagine a voting rule stating that when a litigant challenges agency action as inconsistent with an organic statute, the agency would prevail unless the judges, asking simply what the best interpretation of the statute is, vote to overturn the agency by supermajority vote—say, by a 6-3 vote on the Supreme Court, or by a 3-0 vote on an appellate panel.
Judges, lawyers, and commentators often develop doctrinal solutions to institutional problems. However, legal doctrine is rarely the only type of solution to an institutional problem, and it is not always the best solution. An alternative is to change the rules that govern the composition, powers, or voting rules of relevant institutions. We call these “hard” solutions, in contrast to “soft” or doctrinal solutions. The relative costs and benefits of soft solutions, on the one hand, and hard solutions, on the other, vary across contexts and over time. Our point is not that hard solutions are always superior, for they are not. What we do suggest is that hard solutions prove superior in many domains, yet are frequently overlooked.
Although the respective costs and benefits of hard and soft rules will vary with context, certain regularities make tentative generalizations possible. Consider, abstractly, the problem of partial deference, by which we mean the recurring situation in which it would be good or right for institution A to defer to institution B on the resolution of legal question Q, but only if institution B’s decision is not clearly wrong or unreasonable. A problem that arises in many situations of partial deference is that the triggering conditions for deference are vague or imprecise. What is “clear” or “unreasonable” to one judge is not “clear” or “unreasonable” to another. Conceptually, it is not clear what “clear” means. Psychologically, it is difficult for judges and other decision-makers to avoid collapsing their views of what is reasonable into their views about what is the best legal answer, thereby defining any answer that it not best as unreasonable. Motivationally, deference rules based on vaguely defined triggering conditions allow scope for ideological and even partisan biases. The basic problem in such cases is that individual decision-makers are charged with internalizing a legal norm of deference that is conceptually ill-defined, and that cuts hard against both their individual judgments of what is best and their biases and prejudices.
A shift from a soft legal norm to a hard institutional solution would help solve these problems. Imagine a supermajority voting rule under which institution A defers to institution B unless two-thirds of the members of A believe that B is wrong on the merits. Each decision-maker asks simply what legal answer is best and votes accordingly. Deference is an emergent property of the aggregate vote, rather than of individual decisions. Conceptually, there is no need for decision-makers to develop a theory about what counts as a “clear” (as opposed to merely correct or incorrect) legal answer. Psychologically, there is no requirement that the decision-maker simultaneously hold in her mind two conflicting legal standards, so the cognitive load is greatly reduced. Motivationally, each judge may be biased or prejudiced in some sense, yet assuming some diversity of preferences the biases will be washed out at the aggregate level, with deference enforced by the voting rule. Moreover, it is easier for voters or other principals to monitor the behavior of their agents, the decision-makers in institution A, because those agents have less room to maneuver in a supermajority voting rule system. Rather than concealing their biased votes by claiming that institution B’s decision is or is not clearly wrong or unreasonable—claims whose second-order character makes them inherently costly for outside observers to evaluate—the decision-makers must now simply state their understanding of the correct legal answer.
Further, hard solutions are easier to calibrate than are soft solutions. Suppose, in operation, that an internalized norm of deference produces too much or too little deference from institution A to institution B on legal question Q, with “too much” or “too little” defined by some extrinsic theory. How can the level of deference be adjusted up or down? Where a soft solution is in place, adjustment is difficult and imprecise. The problem is that verbal formulae are typically too crude to capture the fine shades of difference that are needed to tweak deference rules in either direction. A voting rule can be more precisely calibrated. If a 6-3 supermajority rule produces too little deference, a 7-2 rule can be substituted. The calibration will still be imperfect, but comparatively, adjustments in voting rules will be less lumpy and more fine-grained than adjustments that rely on manipulating linguistic formulations in judicial doctrine.
A corollary of the last point is that hard solutions generally increase legal certainty. Suppose that a particular doctrinal formulation produces a particular level of deference, D. This level, D, is a kind of expected value, with variance around that value; some courts applying the doctrine will exceed D; some will fall short; and some courts will exceed D in one period while falling short in another period. A voting rule will produce greater certainty about deference, holding the expected level of deference constant. To the extent that reducing variance is a benefit for actors in the legal system, a voting rule may well be preferable to a doctrinal solution.
If deference to administrative agencies is desirable, then a framework that institutionalizes deference through voting rules is at least as good as doctrinal deference on many dimensions of comparison, and clearly better on some dimensions. Although we do not minimize the challenges of implementation, none seems sufficiently powerful that we should ignore the benefits from switching to Chevron as a voting rule.
Jacob E. Gersen is Assistant Professor of Law, University of Chicago Law School. Adrian Vermeule is Professor of Law, Harvard Law School.
Preferred Citation: Jacob E. Gersen & Adrian Vermeule, Improving Deference: Chevron as a Voting Rule, 116 Yale L.J. Pocket Part 235 (2007), http://thepocketpart.org/2007/01/25/gersen_vermeule.html.
Editor's Note: Improving Deference: Chevron as a Voting Rule accompanies Jacob E. Gersen & Adrian Vermeule, Chevron as a Voting Rule , 116 Yale L.J. 676 (2007).