Making Our Democracy Work

by Breyer, Stephen

  Some argue that applying these different standards is a psychological impossibility. A distinguished circuit judge, Harold Leventhal, for example, once facetiously wrote that he “thought” he had found the case “dreamed of by law school professors,” a case where he could “conscientiously” distinguish among the standards, upholding an agency’s finding of fact because it was supported by “substantial evidence,” even though, had that finding been made by a district judge, he would have struck it down as “clearly erroneous.” But he decided on second thought that the finding was not supported by “substantial evidence” either.8

  Judge Leventhal’s reaction is an overstatement. Judges are able to apply different standards—at least to some degree. When they review a lower-court judge’s decision about what the law means, they can simply ask themselves, “Is it right?” When they review a jury decision, they can ask, “Am I completely certain the jury is wrong, to the point where no one could sensibly come to that conclusion?” A reviewing judge can also think, “I wouldn’t have come to that conclusion, but I can see how someone else might.”

  The matter of different standards for review is often better expressed in terms of degree, not kind. To what extent does the reviewing judge give the other decision maker leeway to come to a decision that the judge, on his own, would likely think wrong? A judge who grants that other decision maker at least some leeway in respect to a decision has adopted an attitude of deference.

  Many administrative law questions—particularly those that define the relation between judges and administrators—can be put in terms of deference: How much deference should the reviewing court grant? For example, should it give the agency the benefit of any doubt, thereby coming close to presuming the correctness of the agency’s decision? Should it instead review the agency’s decision from scratch, giving the agency little or no benefit of any doubt? Administrative lawyers would describe the first attitude as one of strong deference to the agency, the second as one of no deference at all. Should a court give an agency deference? When should it do so? How much deference should it give?9 These are key questions defining relationships between the courts and much of the executive branch.

  DURING THE NEW Deal many administrative law experts supported greater deference. They believed that agency administrators used expertise to determine “scientifically” such matters as the proper level of railroad rates or when an agency should suppress competition on the ground it was “destructive.” At the same time, they believed that courts, often hostile to regulation, would prove too willing to substitute their own views for those of the administrators. Today, however, the public has less confidence in agency expertise. Political appointees, often not experts, are normally responsible for managing agencies and determining policy. And policy often reflects political, not simply “scientific,” considerations. Agency decisions will also occasionally reflect “tunnel vision,” an agency’s supreme confidence in the importance of its own mission to the point where it leaves common sense aside. At the same time, courts no longer seem particularly hostile to regulation as a matter of principle. Hence, the public now relies more heavily on courts to ensure the fairness and rationality of agency decisions.

  Expertise is still relevant, however, to the question of how much deference to give—although “comparative expertise” is a better term. Courts will exercise relatively more control over issues within their expertise while according agencies relatively more leeway (but not unqualified deference) as to issues within theirs. How the Court has applied these principles to two types of agency decisions—those relating to policy and to the interpretation of statutes—is the question to which I now turn.

  REVIEWING AGENCY DECISIONS OF POLICY

  CONSIDER MORE SPECIFICALLY how comparative expertise helps determine deference in respect to agency policy decisions. In reviewing such a decision—for instance, about what standards to impose on tire manufacturers in order to ensure automobile tire safety—a court often must answer a legal question: Is the agency’s tire safety decision “arbitrary, capricious, an abuse of discretion”? A realistic appraisal of comparative expertise would start with an understanding of the kinds of problems facing the agency and how agencies would typically go about solving such problems.10

  An agency staff formulating rules for automobile tire safety might begin with little expert knowledge. Nonetheless, it has time to learn and will likely research the subject matter for months, perhaps longer. The staff can consult with outside experts, learning from those experts even where they have competing interests. It can ask for public comments on its initial efforts and revise those efforts accordingly. It can seek information and reactions from colleagues at other agencies. In a word, it can develop subject-matter expertise.

  At the same time, the agency staff must make technical decisions and write technical standards based on what they have learned. To do so, they must decide such technical issues as whether to employ cost-benefit analyses or whether to base standards on product design or product performance. And to be effective, they may also have to take account of the views of those who favor or oppose their work, perhaps writing standards that reflect political compromise.

  By way of contrast, consider the judge’s expertise in such matters. Judges have little time to spend on any one case, such as a case in which a party contests the reasonableness of the agency’s tire safety regulations. They deal with a record that rarely reflects all that the agency or its staff had to consider. They cannot look for information beyond that record. They must respond to the arguments of the lawyers. They do not necessarily have much political experience.

  It is not surprising, then, that courts, recognizing the institutional differences, find agency policy decisions “arbitrary, capricious, an abuse of discretion” only in rare and clear cases. Courts have struck down, for example, a National Labor Relations Board election rule that allowed officials to buy drinks for the voters on Election Day—a policy that reflected considerations beyond the area of the labor board’s special expertise. But ordinarily, courts, while insisting on proper procedures, will allow agencies considerable leeway. Courts very much defer to agencies when they review agency determinations of policy.11

  REVIEWING AGENCY INTERPRETATION OF STATUTES

  A DIFFICULT AND important problem involving judicial attitude arises when a court reviews an agency decision interpreting a statute. Should a court ever defer to an agency’s decision about the meaning or scope of statutory language? Or should a court always decide what a statute means uninfluenced by the agency’s interpretation? It may seem surprising that courts sometimes take the former approach.

  Why would courts ever defer to an agency’s interpretation of a statute? Statutory interpretation is a basic judicial job that courts perform day in and day out using a well-developed set of tools. If ever courts have comparatively more expertise, isn’t this the place? But let’s imagine that Congress writes a statute that specifically grants to an agency leeway to fill in the blank. Suppose, for example, Congress writes a labor statute that uses the term “employee” and adds “the labor board shall have the power to determine in accordance with the objectives of this statute what kinds of employees fall within this term.” In that case, Congress has given to the agency the power to write a regulation that takes effect as a law. If a court is to maintain a workable relationship with Congress, it must respect that decision. Now suppose that Congress does the same thing implicitly. That is, it delegates to the agency the power to define “employee,” but not by explicitly delegating this power in the statute. A court should similarly respect Congress’s implicit decision.12

  Consider the vast number of statutory provisions that Congress writes to handle the business of government. Many of these provisions concern matters of detail, some quite technical, that are important for the operation of the program. Agency officials are likely to understand these details, but courts may not. Indeed, Congress may have delegated power to the agency in the first place be
cause it, too, lacks this expertise. If so, the court will likely defer to a reasonable agency interpretation of the statutory provision.13

  But how does a court know whether Congress intended the court to defer to a reasonable agency interpretation of a statute? Congress is normally silent on the subject, delegating interpretive power to the agency implicitly if at all. Furthermore, Congress could not have taken from the courts all their ordinary power to interpret statutes. Nor would Congress likely have intended to give the agencies free rein to interpret statutes in ways that, for example, diminish the limits on their delegated powers. Hence the problem: When should courts, recognizing an agency’s comparative expertise, defer, or not defer, to the agency’s interpretation of the statute?

  The question is important. It embodies a modern democratic dilemma. No one doubts that the conditions of modern life require Congress to grant agency officials broad authority to decide many questions that affect our daily lives—for example, whether gasoline can contain lead, whether power plants must eliminate sulfur dioxide, whether garbage must be recycled, whether telemarketers can interrupt families during dinner, whether shippers must pay higher charges to railroads, or whether interest rate costs must be fully disclosed. But how much authority should a legislature delegate?

  In determining how much authority Congress did delegate in a particular statute, a court must balance two countervailing concerns. On the one hand, the court must not recognize more agency authority than Congress intended, because doing so would unnecessarily limit the citizens’ ballot box control over government actions that importantly affect their lives. On the other hand, the court must not recognize less authority than Congress intended, because doing so would prevent citizens from securing basic objectives that they cast their votes to achieve, say a cleaner environment or greater consumer protection. The public cannot achieve its military objectives if Congress is required to enact a statute that tells the army in detail how to take a particular hill—nor is Congress likely to write such a statute. Statutes that tell administrators precisely which pollutants to regulate, and where and how to regulate them, can prove similarly counterproductive. The legislator who tries to become a super-detailed air pollution regulator can easily end up with dirtier air.

  The Court considered the deference question in a well-known case, Chevron v. National Resources Defense Council. The case involved an environmental statute that said the Environmental Protection Agency (EPA) must regulate “new or modified stationary sources” of pollution in regions of the country that had not yet met the statute’s clean air goals. The EPA had developed a system of regulation that in effect treated each machine that emitted a pollutant as a separate “source” that must meet a specified standard. The EPA then changed its system so that it no longer treated each machine as a separate source. Instead, hoping to achieve greater efficiency, the EPA placed an imaginary “bubble” over a factory and treated all the emissions within the bubble as if they came from a single source. This meant a company could keep some dirty machines in operation as long as it offset the resulting emissions by using cleaner machines elsewhere. The question was whether the EPA could apply the statutory word “source” to include all the machines within the bubble taken together, and the Court held that it could do so.14

  In this case the Court set forth a general rule describing when courts should defer to an agency’s interpretation of a statute. It said that if the answer to the statutory question is “clear,” a court should not defer to the agency’s interpretation of the statute; rather, it should provide the answer irrespective of what the agency says. But if the answer is not clear—where, for example, the statute itself is “silent or ambiguous”—then courts should assume that Congress intended to delegate to the agency the power to interpret the statute, and they should defer to (and uphold) a “reasonable interpretation made by the administrator of an agency.”15

  The deference rule has not completely resolved the problem, however. Taken literally, it would give agencies authority to resolve virtually all statutory ambiguities. The Court has not permitted this result, because the deference question arises in respect to so many different programs involving so many different statutory provisions, potentially applicable to so many different kinds of circumstance, raising so many different administrative problems, that a single formula about deference cannot work every time a statute is ambiguous. Thus, the Court has treated Chevron’s rule not as a universally applicable formula but as a guiding rule of thumb. When there are good reasons to think Congress would not have wanted the Court to defer to an agency interpretation, the Court has not done so.

  In 2007, for example, the Court considered a critically important environmental question. Does the Clean Air Act give the EPA the authority to regulate greenhouse gases, such as carbon dioxide? The statute said that the EPA should regulate “any air pollutant” that “endanger[s] public health or welfare.” The statute defines “air pollutant” as including “any air pollution agent … including any physical, chemical … substance … emitted into … the ambient air.” The EPA had interpreted the statutory words “air pollutant” as excluding greenhouse gases, but the Court by a narrow majority reversed that determination. In holding that the statute did cover greenhouse gases, the Court did not defer to the agency’s own interpretation of the statute. Even though the statutory word “any” (in the phrase “any air pollution agent”) created potential ambiguity, the Court thought that Congress would not have intended to delegate to the agency the legal power to interpret the statute to exclude gases that were major contributors to global warming. The decision is also consistent with the view that Congress would not have wanted to grant the agency the power to decide by itself such an important general policy question.16

  In another, less important case the Court made clear that Chevron does not set forth an absolute deference rule. A statute authorizes the U.S. Customs Service to “fix” according to Treasury Department regulations the “final classification and rate of duty applicable” to imported merchandise. A customs service officer, applying a Treasury Department regulation, classified three-ring binder “day planners” among “diaries,” placing them in a high-duty category rather than among “other” items in a low-duty category. In deciding this case, the Court held that the reviewing court, the U.S. Court of International Trade, should not defer to a customs officer’s interpretation of the relevant regulations. The Court was convinced that Congress would not have intended that deference because of the large number of customs officers, the large number of such rulings, the speed and informality with which they were made, and the presence of a specialized reviewing court.17

  What explains these decisions? Why does the Court sometimes defer to an agency’s interpretation of a statute and sometimes not? Why does Chevron not always work as an explanation? The answer begins with the democratic dilemma mentioned earlier. To defer to an agency’s view is to give the agency power to say what the law means. In principle, Congress, elected by the people, should decide how much power the agency should have. But what are courts to do when Congress does not say? As in all statutory matters, the answer will depend on the statute’s purposes and context, and particularly on matters of comparative expertise.

  It is more reasonable to believe that Congress (had it considered the matter) would have wanted the reviewing court to defer to the agency’s views if an agency has special expertise regarding the legal question, if the question concerns detailed matters of the agency’s program or its administration, if the legal question has little general importance, if the agency has considered the matter with greater care, and if the statute’s language is ambiguous.

  Congress (had it considered the matter) is more likely to have thought that the reviewing court possessed the relevant expertise (at least comparatively speaking) and did not intend deference to agency views, however, when the agency does not fully consider the question, the question involves important general matters of policy, or an answer is likely
to clarify, illuminate, or stabilize a broad area of law. In both sets of circumstances, looking at the statutory purposes and context and the comparative expertise of the agency and the Court produces a decision that facilitates, rather than impedes, the working of the statute in the real world.

  In the EPA case the breadth and importance of the legal question at issue seemed more significant than the fact of greater EPA technical expertise in respect to carbon dioxide. The Court could reasonably think that the relevant expertise needed to answer the question was primarily legal, not administrative, and that the agency ruling misinterpreted Congress’s intent. The Court was (relatively) better positioned to consider the purposes of the statute and the related consequences of excluding or including greenhouse gases.

  In the U.S. Customs Service case, the Court noted that the service’s expertise argued in favor of deference. But the officers’ informal ruling letters were “churned out at a rate of 10,000 a year at [the] agency’s 46 scattered offices,” which suggests those letters were done quickly without too much concern for their consistency with statutes, regulations, or one another. Moreover, the reviewing court, the U.S. Court of International Trade, not just the customs officers, had expertise in these matters. Hence, the Court concluded that Congress could not have intended deference.18

  THE COURT, INTERPRETING congressional silence, has worked out a practical system of deference. Using comparative expertise as a touchstone, courts by and large defer where, comparatively speaking, agencies are likely better able to solve the problem. They do not defer where, comparatively speaking, courts are likely better able to solve the problem. This approach is consistent with the Constitution’s democratic aims. It recognizes that decisions about how much authority to delegate to an agency, like other statutory matters, rest ultimately in the hands of Congress. And Congress is responsible to voters. But importantly, those voters may be unaware of the details and may well only know whether Congress’s statutes are “working out well.”