Chevron Step One-and-a-Half

Introduction

The Supreme Court’s decision in Chevron U.S.A. Inc v Natural Resources Defense Council, Inc1 has created a cottage industry in choreography. Justice John Paul Stevens’s opinion introduced the famous Chevron two-step.2 Thereafter, Professors Thomas Merrill and Kristin Hickman identified a Chevron “step zero”—“the inquiry that must be made in deciding whether courts should turn to the Chevron framework at all.”3 (Some subdivide Step Zero into two steps of its own, creating a four-step test.)4 A quarter century after the Chevron decision, Professors Matthew C. Stephenson and Adrian Vermeule declared that in fact “Chevron has only one step”: “whether the agency’s construction is permissible as a matter of statutory interpretation.”5 Some judges now view Chevron as “a three-step inquiry,”6 while others suggest that the number of steps probably does not matter much in practice.7 Little wonder, then, that others throw up their hands (or their feet?) and dismiss the entire step-defining exercise.8

One might infer from this choreographic confusion that we now have too many formulations of Chevron, with no need for another. One might also draw the conclusion that a doctrine designed to simplify judicial review of agency statutory interpretations has instead had the opposite effect. And yet one might infer from this Article’s title that the authors, rather than trying to streamline Chevron, are scheming to propose yet another step in the Chevron shuffle—or, more precisely, a half step. One might ask, quite fairly, whether the Chevron dance really needs another move. (One might also groan that we already have taken the dance metaphor too far.)9

Although we are mindful of Chevron fatigue, the fact remains that neither the one-step, the two-step, the three-step, nor the four-step formulation of Chevron captures an important doctrinal development that has occurred in the federal courts. In the classic Chevron two-step, the court asks at Step One “whether Congress has directly spoken to the precise question at issue”; if the answer is negative, then the court proceeds to Chevron Step Two and asks “whether the agency’s answer is based on a permissible construction of the statute.”10 The DC Circuit and a handful of other courts, however, now do something different. After deciding that the statute is ambiguous but before deciding whether the agency’s construction is permissible, these courts ask a separate question: whether the agency itself recognized that it was dealing with an ambiguous statute. In these courts, a misstep at this intermediate stage can be fatal to an agency’s cause: the court will remand—sometimes with vacatur—if the agency claimed that the statute is clear but the court concludes it is not. In other words, the agency will lose if it mistakenly says that the issue can be resolved at Chevron Step One while the court determines that it should be resolved at Chevron Step Two.

One might call this move “the Prill doctrine” in honor of Prill v National Labor Relations Board,11 the DC Circuit case from 1985 that is sometimes cited as the rule’s origin.12 Or perhaps one might call it “the Negusie doctrine” in honor of Negusie v Holder,13 a 2009 case in which the Supreme Court arguably applied the rule as well (though, as discussed below, we think the Negusie rule is of a slightly different nature).14 We choose to call it “Chevron Step One-and-a-Half,” because—well—that is what it is: a way station between Chevron Step One and Chevron Step Two. Whatever one calls it, however, we should recognize it for what it is and ask why it exists. This Article attempts to do just that.

While Chevron Step One-and-a-Half has up until now been a doctrine without a name, it is nonetheless a doctrine with skeptics. Perhaps chief among them is now–Chief Justice John Roberts. In a concurring opinion that he wrote while on the DC Circuit, then-Judge Roberts criticized his colleagues for invoking Prill and its progeny.15 To be sure, Roberts said that he had “no quarrel with the basic proposition . . . that when an agency erroneously concludes that a statutory interpretation is required by Congress, we should remand to give the agency an opportunity to interpret the statute in the first instance.”16 But he argued that the doctrine ought not apply unless there is “real and genuine doubt concerning what interpretation the agency would choose” if the agency were aware of the ambiguity.17 In all other cases, Roberts said, the doctrine we dub Chevron Step One-and-a-Half “outstrips its rationale”18 and “convert[s] judicial review of agency action into a ping-pong game.”19 Professor Nicholas Bagley, in turn, has urged that such “real and genuine doubt” will be vanishingly rare. “The very fact that an agency has read the statute in a particular way,” wrote Bagley, is itself strong “evidence” that the agency “prefers the interpretation it adopted to the one that it did not adopt.”20 And quite often that probative evidence will be backed up by a representation—in writing and signed by agency lawyers—stating that the agency would stay the course on remand even if the reviewing court were to conclude that the statute is susceptible of multiple meanings.21 In light of this reality, the chief justice and Bagley would have Chevron Step One-and-a-Half be invoked infrequently rather than become a standard part of the Chevron analysis.22

Indeed, regardless of whether one subscribes to the view of the chief justice and Bagley, one might wonder why Chevron Step One-and-a-Half is ever invoked. That is, why would an agency ever insist that a statute is unambiguous? From an agency’s perspective, disclaiming ambiguity seems like a self-inflicted wound, and an easily avoided self-inflicted wound at that. In fact, if agencies seek to enhance their own autonomy (as some common accounts of agency motivation maintain23 ), one might expect them always to argue that the statute is ambiguous so long as a nonfrivolous argument for ambiguity exists. After all, under Chevron, ambiguity acts as a grant of discretion,24 and discretion is power.25 Why would an agency disavow discretion it could credibly claim?

Yet Chevron Step One-and-a-Half remands have occurred in dozens of cases—including in high-profile matters.26 This means that agencies are denying (or at least declining to acknowledge) that the relevant statute is ambiguous, even when nonfrivolous arguments in favor of ambiguity are available (arguments that a court ultimately concludes are meritorious). So what is causing these agencies to stumble on the path to Step Two? We suggest several reasons why agencies might maintain that a statute is unambiguous even though such an assertion disadvantages the agency in litigation. The first, and perhaps most mundane, is agency ignorance: agency lawyers may simply be unaware of the Chevron Step One-and-a-Half doctrine. But while ignorance might be to blame the first time that an agency encounters the doctrine, the explanation is less plausible as time goes on, especially for agencies that have had actions remanded by the DC Circuit on Chevron Step One-and-a-Half grounds on multiple occasions.27 A second reason is agency ambivalence: The agency might believe that a particular result is statutorily compelled while also being unconvinced that it is an optimal policy. The agency might decide that if a court can find ambiguity when the agency could not, then the agency would be happy to consider adopting an alternative approach. We think that this too is a plausible explanation in some cases, though not in all.

We suggest here that even when agency actors are aware of the Chevron Step One-and-a-Half doctrine, and even when they have a firm preference for a particular interpretation, they sometimes may have strategic reasons to say that a statute is unambiguous despite the potentially negative litigation consequences. One such strategic motivation involves intra-agency politics: agency lawyers who prefer a particular outcome might claim that their preferred outcome is statutorily ordained, anticipating that nonlawyers within the agency will be ill-equipped to contest that claim. A second strategic motivation involves intra–executive branch politics: an agency might claim that a particular result is statutorily compelled so as to avoid having to convince the White House’s Office of Information and Regulatory Affairs (OIRA) that the agency’s preferred outcome is cost justified relative to feasible alternatives. A third strategic motivation involves interbranch politics: an agency might seek to shirk blame for an unpopular policy by claiming that responsibility lies with Congress. Fourth, an agency might maintain that a statute unambiguously points one way in an effort to prevent future administrations from choosing a different route.

After laying out the various reasons why Chevron Step One-and-a-Half cases might continue to arise, we consider whether these reasons justify the doctrine’s existence. Despite the hostility that the doctrine faces from some quarters, we suggest that Step One-and-a-Half produces potentially important benefits. In particular, this half step advances the values that motivate (and help justify) Chevron in the first place. If agencies are entrusted with discretionary power on the grounds that they are more accountable than courts, then judicial review should encourage agencies to take responsibility for their decisions. Chevron Step One-and-a-Half can serve to encourage this accountability in an administrable way. It can also help to ensure that deference is reserved for cases in which agencies employ the expertise that they (at least ostensibly) have and that courts do not have. To be sure, the doctrine imposes costs as well, in the form of judicial remands and further litigation. Neither the benefits nor the costs are easily quantifiable, and we cannot confidently say whether the net welfare effects of the doctrine are positive or negative. What we can say, though, is that the potential benefits of Chevron Step One-and-a-Half have been overlooked so far. This Article seeks to bring those potential benefits to the fore.

This Article concludes by considering the choices courts must make when applying the Chevron Step One-and-a-Half doctrine. For example, should the doctrine be triggered whenever an agency fails to acknowledge that a statutory provision might be ambiguous—or only when the agency affirmatively states that the statutory provision is clear? Our proffered justifications for the doctrine might counsel in favor of the former, but the DC Circuit tends to lean toward the latter view. Likewise, should the application of Chevron Step One-and-a-Half result in a remand to the agency—or should it also result in vacatur of the agency’s rule? Remand without vacatur reduces Chevron Step One-and-a-Half’s costs, and yet those “costs” are arguably the doctrine’s virtues—the costs of Chevron Step One-and-a-Half potentially deter agencies from hiding their cards.

We realize, of course, that Chevron Step One-and-a-Half further complicates the Chevron analysis. The simpler version of Chevron set out in administrative law casebooks, however, does not descriptively reflect what is happening in the nation’s “administrative law court.”28 We hope that by naming the Chevron Step One-and-a-Half doctrine, explaining how it might be justified, delineating its contours, and describing its applications, we can cut through some of the complexity.

* * *

This Article proceeds as follows. Part I introduces Chevron Step One-and-a-Half with an illustrative example. Part II describes the origins of this doctrine and explains why Step One-and-a-Half is consistent with, but nonetheless distinct from, other administrative law doctrines like hard-look review and the bar on post hoc rationalizations. Part III then addresses a key puzzle presented by Step One-and-a-Half: Why do these cases continue to arise? The answer, we submit, is central to why Chevron Step One-and-Half can be a beneficial doctrine. Part IV then tentatively defends Step One-and-a-Half from its critics by demonstrating the connection between the doctrine and Chevron’s accountability and expertise justifications; whatever one thinks of the doctrine, it is a mistake to focus exclusively on its costs while ignoring its benefits. Finally, Part V considers some of the difficult choices that courts must make when applying Chevron Step One-and-a-Half in concrete cases.

I. On Prill and Polar Bears

We begin by describing Chevron Step One-and-a-Half’s application in a particular case—not because it is an exceptional example of the doctrine, but because it is an entirely ordinary example.29 The Endangered Species Act of 197330 (ESA) defines an “endangered species” as “any species which is in danger of extinction throughout all or a significant portion of its range.”31 The Act does not address, however, whether a species is “endangered” when the possibility of extinction is far off in the future. (Or at least, the Act arguably does not address that question; more on this below.) The question is particularly significant with respect to polar bears, whose very survival may be affected by losses of sea ice over the next century.32

The US Fish and Wildlife Service, which administers the ESA with respect to land animals,33 concluded in a May 2008 final rule that polar bears are “threatened” but not “endangered.”34 (The term “threatened species” in the ESA refers to “any species which is likely to become an endangered species within the foreseeable future throughout all or a significant portion of its range.”)35 While acknowledging that projected changes in future sea ice conditions pose a danger to the polar bear’s survival in the long run, the Service contended that the “plain language” of the ESA showed that an “endangered” species must face a “substantial and immediate” risk of extinction, whereas a “threatened” species is one that faces a “less imminent” danger.36 Shortly after the Service issued its final rule, complete with this temporal distinction, the Center for Biological Diversity and other environmental groups sued the Service in federal district court in the District of Columbia for failing to classify polar bears as endangered.37 The environmental groups argued that the Service had misinterpreted the ESA by reading an imminence requirement into the definition of endangered.38

Readers familiar with foundational principles of administrative law might expect that this case would be resolved on the basis of the Chevron doctrine. And indeed, that is where the district court began:

The framework for reviewing an agency’s interpretation of a statute that the agency is charged with administering is set forth in Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc. The first step in this review process is for the court to determine whether Congress has directly spoken to the precise question at issue. If the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress. . . . If the court concludes that the statute is either silent or ambiguous with respect to the precise question at issue, the second step of the court’s review process is to determine whether the interpretation proffered by the agency is based on a permissible construction of the statute.39

The district court then examined the “text, structure, and legislative history of the ESA” and determined that “the statute is silent or ambiguous with respect to the specific issue” in the case: whether an endangered species must be in danger of imminent extinction.40 And recall what the court had just said about Chevron: if the statute is “silent or ambiguous” on the specific issue, then the court proceeds to the second step and determines whether the agency’s interpretation of the statute is “permissible.” One might assume, then, that the district court would proceed to the second step of the Chevron analysis. But that assumption would be incorrect.

The district court did not reach Chevron Step Two. (Or, at least, it did not reach Chevron Step Two for another year.) Instead, it said:

Upon finding the definition of an endangered species to be ambiguous, the Court would normally be required to defer to any permissible agency construction of the statute under step two of the Chevron analysis. In this case, however, there is no permissible construction to which the Court can defer. . . . [The Fish and Wildlife Service] relies exclusively on a plain-meaning interpretation of the ESA. As Chevron step 2 deference is reserved for those instances when an agency recognizes that the Congress’s intent is not plain from the statute’s face, this Court is precluded from according the agency’s interpretation deference under Chevron.41

The district court’s key analytical move was to say that deference at Chevron Step Two is reserved for instances in which the agency “recognizes” the ambiguity in the statute. Significantly, the district court cited no Supreme Court precedent for this proposition.42 Instead, it relied on a DC Circuit precedent, Peter Pan Bus Lines, Inc v Federal Motor Carrier Safety Administration,43 which itself invoked a line of DC Circuit cases dating back to the 1985 case Prill.44 The district court then remanded the case back to the Fish and Wildlife Service for the agency to decide whether to adopt the same imminence requirement as a matter of its own discretion.45

The rest of the story is perhaps predictable. Following the district court’s remand order, the Fish and Wildlife Service submitted a memorandum of “supplemental explanation” stating that even if the statutory definition of endangered is ambiguous, it still hewed to the view that the classification should be limited to species facing an immediate risk of extinction.46 In a section titled “The Policies and Purposes of the ESA,” the Service explained that “species currently on the brink of extinction . . . generally need stringent protection,” which the endangered classification provides.47 But

[f]or species not yet on the brink of extinction, particularly for those that have yet to experience any notable decline in numbers or range, [the “threatened” classification] offers the flexibility to fashion restrictions according to the needs of the species, which reflects the generally longer time frames available to test differing conservation strategies.48

The case went back to the district court, which concluded that “the agency’s Supplemental Explanation sufficiently demonstrates that the Service’s definition of an endangered species, as applied to the polar bear, represents a permissible construction of the ESA and must be upheld under step two of the Chevron framework.”49 The DC Circuit upheld the district court’s ruling,50 and the polar bear is still listed as threatened but not endangered.51

While the outcome of the polar bear post-remand saga may not be surprising, much else about the story is puzzling. For one: Why did the Chevron Step One-and-a-Half question arise in the first place? Why didn’t the Fish and Wildlife Service insert a disclaimer in its listing rule along the following lines: “We think the plain meaning of ‘endangered’ is that the danger of extinction must be imminent, but even if we are wrong on that score and the statute is ambiguous, we would arrive at the same result in the exercise of our discretion”? This sort of disclaimer does not appear to be uncommon,52 and yet the Service omitted it here. Why? It is easy to understand how a Step One-and-a-Half issue might arise if no one had ever thought about this issue before, but the DC Circuit has been applying this doctrine for three decades. Prill, after all, was decided in 1985 and has been applied many times in the intervening years.53 Wouldn’t one expect that agencies, after decades of litigation experience, would get the message and start to insert disclaimers of this sort in their rules as a matter of course?

And on a normative note: What use does Chevron Step One-and-a-Half serve? In the polar bear case, did the district court really anticipate that, on remand, the Fish and Wildlife Service would say: “Gee, now that we know that the statute is ambiguous, we think that the term ‘endangered’ should apply to risks far off in the future and polar bears should be listed as ‘endangered’ going forward”? If the Service had wanted to list polar bears as “endangered,” presumably it would have found in the statute the ambiguity so apparent to the district court. Agency officials and lawyers, after all, are pretty good at searching for ambiguity and, indeed, are often willing to press aggressive arguments in favor of it.54 So the district court surely suspected that the statutory interpretation adopted by the agency also reflected the agency’s view of the best policy. Did it warrant an extra round of litigation just so that the Fish and Wildlife Service would say so in more explicit terms?

Especially in light of this example, we understand why some readers might conclude that Chevron Step One-and-a-Half is nonsense twice over—nonsense in that no sensible agency should ever find itself ensnared by the doctrine, and nonsense in that the doctrine itself accomplishes absolutely nothing. And yet we resist those conclusions. In Part III, we explain why a rational agency (or rational actors within the agency) might deliberately choose to argue that a statute is unambiguous without attaching any disclaimer (that is, without arguing in the alternative that even if the statute is ambiguous, the agency’s construction is permissible). And in Part IV, we explain why in our view Chevron Step One-and-a-Half can produce benefits even if, in the mine-run of cases, a remand to the agency will result in the agency spitting back the same rule with cosmetic changes to the preamble. But before doing so, we seek to situate Chevron Step One-and-a-Half in its doctrinal context.

II. Doctrinal Antecedents of Chevron Step One-and-a-Half

To appreciate Chevron Step One-and-a-Half, it is necessary to understand three administrative law doctrines (doctrines no doubt familiar to many readers): Chevron, Securities and Exchange Commission v Chenery Corp55 (“Chenery I”), and Motor Vehicle Manufacturers Association of the United States, Inc v State Farm Mutual Automobile Insurance Co.56 We argue that Chevron Step One-and-a-Half is consistent with each of these doctrines but not necessarily dictated by any one of them.

A. Chevron

Chevron U.S.A. Inc v Natural Resources Defense Council, Inc is by most measures the most frequently cited case in administrative law.57 It is certainly the most familiar to administrative law students.58 And at first blush, it is among the most straightforward. Chevron calls on courts to apply a two-step framework when reviewing an agency’s interpretation of a statute that it administers. The Court in Chevron articulated the two steps as follows:

When a court reviews an agency’s construction of the statute which it administers, it is confronted with two questions. First, always, is the question whether Congress has directly spoken to the precise question at issue. If the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress. If, however, the court determines Congress has not directly addressed the precise question at issue, the court does not simply impose its own construction on the statute, as would be necessary in the absence of an administrative interpretation. Rather, if the statute is silent or ambiguous with respect to the specific issue, the question for the court is whether the agency’s answer is based on a permissible construction of the statute.59

The Court—per Justice Stevens—attempted to justify this deference in at least three ways: political accountability, comparative expertise, and implied delegation, with the first two being the most important in Stevens’s analysis. As to accountability, Stevens reasoned that when a statute is ambiguous, it is better that a politically accountable agency rather than a politically isolated court determine what the statute means.60 At the same time (and, indeed, in the same breath), Stevens emphasized the agency’s expertise relative to judges, who “are not experts in the field.”61 Finally, Stevens (tentatively) advanced the theory that ambiguity constitutes an “implicit” delegation from Congress to the relevant agency to “fill [the] gap” in the statutory framework.62 The upshot of the Chevron Court’s analysis is a “counter-Marbury” rule of interpretation: rather than saying “what the law is,”63 courts confronted with ambiguous language should defer to the “technical expertise and political accountability” of the agency charged with administering the statute.64

While Stevens framed the Chevron inquiry in two-step terms, several scholars have argued that there is a preliminary step implicit in the Chevron analysis, at which the court decides whether to apply the Chevron framework at all. Professors Merrill and Hickman have called this “Chevron’s ‘step zero.’”65 Although Step Zero is complicated (indeed, it may have substeps of its own, hence the suggestion that Chevron has four steps66 ), the gist of this idea is that if Chevron is based (even in part) on a theory of implied delegation, then it is reasonable to think that there are some categories of interpretations for which Congress would not want the agency calling the shots. For example, a court may decline to defer at Step Zero because, as in United States v Mead Corp,67 the procedures used by the agency were so lacking in formality that it is improbable to think Congress intended the resulting interpretations to receive deference.68 Or a court may deny deference at Step Zero because, as in Food and Drug Administration v Brown & Williamson Tobacco Corp,69 the policy is so important that it is unlikely that Congress wanted it resolved by an agency.70 The latter version of Step Zero recently received a boost from Chief Justice Roberts in King v Burwell,71 the 2015 case about whether Affordable Care Act72 tax credits should be available to individuals who enroll in insurance plans through federal rather than state-run exchanges.73 Writing for a six-justice majority, the chief justice declined to defer to the Internal Revenue Service’s interpretation of the statute, reasoning that the matter was of such “deep ‘economic and political significance’” and so “central to this statutory scheme” that “had Congress wished to assign that question to an agency, it surely would have done so expressly.”74

One can see how Chevron Step Zero might be derived from the Court’s decision in Chevron. It is not immediately obvious, though, that Chevron Step One-and-a-Half follows from Chevron itself. Stevens’s opinion in Chevron never suggests that deference depends on whether an agency recognizes ambiguity in the statute.75 Chevron simply says that when Congress has delegated interpretive authority to an agency, “a court may not substitute its own construction of a statutory provision for a reasonable interpretation made by the administrator of an agency.”76 Nowhere does the Court even hint that the administrator of an agency must recognize that her “reasonable” interpretation is not the only permissible interpretation. While we argue below that Step One-and-a-Half is consistent with the core purposes of Chevron, it would be difficult to argue that Step One-and-a-Half is prescribed by Chevron.

But even if Chevron Step One-and-a-Half cannot be derived directly from Chevron, new insights about Chevron can indeed be derived from Chevron Step One-and-a-Half. First, consider the claim by Professors Stephenson and Vermeule—noted above—that Chevron has only one step.77 Stephenson and Vermeule assert that the “single question” in every Chevron case is “whether the agency’s construction is permissible as a matter of statutory interpretation.”78 For at least a fleeting moment, the Supreme Court seemed in accord with Stephenson and Vermeule’s one-step formulation.79 More recent Supreme Court opinions, however, describe Chevron in two-step terms.80 What Chevron Step One-and-a-Half teaches us is that, at least in any court that applies the doctrine, the difference between Step One and Step Two very much matters. A court may believe that an agency’s interpretation of a statute is permissible and yet still flunk the agency on Chevron Step One-and-a-Half grounds because the agency failed to describe its interpretation in Step Two terms. In this respect, Chevron does indeed have (at least) two steps—and any agency that disagrees will have its hat handed to it in the DC Circuit.

Not only does the existence of Chevron Step One-and-a-Half demonstrate that Stephenson and Vermeule’s consolidation of the Chevron inquiry fails to fit with DC Circuit doctrine, but it also offers direction for the order in which the inquiry should occur. Professor Richard Re has argued that courts should have the option of proceeding with the Chevron inquiry in reverse: “[O]ptional two-step Chevron would first ask the reasonableness question, and then it would give courts discretion to ask a second question regarding mandatoriness.”81 Re further argued that the optional two-step formulation of Chevron is consistent with current practice in the federal courts of appeals.82 Chevron Step One-and-a-Half suggests that, at least in the DC Circuit, courts do not have the option of resolving Chevron cases on Step Two grounds alone—at least not in cases in which the court ultimately upholds the agency’s interpretation. If a petitioner or plaintiff seeking review of an agency rule brings a Chevron Step One-and-a-Half challenge in the DC Circuit or DC federal district court, the appellate panel or district judge does not have the option of holding that the agency’s interpretation is reasonable and calling it a day. To respond to the Chevron Step One-and-a-Half challenge, the panel would have to determine whether the agency acknowledged ambiguity in the statute; if not, then the panel or district judge would have to determine whether the agency’s characterization of the statute as unambiguous is correct. To be sure, an agency that jumps straight to Step Two without expressing any opinion on the statute’s ambiguity (or lack thereof) might survive Chevron Step One-and-a-Half (more on this in Part V.A). But a court that jumps straight to Step Two in a case in which a Step One-and-a-Half challenge has been raised would be acting inconsistently with DC Circuit doctrine.

B. Chenery I

Four decades Chevron’s senior, the Chenery I rule is another “hoary principle of administrative law.”83 And like Chevron, the Chenery I rule is easy to state but not always to apply: when a court reviews an agency decision, “[t]he grounds upon which an administrative order must be judged are those upon which the record discloses that its action was based.”84 In other words, if the agency proffers a bad reason for its decision, a court will not uphold the decision just because the agency could have offered a good reason. In fact, this rule applies even if the agency’s own lawyers later offer that very good reason to the court during judicial review.85 Instead, so long as the agency has any discretion in the matter (for there is no reason to remand otherwise86 ), what counts is what the agency said at the time of its decision.

The facts of Chenery I provide a good example of how this works. In Chenery I, a company proposed a reorganization. The Securities and Exchange Commission agreed to allow the reorganization, “but also ordered that ‘preferred stock’ acquired by certain ‘officers, directors, and controlling stockholders’ while reorganization plans were before the Commission could not ‘participate in the reorganization on an equal footing with all other preferred stock.’”87 The SEC justified its decision by invoking common-law principles as articulated by courts. The Supreme Court concluded, however, that the common-law principles relied on by the SEC were inapplicable to this situation—the agency misunderstood the common law.88 Instead of deciding whether the agency’s decision could be sustained on other grounds, the Court remanded to the agency.89 (Much like in the polar bear litigation, the agency responded on remand with new justifications for its old position, and the Supreme Court upheld the agency’s decision the second time around.)90

The Chenery I principle is often invoked in modern administrative law (though not quite as often invoked as Chevron).91 It played a prominent role in Michigan v Environmental Protection Agency,92 a 2015 decision in which the Supreme Court rejected EPA’s claim that it could regulate mercury emissions from coal power plants.93 The DC Circuit likewise invokes Chenery I on a regular basis when an agency puts forth a new rationale for an action in litigation that the agency did not cite in the first instance.94 To be sure, as noted above, Chenery I does not require a remand when the agency’s decision is the only one the law allows, and sometimes (though rarely) an agency may offer a post hoc justification.95 But as a general rule, Chenery I requires an agency’s decision to clear two hurdles: it must be lawful and the reason why it is lawful must be the one given by the agency itself at the time it announced the decision. Otherwise, the decision cannot stand, even if the same outcome could be permissible.

On one view, Chevron Step One-and-a-Half is simply an application of Chenery I.96 If an agency initially says that it is adopting a particular position because it believes that the statute so requires, a court cannot uphold the agency’s action on the ground that the statute is ambiguous but the agency’s interpretation is permissible. The agency, after all, gave the “wrong” reason for its decision.

Yet the path from Chenery I to Chevron Step One-and-a-Half may not be quite that straightforward, depending on how broadly one reads Chenery I. Consider again the polar bear example. Imagine two potential arguments that the Fish and Wildlife Service could make in the preamble to its listing rule:

1. Congress clearly intended for the term “endangered” to include an imminence requirement. The legislative history says that Congress sought to establish “two levels of protection”—“endangered” and “threatened”—with the two levels distinguished by whether the danger of extinction is “imminent.”97
2. The text of the ESA is ambiguous as to whether “endangered” includes an imminence requirement. Nonetheless, we believe that such a requirement best carries out congressional intent. The legislative history says that Congress sought to establish “two levels of protection”—“endangered” and “threatened”—with the two levels distinguished by whether the danger of extinction is “imminent.”98

Assume, for present purposes, that the court considers legislative history to be a relevant consideration, at least at Chevron Step Two.99 Would there really be a Chenery I problem with upholding the agency’s listing decision, whether the agency’s proffered rationale is (a) or (b)? In both scenarios, the agency has cited the legislative history as a reason for its decision, and so Chenery I (arguably100 ) allows the court to rely on that reason on review. And yet in scenario (a), the agency might well run into a Chevron Step One-and-a-Half problem. Unlike in scenario (b), the agency in scenario (a) does not acknowledge ambiguity. Quite the contrary, the agency maintains that congressional intent is clear. To be sure, the agency relies on legislative history rather than statutory text, but the DC Circuit has said under similar circumstances that if an agency does not acknowledge ambiguity, then it is undeserving of Chevron deference.101

To put the point in slightly different terms, imagine an agency says, “The statute unambiguously means X for reasons A, B, and C.” Chenery I clearly prevents the agency from pivoting in litigation and arguing that the statute should be interpreted to mean X for reasons D, E, and F. It is not so obvious, though, that Chenery I stands in the way of a court saying: “We believe that the statute is ambiguous, but reasons A, B, and C are all perfectly good reasons for interpreting the statute to mean X, and so we affirm the agency.” That latter disposition may or may not be consistent with Chenery I, depending on how broadly one reads Chenery I. But no matter how one reads Chenery I, the latter justification would definitely run afoul of Chevron Step One-and-a-Half.

C. State Farm

If Chevron Step One-and-a-Half is not implicit in Chevron and is not (necessarily) required by Chenery I, might it be derived from the State Farm doctrine? State Farm holds that an agency must “articulate a satisfactory explanation for its action”—it must show, in other words, that its action was “the product of reasoned decisionmaking.”102 Perhaps we can say that an agency flunks the “satisfactory explanation” requirement—and therefore regulates unreasonably—when it says that the statute is unambiguous even though the statute is susceptible to multiple meanings.

Maybe—but we are doubtful. The fact that an agency thinks Congress’s intent is clear while the court thinks the statute is ambiguous does not mean that the agency’s view is unreasonable. As we allude to above, some courts consider legislative history at Chevron Step One: on this view, legislative history can render a statute unambiguous under certain circumstances.103 Others consider legislative history only at Step Two.104 An agency and a court might disagree as to whether to resolve a statutory interpretation question at Step One or Step Two, but that does not mean that either holds an unreasonable view, much less that the agency has employed a faulty process or ignored key parts of the problem.

More generally, one can conceptualize statutory clarity as a spectrum (say, from zero to one, with zero being totally ambiguous and one being completely clear).105 The statute specifying that presidential elections must be held “on the Tuesday next after the first Monday in November”106 might qualify as a 0.99 on the scale, while the Federal Communications Commission’s power over assignment of station licenses (such assignments are allowed if the Commission determines “the public interest, convenience, and necessity will be served”107 ) might land at 0.01. We might further imagine that any statutory provision 0.5 or higher is “unambiguous” for Chevron purposes (that is, the case will be resolved at Step One), and any statutory provision below 0.5 on the clarity scale is “ambiguous” (that is, the case can proceed to Step Two). We might also imagine that intelligent interpreters acting in good faith will sometimes differ in their assessments, at least over a limited range. For instance, one might rate a provision a 0.3 on the clarity scale while another equally intelligent, equally well-intentioned interpreter might rate the same provision a 0.2 or a 0.4 (though probably not a 0.9).

From this perspective, an agency’s statement that a statutory provision is unambiguous may sometimes show that the agency’s decision is not the “product of reasoned decisionmaking,”108 but not always. It might be unreasonable for the FCC to say its license assignment power is unambiguous. But an agency might reasonably conclude that a statutory provision is a 0.51 while the court sees it as a 0.49. In the latter scenario, the agency would have a Chevron Step One-and-a-Half problem (assuming it does not insert the requisite disclaimer), but there would be no obvious State Farm issue. To be sure, we are not saying that 0.5 is the right threshold, nor that such fine distinctions are necessarily possible. The important point is that an agency’s decision can be the product of “reasoned decisionmaking,” the hallmark of State Farm, yet still fail Chevron Step-One-and-a-Half.

Granted, there are affinities between Chevron Step-One-and-a-Half and State Farm: both doctrines look to the content of an agency’s explanation for adopting a particular position. But not every case in which the agency runs into a Chevron Step-One-and-a-Half problem is also a case in which the agency would flunk State Farm review (and, of course, vice versa).

D. Prill

Despite not being compelled by Chevron, Chenery I, or State Farm, Chevron Step-One-and-a-Half is a firmly entrenched principle of DC Circuit law. Its origin is traceable at least as far back as the DC Circuit’s 1985 decision in Prill, which announced the following rule:

[A]n agency regulation must be declared invalid, even though the agency might be able to adopt the regulation in the exercise of its discretion, if it was not based on the agency’s own judgment but rather on the unjustified assumption that it was Congress’ judgment that such a regulation is desirable.109

In operation, this means the court will “strike[ ] down an agency interpretation, not because the interpretation was inconsistent with the statute, but because the agency wrongly assumed that a particular interpretation was commanded by the statute.”110 Thus, “[d]eference to an agency’s statutory interpretation ‘is only appropriate when the agency has exercised its own judgment,’ not when it believes that interpretation is compelled by Congress.”111

The troubling facts of Prill illustrate the doctrine. A truck driver named Kenneth Prill was fired for complaining “about the unsafe condition of a company truck and trailer.”112 Even though the truck was unsafe (indeed, state officials agreed with him), Prill was let go “because company officials decided that they could not have him ‘calling the cops all the time.’”113 Upset, Prill sought relief from the National Labor Relations Board (NLRB), alleging that his firing was an unfair labor practice.114 Relevant here, an administrative law judge agreed that Prill’s safety complaint was protected under § 7 of the National Labor Relations Act115 (NLRA) as a “concerted activit[y],”116 but the NLRB disagreed because, on its view of the NLRA, “an employee’s conduct is not ‘concerted’ unless it is ‘engaged in with or on the authority of other employees, and not solely by and on behalf of the employee himself.’”117

This issue then came to the DC Circuit, which concluded that the NLRB was wrong to say that it lacked discretion to consider Prill’s activity “concerted.” The agency, however, urged that its decision rejecting Prill’s position should nonetheless be upheld because the NLRB has “broad authority to construe the NLRA.”118 The DC Circuit would have none of it, explaining that “judicial deference is not accorded a decision of the NLRB when the Board acts pursuant to an erroneous view of law and, as a consequence, fails to exercise the discretion delegated to it by Congress.”119 Because the NLRB was wrong to conclude that its interpretation of the statute was “mandated” by the statute,120 the court remanded “under the principles of [Chenery I], so that the Board may reconsider the scope of ‘concerted activities’ under section 7.”121 Because the agency’s mistake might have had some “bearing on the procedure used or the substance of decision reached,” the NLRB’s decision could not stand.122

Prill, which was argued and decided just months after Chevron, never cites Stevens’s now-canonical opinion; in subsequent cases, however, the DC Circuit has made the link between Chevron and Prill clearer. An illustrative example is Peter Pan Bus Lines, Inc v Federal Motor Carrier Safety Administration. Peter Pan concerned two competing budget bus lines: Peter Pan Bus Lines and Fung Wah Transportation, Inc (known to a generation of East Coast college students as “the Chinatown bus”). Fung Wah filed an application with the Federal Motor Carrier Safety Administration (FMCSA) for a license to operate a passenger bus line between New York and Boston.123 Peter Pan objected to Fung Wah’s application because, Peter Pan said, “Fung Wah was unwilling or unable to comply with the requirements of the regulations [the Department of Transportation] has promulgated under the Americans With Disabilities Act.”124 The agency, however, concluded that the licensing statute did not allow it to consider Fung Wah’s alleged Americans with Disabilities Act125 (ADA) violations as part of the licensing proceeding.126

As in Prill, the DC Circuit disagreed with the agency’s view of the statute.127 In the court’s view, the licensing statute did not unambiguously preclude the FMCSA from considering the ADA violations as part of the licensing proceeding. As Judge Karen LeCraft Henderson wrote for the panel:

Under the Chevron two-step, we stop the music at step one if the Congress has directly spoken to the precise question at issue. . . . But if the statute is silent or ambiguous, we dance on and, at step two, defer to the Commission’s interpretation if it is based on a permissible construction of the statute. . . . In rejecting Peter Pan’s argument that [the licensing statute allowed for consideration of alleged ADA violations], the FMCSA unequivocally declared: “This interpretation is not consistent with the plain language of the statute and the legislative history.” . . . To the contrary, we find the text of the statute to be ambiguous. . . . We therefore cannot uphold the FMCSA’s interpretation under step 1 of Chevron. Nor may we review it under step 2. . . . We must therefore remand for the FMCSA to interpret the statutory language anew.128

Peter Pan may be the clearest statement of how the Chevron Step-One-and-a-Half doctrine fits into the broader Chevron framework. In terms of its outcome, however, Peter Pan is not an outlier. To the contrary, the DC Circuit regularly invokes Prill and its progeny in Chevron cases—or at least relatively regularly. As Professor Bagley has noted, “[e]xamples abound” of the DC Circuit applying (and, in Bagley’s view, misapplying) the Prill framework.129

Other courts have also recognized Chevron Step-One-and-a-Half—including arguably the US Supreme Court. In Negusie v Holder, the justices confronted the “persecutor bar” in immigration law, which precludes anyone from receiving refugee status who has “ordered, incited, assisted, or otherwise participated in the persecution of any person on account of race, religion, nationality, membership in a particular social group, or political opinion.”130 The Board of Immigration Appeals (BIA), following what it believed judicial precedent to require, concluded “that the persecutor bar applies even if the alien’s assistance in persecution was coerced or otherwise the product of duress.”131 Although agreeing that the BIA would be eligible for Chevron deference on this question in the ordinary course, the Supreme Court did not defer to the agency’s interpretation because, contrary to the agency’s claim, the question was not foreclosed by judicial precedent.132 Before the Supreme Court, lawyers for the BIA argued that even if the agency’s interpretation was not mandated, it was at least “reasonable and thus controlling.”133 Justice Anthony Kennedy, however, writing for the Court, rejected the applicability of Chevron, using language reminiscent of Chevron Step-One-and-a-Half: “The BIA deemed its interpretation to be mandated[,] . . . and that error prevented it from a full consideration of the statutory question here presented.”134 With a favorable citation to the DC Circuit, the Court remanded to “[t]he agency [to] confront the same question free of this mistaken legal premise.”135

Negusie goes a fair bit of the way toward establishing Chevron Step-One-and-a-Half as the law of the (entire) land. But it does not go all of the way. For one thing, the context of Negusie was peculiar. The agency did not engage in statutory interpretation as much as it engaged in judicial-decision interpretation. That may present a different type of analysis (whether it should is a question for another day). Likewise, Negusie’s analysis is quite cursory. If the Court was, in fact, adopting Chevron Step-One-and-a-Half, it did not really explain why it was making such an important doctrinal move, much less did it explain the scope of the doctrine. Given this cursory analysis, it is hardly surprising that lower courts do not appear to have taken the justices’ hint, if in fact the justices were giving such a hint.136

Indeed, Chevron Step-One-and-a-Half has not been adopted everywhere—even after Negusie. For instance, it appears that neither the Second nor the Third Circuit has “yet addressed the question of whether [Chevron] deference is only appropriate when the agency has exercised its own judgment, not when it believes that its interpretation is compelled by Congress.”137 Before Negusie, the Sixth Circuit deferred to the Department of Agriculture at Chevron Step Two even while holding that the Department had incorrectly concluded that its interpretation was compelled by Congress.138 The Seventh Circuit seemed to take a different approach before Negusie: after holding that the BIA had incorrectly concluded it lacked discretion, the Seventh Circuit did not remand the case to the BIA but instead decided the issue itself de novo.139 Post-Negusie, the Seventh Circuit has not decided whether this remains good law. The Ninth Circuit appeared to adopt the doctrine in a 2013 case, but has not applied it since.140 And the Tenth and Federal Circuits have cited DC Circuit case law to the effect that an agency is ineligible for deference at Chevron Step Two if it believes that the statute is clear, though those courts have yet to remand a case on Step One-and-a-Half grounds.141

III. The Puzzle of Chevron Step One-and-a-Half

In the previous Part, we discussed the murky doctrinal origins of Chevron Step One-and-a-Half. An even more perplexing puzzle is why these cases continue to crop up. The first few Chevron Step One-and-a-Half cases are easy enough to account for: in Prill and other early cases, the agency might not have known what was coming. But it is not 1985 anymore. Over time, we should expect agencies to become accustomed to Step One-and-a-Half and to write their rules and orders to avoid being tripped up by the doctrine. After all, sometimes it may not take much effort to insulate a decision from any Chevron Step One-a-and-Half problem—a disclaimer combined with some analysis of why the agency’s interpretation is good policy may be enough.142

And yet far from falling into desuetude, Chevron Step One-and-a-Half remains a vital doctrine—at least in the DC Circuit and the DC federal district court (with rarer appearances beyond143 ). What might account for the continued appearance of Chevron Step One-and-a-Half cases more than thirty years after Prill? Why haven’t agencies adapted by writing their rules with an eye toward sidestepping any potential Chevron Step One-and-a-Half issues?144

To be sure, some have. Sometimes an agency will argue that its reading is compelled by the statute while adding—in the alternative—that it would reach the same conclusion as a matter of discretion.145 Yet this practice, while apparently not uncommon, is not universal. And while the costs of drafting such a disclaimer are nonzero (especially because an agency may still run into a State Farm problem if it fails to justify its policy choice), an agency can circumvent Chevron Step One-and-a-Half as long as it clearly states that it would adopt the same policy even if it had another option.

Setting the drafting costs aside, why else might an agency say that its decision is compelled by Congress, without also saying that it would reach the same result as an exercise of discretion? We suggest six hypotheses.146

A. Agency Ignorance

Perhaps the likeliest answer is that, more than thirty years after Prill, many agency lawyers remain unaware of the Chevron Step One-and-a-Half doctrine. They probably would not have learned about it in law school, even if they graduated after the doctrine emerged in the mid-1980s. After all, no mention of Prill appears in leading administrative law casebooks.147 While survey evidence indicates that agency drafters are generally aware of Chevron, their knowledge of other administrative law doctrines is not nearly as widespread (for example, 39 percent of drafters surveyed said they were unaware of Mead, and 47 percent said they were unaware of Bowles v Seminole Rock & Sand Co148  / Auer v Robbins149 ).150 We have no similar data with respect to Chevron Step One-and-a-Half, but we expect that in at least some cases, the agency lawyers drafting rules and orders are simply unaware of the doctrine. Ignorance, however, is a less likely explanation for agencies that appear regularly before the DC Circuit.

B. Agency Ambivalence

Another possibility is that agency leaders—or the lawyers who work under them—genuinely believe that the relevant statutory provision is unambiguous but are less than enthusiastic about that result as a policy matter. To return to the polar bear example, Fish and Wildlife Service officials might believe that the term “endangered” implies an imminent risk of extinction, but the officials might wish to list the polar bear as an endangered species if they could. Alternatively, agency officials might believe that the term “endangered” implies imminence but might be uncertain as to how they would resolve the issue if freed from statutory shackles. In either scenario, agency officials might decide to argue that the term “endangered” unambiguously means imminent risk of extinction, and if a reviewing court decides otherwise, then they will reverse (or at least reconsider) their position if and when they have to.

The first version of the “ambivalence” story—agency officials believe the statute unambiguously means X but wish they could do Y—strikes us as a strained explanation for the continued appearance of Chevron Step One-and-a-Half cases. Let’s say that the statutory provision either (i) unambiguously means X or (ii) is ambiguous and could mean X or Y. Let’s also say that agency officials would prefer result Y. If there is a credible argument that the statute is ambiguous, then we would expect agency officials who desire result Y to make that argument. For this first version of the ambivalence story to be true, there would have to be an argument as to why the statutory provision is ambiguous that the court recognizes but the agency officials do not. This is not impossible—agency officials sometimes overlook viable arguments in their favor—but we have trouble believing that this a frequent reason for Chevron Step One-and-a-Half remands. In a world in which agency officials are able to spot “ambiguities” that courts reject as nonexistent or borderline frivolous,151 it is more than a bit strange to think that agency officials are unable to spot ambiguities that support the agency’s preferred policy and that the reviewing court can nonetheless identify.

Note, moreover, that in the first scenario—agency officials believe that the statute unambiguously means X but wish it were not so—the agency has an option other than simply failing to argue in the alternative. The agency can say, “We think the statute unambiguously means X, but we think that Y is the better policy.” Indeed, one can find quite a few examples in the Federal Register of agencies taking this tack.152 This is similar to what judges do when they criticize a law but state that they nonetheless must apply it.153 Agencies, to be sure, sometimes may be reluctant to play the blame game for fear of irritating Congress. But if the agency feels strongly that it is being forced to act against the public interest, it can say so. It is perhaps telling that we can find no Chevron Step One-and-a-Half case in which an agency says that it is acting contrary to its view of the public interest because it believes Congress has mandated that result and a reviewing court then concludes that the statute is in fact ambiguous.

The second version of the “ambivalence” story—agency officials believe that the statute unambiguously means X but have yet to decide whether they would choose X or Y if freed from statutory constraints—seems more plausible. This is especially true once we remember that “agencies are a ‘they,’ not an ‘it.’”154 For example, the agency’s general counsel might take the view that the statute unambiguously means X, while other offices (and officers) within the agency are divided as to whether X or Y is the better policy. Under these circumstances, the path of least resistance might be for the agency to say, “We think the statute unambiguously means X,” and then if the court disagrees, competing constituencies within the agency can duke it out over whether the agency should take position X or position Y. In other words, an agency’s failure to argue in the alternative may be part of an internal conflict-avoidance strategy.155

C. Intra-agency Politics

A third reason why Chevron Step One-and-a-Half cases continue to crop up may be that agency general counsels are using statutory interpretation to enhance their own power and the power of their patrons. According to former EPA general counsel E. Donald Elliott, lawyers in the general counsel’s office at an agency can give advice on statutory meaning to other agency officials in one of two forms: (a) lawyers can give “legal advice as a point estimate, e.g., ‘the statute means this,’” or (b) lawyers can describe the “policy space” that the statute allows—that is, the “range of permissible interpretive discretion, within which a variety of decisions that the agency might make would be legally defensible to varying degrees.”156 Option (a) means much more power for the general counsel, who then gets to define the agency’s position. Even if she selects option (a), the general counsel’s word is not necessarily the last, but other agency actors are much more likely to defer to the general counsel’s position if she says, “You must do X or we will lose in court,” than if she says, “You can do X or Y, and I think X is the better choice.”157

To be sure, the general counsel could try to thread the needle by telling other agency officials: “We must do X or else we will lose in court, but it would help us win in court if we say in our preamble, ‘We would do X even if we had the option to do Y.’” But such a tack would, we assume, raise eyebrows, especially if some officials in the agency most assuredly would not choose X if Y were an option and do not see why they should have to say otherwise in support of a policy with which they disagree. The larger point, moreover, is that preambles are written for multiple audiences—and not just for courts. One of those other audiences comprises agency officials beyond the general counsel’s office. A general counsel familiar with Chevron Step One-and-a-Half might decide that the added litigation risk from failing to argue in the alternative is worth it in order to maintain control of policy-making within the agency. In other words, agencies are “coalitions of diverse participants who have somewhat different motives” and who sometimes work at cross-purposes.158

Dynamics of this sort might be particularly likely in an agency at which the preferences of the general counsel diverge from other actors’. At the Equal Employment Opportunity Commission (EEOC), for example, the president appoints the general counsel, but some of the commissioners may have been appointed by a prior president or be from a different political party.159 After the White House changes partisan hands, the general counsel may be an appointee of the new president while a majority of the Commission is composed of members of the other party. The general counsel may have an incentive to say, “The statute unambiguously means X,” while knowing full well that a good chunk of the Commission would prefer position Y. Of course, the commissioners of the opposing party may challenge the general counsel’s statutory interpretation, but the general counsel may decide that her best bet of winning the X-versus-Y fight is to say that the statute requires outcome X.160 And note that the relevant actor need not be the general counsel: at agencies in which other offices wield control over the regulation-writing process, staffers in those offices may state that an interpretation is statutorily compelled as a way to exert influence over the final product.161

D. Intra–Executive Branch Politics

Just as agencies are a “they,” not an “it,” so too for the executive branch more broadly.162 And these different parts of the executive branch may have different missions and goals.163 In some cases, these intra–executive branch conflicts may lay the foundation for Chevron Step One-and-a-Half cases. For instance, an agency that foresees disagreement with OIRA as to a particular policy may have incentives to argue that the agency’s preference is compelled by Chevron Step One, and so it cannot do what OIRA would prefer. (Of course, the White House also is a “they,” not an “it,” so the real world is even more complicated than this discussion lets on.164 For purposes of our analysis here, however, it is sufficient to lump the White House and OIRA together.)

For readers unfamiliar with the process of presidential review, a bit of background may be helpful.165 In recent decades, the White House has attempted to assert greater control over the regulatory efforts of agencies.166 One of the key ways it has done so is through a series of executive orders vesting OIRA with the responsibility for reviewing agency actions. Among the first things that President Ronald Reagan did in office was issue Executive Order 12291, which required agencies to submit proposed and final rules to the Office of Management and Budget (wherein OIRA is housed).167 The order also required agencies to prepare a regulatory impact analysis for “major rules,” assessing the benefits and costs of the contemplated action.168 Thereafter, later presidents issued a number of other additional executive orders on the subject—including President Bill Clinton’s Executive Order 12866,169 which retained “the most important features of President Reagan’s oversight system” and further formalized OIRA’s regulatory review role,170 as well as President Barack Obama’s Executive Order 13563, which “reaffirm[ed]” and supplemented Clinton’s order.171 While President Donald Trump has issued a number of new executive orders addressing the regulatory process,172 he has thus far left the OIRA review infrastructure established by his predecessors in place.173

Relevant to Chevron Step One-and-a-Half, Executive Order 13563 declares that “to the extent permitted by law, each agency must . . . select, in choosing among alternative regulatory approaches, those approaches that maximize net benefits.”174 The italicized language is important. If an agency concludes that approach Y would maximize net benefits but the statute allows only approach X, then Executive Order 13563 permits the agency to proceed with approach X.175 If the agency concludes, though, that the statute allows either X or Y, then Executive Order 13563 generally requires the agency to explain why X maximizes net benefits or else to choose Y. Thus, if an agency prefers X, it potentially has an incentive to say that the statute unambiguously requires X.

To see how the strictures of Executive Order 13563 might give rise to Chevron Step One-and-a-Half cases, imagine the following scenario: EPA prefers approach X over approach Y but strongly suspects that the OIRA administrator thinks that approach Y maximizes net benefits. (These sorts of disagreements are not uncommon, especially between EPA and the White House.)176 EPA can either (i) try to convince the OIRA administrator that the statute requires approach X or (ii) try to convince the OIRA administrator that in fact X is the benefit-maximizing course. EPA may decide that (i) is the more promising strategy—perhaps because the OIRA administrator’s views on cost-benefit analysis are more fixed than her views on statutory interpretation.177 Accordingly, EPA may represent to OIRA that it is choosing X because the statute requires that result, even though in reality the agency believes the statute is ambiguous.

OIRA may still disagree with EPA’s statutory interpretation and instruct the agency to adopt approach Y. If so, there may not be a Chevron Step-One-and-a-Half scenario by the time the issue comes to court. And in theory, EPA could tell OIRA that the statute requires X while also inserting a statement into its preamble that the agency would choose X even if the statute were ambiguous. But we imagine that the latter course once more would be difficult to pass by OIRA—especially if OIRA most certainly would not approve of X if it had a choice. Again, the agency’s challenge is that it has to write a preamble for multiple audiences. What the agency needs to say to get past OIRA review may not be the same as what it would say if it were strictly maximizing its chances of winning in court.

E. Interbranch Politics

Agency leaders, like anyone else,178 also presumably like to take credit for popular policies and avoid being blamed for unpopular ones. This may be true for personal reasons (the agency head cares about her own legacy and post-office career prospects), institutional reasons (the agency head does not want her institution to be dragged through the mud in the media or in congressional hearings), or partisan reasons (the agency head wants her actions to reflect well on the president who appointed her). When an agency head feels that, for whatever reason, a policy is the best course even though it is unpopular, she might say, “Don’t look at us; we’re just doing what Congress required.” That is, an agency may announce in a preamble to a regulation that “the statute unambiguously means X”—with no argument in the alternative—because outcome X is widely disliked among favored constituents or the general public, and the agency head, who nonetheless thinks X is the right policy, wants both for X to go into effect and to pass the buck for it.

Here, return again to the polar bear example. Prior to the DC federal district court’s November 2010 remand decision,179 the Fish and Wildlife Service could have said that it would adopt the same interpretation of “endangered” whether or not Congress compelled that interpretation. As it turns out, that would have been a useful weapon for the agency’s lawyers to wield in the inevitable litigation. But the agency said no such thing, at least until it was forced to do so by the district court. Might the Service’s reluctance to take ownership of the interpretation until it was forced to do so be explained by the fact that many in the environmental law community wanted the agency to take a more aggressive position?180 The agency, already under attack by groups who could be political allies, quite likely had an incentive to say, in effect, “Don’t blame us; we were just following the law,” rather than accept responsibility for a controversial policy decision.

To be sure, a buck-shifting agency could say, “Don’t look at us; Congress made us do X. And oh, by the way, if Congress hadn’t made us do X, we would have done X anyway.” But for obvious reasons, we doubt that this would be an effective blame-minimization strategy. Thus, the agency leader may face an unenviable choice: (a) minimize blame by attributing X entirely to Congress, or (b) minimize litigation risk while taking some ownership of the X result. When agency heads take the former course, they run the risk of a Chevron Step One-and-a-Half remand. But if X is sufficiently unpopular, or if the prospect of litigation is sufficiently unlikely or distant, then that may be a risk they are willing to take.181

F. Interadministration Politics

Finally, Chevron Step One-and-a-Half cases may sometimes arise because agency leaders are trying to bind their successors.182 Say that an agency head prefers policy X but is worried that a future administration may choose Y. The agency head may want a rule’s preamble to say that “the statute unambiguously means X”—without any argument in the alternative—for two reasons. First, the agency head may worry that arguing in the alternative will alert future agency leaders to the possibility of choosing a policy other than X. We doubt the efficacy of this strategy, though: if the successor is reasonably sophisticated, then the successor will presumably notice that the statutory language is not so clear-cut.

Second, and more effectively, the agency head may realize that the best way of preserving policy X is to secure a court decision holding that the statute unambiguously means X. As the Supreme Court stated in National Cable & Telecommunications Association v Brand X Internet Services,183 “A court’s prior judicial construction of a statute trumps an agency construction otherwise entitled to Chevron deference only if the prior court decision holds that its construction follows from the unambiguous terms of the statute and thus leaves no room for agency discretion.”184 That is, if a court says that the statute unambiguously means X as a Chevron Step One matter, then the agency later in time lacks the option of changing its policy to Y (unless, of course, it also can convince a later-in-time court to overrule the earlier court’s decision—itself a particularly difficult task in light of the “enhanced force” of stare decisis in statutory interpretation cases185 ). Yet if the agency offers an argument in the alternative, then a reviewing court could more easily sidestep the Chevron Step One issue and jump straight to Chevron Step Two.186 In that event, the agency head would not get what she is hoping for: a precedent on the books saying that X is the agency’s only option.

By maintaining that the statute unambiguously means X and refusing to argue in the alternative, the agency forces—or at least pushes—a court to decide the Chevron Step One question. And if the agency thinks that it is likely to face a relatively sympathetic court in the near term, then this strategy may be a rational one. Even if the composition of the relevant court changes over time, the holding by today’s court that the statute unambiguously means X carries stare decisis effect. The agency’s refusal to argue in the alternative may thus be part of a bet with a large potential upside. If the agency faces what it perceives to be a sympathetic court in the present period, then pressing that court toward a Chevron Step One resolution may mean that the current agency’s views of good policy will end up being locked in, even if the agency’s personnel (and policy preferences) were to change, and even if the court’s personnel and preferences were to change. That is a valuable asset—one almost as good as an act of Congress enshrining the current agency’s preference.187

And note that while the potential upside is large, the downside may be minimal. If a court remands the rule to the agency on Chevron Step One-and-a-Half grounds, then the agency always has the option of coming back and saying that it chooses X as a matter of discretion. Or, at least, almost always: near the end of an administration, there may not be time for the agency to supplement its initial preamble in the event of a Chevron Step One-and-a-Half remand. Then again, if the administration is in its waning days and its successor is unsympathetic to policy X, then policy X may not be long for this world anyway. Thus, one can see why this “bet” is potentially attractive: At worst, the agency that adopts this strategy allows a court to strike down a rule that would have been rescinded by the next administration in any event. At best, the agency wins a Chevron Step One ruling that binds subsequent administrations for the foreseeable future.

IV. The Benefits of Chevron Step One-and-a-Half

The previous Part listed several hypotheses as to why Chevron Step One-and-a-Half cases might continue to arise. And we have argued that, at least sometimes, agency actors may have an incentive to maintain that a statute is unambiguous even when such a claim harms the agency in litigation. On this view, the fact that Chevron Step One-and-a-Half cases continue to arise is hardly nonsensical. Agencies may have good strategic reasons for doing what they do.

Even so, an important question lingers: Does the Chevron Step One-and-a-Half doctrine make sense? Professor Bagley, for instance, has argued not: “the norm in Prill cases” is “needless punishment” for agencies guilty of nothing more than a foot fault.188 Indeed, wrote Bagley:

Rigid adherence to Prill won’t make agencies better at spotting latent ambiguities. In all likelihood, agencies will carry on much as they would in the absence of Prill, deaf to its marginal incentive effects. To the extent agencies do pay attention, their decisions will become bloated with boilerplate legal analysis. Chenery does not demand—and should not be read to demand—that kind of waste.189

We are sympathetic toward Bagley’s position insofar as he argues that “Chenery does not demand” that Chevron deference be limited to cases in which the agency acknowledges ambiguity. As we observed above, Chevron Step One-and-a-Half is compatible with, but perhaps not compelled by, Chenery I. And yet we are not convinced that the doctrine is a “waste.” In particular, here we show that Chevron Step One-and-a-Half can serve a useful purpose—or, rather, that it can serve useful purposes, with the purpose depending on the different reasons why agencies might insist that ambiguous statutes are clear. Whether these benefits outweigh the costs of additional remands and further litigation is a harder question—one that we cannot claim to resolve here. What we can say, though, is that the purposes served by Chevron Step One-and-a-Half ought not be ignored in the calculus.

A. Chevron Step One-and-a-Half as a Response to Agency Ignorance

The case for Chevron Step One-and-a-Half is perhaps the hardest to make when the issue arises due to the agency’s ignorance of the doctrine. Imagine that an agency genuinely prefers position X over position Y but says that it is adopting position X because “the statute compels X,” when in fact the statute merely permits X but could also be read to mean Y. In this scenario, what good does it do for the court to remand to the agency just so that the agency can promptly turn around and jot off a supplemental document explaining why it hews to position X as an exercise of discretion?

Even in this most difficult scenario, however, we can think of at least two reasons for applying the Chevron Step One-and-a-Half doctrine. First, even if the agency general counsel understands that the statute can be read to mean X or Y—and even if the statement that “the statute compels X” was simply an ill-advised way to explain that X is the agency’s chosen policy—one might be concerned that other actors within the agency lack the same knowledge as the general counsel. Other agency officials might believe that the general counsel’s statement—“the statute compels X”—is an indication that anything other than X would be unlawful. Indeed, it strikes us as unlikely that (a) the general counsel would lack the administrative law knowledge to understand that the statement “the statute compels X” sets the agency up for a Chevron Step One-and-a-Half problem, but (b) agency actors outside the general counsel’s office (including nonlawyers) would be so sophisticated on matters of administrative law as to understand that the statement “the statute compels X” actually means that X and Y are both options. Put differently, our worry is that even if the agency’s Chevron Step One-and-a-Half error was entirely innocent, actors within the agency will fail to understand that the statute actually allows option Y. Under these circumstances, Chevron Step One-and-a-Half serves to ensure that the general counsel does not short-circuit the agency’s internal policy-making process with an erroneous assertion that the statute is clear.

Even if one is not persuaded by this first justification, however, the second is perhaps stronger: instances of agency ignorance will be difficult to distinguish from instances of agency ambivalence or strategic maneuvering. As we argue below, the argument in favor of Chevron Step One-and-a-Half is strongest under those latter circumstances. And applying Chevron Step One-and-a-Half in cases of agency ignorance may be necessary so that the court does not also uphold the agency’s rule in cases of ambivalence or strategic maneuvering.

Concededly, Chevron Step One-and-a-Half remands in cases of agency ignorance may lead to waste in some instances. It is entirely possible that the X-versus-Y decision was fully vetted by the relevant actors within the agency, and that the general counsel—failing to understand the Chevron Step One-and-a-Half doctrine—approved a preamble that characterized the choice of X as compelled by Congress. Yet we would hope (and, indeed, expect) that such an occurrence would be a one-off event, and that agency lawyers would quickly learn to draft their preambles more carefully. If Chevron Step One-and-a-Half cases recur (and especially if the same agency runs afoul of the doctrine on multiple occasions), then one should begin to suspect that agencies are failing to acknowledge ambiguity for reasons other than innocence or ignorance. If so, then—as discussed below—the case for Chevron Step One-and-a-Half grows significantly stronger.

B. Chevron Step One-and-a-Half as a Response to Agency Ambivalence

The second possible cause of Chevron Step One-and-a-Half cases that we mentioned above is agency ambivalence: while agency officials believe that the statute compels X, those officials have yet to decide whether they would prefer X or Y if freed from statutory constraints (or, perhaps, they actually prefer Y). The case for Chevron Step One-and-a-Half under these conditions is straightforward: If the agency selected policy X because it misinterpreted the relevant statute, and if the agency would not (or might not) hew to X if liberated from its misapprehension, then why allow the agency’s choice of X to carry the force of law? Even then-Judge Roberts, a Chevron Step One-and-a-Half skeptic, acknowledged that a remand is appropriate in these circumstances:

[W]hen an agency erroneously concludes that a statutory interpretation is required by Congress, we should remand to give the agency an opportunity to interpret the statute in the first instance. That course is consistent with principles of Chevron deference, and with the respect due Congress’s delegation of interpretive authority to the agency.190

Bagley has argued that cases of this sort are “rare.”191 But they are not unheard of. In 1997, the Federal Communications Commission considered a provision of the Communications Act,192 as amended by the Telecommunications Act of 1996,193 that temporarily forbid any “Bell operating company” from acquiring an “alarm monitoring service entity”;194 the Commission concluded that this term was “unambiguous” and that the prohibition applied only to the takeover of an entity with “independent legal existence.”195 Later that year, the DC Circuit remanded the matter to the FCC on Chevron Step One-and-a-Half grounds: in the court’s view, the term “alarm monitoring service entity” was ambiguous, and the statutory prohibition might (though need not) apply to the acquisition of an unincorporated operating division as well.196 Given the chance at a do-over, the FCC turned on a dime and said that an unincorporated operating division was indeed an “alarm monitoring service entity” for purposes of the provision.197

Somewhat surprisingly, the same pattern repeats itself with respect to two other FCC determinations from the same year: 1997. In one case, the FCC determined that § 254(g) of the Telecommunications Act—which requires certain providers of telecommunications services to offer uniform rates to customers across states198 —applied to providers of commercial mobile radio services.199 The DC Circuit later vacated the FCC’s order because the agency erroneously believed that § 254(g) unambiguously applied to commercial mobile radio services providers.200 The FCC then reversed itself: “Exercising our authority to interpret this ambiguous provision,” the Commission said, “we find that the approach more faithful to the spirit of the statutory rate integration requirement is that . . . the requirement does not apply to [commercial mobile radio services] providers.”201 Also in 1997, the FCC determined that it was barred by Congress from collecting a regulatory fee from the congressionally chartered satellite communications provider Comsat.202 Another operator of telecommunications satellites, PanAmStat, challenged Comsat’s exemption; the DC Circuit determined that “the FCC was mistaken in its conclusion that the statute compelled an exemption for Comsat”;203 and on remand, the FCC, exercising its newfound discretion, decided to charge Comsat $1.6 million in fees.204

Moreover, Chevron Step One-and-a-Half remands can serve important purposes even when they do not spur the agency to reverse course 180 degrees. For example, a remand might (and sometimes does205 ) lead an agency to make substantive alterations to a prior rule without changing direction entirely. And even if an agency reaches the same result on remand, while doing so it may also gain additional knowledge that will have spillover effects on other policies. For instance, imagine an agency that is ambivalent about its preferred policy when it first announces its view that the statute is unambiguous (for example, it tentatively thinks that if it had discretion, it would choose X, but it is not sure). Forcing the agency to drill down and really decide for sure that it wants X rather than Y increases the agency’s knowledge of what it is regulating, and so also its expertise. And the fact that the agency has been given this discretion suggests, under the implied delegation theory of Chevron, that Congress wants the agency to undertake that learning process rather than simply free ride on the statute.206 To a reviewing court (or a skeptical professor), it may look like nothing has changed—the agency ended up in the same place it began. Only now the agency knows for certain what it thinks because it has done its homework.207

Even so, we obviously agree with Bagley that, in most cases, an agency that says in the first instance that “the statute compels X” will, following a Chevron Step One-and-a-Half remand, adopt position X as an exercise of discretion. Yet we doubt that courts can separate out these cases from ones in which a remand will indeed cause the agency to change course, either with regard to the same rule or with regard to another rule altogether. Reliance on the agency’s representations in litigation is itself an unreliable approach: the positions adopted by attorneys who represent the agency in court—oftentimes Justice Department lawyers rather than agency employees—may not reflect the views of all the agency actors who might have the opportunity to weigh in on remand.208 In any event, even if agency leaders do sign off on the litigators’ strategy, the agency’s learning process would still be short-circuited, because the process of signing off on a lawyer’s call is not the same as deliberately considering a policy through a more structured rulemaking process.

C. Chevron Step One-and-a-Half as a Response to Strategic Behavior

In Part III, we suggested that an agency’s failure to argue in the alternative that it would arrive at the same position as an exercise of discretion does not necessarily reflect ignorance or ambivalence. We proffered four potential strategic explanations—rooted in intra-agency, intra–executive branch, interbranch, and interadministration politics—that might account for the observed phenomenon. Here, we argue that if indeed Chevron Step One-and-a-Half cases arise because of any of these first three strategic reasons, then the justification for the doctrine is especially sound. (We address the possibility of interadministration strategic behavior in Part IV.D.)

Start with the first of these strategic explanations: an agency’s general counsel may maintain that the statute compels X so as to exert greater control over the intra-agency decision-making process. For adherents to the implied delegation theory of Chevron, this is an obvious problem. Congress generally delegates authority to agency leaders or to the agency as a whole,209 not to the agency general counsel. We struggle to see any reason why it is normatively desirable—or acceptable—for lawyers within an agency to use their privileged positions to impose policy preferences to which other agency actors might object. Chevron Step One-and-a-Half helps deter this sort of behavior, although in an admittedly messy way: it imposes a cost on the agency as a whole (the cost of responding to a remand) when actors within the agency attempt to paper over the possibility that the statute allows for other options.

The notion of agency lawyers asserting that a statute is unambiguous in order to squeeze other actors out of the decision-making process is equally problematic—if not more so—under the expertise account of Chevron. The Supreme Court has said that “practical agency expertise is one of the principal justifications behind Chevron deference”;210 such practical (and often technical) expertise is likely to be found in parts of the agency beyond the general counsel’s office. Courts have reason to doubt that such expertise will be brought to bear on a policy question when the general counsel comes out of the gate saying that a particular result is legally required.

Next, consider the possibility that Chevron Step One-and-a-Half cases arise as a by-product of agencies attempting to avoid reversal by OIRA. Critics of OIRA’s role in the regulatory process might be willing to countenance this outcome.211 Adherents to the accountability theory of Chevron, however, will find it perturbing. The accountability theory of Chevron is based on the twin premises that agencies are accountable to the chief executive and that the chief executive is accountable to voters. OIRA review of agency action justifies the first of these twin premises.212 Agency circumvention of OIRA review undermines that same premise. The more we take the accountability discussion in Chevron seriously, the stronger the argument for Chevron Step One-and-a-Half becomes.

In a similar vein, adherents to the accountability theory of Chevron might be concerned about agency preambles attributing to Congress policy decisions that agency officials have made themselves. If the theory undergirding Chevron is that voters should be the judges of the executive branch’s policy choices, then presumably the executive branch should have to take ownership of those policy choices so that voters know whom to blame (and to credit). To be sure, we doubt whether citizens are consulting the Federal Register along with the League of Women Voters’ Guide before they head to the polls.213 And yet accountability theories need not rely on electoral accountability as the exclusive transmission belt. We might take a more pluralistic approach and imagine interest groups, media organizations, and other sophisticated actors attributing credit and blame across branches based in part on agencies’ characterizations of their own degrees of freedom.214 If this is the case, then an agency’s assertion that the relevant statute is unambiguous may serve an accountability-deflecting function. Granted, this argument depends on those who evaluate agencies (1) being sophisticated enough to distinguish, based on a preamble or similar document, whether the agency believes its action to be compelled by Congress, but (2) not being so sophisticated as to arrive at their own conclusion as to whether the agency’s action is congressionally compelled. These two conditions may strike some readers as implausible. Yet when the stakes are high and many people are watching, an agency may be happy for any blame it can avoid, even if it cannot avoid all of it.

For all these reasons, we might be concerned about an agency (or the rule-writing lawyers within an agency) characterizing a statute as unambiguous when a fairer reading is that the statute affords the agency substantial discretion. And yet there is one more step in the logic before these concerns lead to a justification for Chevron Step One-and-a-Half. Arguably, a court could mitigate at least some of these concerns with a thorough opinion explaining that the relevant statute is ambiguous and yet upholding the agency’s interpretation as permissible. Assuming that the relevant individuals—nonlawyers within the agency, staffers at OIRA, and the external actors assigning credit or blame to Congress or the executive branch—read the opinion as well as the preamble (or read synopses of each), then the court could at least alert these individuals that X is not compelled and that Y remains an option. And an opinion would do so without the disruption of a remand, thus partially addressing Bagley’s and Roberts’s concerns about “waste[ful]” rounds of “ping-pong.”215

There are, however, two reasons to doubt the corrective power of an opinion that identifies a statutory ambiguity but does not remand to the agency. First, the agency’s initial selection of X over Y sets the status quo to X, and the status quo may be a sticky one. Actors within the agency who would prefer Y may be in a weaker position to advocate for their preference once X is in place than when the X-versus-Y decision is incipient. This is not unlike what happens within Congress; because (by design) it is so hard to run the lawmaking gauntlet, there is a heavy status quo bias.216 So too with agencies; in a world of “ossification” of significant rules,217 it can be difficult to do all the work that is required to unwind an administrative decision. It thus is quite unlikely, for instance, that language in a judicial opinion—language not backed up by a judgment—will restart the OIRA review process and force the agency to justify X over Y on the basis of cost-benefit analysis and presidential priorities. In short, an after-the-fact judicial opinion announcing that the relevant statute is ambiguous will not entirely negate the ability of agency officials (and, in particular, agency lawyers) to impose their preferences in ways not contemplated by Chevron.

Second, many agency actions do not result in litigation, and in all but the most exceptional cases, the probability of litigation is less than unity. (And even if the probability of litigation is one, there is no guarantee that the plaintiffs will raise and properly preserve a Chevron Step One-and-a-Half argument.) If an agency can assert that a statute is unambiguous and face only the consequence of a judicial slap on the wrist, then the incentive for agencies (and agency lawyers) to take this tack will be stronger. If agency leaders (or agency lawyers) prefer X and know that there is no real sting from telling intra-agency constituencies, OIRA, and the public that X is the only option allowed by the statute, then—well—why not? Critical language in a judicial opinion may be embarrassing to the agency, but, at the same time, the odds of that language being written are long; there must be litigation (no foregone conclusion), and the court must resolve the case against the agency on Chevron Step One-and-a-Half grounds (as opposed to on any other outcome-determinative issue). If judicial review amounts to nothing more than a chance—not even a certainty—of harsh words, it is easy to imagine the cost-benefit analysis cutting against acknowledging statutory ambiguity. Hence one of the key benefits of Chevron Step One-and-a-Half: remands under Step One-and-a-Half impose a cost on agencies that is borne in part by the agency leaders and lawyers likely to make the decision whether to acknowledge ambiguity in the preamble.

So what Roberts referred to as “ping-pong” and Bagley considered a “waste” might instead be a necessary element of an optimal deterrence approach: yes, Chevron Step One-and-a-Half remands impose costs on agencies, but those are costs meant to discourage agencies from ascribing policy choices to Congress that Congress has delegated to the agencies. This point is especially forceful when one remembers that agencies and courts are repeat players. Given this dynamic, a remand in one case may very well have beneficial systemic effects for a host of other cases.218

This raises the question: How costly should Chevron Step One-and-a-Half remands be? When an agency says in the preamble to a rule that the statute compels X rather than Y and the court concludes that X and Y are both permissible interpretations of the statute, then the court could:

1. Do nothing so long as the agency represents in litigation that it in fact prefers X (the Roberts/Bagley approach);
2. Remand to the agency but leave the rule in place for the time being;
3. Remand to the agency and vacate the rule; or
4. Impose Y as a penalty for the agency’s failure to acknowledge ambiguity.

We explore in Part V.B the advantages and disadvantages of each of these options. But before addressing the question of remedy in Chevron Step One-and-a-Half cases, we take up the last of the strategic reasons why agencies might put themselves in a Chevron Step One-and-a-Half box: the “interadministration politics” possibility.

D. Chevron Step One-and-a-Half as an Invitation for Strategic Agency Behavior

We noted above that agencies might insist that a statute unambiguously means X so as to secure a holding that the statute indeed means X (and only X). If the agency can get a reviewing court to say that the statute means X, that locks X in place against modification by future agencies.219 Of course, this gambit might, in some cases, result in a remand should the court conclude that the statute could mean X or Y. And yet, in one sense, Chevron Step One-and-a-Half actually aids the agency’s strategy in these sorts of cases, because the doctrine prevents the reviewing court from skipping to Chevron Step Two.

To elaborate: Imagine that a statute might mean X or Y, the agency prefers X over Y, and the agency fears that a future administration might prefer Y over X. The agency also thinks that the relevant reviewing court is generally minimalist, and so will rule (in the agency’s favor) on Chevron Step Two grounds if given the option. Finally, the agency thinks that the reviewing court, as currently composed, is likely to rule in favor of the agency at Chevron Step One today if forced to confront the Step One issue—but that if a future administration switches to position Y, the reviewing court later in time might not strike down the future agency’s action on Chevron Step One grounds. (This could be because the agency believes that the composition of the court might change, or because the agency believes that, all else equal, the court is more likely to go along with the agency’s position than go against it.)

Under these circumstances, the agency might want to force a Chevron Step One decision today. But in the absence of the Chevron Step One-and-a-Half doctrine, the agency would have trouble doing so. Even if the agency insists that the statute unambiguously means X, a court in a world without Step One-and-a-Half would have the option of skipping straight to Chevron Step Two. And that, of course, would thwart the agency’s entrenchment strategy.

Should the possibility that Chevron Step One-and-a-Half might facilitate this sort of strategic behavior curb our enthusiasm for the doctrine? Perhaps—but probably not by much. While we believe that some amount of strategic behavior by agencies might explain some number of Chevron Step One-and-a-Half cases, the amount of plotting involved in this last scenario strikes us as House of Cards-esque. This is not to rule out the possibility that Frank Underwood types might be found in some agency general counsel’s offices (and, indeed, we suspect that agency lawyers are engaged in strategic maneuvering to some extent, especially for high-profile decisions220 ). But although it is imperfect, we think the better response to this last gambit would be for the reviewing court to reject the agency’s attempt to force a Chevron Step One ruling unless the court believes that the statute is in fact unambiguous—in which case a ruling to that effect strikes us as entirely appropriate.

E. Chevron Step One-and-a-Half as an Invitation for Strategic Judicial Behavior

Our analysis would not be complete without mentioning one last concern: the possibility that Chevron Step One-and-a-Half might add to the power of courts, potentially in undesirable ways. So far, our analysis has discussed the strategic reasons (among others) why an agency may say that a statute is unambiguous while a court later concludes that the same statute is ambiguous. But might not Chevron Step One-and-a-Half open up new strategic opportunities for the court as well? Imagine that the agency prefers to read a statute to mean X, while a court prefers for the same statute to mean Y. If the agency says that the statute unambiguously means X, then without Chevron Step One-and-a-Half, the court would have two statutory interpretation options: (1) uphold the agency’s interpretation or (2) say that the agency’s interpretation is not a “permissible” reading.221 The second option may be costly for the court, perhaps because the claim that X is an impermissible reading of the statute is difficult to make with a straight face (and thus the assertion damages the court’s reputation and perceived legitimacy). Chevron Step One-and-a-Half gives the court a third option: say that the statute is ambiguous, and thus that the matter must be remanded to the agency. This third option may be a low-cost course from the court’s perspective, because what constitutes ambiguity can itself be ambiguous;222 the reputational consequences for the court are thus less severe than the ramifications of option (2). In this way, Chevron Step One-and-a-Half gives a court with ideological preferences at odds with an agency’s an easy way to throw sand in the agency’s gears.

We are mindful of this concern and understand why those who are skeptical of judicial review may believe that this risk cuts against enforcement of Chevron Step One-and-a-Half. That said, we are not especially moved. If an agency, for instance, is worried about this sort of strategic judicial behavior, it can always argue in the alternative (that is, both assert that the statute is unambiguous and also explain why it would reach the same result even if the statute were ambiguous). In other words, although this concern about judges should not be understated, it also should not be overstated; agencies have a “self-help” mechanism, even though they do not always opt to use it. Moreover, this third option availed to courts by Chevron Step One-and-a-Half is a strategy of questionable efficacy: if the agency prefers reading X to reading Y, it still can adopt reading X as an exercise of interpretive discretion on remand. Perhaps a court in the last few months of an administration may view the third option as a way to run out the clock, in the hope that a different administration—with ideological preferences closer to the court’s—will take the reins before the remand proceedings are complete. We should note, though, that none of the Chevron Step One-and-a-Half cases from the DC Circuit identified in our Online Appendix plausibly fit this pattern.223

V. Implementing Chevron Step One-and-a-Half

Thus far, we have argued that Chevron Step One-and-a-Half is a distinct doctrine—consistent with, but not dictated by, the justifications proffered in (and for) Chevron itself. We have also argued that Chevron Step One-and-a-Half cases may continue to arise even if agency lawyers are aware of the doctrine: there are reasons rooted in ambivalence as well as strategy why agency actors may insist that a statute is unambiguous even though such insistence works to the agency’s disadvantage in litigation. Finally, we have argued that Chevron Step One-and-a-Half is an appropriate response to agency protestations of statutory clarity—though with the consequence that in some cases the doctrine may give agencies the option of forcing a judicial decision at Chevron Step One.

Here we address three ancillary questions. First, in what circumstances should Chevron Step One-and-a-Half be triggered? Second, what sort of remedy is appropriate in Chevron Step One-and-a-Half cases? And third, even if Chevron Step One-and-a-Half is sensible prudentially, can it be justified jurisprudentially? Our views on the first two questions remain tentative—and the DC Circuit itself has yet to arrive at firm answers to either. As to the third question, we acknowledge that Chevron Step One-and-a-Half is—for better or worse—“administrative common law,” without a firm anchor in the text of the Administrative Procedure Act224 (APA). And yet so too (and no less) is Chevron itself. So long as we have a Chevron doctrine, we think that Chevron Step One-and-a-Half plays an important role in preventing the Chevron framework from being flipped on its head.

A. Ambiguity about Ambiguity

The easiest Chevron Step One-and-a-Half cases are the ones in which an agency says that “the statute unambiguously means X” or says that “the statute is ambiguous—it allows both X and Y—and we choose X as a matter of discretion.” Such cases are not uncommon.225 Nor, however, are such cases universal: oftentimes an agency’s preamble or decision will lie somewhere between these two extremes.226

One possible response is for the court to apply a “magic words” approach: unless the agency explicitly says that the statute is ambiguous, or explicitly says that it would reach the same result if the statute were ambiguous, then the court will not afford deference to the agency at Chevron Step Two. At the other extreme, the court could apply a “benefit of the doubt” approach: unless the agency explicitly says that the statute is unambiguous, then the court will assume that the agency is exercising its discretion. A third possibility is a “mind reading” tack: if the agency is ambiguous as to whether it thinks the statute is ambiguous, then the court will do its best to divine the agency’s view. A fourth route is an “ask the attorney” approach: when in doubt, the court can ask the agency’s lawyer at oral argument whether the agency would hew to its position if the statute were ambiguous.

We think there is much to be said for the “magic words” approach, which would encourage the authors of agency preambles to make clear to other agency officials, OIRA, and the public when the agency is exercising policy discretion. We think there is much less to be said for the “ask the attorney” approach, which would do little to deter the sort of strategic behavior discussed above. Agencies (and agency lawyers) would then be able to say in their preambles that a particular policy is compelled by Congress, and only in the event of litigation (and, even then, only in the event that a Chevron Step One-and-a-Half argument arises) would the agency have to take ownership of its choice.

Ultimately, the choice among triggering rules comes down to how we want to allocate interpretive and explanatory burdens. The “magic words” approach imposes a burden on preamble writers to say at the outset that the agency is exercising its discretion. The “benefit of the doubt” approach imposes a burden on other actors—within the agency, elsewhere within the executive branch, and outside the government—to understand that anything other than an unambiguous statement of unambiguity amounts to an exercise of discretion. The “mind reading” tack places the burden on the court to divine the agency’s intent—but also on the other actors within and outside the agency to determine whether the agency is or is not exercising interpretive discretion. The “ask the attorney” approach places an even heavier burden on the other actors within and outside the agency, who now must divine the agency’s intent not on the basis of the past statements, but on the basis of future ones.

In some cases, the DC Circuit appears to follow something like the “benefit of the doubt” approach. For instance, in Braintree Electric Light Department v Federal Energy Regulatory Commission,227 the court rejected a Chevron Step One-and-a-Half attack by stating that so “long as the text is ambiguous and the agency does not insist that it is clear, a reasonable interpretation will warrant our deference.”228 This raises the obvious concern that the preamble to an agency rule may suggest to nonlawyers within and outside the agency that the statutory language is clear, while still leaving the agency wiggle room in the event of a Chevron Step One-and-a-Half challenge. In other cases, the DC Circuit appears to impose a higher bar: for example, the court in Peter Pan said that “Chevron step 2 deference is reserved for those instances when an agency recognizes that the Congress’s intent is not plain from the statute’s face.”229 Our modest suggestion is that the DC Circuit should clarify the ambiguity as to how it will treat ambiguity about ambiguity. If the “benefit of the doubt” approach is the governing regime, then relevant actors within and outside the agency should know that any time the agency does not say that the statute is clear, the agency is eligible for Chevron Step Two deference. Such a bright-line rule—clearly stated by the courts—would go some way toward reducing opportunities for agency actors to suggest one thing in the rulemaking process and another in litigation.

B. Choosing a Remedy

A second question in the implementation of Chevron Step One-and-a-Half is the choice of remedy: If an agency runs afoul of the Step One-and-a-Half doctrine, should a reviewing court (1) let it go, (2) remand without vacatur, (3) remand with vacatur, or (4) impose the opposite of the agency’s preferred rule? For all the reasons explained above, we think option (1) is a bad one. Choosing between options (2) and (3), however, is more complicated.

Administrative remedies have been largely overlooked in the scholarly literature—to the detriment of the field.230 In short, § 706 of the APA says that “[t]he reviewing court shall . . . hold unlawful and set aside agency action” that is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.”231 Despite this language, the DC Circuit has concluded—albeit not without controversy232 —that not all unlawful agency actions must be vacated. Instead, sometimes a court will remand without vacating under what has come to be known as the Allied–Signal doctrine, so named for the 1993 case (Allied–Signal, Inc v United States Nuclear Regulatory Commission233 ) in which it was articulated.

Under the Allied–Signal doctrine, “[t]he decision whether to vacate depends on ‘the seriousness of the order’s deficiencies (and thus the extent of doubt whether the agency chose correctly) and the disruptive consequences of an interim change that may itself be changed.’”234 The facts of Allied–Signal illustrate how the doctrine works. The US Nuclear Regulatory Commission promulgated a rule to recoup some of the costs incurred when the agency provides services to regulated parties. Allied-Signal challenged that regulation as arbitrary and capricious and contrary to the statute because the agency did not adequately consider the ability of certain types of companies to “pass through” charges to consumers.235 The Commission had exempted certain nonprofit institutions from the rule because they could not readily pass through charges to consumers, but had not considered whether other entities were similarly restricted—which rendered the rule a violation of the “reasoned decision-making” requirement.236 The DC Circuit concluded, however, that vacatur was not required. After all, it was possible that the Commission could “explain how the principles supporting an exemption for educational institutions do not justify a similar exemption” for other types of entities.237 And “[a]t the same time, the consequences of vacating may be quite disruptive,” especially because “the Commission would need to refund all . . . fees collected from those [entities]; in addition it evidently would be unable to recover those fees under a later-enacted rule.”238 The Allied–Signal doctrine has also been adopted beyond the DC Circuit.239 The Supreme Court, however, has not addressed its legality.240

The Allied–Signal doctrine could be combined with Chevron Step One-and-a-Half. For instance, in Coalition for Common Sense in Government Procurement v United States,241 the court confronted a Department of Defense regulation involving “Federal Ceiling Prices” (that is, the highest amount federal agencies can pay contractors) that the agency claimed was mandated by Congress.242 The court disagreed with the agency that the agency did not have discretion and, under Prill, concluded that the court could not determine whether the agency’s regulation could nonetheless be upheld.243 Rather than vacate the regulation, however, the court applied Allied–Signal to conclude that the regulation should be remanded only.244 Likewise, in the polar bear illustration presented above, the court concluded that the agency wrongly disclaimed discretion, but rather than vacate the agency’s rule, the court simply remanded for additional proceedings.245

There is much to be said for this sort of approach—making an agency start the regulatory process anew imposes massive costs, and perhaps the disciplining effects of Chevron Step One-and-a-Half can be achieved without the need for such costs, at least sometimes. Thus, a reviewing court, as a matter of its equitable judgment, could determine whether the case-specific costs or the broader institutional values should win out. The problem, however, is that courts may not be well positioned to strike that balance, especially because much of what happens during the rulemaking process occurs in a “black box.”246 Hence, it may be difficult to know just how innocent a mistake is.

Moreover, the costs imposed by vacatur are arguably part of the deterrence effect of Chevron Step One-and-a-Half. Indeed, courts could go a step further and establish a rule of the following sort: if a statute could mean X or Y and the agency insists that the statute unambiguously means X, then the penalty for the agency’s failure to acknowledge ambiguity is that the court will impose Y. We anticipate that under such a regime, agencies would much more readily acknowledge their own discretion. Indeed, the total costs of Chevron Step One-and-a-Half remands may decline as such remands become rarer and rarer.247 Yet this “super penalty” approach makes much more sense on a whiteboard than in the real world. If an agency promulgates an eighty-page rule on a complicated subject based on its view that a statute means X, would a court be competent to write its own version of the rule on the view that the statute means Y? We are doubtful.248

At bottom, we think the remedy question defies easy answer. Our intuition is that, given the high costs of vacatur, a context-dependent approach makes the most sense here—with vacatur reserved for cases that cannot be chalked up to innocent agency error. And yet we acknowledge that the high costs of vacatur might also counsel in favor of invoking that remedy more readily so as to achieve optimal deterrence.

C. Chevron Step One-and-a-Half as Administrative Common Law

At last we come to a question that arguably should have come much earlier: Even if Chevron Step One-and-a-Half makes sense as a prudential matter, can courts justify the doctrine as a jurisprudential matter? After all, Chevron Step One-and-a-Half is a judge-made doctrine, not derived from the text of the APA or any other statute. Do courts have the authority to make up a doctrine like this on their own?

The reality is that sometimes for these higher-order questions of administrative law, textual sources do not have much to say. For instance, although both Chenery I and State Farm could, in a pinch, be tied to the APA’s bar on arbitrary and capricious agency action, the nexus is not self-evident. At the same time, however, what agencies are allowed to do has also expanded. Chevron itself, for instance, is hardly compelled by any statutory source, and, indeed, it may be in some tension with the APA.249 In short, one cannot be asymmetric about these things: if deference doctrines are allowed to stand (a question we leave for another day), then doctrines that help police those deference doctrines must also be allowed to expand. There no doubt is a bit of “administrative common law” in Chevron Step One-and-a-Half, but that alone is not necessarily a reason to reject the doctrine.250 If agencies are to be trusted with Chevron powers for reasons of political accountability and expertise, then Chevron Step One-and-a-Half—which forces agencies to take accountability for their actions and pushes them to employ their expertise—protects the integrity of the Chevron framework.

Our tentative defense of Chevron Step One-and-a-Half still leaves the doctrine open to attack on grounds that it violates the principle that reviewing courts are “generally not free” to impose procedural requirements on agencies unless those requirements stem from the APA or some other statutory source.251 This principle—known as the Vermont Yankee rule for the case, Vermont Yankee Nuclear Power Corp v Natural Resources Defense Council, Inc,252 in which it was announced—would seem to be in some tension with Chevron Step One-and-a-Half. After all, the requirement that an agency acknowledge ambiguity in the relevant statute as a prerequisite for deference at Chevron Step Two might be described as an additional procedural requirement ginned up by the DC Circuit and found nowhere in any statutory source.

We have no knockdown rebuttal to the Vermont Yankee objection, though we do have at least two responses. First, Chevron Step One-and-a-Half is at least arguably justified on the basis of Chenery I, which predates Vermont Yankee by several decades.253 And second, Chevron Step One-and-a-Half is at least arguably a logical extension of Chevron’s holding that courts should defer in instances when an agency enjoys a relative advantage in terms of electoral accountability and expertise. If judges can decide that agencies will receive deference under certain circumstances, then presumably the same judges can decide in what circumstances such deference ought not to be forthcoming. Our modest suggestion is that so long as they exist, deference-limiting doctrines should be fitted to the values that motivated Chevron in the first place. And by our lights, Chevron Step One-and-a-Half meets that criterion.

Conclusion

Chevron Step One-and-a-Half is an important element of judicial review of agency statutory interpretations in the DC Circuit—and there are at least signs of it taking hold in courts beyond the District of Columbia’s “ten Miles square.”254 And yet so far the doctrine has largely escaped academic attention: casebooks and (with few exceptions) commentators have passed over the gap between Chevron Step One and Chevron Step Two. We seek to fill this void by presenting the first comprehensive treatment of Chevron Step One-and-a-Half. We ask: Why does the doctrine exist, and are Chevron Step One-and-a-Half’s critics correct that the doctrine’s application “outstrips its rationale”?255

On our view, before one can answer those questions, it is essential to confront the puzzle at the heart of Chevron Step One-and-a-Half: Why do these cases still arise? One might have expected that Prill would be a one-off, and that agencies would quickly learn to insulate their rules from Step One-and-a-Half attacks. And yet the last thirty years have proven that expectation to be incorrect. Perhaps the reason is that agencies are slow to learn (in which case they would have to be very slow indeed!). We believe, though, that any explanation sounding in agency ignorance is incomplete. Agency actors have strategic reasons to assert that statutes are unambiguous even when such assertions increase the likelihood of a remand. Central to our argument is the fact that reviewing courts are not the only relevant veto players in the administrative process: actors within and outside the agency are, too. Moreover, agency actors are not trying to maximize only their probability of success in litigation; they also may care about the allocation of credit and blame between themselves and Congress, as well as the durability of their policies once their successors grab the reins.

We are, of course, not the first to observe that agencies are themselves multifaceted bodies, that agencies are concerned not only about courts but also about OIRA, or that agencies seek to entrench their preferred policies so that their work is not undone by successors. We are, however, the first to explain how these considerations interact with Chevron Step One-and-a-Half. And these strategic considerations not only help to explain why Chevron Step One-and-a-Half cases continue to arise, but also may help to justify the doctrine’s continued existence.

• 1467 US 837 (1984).
• 2See id at 842–43.
• 3Thomas W. Merrill and Kristin E. Hickman, Chevron’s Domain, 89 Georgetown L J 833, 836 (2001). See also Cass R. Sunstein, Chevron Step Zero, 92 Va L Rev 187, 191 n 19 (2006) (crediting Merrill and Hickman for the article’s title).
• 4See, for example, William S. Jordan III, Judicial Review of Informal Statutory Interpretations: The Answer Is Chevron Step Two, Not Christensen or Mead, 54 Admin L Rev 719, 725 (2002) (“[T]he Court erected a new four step test to replace what we once knew as the Chevron two step.”).
• 5Matthew C. Stephenson and Adrian Vermeule, Chevron Has Only One Step, 95 Va L Rev 597, 598 n 4, 599 (2009). But see Kenneth A. Bamberger and Peter L. Strauss, Chevron’s Two Steps, 95 Va L Rev 611, 624–25 (2009) (rejecting Stephenson and Vermeule’s argument); Richard M. Re, Should Chevron Have Two Steps?, 89 Ind L J 605, 608 (2014) (same).
• 6See Alaska Wilderness League v Jewell, 788 F3d 1212, 1217 (9th Cir 2015) (“[U]nder Chevron . . . we engage in a three-step inquiry when reviewing an agency’s interpretation of a statute.”); Restrepo v Attorney General of United States, 617 F3d 787, 792 (3d Cir 2010) (“When confronted with a potential Chevron application, we administer a three-step analysis.”).
• 7See, for example, Carter v Welles–Bowen Realty, Inc, 736 F3d 722, 731 (6th Cir 2013) (Sutton concurring) (arguing that “[i]f you believe that Chevron has two steps, you would” reach a result one way, and “[i]f you believe that Chevron has only one step,” you would reach the same result another way).
• 8See, for example, Brett M. Kavanaugh, Book Review, Fixing Statutory Interpretation, 129 Harv L Rev 2118, 2154 (2016):

  In short, the problem with certain applications of Chevron, as I see it, is that the doctrine is so indeterminate—and thus can be antithetical to the neutral, impartial rule of law—because of the initial clarity versus ambiguity decision. . . . [W]e need to consider eliminating that inquiry as the threshold trigger.

  See also Jack M. Beermann, End the Failed Chevron Experiment Now: How Chevron Has Failed and Why It Can and Should Be Overruled, 42 Conn L Rev 779, 834 n 218 (2010) (“The number of steps in Chevron in any given case may turn out to depend on who writes the Court’s opinion.”).

• 9But others have taken it further. See, for example, Northpoint Technology, Ltd v Federal Communications Commission, 412 F3d 145, 151 (DC Cir 2005) (Henderson) (“[U]nder the Chevron two-step, we stop the music at step one if the Congress has directly spoken to the precise question at issue. . . . But if the statute is silent or ambiguous, we dance on . . . [to] step two.”) (quotation marks omitted); Braintree Electric Light Department v Federal Energy Regulatory Commission, 667 F3d 1284, 1288 (DC Cir 2012) (Garland) (“[T]he Chevron two-step is a dance for the court.”).
• 10Chevron, 467 US at 842–43.
• 11755 F2d 941 (DC Cir 1985).
• 12See id at 942. See also Nicholas Bagley, Remedial Restraint in Administrative Law, 117 Colum L Rev 253, 296–301 (2017) (discussing “Prill cases,” “Prill claims,” and “Prill violation[s]”).
• 13555 US 511 (2009).
• 14See notes 130–41 and accompanying text.
• 15See PDK Laboratories Inc v United States Drug Enforcement Administration, 362 F3d 786, 808–09 (DC Cir 2004) (Roberts concurring in part and concurring in the judgment).
• 16Id at 808 (Roberts concurring in part and concurring in the judgment).
• 17Id (Roberts concurring in part and concurring in the judgment).
• 18Id at 809 (Roberts concurring in part and concurring in the judgment).
• 19PDK Laboratories, 362 F3d at 809 (Roberts concurring in part and concurring in the judgment), quoting Time, Inc v United States Postal Service, 667 F2d 329, 335 (2d Cir 1981).
• 20Bagley, 117 Colum L Rev at 297 (cited in note 12).
• 21See id at 297–98.
• 22See PDK Laboratories, 362 F3d at 809 (Roberts concurring in part and concurring in the judgment); Bagley, 117 Colum L Rev at 301 (cited in note 12).
• 23See, for example, John C. Coffee Jr, The Future as History: The Prospects for Global Convergence in Corporate Governance and Its Implications, 93 Nw U L Rev 641, 702 (1999) (“[T]he occasions are rare on which a regulatory agency has voluntarily ceded control without some compelling need that required such a surrender. The usual assumptions of political science are that public agencies act to maximize their powers, just as private firms seek to maximize revenues or profits.”). See also Talk America, Inc v Michigan Bell Telephone Co, 564 US 50, 69 (2011) (Scalia concurring) (“By contrast, deferring to an agency’s interpretation of its own rule encourages the agency to enact vague rules which give it the power, in future adjudications, to do what it pleases.”).
• 24See Peter L. Strauss, “Deference” Is Too Confusing—Let’s Call Them “Chevron Space” and “Skidmore Weight”, 112 Colum L Rev 1143, 1145 (2012) (explaining that Chevron “empower[s]” agencies “to act in a manner that creates legal obligations”).
• 25See, for example, Burlington Truck Lines, Inc v United States, 371 US 156, 167 (1962).
• 26See, for example, Teva Pharmaceuticals USA, Inc v Food & Drug Administration, 441 F3d 1, 4–5 (DC Cir 2006) (declaring in a case involving a major cholesterol drug that “[w]e therefore generally remand for an agency to make the first interpretation of an ambiguous statutory term when it has failed to do so previously”).
• 27The Federal Communications Commission, the Department of Health and Human Services, the Department of Transportation, and the National Labor Relations Board are all repeat losers before the DC Circuit in Chevron Step One-and-a-Half cases. See Online Appendix.
• 28The Contribution of the D.C. Circuit to Administrative Law, 40 Admin L Rev 507, 509 (1988) (quoting remarks by Judge Patricia Wald).
• 29The DC Circuit has ruled against an agency on Chevron Step One-and-a-Half grounds in each of the following cases:

  NextEra Desert Center Blythe, LLC v Federal Energy Regulatory Commission, 2017 WL 1228577, *3–4 (DC Cir);

  United States v Ross, 2017 WL 728040, *4–5 (DC Cir);

  Noble Energy, Inc v Salazar, 671 F3d 1241, 1245–46 (DC Cir 2012);

  PSEG Energy Resources & Trade LLC v Federal Energy Regulatory Commission, 665 F3d 203, 208–10 (DC Cir 2011);

  United States Postal Service v Postal Regulatory Commission, 640 F3d 1263, 1268 (DC Cir 2011);

  Prime Time International Co v Vilsack, 599 F3d 678, 683 (DC Cir 2010);

  Labor, Mine Safety and Health Administration v National Cement Co of California, 494 F3d 1066, 1073–75 (DC Cir 2007);

  Menkes v Department of Homeland Security, 486 F3d 1307, 1313–14 (DC Cir 2007);

  Peter Pan Bus Lines, Inc v Federal Motor Carrier Safety Administration, 471 F3d 1350, 1353–54 (DC Cir 2006);

  Teva Pharmaceuticals USA, Inc v Food & Drug Administration, 441 F3d 1, 4–5 (DC Cir 2006);

  PDK Laboratories Inc v United States Drug Enforcement Administration, 362 F3d 786, 797–98 (DC Cir 2004);

  Arizona v Thompson, 281 F3d 248, 259–60 (DC Cir 2002);

  ITT Industries, Inc v National Labor Relations Board, 251 F3d 995, 1004–05 (DC Cir 2001);

  Jacoby v National Labor Relations Board, 233 F3d 611, 617 (DC Cir 2000);

  GTE Service Corp v Federal Communications Commission, 224 F3d 768, 775–76 (DC Cir 2000);

  Transitional Hospitals Corp of Louisiana, Inc v Shalala, 222 F3d 1019, 1029 (DC Cir 2000);

  PanAmSat Corp v Federal Communications Commission, 198 F3d 890, 896–97 (DC Cir 1999);

  Sea–Land Service, Inc v Department of Transportation, 137 F3d 640, 646–47 (DC Cir 1998);

  Alarm Industry Communications Committee v Federal Communications Commission, 131 F3d 1066, 1071–72 (DC Cir 1997);

  City of Los Angeles Department of Airports v United States Department of Transportation, 103 F3d 1027, 1033–34 (DC Cir 1997);

  Cajun Electric Power Cooperative, Inc v Federal Energy Regulatory Commission, 924 F2d 1132, 1135–37 (DC Cir 1991);

  American Petroleum Institute v United States Environmental Protection Agency, 906 F2d 729, 740 (DC Cir 1990) (per curiam);

  Baltimore and Ohio Railroad Co v Interstate Commerce Commission, 826 F2d 1125, 1129 (DC Cir 1987);

  Phillips Petroleum Co v Federal Energy Regulatory Commission, 792 F2d 1165, 1169–70, 1172 (DC Cir 1986);

  Prill, 755 F2d at 942.

  The list is nonexhaustive and does not include cases in which the DC federal district court invoked Chevron Step One-and-a-Half and the agency did not press an appeal before the DC Circuit. See, for example, American Petroleum Institute v Securities and Exchange Commission, 953 F Supp 2d 5, 13 (DDC 2013); International Swaps and Derivatives Association v United States Commodity Futures Trading Commission, 887 F Supp 2d 259, 280–82 (DDC 2012); Coalition for Common Sense in Government Procurement v United States, 671 F Supp 2d 48, 55–56 (DDC 2009). Nor does it include cases in other circuits applying the doctrine. See notes 137–41 and accompanying text.

  Note, moreover, that the total number of cases in which an agency interpretation fails at Chevron Step Two is quite small. Reviewing all published Chevron decisions from the courts of appeals in 2011, Professor Richard Re identified only two instances in which the court ruled against the agency specifically at Step Two. See Re, 89 Ind L J at 638, 640 (cited in note 5). Professors Kent Barnett and Christopher J. Walker identified fewer than five Step Two invalidations per year in a review of circuit court cases between 2003 and 2013, with the DC Circuit accounting for less than a fifth of all Chevron cases in the courts of appeals. See Kent Barnett and Christopher J. Walker, Chevron in the Circuit Courts, 115 Mich L Rev *25–26, 31–32 (forthcoming 2017), archived at http://perma.cc/DG5V-SH82.

• 30Pub L No 93-205, 87 Stat 884, codified as amended at 16 USC § 1531 et seq.
• 31ESA § 3(4), 87 Stat at 885, codified at 16 USC § 1532(6).
• 32See Arthur Neslen, Climate Change Is ‘Single Biggest Threat’ to Polar Bear Survival (The Guardian, Nov 18, 2015), archived at http://perma.cc/Z57U-EDF5.
• 33On the division of responsibility between the Interior Department’s Fish and Wildlife Service and the Commerce Department’s National Marine Fisheries Service, see Hawksbill Sea Turtle v Federal Emergency Management Agency, 126 F3d 461, 470–71 (3d Cir 1997).
• 34Department of the Interior, Fish and Wildlife Service, Endangered and Threatened Wildlife and Plants; Determination of Threatened Status for the Polar Bear (Ursus maritimus) throughout Its Range, 73 Fed Reg 28212, 28212 (2008), amending 50 CFR Part 17.
• 35ESA § 3(15), 87 Stat at 886, codified at 16 USC § 1532(20).
• 36In re Polar Bear Endangered Species Act Listing and § 4(d) Rule Litigation, 748 F Supp 2d 19, 24, 26 (DDC 2010).
• 37A number of other plaintiffs, including the state of Alaska and Safari Club International, sued the Service arguing that polar bears did not meet the definition of “threatened.” Id at 20–21.
• 38See Third Amended Complaint of Plaintiffs Center for Biological Diversity, Natural Resources Defence Council and Greenpeace, Inc, In re Polar Bear Endangered Species Act Listing and § 4(d) Rule Litigation, Misc Action No 08-0764, *28 (DDC filed Mar 13, 2009) (claiming that “[t]he Secretary’s failure to list the polar bear as ‘endangered’” was in part because he “applied the wrong legal standards in making his determination”).
• 39In re Polar Bear, 748 F Supp 2d at 24–25 (citations and quotation marks omitted).
• 40Id at 28–29 (quotation marks omitted).
• 41Id at 29 (citation and quotation marks omitted).
• 42See id.
• 43471 F3d 1350 (DC Cir 2006). See also In re Polar Bear, 748 F Supp 2d at 29, citing Peter Pan, 471 F3d at 1354.
• 44Peter Pan, 471 F3d at 1354.
• 45In re Polar Bear, 748 F Supp 2d at 29–30.
• 46See generally Memorandum: Supplemental Explanation for the Legal Basis of the Department’s May 15, 2008, Determination of Threatened Status for Polar Bears, In re Polar Bear Endangered Species Act Listing and § 4(d) Rule Litigation, Misc Action No 08-0764 (DDC filed Dec 22, 2010).
• 47Id at *7–8.
• 48Id at *8.
• 49In re Polar Bear Endangered Species Act Listing and § 4(d) Rule Litigation, 794 F Supp 2d 65, 90 (DDC 2011).
• 50See In re Polar Bear Endangered Species Act Listing and Section 4(d) Rule Litigation, 709 F3d 1, 3 (DC Cir 2013).
• 51Polar Bear (Ursus maritimus) (US Fish and Wildlife Service), archived at http://perma.cc/7ZEG-6RZ9.
• 52See, for example, Department of Commerce, National Oceanic and Atmospheric Administration, Endangered and Threatened Species; Identification of 14 Distinct Population Segments of the Humpback Whale (Megaptera novaeangliae) and Proposed Revision of Species-Wide Listing, 80 Fed Reg 22304, 22348 (2015), amending 50 CFR Parts 223–24 (“To the extent it may be said that the statute is ambiguous as to precisely how the updated listings should replace the original listing in such circumstances, we provide our interpretation of the statutory scheme.”); Department of Veterans Affairs, Health Care for Homeless Veterans Program, 76 Fed Reg 52575, 52577 (2011), amending 38 CFR Part 63 (“Even if the statute is ambiguous, our interpretation that it applies to veterans who are homeless and have a serious mental illness is consistent with Congress’ intent.”); Department of Health and Human Services, Centers for Medicare & Medicaid Services, Medicare Program; Payment Policies under the Physician Fee Schedule and Other Revisions to Part B for CY 2009; E-prescribing Exemption for Computer-Generated Facsimile Transmissions; and Payment for Certain Durable Medical Equipment, Prosthetics, Orthotics, and Supplies (DMEPOS), 73 Fed Reg 69726, 69911 (2008), amending various sections of CFR Title 42 (“First, we continue to believe that our interpretation is consistent with the plain language of the Medicare statute, and alternatively, if the statute is ambiguous to this point, we believe our interpretation best captures the Congress’ intent and is a reasonable and permissible interpretation.”); Department of Health and Human Services, Health Care Financing Administration, Medicare Program; Changes to the Hospital Inpatient Prospective Payment Systems and Fiscal Year 1998 Rates, 63 Fed Reg 26318, 26346 (1998), amending various sections of CFR Title 42:

  Taking into consideration the statutory language, the statutory scheme, and the legislative history, we believe the best reading of the statute enacted by the Congress is that we should calculate a single number for hospitals within each class and not apply a wage adjustment. We believe that, in any event, the Secretary’s policy is consistent with the statute and is reasonable.

• 53See note 29.
• 54See, for example, Wachtel v Office of Thrift Supervision, 982 F2d 581, 585 (DC Cir 1993) (rejecting the agency’s argument in favor of ambiguity as “almost frivolous”).
• 55318 US 80 (1943).
• 56463 US 29 (1983).
• 57By most measures, but not by all: Harlow v Fitzgerald, 457 US 800 (1982), is arguably an “admin law” case, and it has been cited more frequently than Chevron. See Chris Walker, Most Cited Supreme Court Administrative Law Decisions (Notice & Comment, Oct 9, 2014), archived at http://perma.cc/J26H-QVBW. And if one counts Daubert v Merrell Dow Pharmaceuticals, Inc, 509 US 579 (1993), and we don’t, Daubert comes out on top. Walker, Most Cited Supreme Court Administrative Law Decisions (cited in note 57).
• 58See Ronald M. Levin, The Anatomy of Chevron: Step Two Reconsidered, 72 Chi Kent L Rev 1253, 1253 (1997) (noting that Chevron “pervades the administrative law courses and lodges in the consciousness of impressionable students”).
• 59Chevron, 467 US at 842–43 (citations omitted).
• 60See id at 865–66:

  While agencies are not directly accountable to the people, the Chief Executive is, and it is entirely appropriate for this political branch of the Government to make such policy choices—resolving the competing interests which Congress itself either inadvertently did not resolve, or intentionally left to be resolved by the agency charged with the administration of the statute in light of everyday realities.

• 61Id at 865.
• 62Id at 843–44. To be sure, Stevens did not claim that Congress intentionally delegated such gap-filling authority to agencies. Rather, he stated that whether Congress did so intentionally or not, the Court would still infer such a delegation, presumably so that the question would be resolved by accountable and expert agencies rather than unelected and generalist judges. See id at 865:

  Perhaps [Congress] consciously desired the Administrator to strike the balance at this level, thinking that those with great expertise and charged with responsibility for administering the provision would be in a better position to do so; perhaps it simply did not consider the question at this level; and perhaps Congress was unable to forge a coalition on either side of the question, and those on each side decided to take their chances with the scheme devised by the agency. For judicial purposes, it matters not which of these things occurred.

  Many thus have dubbed this “implied delegation” a legal fiction. See, for example, John F. Manning, Chevron and the Reasonable Legislator, 128 Harv L Rev 457, 458 (2014). Today, however, there is more reason to think that Congress legislates against the backdrop of Chevron, see Abbe R. Gluck and Lisa Schultz Bressman, Statutory Interpretation from the Inside—an Empirical Study of Congressional Drafting, Delegation, and the Canons: Part I, 65 Stan L Rev 901, 995–97 (2013), though there is also evidence suggesting that some members of Congress do not want courts to apply Chevron, see Separation of Powers Restoration Act of 2016, HR 4768, 114th Cong, 2d Sess (July 13, 2016) (calling for de novo review of agency actions). For our purposes, it matters not whether Chevron reflects congressional intent. In either event, Chevron is still a decision of the Supreme Court entitled to stare decisis weight.

• 63Cass R. Sunstein, Beyond Marbury: The Executive’s Power to Say What the Law Is, 115 Yale L J 2580, 2584 (2006), quoting Marbury v Madison, 5 US (1 Cranch) 137, 177 (1803).
• 64Sunstein, 115 Yale L J at 2583 (cited in note 63).
• 65See Merrill and Hickman, 89 Georgetown L J at 836 (cited in note 3).
• 66See, for example, Jordan, 54 Admin L Rev at 725 (cited in note 4).
• 67533 US 218 (2001).
• 68Id at 231.
• 69529 US 120 (2000).
• 70Id at 159–60.
• 71135 S Ct 2480 (2015).
• 72Patient Protection and Affordable Care Act, Pub L No 111-148, 124 Stat 119 (2010).
• 73King, 135 S Ct at 2485.
• 74Id at 2489, quoting Utility Air Regulatory Group v Environmental Protection Agency, 134 S Ct 2427, 2444 (2014), quoting Brown & Williamson, 529 US at 160.
• 75Indeed, one might question whether the agency in Chevron fully acknowledged its own discretion under the Clean Air Act. See Environmental Protection Agency, Requirements for Preparation, Adoption and Submittal of Implementation Plans and Approval and Promulgation of Implementation Plans, 46 Fed Reg 50766, 50767 (1981), amending 40 CFR Parts 51–52:

  Congress expressly provided that states are to play the primary role in pollution control. It also intended that states retain the maximum possible flexibility to balance environmental and economic concerns in designing plans to clean up nonattainment areas. Today’s action follows this mandate by allowing states much greater flexibility in developing their nonattainment area NSR programs and attainment demonstrations.

  (citations omitted).

• 76Chevron, 467 US at 844.
• 77See generally Stephenson and Vermeule, 95 Va L Rev 597 (cited in note 5).
• 78Id at 599.
• 79See Entergy Corp v Riverkeeper, Inc, 556 US 208, 218 n 4 (2009) (“[S]urely if Congress has directly spoken to an issue then any agency interpretation contradicting what Congress has said would be unreasonable.”).
• 80See Encino Motorcars, LLC v Navarro, 136 S Ct 2117, 2124 (2016):

  In the usual course, when an agency is authorized by Congress to issue regulations and promulgates a regulation interpreting a statute it enforces, the interpretation receives deference if the statute is ambiguous and if the agency’s interpretation is reasonable. This principle is implemented by the two-step analysis set forth in Chevron.

  See also King, 135 S Ct at 2488–89 (“When analyzing an agency’s interpretation of a statute, we often apply the two-step framework announced in Chevron. . . . In extraordinary cases, however, there may be reason to hesitate before concluding that Congress has intended such an implicit delegation. This is one of those cases.”) (quotation marks, citation, and paragraph break omitted). Note, though, that in both of these cases, the Court ultimately declined to apply the Chevron framework, resolving the matter at what might be described as Chevron Step Zero, see Kristin E. Hickman, The (Perhaps) Unintended Consequences of King v. Burwell, 2015 Pepperdine L Rev 56, 57, or Chevron Step 0.5, see Daniel Hemel and Michael Pollack, Chevron Step 0.5 (Whatever Source Derived, June 23, 2016), archived at http://perma.cc/M4M2-WM7P (discussing Encino Motorcars).

• 81Re, 89 Ind L J at 619 (cited in note 5).
• 82Id at 641.
• 83National Association of Home Builders v Defenders of Wildlife, 551 US 644, 684 (2007) (Stevens dissenting). See also generally Kevin M. Stack, The Constitutional Foundations of Chenery, 116 Yale L J 952 (2007) (emphasizing the constitutional significance of the doctrine).
• 84Chenery I, 318 US at 87.
• 85See Merrick B. Garland, Deregulation and Judicial Review, 98 Harv L Rev 505, 526 n 106 (1985), quoting American Textile Manufacturers Institute, Inc v Donovan, 452 US 490, 539 (1981) (“A corollary of this rule is that ‘the post hoc rationalizations’ of the agency or its counsel ‘cannot serve as a sufficient predicate for agency action.’”).
• 86See, for example, Morgan Stanley Capital Group Inc v Public Utility District No 1 of Snohomish County, 554 US 527, 544–45 (2008):

  We will not uphold a discretionary agency decision where the agency has offered a justification in court different from what it provided in its opinion. But FERC has lucked out: The Chenery doctrine has no application to these cases, because we conclude that the Commission was required [to act as it did].

  (citation omitted).

• 87Aaron L. Nielson, Beyond Seminole Rock, 105 Georgetown L J 943, 958 (2017), quoting Chenery I, 318 US at 81.
• 88See Chenery I, 318 US at 89–90.
• 89Id at 95.
• 90See Securities & Exchange Commission v Chenery Corp, 332 US 194, 199, 209 (1947) (“Chenery II”). See also Nielson, 105 Georgetown L J at 959–60 (cited in note 87).
• 91Walker, Most Cited Supreme Court Administrative Law Decisions (cited in note 57).
• 92135 S Ct 2699 (2015).
• 93See id at 2710 (“This line of reasoning contradicts the foundational principle of administrative law that a court may uphold agency action only on the grounds that the agency invoked when it took the action.”). See also Citizens to Preserve Overton Park, Inc v Volpe, 401 US 402, 420 (1971) (stating that “post hoc rationalization[s] . . . must be viewed critically”) (quotation marks omitted).
• 94See, for example, National Labor Relations Board v Southwest Regional Council of Carpenters, 826 F3d 460, 465 (DC Cir 2016) (“Even if the Board had explained the relevance of these alleged factual differences, we cannot address this argument because it did not appear in the Board’s orders below.”).
• 95See, for example, Camp v Pitts, 411 US 138, 142–43 (1973) (per curiam) (allowing post hoc justifications when “there was such failure to explain administrative action as to frustrate effective judicial review”). But see Florida Power & Light Co v Lorion, 470 US 729, 744 (1985) (concluding that, even in such circumstances, remanding the case is often preferable).
• 96This was certainly the view of the majority in Prill itself. See Prill, 755 F2d at 948 (“We think that the teachings of Chenery are plainly implicated in this case.”).
• 97Endangered Species Act of 1973, S Rep No 93-307, 93d Cong, 1st Sess 3 (1973).
• 98Id.
• 99See, for example, Fournier v Sebelius, 718 F3d 1110, 1123 (9th Cir 2013) (holding that, at Chevron Step Two, “legislative history permissibly may be considered”).
• 100Of course, one could read Chenery I in a more maximalist fashion, such that scenario (a) would not be upheld because the agency did not acknowledge its discretion, just as if the agency had not acknowledged its discretion for reasons that do not sound in statutory interpretation. Even though the agency said its reading was consistent with the legislative history, by definition it could have reached another result despite the legislative history. Otherwise, this could not be a Chevron Step Two case. It is enough here to observe that it is not obvious that Chenery I should be read in such a maximalist fashion, though for what it is worth, one of the authors (Nielson) subscribes to a maximalist view, while the other (Hemel) is not yet persuaded. Perhaps tellingly, courts also appear to vary widely on how they understand Chenery I. See Bagley, 117 Colum L Rev at 306–07 (cited in note 12).
• 101See PDK Laboratories Inc v United States Drug Enforcement Administration, 362 F3d 786, 794 (DC Cir 2004) (refusing to grant Chevron deference when an agency official “cited ‘the legislative history’” of the statute in question “[a]s one of his reasons for thinking the statute clear”).
• 102State Farm, 463 US at 43, 52.
• 103See, for example, National Electrical Manufacturers Association v United States Department of Energy, 654 F3d 496, 504–05 (4th Cir 2011) (“[W]e have described legislative history as one of the traditional tools of interpretation to be consulted at Chevron’s step one.”).
• 104See, for example, Village of Barrington, Illinois v Surface Transportation Board, 636 F3d 650, 666 (DC Cir 2011) (“Although we would be uncomfortable relying on such legislative history at Chevron step one, we think it may appropriately guide an agency in interpreting an ambiguous statute.”).
• 105For a similar take, see Kavanaugh, Book Review, 129 Harv L Rev at 2137–38 (cited in note 8).
• 1063 USC § 1. Or so we would think. Perhaps one might argue that “the Tuesday next” means “next Tuesday” as opposed to “this Tuesday,” so if the first Monday in November is November 7, then “the Tuesday next” is November 15.
• 10747 USC § 310(d).
• 108State Farm, 463 US at 52.
• 109Prill, 755 F2d at 948 (quotation marks and brackets omitted), quoting Planned Parenthood Federation of America, Inc v Heckler, 712 F2d 650, 666 (DC Cir 1983) (Bork concurring in part and dissenting in part).
• 110Stuart Buck, Salerno vs. Chevron: What to Do about Statutory Challenges, 55 Admin L Rev 427, 472 (2003). See also Charles H. Koch Jr, Judicial Review of Administrative Discretion, 54 Geo Wash L Rev 469, 482 n 54 (1986):

  [T]he court found that the National Labor Relations Board (NLRB) failed to recognize that the broad standard conferred executing discretion and that the NLRB acted improperly by not exercising it. . . . [T]he court was perfectly correct in instructing the agency of the boundaries of that discretion and compelling it to exercise that discretion according to the breadth of the legislative standard.

• 111Arizona v Thompson, 281 F3d 248, 254 (DC Cir 2002), quoting Phillips Petroleum Co v Federal Energy Regulatory Commission, 792 F2d 1165, 1169 (DC Cir 1986).
• 112Prill, 755 F2d at 942.
• 113Id.
• 114Id.
• 11549 Stat 449 (1935), codified as amended at 29 USC § 151 et seq.
• 116See 29 USC § 157 (“Employees shall have the right . . . to engage in [ ] concerted activities for the purpose of collective bargaining or other mutual aid or protection.”).
• 117Prill, 755 F2d at 942.
• 118Id.
• 119Id.
• 120Id at 948 (concluding, inter alia, “that the Board erred in assuming that the NLRA mandates its present interpretation of ‘concerted activities’”).
• 121Prill, 755 F2d at 942 (citation omitted).
• 122Id at 948. See also Phillips Petroleum, 792 F2d at 1169 (“This deference, however, is only appropriate when the agency has exercised its own judgment. When, instead, the agency’s decision is based on an erroneous view of the law, its decision cannot stand.”). On remand, the NLRB adhered to its position that Prill’s activity was not “concerted”: it acknowledged that its interpretation of the NLRA “is not actually required by the NLRA but rather is most responsive to the central purposes for which the Act was created.” Prill v National Labor Relations Board, 835 F2d 1481, 1483 (DC Cir 1987) (quotation marks omitted). The DC Circuit affirmed the NLRB’s ruling the second time around. Id at 1482.
• 123Peter Pan, 471 F3d at 1352.
• 124Id at 1351–52.
• 125Pub L No 101-336, 104 Stat 327 (1990), codified as amended at 42 USC § 12101 et seq.
• 126See Peter Pan, 471 F3d at 1352–53.
• 127See id at 1353–55.
• 128Id at 1353–54 (brackets, paragraph breaks, and quotation marks omitted).
• 129Bagley, 117 Colum L Rev at 300 & n 326 (cited in note 12) (providing examples). See also note 29 (providing more examples).
• 1308 USC § 1101(a)(42).
• 131Negusie, 555 US at 514.
• 132See id at 517–18.
• 133Id at 521, quoting Brief for the Respondent, Negusie v Mukasey, Docket No 07-499, *11 (US filed Aug 15, 2008) (available on Westlaw at 2008 WL 3851621).
• 134Negusie, 555 US at 521.
• 135Id at 516. See also id at 523, quoting Cajun Electric Power Cooperative, Inc v Federal Energy Regulatory Commission, 924 F2d 1132, 1136 (DC Cir 1991) (“If an agency erroneously contends that Congress’ intent has been clearly expressed and has rested on that ground, we remand to require the agency to consider the question afresh in light of the ambiguity we see.”) (brackets omitted).
• 136The Supreme Court arguably indicated its endorsement of the Chevron Step One-and-a-Half doctrine long before Negusie, in Federal Communications Commission v RCA Communications, Inc, 346 US 86 (1953). The relevant statute directed the Federal Communications Commission to grant radio licenses whenever it “shall determine that public interest, convenience, or necessity would be served by the granting thereof.” Radio Act of 1927 § 11, 44 Stat 1162, 1167. The Commission took the position that “competition is in the public interest where competition is reasonably feasible.” RCA Communications, 346 US at 89 (quotation marks omitted). The Supreme Court ordered a lower court to remand the matter to the Commission. As Justice Felix Frankfurter wrote for the Court, “[T]he Commission professed to dispose of the case merely upon its view of a principle which it derived from the statute and did not base its conclusion on matters within its own special competence.” Id at 91. Because the Court did not think that the statute unambiguously stood for the principle that the Commission had derived, the Court ordered the Commission to reconsider its licensing decision, bringing its “own experience” to bear. Id at 91, 98.

  The Supreme Court’s decision in RCA Communications has all the essential features of a Chevron Step One-and-a-Half remand: the agency professed that the statute was unambiguous; the Court concluded that the agency in fact enjoyed discretion; and so the Court remanded the matter to the agency for a do-over. Yet RCA Communications preceded Chevron by more than three decades. To call it the genesis of the Chevron Step One-and-a-Half doctrine might be placing the cart before the horse. At the very least, though, RCA Communications suggests that the rule articulated in Prill has antecedents that extend much further back than Prill—indeed, much further back than Chevron itself.

• 137Martinez v Attorney General of United States, 693 F3d 408, 412 (3d Cir 2012) (quotation marks and brackets omitted). See also Murray v UBS Securities, LLC, 2013 WL 2190084, *6 (SDNY) (“The Second Circuit has not yet addressed the question of whether deference is only appropriate when the agency has exercised its own judgment, not when it believes that its interpretation is compelled by Congress.”) (quotation marks and brackets omitted).
• 138See Lansing Dairy, Inc v Espy, 39 F3d 1339, 1354–55 (6th Cir 1994) (“Because, as discussed above, the language of the Act and the legislative history are not unambiguous and the Secretary’s interpretation is neither unreasonable nor in conflict with Congress’ intent, we are bound by Chevron to uphold the Secretary’s interpretation.”).
• 139See Escobar Barraza v Mukasey, 519 F3d 388, 391 (7th Cir 2008):

  [W]e do not give any special weight to the agency’s construction. . . . That’s not because we think the statute unambiguous. . . . But the Board (acting by a single member) asserted that the statute is clear, and by forswearing any exercise of administrative discretion the Board also disabled its counsel from invoking the principle of Chevron.

  But see Madison Gas & Electric Co v United States Environmental Protection Agency, 25 F3d 526, 529 (7th Cir 1994):

  The EPA knows more about it than we do and under Chevron has the primary responsibility for interpreting undefined terms. But it has not furnished a reasoned ground for denying these utilities’ requests for additional allowances and it therefore must, if it wants to adhere to its denial, try again.

• 140Gila River Indian Community v United States, 729 F3d 1139, 1149 (9th Cir 2013) (“We conclude that the Secretary’s interpretation warrants no deference because it rests on a mistaken conclusion that the language has a plain meaning.”).
• 141See American Federation of Government Employees, Local 1592 v Federal Labor Relations Authority, 836 F3d 1291, 1295–96 (10th Cir 2016), citing PDK Laboratories, 362 F3d at 798, and Peter Pan, 471 F3d at 1354; National Organization of Veterans’ Advocates, Inc v Secretary of Veterans Affairs, 314 F3d 1373, 1378–79 (Fed Cir 2003), citing Transitional Hospitals Corp of Louisiana, Inc v Shalala, 222 F3d 1019, 1029 (DC Cir 2000).
• 142See, for example, Environmental Protection Agency, Prevention of Significant Deterioration and Title V Greenhouse Gas Tailoring Rule, 75 Fed Reg 31514, 31528 (2010), amending various sections of CFR Title 40 (“We believe that this approach is both compelled by the statute and reflects the preferable policy approach.”); Local Union 1261, District 22, United Mine Workers of America v Federal Mine Safety and Health Review Commission, 917 F2d 42, 43 (DC Cir 1990) (upholding an agency decision despite disagreement over whether the statute was “clear” because the agency also “adequately stated the practical and policy considerations ultimately motivating its interpretation”).
• 143See text accompanying notes 140–41.
• 144We refer to “rules” in text but note that Chevron Step One-and-a-Half applies with equal force to agency statutory interpretations in adjudicative proceedings. See, for example, Alarm Industry Communications Committee v Federal Communications Commission, 131 F3d 1066, 1072 (DC Cir 1997); City of Los Angeles Department of Airports v United States Department of Transportation, 103 F3d 1027, 1034 (DC Cir 1997); Baltimore and Ohio Railroad Co v Interstate Commerce Commission, 826 F2d 1125, 1129 (DC Cir 1987).
• 145See note 52.
• 146Chevron Step One-and-a-Half can be applied to both agency rulemakings and agency adjudications—it applies whenever an agency does not recognize that it has interpretative discretion. For purposes of this Article, our analysis does not turn on this distinction. In the future, however, this may be a fruitful avenue of research. For instance, perhaps agencies are more “strategic” in rulemakings than adjudications, or vice versa. Such questions, while important, are beyond the scope of this Article.
• 147See generally Stephen G. Breyer, et al, Administrative Law and Regulatory Policy: Problems, Text, and Cases (Wolters Kluwer Law & Business 7th ed 2011); Jerry L. Mashaw, et al, Administrative Law: The American Public Law System; Cases and Materials (West Academic 7th ed 2014).
• 148325 US 410 (1945).
• 149519 US 452 (1997).
• 150See Christopher J. Walker, Inside Agency Statutory Interpretation, 67 Stan L Rev 999, 1019 (2015).
• 151See, for example, Alarm Industry Communications Committee, 131 F3d at 1068 (rejecting an agency interpretation as “senseless”).
• 152See, for example, Library of Congress, Copyright Office, Copyright Arbitration Royalty Panels, 59 Fed Reg 63025, 63032 (1994), amending various sections of CFR Title 37 (“The method of payment of the CARPs is a problem, especially given our lack of statutory authority to pay the arbitrators directly. Unfortunately, the legislative solution discussed in the Interim Regulations . . . failed for this Congress.”); National Credit Union Administration, Truth in Savings, 58 Fed Reg 50394, 50430 (1993), amending various sections of CFR Title 12:

  The Board has been very concerned over the potential adverse effect that TISA and part 707 will have on small credit unions that have insufficient computer capacity to provide the statement disclosures required under § 707.6. . . . Unfortunately, Congress, when it enacted TISA, gave the Board little leeway to mitigate the statute’s impact on such credit unions.

  See also Department of Justice, Redress Provisions for Persons of Japanese Ancestry, 54 Fed Reg 34157, 34158 (1989), amending 28 CFR Part 74:

  Unfortunately, however, section 108(2) of the Civil Liberties Act of 1988 limits the definition of an “eligible individual” specifically to “any individual of Japanese ancestry.” Indeed, the focus throughout the Act is on those of Japanese ancestry and the discrimination they suffered based on their race. In light of the specificity with which Congress has spoken and its focus on the racial discrimination suffered, it must be concluded that the statute authorizes compensation be paid only to those of Japanese ancestry, and not to those who are of non-Japanese ancestry but who were nevertheless interned.

  . . . Therefore, the Department will submit legislation to the Congress to amend the Civil Liberties Act of 1988 to render eligible those non-Japanese family members who suffered the effects of the government’s internment policy by accompanying their spouses or children of Japanese ancestry through the evacuation and internment process.

• 153See, for example, Meshal v Higgenbotham, 804 F3d 417, 431 (DC Cir 2015) (Kavanaugh concurring) (“If I were a Member of Congress, I might vote to enact a new tort cause of action to cover a case like Meshal’s. But as judges, we do not get to make that decision.”); Caterpillar, Inc v International Union, United Automobile, Aerospace and Agricultural Implement Workers of America, 107 F3d 1052, 1066 (3d Cir 1997) (en banc) (Alito dissenting) (“If I were a legislator, I would not vote to criminalize the payments to grievance chairmen that are at issue here. . . . Our job, however, is to interpret Section 302 as it is written.”) (paragraph break omitted).
• 154See Elizabeth Magill and Adrian Vermeule, Allocating Power within Agencies, 120 Yale L J 1032, 1036–38 (2011).
• 155To be sure, the officials within the agency who favor position Y might also challenge the general counsel’s view that the statute unambiguously means X. On agency general counsels as “keepers of what the statute means,” see E. Donald Elliott, Chevron Matters: How the Chevron Doctrine Redefined the Roles of Congress, Courts and Agencies in Environmental Law, 16 Vill Envir L J 1, 11 (2005).
• 156Id at 11–12.
• 157On the role of general counsels within administrative agencies, see Magill and Vermeule, 120 Yale L J at 1059–61 (cited in note 154); Jennifer Nou, Intra-agency Coordination, 129 Harv L Rev 421, 444–47 (2015).
• 158James Q. Wilson, The Politics of Regulation, in James Q. Wilson, ed, The Politics of Regulation 357, 373–74 (Basic Books 1980). See also Magill and Vermeule, 120 Yale L J at 1038 (cited in note 154) (“[T]he basic points are simple: agencies contain identifiable constituencies that affect policymaking, and these constituencies can, and do, come into conflict over the proper functioning of the agency.”).
• 159See 42 USC § 2000e-4(a), (b)(1).
• 160Of course, a Chevron Step One-and-a-Half issue arises only if the EEOC passes Chevron Step Zero. Some EEOC regulations—but not all—are considered Chevron eligible. Compare Federal Express Corp v Holowecki, 552 US 389, 395 (2008) (“The [EEOC] has statutory authority [under the Age Discrimination in Employment Act] to issue regulations, . . . and when an agency invokes its authority to issue regulations, which then interpret ambiguous statutory terms, the courts defer to its reasonable interpretations.”), with Equal Employment Opportunity Commission v Arabian American Oil Co, 499 US 244, 256–58 (1991) (reasoning that there is no Chevron deference to EEOC guidelines under Title VII).
• 161For the influence of career staffers over agency rules, see generally Daniel E. Walters, Litigation-Fostered Bureaucratic Autonomy: Administrative Law against Political Control, 28 J L & Polit 129 (2013).
• 162See Neomi Rao, Public Choice and International Law Compliance: The Executive Branch Is a “They,” Not an “It”, 96 Minn L Rev 194, 197 (2011).
• 163See, for example, Daniel A. Farber and Anne Joseph O’Connell, The Lost World of Administrative Law, 92 Tex L Rev 1137, 1138–40 (2014) (explaining how agencies and the White House may disagree on policy).
• 164See Cass R. Sunstein, The Office of Information and Regulatory Affairs: Myths and Realities, 126 Harv L Rev 1838, 1840, 1854, 1858 (2013).
• 165For a more comprehensive overview, see generally Jennifer Nou, Agency Self-Insulation under Presidential Review, 126 Harv L Rev 1755 (2013).
• 166See Elena Kagan, Presidential Administration, 114 Harv L Rev 2245, 2248 (2001).
• 167Executive Order 12291 § 3 (1981), 3 CFR 127, 128–30. Reagan later issued Executive Order 12498, which built upon the earlier order. See Executive Order 12498 (1985), 3 CFR 323.
• 168See Executive Order 12291 § 3(a), (d), 3 CFR at 128–29.
• 169See Executive Order 12866 (1993), 3 CFR 638.
• 170Kagan, 114 Harv L Rev at 2285 (cited in note 166).
• 171Executive Order 13563 § 1(b) (2011), 3 CFR 215, 215.
• 172See Executive Order 13771 § 1, 82 Fed Reg 9339, 9339 (2017) (providing that for every new regulation issued by an executive agency, two existing regulations must be “identified for elimination”); Executive Order 13777 §§ 2–3, 82 Fed Reg 12285, 12285–86 (2017) (directing each executive agency to designate an agency official as a “Regulatory Reform Officer” and to convene a “Regulatory Reform Task Force”).
• 173See Executive Order 13777 § 2(i)–(ii), 82 Fed Reg at 12285 (cited in note 172) (making specific reference to Clinton’s Executive Order 12866 and Obama’s Executive Order 13563).
• 174Executive Order 13563 § 1(b), 3 CFR at 215 (emphasis added).
• 175See Lisa Heinzerling, Statutory Interpretation in the Era of OIRA, 33 Fordham Urban L J 1097, 1100 (2006) (“From the beginning, [ ] it has been clear that, in reviewing the regulatory initiatives of its sister agencies, OIRA may not interfere with the agencies’ compliance with statutory directives.”).
• 176See, for example, Lisa Heinzerling, Inside EPA: A Former Insider’s Reflections on the Relationship between the Obama EPA and the Obama White House, 31 Pace Envir L Rev 325, 348 (2014):

  OIRA, in any event, lavishes skeptical attention on EPA’s estimates of regulatory costs. . . . And in my experience, OIRA personnel keep an eagle eye on EPA—on its public announcements, website, etc.—to make sure EPA does not sneak something past it. From OIRA’s perspective, the system appears to work: EPA receives more sustained attention from OIRA than any other federal agency. Most often, EPA is the agency with the largest number of rules under review at OIRA.

• 177Indeed, it is worth noting that several OIRA administrators in the past have been nonlawyers, including James C. Miller III (1981), Wendy Lee Gramm (1985–1988), John Graham (2001–2006), and Susan Dudley (2007–2009) (though, of course, all have had the counsel of lawyers on OIRA’s staff).
• 178See, for example, Steve R. Johnson, The Dangers of Symbolic Legislation: Perceptions and Realities of the New Burden-of-Proof Rules, 84 Iowa L Rev 413, 477 & n 281 (1999) (“[L]egislators will try to shift blame to the IRS for the embarrassment of Congress’s own making.”).
• 179See generally In re Polar Bear Endangered Species Act Listing and § 4(d) Rule Litigation, 748 F Supp 2d 19 (DDC 2010).
• 180See Activists Slam Salazar’s ESA Decision (Clean Air Report, May 14, 2009) (available on Westlaw at 2009 WLNR 9095136) (“Activists and Democrats are criticizing Interior Secretary Ken Salazar’s just-announced decision to uphold a controversial Bush administration rule on polar bear protections that limits the government’s ability to use the Endangered Species Act (ESA) to regulate greenhouse gases.”). Note that the polar bear listing rule was initially promulgated by the Bush administration, which may have been less reliant on environmentalist constituencies. See Josey Ballenger, Bush’s Choice of EPA Advisers Signals Tilt toward Industry (Center for Public Integrity, Feb 12, 2001), archived at http://perma.cc/3JWG-MM6G. Our political explanation accounts for why the Obama administration stuck to the Bush administration’s course—though not necessarily why the Bush administration framed the statute as unambiguous in the first place. As to the latter point (or, chronologically, the former point in time), see the next Section’s discussion of interadministration politics.
• 181Of course, there are other possible interbranch political dynamics. For instance, perhaps Congress—if it dislikes what the agency has done—may seize on a Chevron Step One-and-a-Half decision to castigate the agency; losing is not popular. Needless to say, how the branches interact is a complicated subject that merits careful examination.
• 182On efforts by agencies to entrench their policy preferences before the handover to a new administration, see generally Nina A. Mendelson, Agency Burrowing: Entrenching Policies and Personnel before a New President Arrives, 78 NYU L Rev 557 (2003).
• 183545 US 967 (2005).
• 184Id at 982.
• 185Kimble v Marvel Entertainment, LLC, 135 S Ct 2401, 2409 (2015).
• 186See, for example, Stryker v Securities and Exchange Commission, 780 F3d 163, 166 (2d Cir 2015) (“We need not, however, decide if Congress clearly intended to bar a whistleblower award to petitioner at Chevron Step 1 because even if Dodd–Frank is ambiguous, we defer to the SEC’s interpretation of Dodd–Frank at Step 2.”).
• 187See generally Aaron L. Nielson, Sticky Regulations, 85 U Chi L Rev (forthcoming 2018), archived at http://perma.cc/BU26-BCV2.
• 188Bagley, 117 Colum L Rev at 300 (cited in note 12).
• 189Id at 301.
• 190PDK Laboratories Inc v United States Drug Enforcement Administration, 362 F3d 786, 808 (DC Cir 2004) (Roberts concurring in part and concurring in the judgment).
• 191Bagley, 117 Colum L Rev at 300 n 326 (cited in note 12).
• 19248 Stat 1064 (1934), codified as amended in various sections of Title 47.
• 193Pub L No 104-104, 110 Stat 56, codified as amended in various sections of Title 47.
• 194Telecommunications Act of 1996 § 275(a), 110 Stat at 105, codified at 47 USC § 275(a).
• 195In the Matter of Enforcement of Section 275(a)(2) of the Communications Act of 1934, as Amended by the Telecommunications Act of 1996, against Ameritech Corp, 12 FCC Rec 3855, 3859 (1997).
• 196Alarm Industry Communications Committee v Federal Communications Commission, 131 F3d 1066, 1068–72 (DC Cir 1997).
• 197See In the Matter of Enforcement of Section 275(a)(2) of the Communications Act of 1934, as Amended by the Telecommunications Act of 1996, against Ameritech Corp, 13 FCC Rec 19046, 19052 (1998).
• 198Telecommunications Act of 1996 § 254(g), 110 Stat at 73, codified as amended at 47 USC § 254(g).
• 199See GTE Service Corp v Federal Communications Commission, 224 F3d 768, 770–71 (DC Cir 2000).
• 200Id at 775–76.
• 201In the Matter of South Seas Broadcasting, Inc, 26 FCC Rec 4164, 4167 (2011).
• 202PanAmSat Corp v Federal Communications Commission, 198 F3d 890, 892–94 (DC Cir 1999).
• 203Id at 896.
• 204Comsat Corp v Federal Communications Commission, 283 F3d 344, 346–47 (DC Cir 2002).
• 205See, for example, Coalition for Common Sense in Government Procurement v United States, 671 F Supp 2d 48, 56 (DDC 2009); Coalition for Common Sense in Government Procurement v United States, 821 F Supp 2d 275, 279 n 4 (DDC 2011).
• 206See, for example, PDK Laboratories, 362 F3d at 797–98 (explaining that Chevron Step Two requires that the agency “must bring its experience and expertise to bear in light of competing interests at stake”).
• 207Given “logrolling” within an agency (that is, trading votes across issues), Chevron Step One-and-a-Half may affect outcomes on the ground in other ways. See Jim Rossi, Book Review, Public Choice Theory and the Fragmented Web of the Contemporary Administrative State, 96 Mich L Rev 1746, 1753 & n 32 (1998); Glen Staszewski, Textualism and the Executive Branch, 2009 Mich St L Rev 143, 170. An agency general counsel’s assertion that a statute is unambiguous may take the relevant issue off the table for purposes of intra-agency bargaining, whereas if the issue were the subject of bargaining, the agency’s position not on that particular issue but on other issues might change. Imagine that on Issue 1, the agency chooses between options X and Y, while on Issue 2, the agency chooses between options P and Q. If the agency general counsel says that the statute compels X, then the forces in favor of P may prevail in the P-versus-Q debate. If, however, a court remands on the ground that the statute allows for X or Y, relevant actors within the agency who favor X over Y may agree to support option Q over P in return for others within the agency coming around to the X position. The court’s Chevron Step One-and-a-Half remand on the X-versus-Y question may have no effect on that particular issue, but may sway the outcome from P to Q.
• 208See Sunstein, 126 Harv L Rev at 1855 (cited in note 164).
• 209See David J. Barron and Elena Kagan, Chevron’s Nondelegation Doctrine, 2001 S Ct Rev 201, 238.
• 210Pension Benefit Guaranty Corp v The LTV Corp, 496 US 633, 651–52 (1990).
• 211See generally, for example, Heinzerling, 33 Fordham Urban L J 1097 (cited in note 175).
• 212See Kagan, 114 Harv L Rev at 2376 (cited in note 166).
• 213For a broader critique of accountability arguments in administrative law, see generally Nicholas O. Stephanopoulos, Discounting Accountability (unpublished manuscript, 2016), archived at http://perma.cc/M7GF-URYZ.
• 214See, for example, Glen Staszewski, Reason-Giving and Accountability, 93 Minn L Rev 1253, 1271–77 (2009) (explaining “Partial, Proxy, and Direct Political Accountability” theories).
• 215See text accompanying notes 19 and 189.
• 216See Jacob E. Gersen and Anne Joseph O’Connell, Deadlines in Administrative Law, 156 U Pa L Rev 923, 936–37 (2008) (“The future legislature can always repeal or alter the program, but once regulations have been implemented, some form of status quo bias may make it marginally harder to eliminate them—especially during periods of divided government.”). Consider also John F. Manning, Lawmaking Made Easy, 10 Green Bag 2d 191 (2007) (“The cumbersomeness of the process seems obviously suited to interests that contradict the ‘more is better’ attitude that has come to be almost an unconscious assumption of public law.”).
• 217See generally Richard J. Pierce Jr, Rulemaking Ossification Is Real: A Response to Testing the Ossification Thesis, 80 Geo Wash L Rev 1493 (2012).
• 218So far, we have been operating under the assumption that other actors within the agency are unaware of the Chevron Step One-and-a-Half doctrine. Relaxing this assumption leads to a further argument in Chevron Step One-and-a-Half’s favor. This is because even if other actors within the agency are sufficiently sophisticated to understand the contours of Chevron Step One-and-a-Half, it may still be the case that the agency general counsel, by virtue of the legal expertise at her disposal, remains the agency actor best able to predict how a court will interpret a statute. For this reason, other agency actors may value the general counsel’s opinion as to whether a statute unambiguously means X. But in the absence of Chevron Step One-and-a-Half, the general counsel’s assertion that “the statute unambiguously means X” is cheap talk: other actors within the agency have little reason to credit the general counsel’s assertion, given that the assertion is costless from the general counsel’s perspective. With Chevron Step One-and-a-Half in the background, other agency actors may be more likely to believe the general counsel’s assertion that “the statute unambiguously means X,” because if the reviewing court concludes otherwise, a remand will impose costs on the agency general counsel (for example, the costs of further rulemaking proceedings and potential litigation). Thus, by attaching a cost to the general counsel’s assertion regarding a statute’s unambiguous meaning, the Chevron Step One-and-a-Half doctrine potentially allows for more efficient intra-agency information sharing. That is, Chevron Step One-and-a-Half may allow the general counsel to communicate her interpretation of the statute’s meaning to other agency actors—and to do so credibly.
• 219See Gersen and O’Connell, 156 U Pa L Rev at 936 (cited in note 216) (noting tension between present and future versions of the agency).
• 220See Nielson, 105 Georgetown L J at 979–82 (cited in note 87) (listing examples of strategic behavior).
• 221Chevron, 467 US at 842–43.
• 222See, for example, Kavanaugh, Book Review, 129 Harv L Rev at 2118 (cited in note 8).
• 223One such case, Jacoby v National Labor Relations Board, 233 F3d 611 (DC Cir 2000), was handed down less than six weeks before the end of the Clinton administration and the beginning of the presidency of George W. Bush. We doubt, though, that Jacoby is an example of strategic behavior by a court that was ideologically adverse to the agency and stalling for time. First, the relevant agency—the National Labor Relations Board—is an independent agency whose members serve five-year terms, and Clinton-appointed Democratic members continued to constitute a majority of the Board for months after Bush’s inauguration. (Indeed, the Board did not become majority Republican until January 2002.) See John D. Schulz, Changing Faces (JOC, Oct 20, 2002), online at http://www.joc.com/trucking-logistics/changing-faces_20021020.html (visited Apr 9, 2017) (Perma archive unavailable); Board Members since 1935 (National Labor Relations Board), archived at http://perma.cc/55Y9-CE94. This was thus not a case in which delaying a resolution for a few months until a new administration took power was likely to change the result. Second, Jacoby is a case that does not map easily onto ideological lines: unlike an employee-versus-employer case (in which one might expect the Board’s more liberal members to be more sympathetic toward the employee), and unlike a union-versus-employer case (in which one might expect the Board’s more liberal members to be more sympathetic toward the union), Jacoby was an employee-versus-union case. The NLRB’s decision had the support of one of its two Republican members at the time, see generally Plumbers Local 342 (Contra Costa Electric), 329 NLRB 688 (1999), and the DC Circuit’s unanimous opinion was issued by an ideologically diverse panel including two Reagan appointees (Judge Stephen Williams and Judge David Sentelle) as well as one Clinton appointee (Judge Judith Rogers). Jacoby, 233 F3d at 613.
• 22460 Stat 237 (1946), codified as amended at 5 USC § 551 et seq.
• 225See, for example, Environmental Protection Agency, Air Quality Plans; Florida; Infrastructure Requirements for the 2012 PM2.5 National Ambient Air Quality Standard, 81 Fed Reg 50416, 50418 (2016), amending 40 CFR Part 52:

  EPA therefore believes that while the timing requirement in section 110(a)(1) is unambiguous, some of the other statutory provisions are ambiguous. In particular, EPA believes that the list of required elements for infrastructure SIP submissions provided in section 110(a)(2) contains ambiguities concerning what is required for inclusion in an infrastructure SIP submission.

  See also Securities and Exchange Commission, Disclosure of Payments by Resource Extraction Issuers, 81 Fed Reg 49360, 49386 n 351 (2016), amending various sections of CFR Title 17:

  We acknowledge that the statutory interpretation arguments we identify do not demonstrate an unambiguous Congressional intent to require public disclosure. . . . We believe that, at a minimum, Congress provided the Commission with discretionary authority. As such, based on our assessment of the record evidence and our weighing of the various policy considerations, we have determined to exercise that discretion by requiring public disclosure of each issuer’s annual report on Form SD.

  See also Nuclear Regulatory Commission, Final Procedures for Conducting Hearings on Conformance with the Acceptance Criteria in Combined Licenses, 81 Fed Reg 43266, 43278 (2016) (“[T]he statutory language is not clear and unambiguous.”).

• 226In the polar bear case, for example, counsel for the Fish and Wildlife Service “conceded at oral argument that the agency does not seek deference to its interpretation of the definition of an endangered species under step two of the Chevron test and instead relies exclusively on a plain-meaning interpretation of the ESA.” In re Polar Bear Endangered Species Act Listing and § 4(d) Rule Litigation, 748 F Supp 2d 19, 29 (DDC 2010). The district court noted, though, that absent this concession it would have been difficult to determine whether the Service considered the statute to be clear or whether it considered the statute to be ambiguous and had adopted its interpretation as an exercise of discretion. See id at 29 n 16.
• 227667 F3d 1284 (DC Cir 2012).
• 228Id at 1288–89 (emphasis added). See also Labor, Mine Safety and Health Administration v National Cement Co of California, 494 F3d 1066, 1074 (DC Cir 2007) (focusing on an agency assertion that “the definition of ‘coal or other mine’ plainly includes a road such as the one at issue”); Arizona v Thompson, 281 F3d 248, 253 (DC Cir 2002) (focusing on an agency assertion that “the TANF legislation does not permit it being designated as the primary program”) (ellipses and emphasis omitted).
• 229Peter Pan, 471 F3d at 1354 (emphasis added).
• 230See Bagley, 117 Colum L Rev at 255 (cited in note 12) (noting “systematic inattention to remedial questions” in administrative law).
• 2315 USC § 706.
• 232See, for example, Comcast Corp v Federal Communications Commission, 579 F3d 1, 10 (DC Cir 2009) (Randolph concurring) (“I continue to believe that whenever a reviewing court finds an administrative rule or order unlawful, the Administrative Procedure Act requires the court to vacate the agency’s action.”). See also generally Daniel B. Rodriguez, Of Gift Horses and Great Expectations: Remands without Vacatur in Administrative Law, 36 Ariz St L J 599 (2004).
• 233988 F2d 146 (DC Cir 1993).
• 234Id at 150–51, quoting International Union, United Mine Workers of America v Federal Mine Safety and Health Administration, 920 F2d 960, 967 (DC Cir 1990).
• 235Allied–Signal, 988 F2d at 148–49.
• 236Id at 150.
• 237Id at 151.
• 238Id. See also Ronald M. Levin, “Vacation” at Sea: Judicial Remedies and Equitable Discretion in Administrative Law, 53 Duke L J 291, 374–75 (2003). See also generally Kristina Daugirdas, Note, Evaluating Remand without Vacatur: A New Judicial Remedy for Defective Agency Rulemakings, 80 NYU L Rev 278 (2005).
• 239See Bagley, 117 Colum L Rev at 307 (cited in note 12) (“Other circuits—including the First, Third, Fifth, Eighth, Ninth, Tenth, and Federal—have followed the D.C. Circuit’s lead.”). See also, for example, Natural Resources Defense Council v United States Environmental Protection Agency, 808 F3d 556, 584 (2d Cir 2015) (finding an agency action to be arbitrary and capricious and remanding without vacating); California Communities against Toxics v United States Environmental Protection Agency, 688 F3d 989, 992 (9th Cir 2012) (per curiam) (“A flawed rule need not be vacated.”).
• 240See Bagley, 117 Colum L Rev at 308 (cited in note 12).
• 241671 F Supp 2d 48 (DDC 2009).
• 242Id at 53–54.
• 243See id at 55–56 (explaining that “[h]aving concluded that the statutory language does not speak to precisely how the Department should implement the statute, the Court ordinarily would move to Chevron step two,” but stating that it accepted the plaintiff’s argument that it could not “do so here because the rule’s preamble reveals that the agency mistakenly believed that Congress mandated the requirement of a manufacturer refund”).
• 244See id at 59.
• 245See In re Polar Bear, 748 F Supp 2d at 30:

  [T]he Court hereby remands the Listing Rule to the agency for the limited purpose of providing additional explanation for the legal basis of its listing determination, and for such further action as it may wish to take in light of the Court’s finding that the definition of an “endangered species” under the ESA is ambiguous.

• 246Mark Seidenfeld, Cognitive Loafing, Social Conformity, and Judicial Review of Agency Rulemaking, 87 Cornell L Rev 486, 488, 514–15 (2002).
• 247See generally Gary S. Becker, Crime and Punishment: An Economic Approach, 76 J Polit Econ 169 (1968) (theorizing a model for the optimal choice of punishment in order to deter wrongdoing while minimizing social loss).
• 248Consider too that the higher the cost to the court of a Chevron Step One-and-a-Half remand, the more reluctant the court will be to invoke the doctrine. For that reason, a variant of Chevron Step One-and-a-Half that forces the court to rewrite the agency’s rule on its own may be a less effective deterrent than a milder version.
• 249See, for example, Perez v Mortgage Bankers Association, 135 S Ct 1199, 1211 (2015) (Scalia concurring in the judgment) (“Heedless of the original design of the APA, we have developed an elaborate law of deference to agencies’ interpretations of statutes and regulations.”).
• 250See generally Gillian E. Metzger, Embracing Administrative Common Law, 80 Geo Wash L Rev 1293 (2012).
• 251Vermont Yankee Nuclear Power Corp v Natural Resources Defense Council, Inc, 435 US 519, 524 (1978).
• 252435 US 519 (1978).
• 253See Part II.B. Of course, some may grumble that this defense of Chevron Step One-and-a-Half just kicks the can down the road. After all, why is Chenery I consistent with the APA? See, for example, Bagley, 117 Colum L Rev at 293–95 (cited in note 12) (questioning whether a broad view of Chenery I comports with the APA’s instruction that courts apply harmless-error principles). Yet Chenery I has a very strong claim to stare decisis—it has been the law for more than seventy years. Equally important, it also predates the APA, so presumably Congress legislated against the Chenery I backdrop. See, for example, Cannon v University of Chicago, 441 US 677, 698–99 (1979) (“[O]ur evaluation of congressional action . . . must take into account its contemporary legal context.”).
• 254US Const Art I, § 8, cl 17.
• 255PDK Laboratories Inc v United States Drug Enforcement Administration, 362 F3d 786, 809 (DC Cir 2004) (Roberts concurring in part and concurring in the judgment).