Chevron is Dead; Long Live Chevron

Chevron is Dead; Long Live Chevron

The Supreme Court’s decision in Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc. continues to obsess academics and courts alike. Despite all the attention, however, the “Chevron revolution” never quite happens. This decision, though seen as trans­formatively important, is honored in the breach, in constant danger of being abandoned, and the subject of perpetual confusion and uncertainty. This Essay seeks both to bury and to praise Chevron.

Chevron is not a revolutionary shift of authority from the judiciary to the executive. That Chevron is dead. The Chevron that survives is an appropriate allocation of decisionmaking responsibility among the three branches, relying on the judiciary to enforce congres­sional decisions, but protecting agency authority and discretion where Congress has left the decision to the executive. Long may it reign.

INTRODUCTION

  1. CHEVRON AS SELF-REGULATION
  2. A VIABLE CHEVRON (AND SKIDMORE)
    1. Doctrinal Consequences of the Categorical Model
    2. Skidmore Within Chevron
    3. It’s Not Just “Deference” That Is Too Confusing—A Chevron Lexicon
  3. A CASE STUDY: CHEVRON AND JURISDICTIONAL QUESTIONS
    1. Defining “Jurisdictional” Questions
    2. “Jurisdictional” Questions and the APA

CONCLUSION

 

Introduction

At this point, it takes chutzpah to write about Chevron. 1 Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984). Everyone is sick to death of Chevron, and four gazillion other people have written about it, creating a huge pile of scholarship and precious little left to say. Most daunting, Peter Strauss has contributed significantly to the existing pile. 2 See generally Kenneth A. Bamberger & Peter L. Strauss, Chevron’s Two Steps, 95 Va. L. Rev. 611 (2009); Peter L. Strauss, “Deference” Is Too Confusing—Let’s Call Them “Chevron Space” and “Skidmore Weight”, 112 Colum. L. Rev. 1143 (2012) [hereinafter Strauss, Space/Weight]; Peter L. Strauss, In Search of Skidmore, 83 Fordham L. Rev. 789 (2014) [hereinafter Strauss, In Search of Skidmore]; Peter L. Strauss, One Hundred Fifty Cases per Year: Some Implications of the Supreme Court’s Limited Resources for Judicial Review of Agency Action, 87 Colum. L. Rev. 1093 (1987) [hereinafter Strauss, 150 Cases]; Peter L. Strauss, Overseers or “The Deciders”—The Courts in Administrative Law, 75 U. Chi. L. Rev. 815 (2008) [hereinafter Strauss, Overseers or Deciders]. If Professor Strauss has half-a-dozen articles on a subject, the sensible thing is to get one’s hands on them, read and learn from them, and keep quiet.

Nonetheless, here is an addition to the pile. It comes at a critical moment for the Chevron doctrine. In 2007, Linda Jellum reported Chevron’s “demise.” 3 Linda Jellum, Chevron’s Demise: A Survey of Chevron from Infancy to Senescence, 59 Admin. L. Rev. 725, 772 (2007). “Demise” is a strong word, as Mark Twain famously noted. Still, reports of Chevron’s death seemed to get significant confir­mation at the end of the Supreme Court’s 2014–2015 Term, when the Court decided three important cases that suggested that Chevron’s condition was, if not terminal, at least serious.

First, in King v. Burwell, 4 135 S. Ct. 2480, 2495–96 (2015) (holding insurance exchange set up by Department of Health and Human Services qualified as “exchange established by a state” for purposes of participants’ eligibility for tax credits to cover premium payments). a challenge to an Internal Revenue Service (IRS) rule regarding eligibility for tax credits under the Affordable Care Act, 5 Patient Protection and Affordable Care Act, Pub. L. No. 111-148, 124 Stat. 119 (2010) (codified as amended in scattered sections of 26 and 42 U.S.C.). Chief Justice Roberts perfunctorily rejected the claim that Chevron applied, brushing the case aside like a slightly annoying but unthreat­ening bug. The relevant statutory provision was in the Internal Revenue Code; 6 I.R.C. § 36B (2012). the statute granted the IRS authority to write all necessary regu­lations to implement the provision; 7 Id. § 36B(g) (authorizing IRS to “prescribe such regulations as may be necessary to carry out the provisions of this section”). The IRS already had authority to “prescribe all needful rules and regulations for the enforcement of this title.” Id. § 7805(a). the IRS had, through notice-and-comment rulemaking and in express reliance on that authority, issued a regulation directly addressing the legal question in the case. 8 Health Insurance Premium Tax Credit, 77 Fed. Reg. 30,377, 30,385 (May 23, 2012) (codified at 26 C.F.R. pts. 1, 602). Yet in two succinct paragraphs, the Chief Justice concluded that Chevron simply did not apply—the issue was of such “economic and political significance” that it was inconceivable that Congress would simply have left it to the IRS to resolve. 9 King, 135 S. Ct. at 2488–89. This was a particularly robust application of the “major questions doctrine,” which holds that judicial deference is out of place with regard to hugely significant policy questions—the sort of issues that Congress should, or must, or can be presumed to, decide. 10 Other examples include Utility Air Regulatory Group v. EPA, 134 S. Ct. 2427, 2444 (2014) (“EPA’s interpretation is also unreasonable because it would bring about an enormous and transformative expansion in EPA’s regulatory authority without clear con­gressional authorization.”) and FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 160 (2000) (“[W]e are confident that Congress could not have intended to delegate a decision of such economic and political significance to an agency in so cryptic a fashion.”). Strikingly, the magnitude of the issue did not simply keep the Court in “step one” of Chevron, it induced the Court to jettison Chevron altogether.

Second, in Michigan v. EPA, 11 135 S. Ct. 2699 (2015). the Court rejected the Environmental Protection Agency’s (EPA) reading of a provision of the Clean Air Act. This time, it accepted that Chevron applied, but rejected the EPA’s view as unreasonable, a rare “step two” agency loss. 12 Id. at 2706–07 (reviewing agency interpretation “under the standard set out in Chevron” and holding “EPA strayed far beyond those bounds”). There is nothing remotely deferential about the majority opinion. Moreover, Justice Thomas concurred, writing separately (though for himself only) to argue that the Chevron doctrine is unconstitutional because it cedes to the executive the exclusive judicial authority to “say what the law is.” 13 Id. at 2712 (Thomas, J., concurring) (internal quotation marks omitted) (quoting Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803)).

Third, in Perez v. Mortgage Bankers Ass’n, 14 135 S. Ct. 1199 (2015). Justice Scalia, of all people, revealed some waning of his longstanding and strident enthusiasm for Chevron. In an opinion that was primarily an attack on the distinct doc­trine of deference to agency interpretations of their own regulations, Justice Scalia noted that Chevron is “[h]eedless of the original design of the [Administrative Procedure Act]” 15 Id. at 1211 (Scalia, J., concurring in the judgment). and hinted that perhaps it should be “uprooted.” 16 Id. at 1212 (Scalia, J., concurring in the judgment). This from the Justice generally understood as the firm­est and most ferocious defender of Chevron. 17 See, e.g., Thomas W. Merrill & Kristin E. Hickman, Chevron’s Domain, 89 Geo. L.J. 833, 859–60 (2001) (“Justice Scalia has consistently argued for the broadest possible conception of the scope of the Chevron doctrine, urging, for example, that it applies to agencies that lack the power to render binding legal rulings and to interpretations by agencies issued in opinion letters or in briefs.”).

Some early commenters see these decisions as at least potentially marking a watershed moment, a fundamental shift from a regime of meaningful deference to a reassertion of judicial supremacy. 18 See, e.g., Randolph J. May, Chevron Decision’s Domain May Be Shrinking, The Hill (July 7, 2015), http://thehill.com/blogs/pundits-blog/the-judiciary/247015-chevron-decisions-domain-may-be-shrinking [http://perma.cc/8P9Y-2RUV] (suggesting while “Chevron’s demise” is not “imminent,” recent cases “could portend a diminished [agency] role”); Cass R. Sunstein, The Catch in the Obamacare Opinion, Bloomberg View (June 25, 2015), http://www.bloombergview.com/articles/2015-06-25/the-catch-in-the-obamacare-opinion [http://perma.cc/CQ37-W8PP] (comparing King to Marbury in that Court made strong assertion of judicial power but declined to use that power to counter executive action in case before it). Time will tell, but in fact it has long been the case that deference under Chevron is a principle often honored in the breach. For all the clamor, attention, and citations, 19 It seems an obligation of the form to point out that Chevron is the most cited decision in administrative law. See, e.g., Jack M. Beermann, Chevron at the Roberts Court: Still Failing After All These Years, 83 Fordham L. Rev. 731, 731 (2014) (stating Chevron is most cited Supreme Court administrative law decision); Thomas W. Merrill, The Story of Chevron: The Making of an Accidental Landmark, in Administrative Law Stories 399 & n.2 (Peter L. Strauss ed., 2006) (noting Chevron is most cited administrative law case, though not “the overall citation champion.”); Thomas J. Miles & Cass R. Sunstein, Do Judges Make Regulatory Policy? An Empirical Investigation of Chevron, 73 U. Chi. L. Rev. 823, 823 (2006) (labeling Chevron most cited case in “modern public law”). So: obligation fulfilled. It is sometimes said that it is the most cited U.S. Supreme Court decision, period. See, e.g., Richard J. Pierce, Jr., Administrative Law Treatise 140 (4th ed. 2002) (“Chevron is one of the most important decisions in the history of administrative law. It has been cited and applied in more cases than any other Supreme Court decision in history.”). This is an overstatement; in fact Chevron is miles behind some other decisions. As of September 18, 2015, Chevron had been cited in 13,454 federal cases. See WestlawNext, https://a.next.westlaw.com (last visited Sept. 18, 2015) [hereinafter WestlawNext Search] (conducting search by accessing Chevron case, then Citing References tab, and narrowing list of citing cases by clicking cases and federal jurisdiction in left-hand sidebar). But Erie Railroad Co. v. Tompkins, 304 U.S. 64 (1938), has been cited 17,188 times; Ashcroft v. Iqbal, 556 U.S. 662 (2009), was cited 98,104 times in just six years; and Bell Atlantic v. Twombly, 550 U.S. 544 (2007), has been cited a whopping 120,975 times in about eight years, an order of magnitude more than Chevron. See WestlawNext Search, supra (conducting search with same methodology as for Chevron discussed above). Chevron has had less of an impact than this attention implies. Several scholars have surveyed the case law and reported that judges, especially those on the Supreme Court, do not defer as much as the doctrine seems to require. Rather, they have narrowed the circumstances in which Chevron, by its own terms, applies and invoke Chevron only intermittently in those circumstances. 20 See Jack M. Beermann, The Turn Toward Congress in Administrative Law, 89 B.U. L. Rev. 727, 743 n.65 (2009) [hereinafter Beerman, Turn Toward Congress] (“Chevron no longer has much of an affect [sic] on statutory interpretation.”); William N. Eskridge, Jr. & Lauren E. Baer, The Continuum of Deference: Supreme Court Treatment of Agency Statutory Interpretations from Chevron to Hamdan, 96 Geo. L.J. 1083, 1122 (2008) (“[I]t might be the case that the Court is more likely to cite Chevron and follow its two-step approach when the majority is prepared to uphold the agency action.”); Daniel J. Gifford, The Emerging Outlines of a Revised Chevron Doctrine: Congressional Intent, Judicial Judgment, and Administrative Autonomy, 59 Admin. L. Rev. 783, 833 (2007) (noting emergence of “new model of judicial review of agency inter­pretations” in which “mandatory obligation to defer . . . is limited”); Ann Graham, Searching for Chevron in Muddy Watters: The Roberts Court and Judicial Review of Agency Regulations, 60 Admin. L. Rev. 229, 271 n.257 (2008) (surveying recent Supreme Court decisions and concluding Court has abandoned Chevron and accepts only agency interpretations with which it agrees); Linda D. Jellum, The Impact of the Rise and Fall of Chevron on the Executive’s Power to Make and Interpret Law, 44 Loy. U. Chi. L.J. 141, 173–88 (2012) (describing ways Supreme Court, in recent years, has reclaimed authority it had ceded to executive in Chevron); Deborah N. Pearlstein, After Deference: Formalizing the Judicial Power for Foreign Relations Law, 159 U. Pa. L. Rev. 783, 807–17 (2011) (detailing limited impact and scope of Chevron and arguing against deference to executive inter­pretations in foreign relations); Connor N. Raso & William N. Eskridge, Jr., Chevron as a Canon, Not a Precedent: An Empirical Study of What Motivates Justices in Agency Deference Cases, 110 Colum. L. Rev. 1727, 1740 (2010) (surveying 1,014 post–Chevron Supreme Court cases and finding Court invoked Chevron in only one-third of cases where it should have applied); Brian G. Slocum, The Importance of Being Ambiguous: Substantive Canons, Stare Decisis, and the Central Role of Ambiguity Determinations in the Administrative State, 69 Md. L. Rev. 791, 794–98 (2010) (“[Chevron] is a far less significant doctrine if it does not require significant changes to statutory interpretation methodology, it is routinely ignored by courts, and it does not constrain judicial discretion because it allows judges to decide cases in accordance with their ideological preferences.”).

Two responses to this shortfall are appropriate. The first, the subject of Part I, is: “What did you expect?” Only in the rarest of self-protective settings will courts create and stand by a super-strong hands-off principle. Chevron has never, in practice, amounted to an abdication of the judicial role, if only because judges are not going voluntarily to disarm. The second response, considered in the remainder of this Article, is: “Good.” That is, Chevron should not be seen as a revolutionary decision. It insists on respect for the delegation of policymaking authority to administrative agencies, but it preserves interpretive authority for courts. The question “What should be done?” is for agencies. The question “What did Congress do?” is for courts (although they should take seriously what agencies have to say on the matter).

This Essay proceeds in three parts. Part I describes Chevron as a self-regulating doctrine, one that, for that reason, has not taken over the world and will never do so. Part II lays out a basic understanding of Chevron and of Skidmore. 21 Skidmore v. Swift & Co., 323 U.S. 134 (1994). The argument here is for a reading of Chevron that is both weaker and stronger than that often proposed. On the one hand, courts retain an essential and meaningful role in determining, to use Professor Strauss’s terms, the boundaries of “Chevron space”; 22 See generally Strauss, Space/Weight, supra note 2, at 1143. they have real work to do. On the other hand, in doing that work, the views of the agency can never be ignored; there is no completely agency-free space within which courts interpret statutes that agencies administer. These principles are explicated in part by reviewing the vocabulary of the Chevron doctrine. This Part concludes by placing Chevron in a jurispru­dential framework that draws on the distinction between “interpretation” and “construction.” That distinction has been elaborated in other areas but has not infiltrated administrative law. In the Chevron opinion, its progeny, and the commentary, “interpretation” and “construction” are used as synonyms. But they can be seen as distinct undertakings; one concerns determining semantic meaning, the other applying that meaning to concrete situations. That distinction, or something like it, maps onto and helps elucidate the distinction between the proper roles of the courts and agencies, and the activities of step one and step two. Part III applies the framework developed in Part II to the Supreme Court’s decision in City of Arlington, 23 City of Arlington v. FCC, 133 S. Ct. 1863 (2013). which held that Chevron applies to agency determinations going to their own jurisdiction. The majority and the dissent in that case were both correct, making City of Arlington that rare creature, a unanimous 6-3 decision.

I. Chevron as Self-Regulation

Chevron U.S.A. Inc. v. Natural Resources Defense Council 24 467 U.S. 837 (1984). is the Supreme Court’s most important decision regarding judicial deference to agency views of statutory meaning. It lays out a seemingly straightforward ap­proach to considering whether to uphold an agency’s view of a statute:

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. 25 Id. at 842–43 (footnotes omitted).

In what is always referred to as “step one” (though that term does not appear in the Chevron opinion itself), the reviewing court “employ[s] traditional tools of statutory construction” to determine statutory mean­ing. 26 Id. at 843 n.9. What exactly happens in step two is disputed, though the dominant judicial and academic formulation is that there, the court asks (a) whether the statute clearly precludes the agency’s reading (in which case the inquiry largely, if not completely, overlaps with step one) and (b) whether the agency’s determination is arbitrary and capricious. 27 See, e.g., Astrue v. Capato ex rel. B.N.C., 132 S. Ct. 2021, 2034 (2012) (deferring under step two because relevant agency regulations were “neither arbitrary or capricious in substance, [n]or manifestly contrary to the statute” (alteration in original) (internal quotation marks omitted) (quoting Mayo Found. for Med. Educ. and Research v. United States, 131 S. Ct. 704, 711 (2011))); Mayo, 131 S. Ct. at 711 (“[U]nder [Chevron step two] we may not disturb an agency rule unless it is ‘arbitrary or capricious in substance, or manifestly contrary to the statute.’” (citations omitted)); Ronald M. Levin, The Anatomy of Chevron: Step Two Reconsidered, 72 Chi.-Kent L. Rev. 1253 (1997) (arguing step two tracks arbitrary-and-capricious test).

If Chevron is revolutionary, it is because step two seems to take courts out of the statutory-interpretation game, compelling them to accept the agency’s view of statutory meaning. Contrary to settled understandings, it charges agencies rather than courts with the authority to “say what the law is.” 28 Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803). Hence Cass Sunstein’s much-quoted description of the decision as “a kind of counter-Marbury for the administrative state.” 29 Cass R. Sunstein, Beyond Marbury: The Executive’s Power to Say What the Law Is, 115 Yale L.J. 2580, 2589 (2006) [hereinafter Sunstein, Beyond Marbury].

Doctrinally, the impact and meaning of Chevron will depend on (a) in what circumstances it is triggered, 30 Here the leading case in a complicated body of decisions is United States v. Mead Corp., 533 U.S. 218 (2001), which held that Chevron only applies when Congress has delegated to the agency authority to make rules with the force of law and the agency has acted pursuant to that authority. See generally Merrill & Hickman, supra note 17 (surveying Chevron’s application). Though it predates Mead and is somewhat out of date, Merrill and Hickman’s article, id., remains the most useful and comprehensive survey of exactly when Chevron applies and when it does not. and (b) how capacious step one is, i.e., how quickly a court throws up its hands and defers in step two. 31 This is sometimes referred to as the “how clear is clear?” question. See, e.g., Antonin Scalia, Judicial Deference to Administrative Interpretations of Law, 1989 Duke L.J. 511, 520–21 (“How clear is clear? It is here, if Chevron is not abandoned, that the future battles over acceptance of agency interpretations of law will be fought.”); see also Note, “How Clear is Clear” in Chevron’s Step One?, 118 Harv. L. Rev. 1687, 1698–703 (2005) (arguing in determining “how clear is clear,” reviewing court should consider agency’s institutional advantages, if any). Were Chevron to apply any time an agency had taken a view regarding statutory meaning, and were the court to yield to that view in the face of any statutory uncertainty, the decision would be enormously consequential (and problematic). As discussed above, that has not happened. At least part of the reason is that Chevron is an example of self-regulation.

Self-regulatory regimes rely on entities to oversee themselves. The Chevron doctrine is essentially an example of self-regulation. It is a judicially imposed limitation on the scope of judicial authority, a doctrine through which those in the judging business constrain the activities of the members of their own industry.

Some doubters object that the “self” in “self-regulation” is like the “constructive” in “constructive consent,” or the “quasi” in “quasi-contract”—that is, a synonym for “not.” If you are doing it yourself, it is not regulation. 32 Bos. Consulting Grp., U.S. Securities and Exchange Commission: Organizational Study and Reform 25 (2011), http://www.sec.gov/news/studies/2011/967study.pdf [https://perma.cc/5NJU-CUQ4?type=pdf] (“The most fundamental critique is that self-regulation is not real regulation at all: at best, self-regulation is less effective than government regulation, and at worst, is merely ‘an illusion’ meant to deflect calls for government oversight.”). On this account, self-regulation guarantees under-regulation. That may be an overstatement, but indisputably, the essential risk of self-regulation is excessive laxity. After all, it is a regime in which the regulator is not merely “captured” by regulated interests; regulator and regulatee are one and the same. 33 See, e.g., Douglas C. Michael, Federal Agency Use of Audited Self-Regulation as a Regulatory Technique, 47 Admin. L. Rev. 171, 189–90 (1995) (identifying inadequate enforcement as one of three principal shortcomings of self-regulation).

To some extent, this dubiousness reflects a misunderstanding of the concept; “self-regulation” is not a synonym for “will power.” The princi­ple is not that an individual person or firm will control itself, but rather that an industry will control its members. That will not necessarily work, but it can work in a market setting if the individual firms recognize that their long-term prosperity depends on mutual constraint. As one scholar concluded from a study of self-regulation in the nuclear power and chemical manufacturing industries: “[T]he key to the rise of self-regulation was the industry’s collective perception of itself as a ‘community of fate.’ Each industry’s future prosperity was seen as depending upon its ability to impose collective self-restraint on its members’ profit-seeking activities in the name of public safety.” 34 Saule T. Omarova, Wall Street as Community of Fate: Toward Financial Industry Self-Regulation, 159 U. Pa. L. Rev. 411, 420 (2011) (citation omitted).

If that is the case, then one setting where one would expect self-regulation to fail is where the industry consists of a single monopolist or, relatedly, where there are multiple players but the industry produces a good for which there are no substitutes. The federal judiciary is one such setting; it enjoys a constitutionally established monopoly on adjudication. Therefore, norms of conduct would have to be overwhelmingly powerful for self-regulation to function effectively.

The federal judiciary self-regulates in many settings; indeed, it is the norm. The traditional approach to judicial misconduct, conflicts of interest, and incompetence has been one of self-regulation (at least up to the point where there has been an impeachable offense, at which junc­ture the U.S. Congress may take over). There are powerful argu­ments sounding in judicial independence for this approach. 35 See, e.g., Harry T. Edwards, Regulating Judicial Misconduct and Divining “Good Behavior” for Federal Judges, 87 Mich. L. Rev. 765, 766 (1989) (“[T]he only constitu­tionally permissible way to regulate judicial misconduct and disability that does not involve impeachable action is a system of judicial self-regulation unencum­bered by any form of congressional interference.”). And professional and ethical norms may be so imbued, and reputational capital so pre­cious, that self-regulation may function adequately in this setting. But the risk of underenforcement is real. 36 See generally, Anthony D’Amato, Self-Regulation of Judicial Misconduct Could Be Mis-Regulation, 89 Mich. L. Rev. 609 (1990) (arguing judiciary cannot be left to police itself). Indeed, Congress has recently grown uneasy with the traditional approach. 37 See Dana A. Remus, The Institutional Politics of Federal Judicial Conduct Regulation, 31 Yale L. & Pol’y Rev. 33, 61–70 (2012) (describing new approach in legisla­tion that “undermined judicial regulation” through Judicial Transparency and Ethics Enhancement Act of 2006).

Judicial deference doctrines are also examples of self-regulation. This is clearly true of Skidmore 38 Skidmore v. Swift & Co., 323 U.S. 134, 140 (1944) (instructing lower courts to take into account agency interpretations of statutes because, “while not controlling upon the courts by reason of their authority, [they] do constitute a body of experience and informed judgment to which courts and litigants may properly resort for guidance”). and Auer, 39 Auer v. Robbins, 519 U.S. 452, 461 (1997) (holding when meaning of regulation is in doubt, agency’s interpretation is “controlling unless ‘plainly erroneous or inconsistent with the regulation’” (internal quotation marks omitted) (quoting Robertson v. Methow Valley Citizens Council, 490 U.S. 332, 359 (1989))). which do not purport to rest on congressional command. And, of course, courts defer to actors other than administrative agencies offering legal interpretations. They defer to agencies doing other things (e.g., making enforcement decisions 40 See, e.g., Heckler v. Chaney, 470 U.S. 821, 837–38 (1985) (holding agency decisions not to enforce are presumptively unreviewable). ), they defer to other actors’ legal interpretations (e.g., state courts’ decisions regarding their own law 41 See, e.g., Erie R.R. Co. v. Tompkins, 304 U.S. 64, 78 (1938) (holding federal courts deciding questions of state law must apply law of state in which they sit as determined by courts of that state). ), and they defer to non-agencies who are not interpreting the law (e.g., the presumption of the constitutionality of federal statutes 42 See, e.g., United States v. Morrison, 529 U.S. 598, 607 (2000) (referring to “presumption of constitutionality” and noting “[d]ue respect for the decisions of a coordi­nate branch of Government demands that we invalidate a congressional enactment only upon a plain showing that Congress has exceeded its constitutional bounds”); Neal Kumyar Katyal & Thomas P. Schmidt, Active Avoidance: The Modern Supreme Court and Legal Change, 128 Harv. L. Rev. 2109, 2139 (2015) (“The principle that the Court must presume laws constitutional is as old as judicial review.”); see also Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 41–43 & n.9 (1983) (holding presumption of constitutionality afforded federal statutes is stronger than presumption of regularity afforded agency action being reviewed for arbitrariness). the political question doctrine 43 See, e.g., Nixon v. United States, 506 U.S. 224, 233–36, 238 (1993) (holding federal courts could not review procedures used by U.S. Senate to convict impeached federal judge); Baker v. Carr, 369 U.S. 186, 217 (1962) (cataloging varieties of nonjusticiable cases that have been labeled “political questions”). ). Almost none of these hang on a statutory or constitutional peg; they are prudential doctrines of self-restraint.

Chevron is a self-imposed limitation of this sort. To be sure, the Court and commentators often assert that Chevron is not self-imposed but rests on congressional instructions. 44 See United States v. Mead Corp., 533 U.S. 218, 226–27 (2001) (holding “admin­istrative implementation of a particular statutory provision qualifies for Chevron deference when it appears that Congress delegated authority to the agency generally,” as shown by some “indication of . . . congressional intent”). This theory solves one huge, oft-noted problem, namely that the Administrative Procedure Act (APA), pursuant to which most cases in which Chevron is in play are brought, instructs courts to “decide all relevant questions of law, interpret constitutional and statutory provisions,” and set aside agency action “in excess of statutory jurisdiction, authority, or limitations, or short of statutory right.” 45 5 U.S.C. § 706 (2012). The tension between Chevron and the text of the APA has been often noted. See, e.g., Perez v. Mortgage Bankers Ass’n, 135 S. Ct. 1199, 1211 (2015) (Scalia, J., 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. Never mentioning § 706’s directive that the ‘reviewing court . . . interpret . . . statutory provisions,’ we have held that agencies may authoritatively resolve ambiguities in statutes.” (citing Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 842–43 (1984))); Jack M. Beermann, End the Failed Chevron Experiment Now: How Chevron Has Failed and How and Why It Should Be Overruled, 42 Conn. L. Rev. 779, 788–94 (2010) (arguing first among Chevron’s “intrinsic failings” is its inconsistency with APA); Ronald A. Cass, Vive La Deference? Rethinking the Balance Between Administrative and Judicial Discretion, 83 Geo. Wash. L. Rev. (forthcoming 2015) (manuscript at 29 & n.134) (on file with the Columbia Law Review) (arguing courts should “implement the APA’s review provisions as written,” deferring only on matters within agency discretion and not on questions of statutory meaning); Cynthia R. Farina, Statutory Interpretation and the Balance of Power in the Administrative State, 89 Colum. L. Rev. 452, 472–73 (1989) (noting Chevron approach does not “take section 706 at face value”); Jacob E. Gersen & Adrian Vermeule, Chevron as a Voting Rule, 116 Yale L.J. 676, 689 (2007) (stating theory Congress has delegated interpretive authority to agencies is fiction undercut by APA’s explicit instruction that courts shall decide all questions of law). But it is hard to find anyone who does not consider congressional delegation a fiction. 46 See, e.g., David J. Barron & Elena Kagan, Chevron’s Nondelegation Doctrine, 2001 Sup. Ct. Rev. 201, 212–25 (concluding Chevron does not rest on actual congressional intent); Lisa Schultz Bressman, Reclaiming the Legal Fiction of Congressional Delegation, 97 Va. L. Rev. 2009, 2009 (2011) (“The framework for judicial review of agency statutory interpretation rests on a legal fiction: Congress intends to delegate interpretive authority to federal agencies whenever it fails to resolve clearly the meaning of statutory language.”); Abbe R. Gluck, What 30 Years of Chevron Teach Us About the Rest of Statutory Interpretation, 83 Fordham L. Rev. 607, 613 (2014) (“[I]n contrast to most of the other interpretive rules, there is widespread agreement about Chevron’s source: the Court created the doctrine.”); Ronald J. Krotoszynski, Jr., Why Deference?: Implied Delegations, Agency Expertise, and the Misplaced Legacy of Skidmore, 54 Admin. L. Rev. 735, 749 (2002) (“Chevron deference revolves around the fiction of a congressional delegation . . . .”); Scalia, supra note 31, at 517 (noting “any rule [regarding congressional allocation of interpretive authority] represents merely a fictional, presumed intent”); Sunstein, Beyond Marbury, supra note 29, at 2589–91 (describing proposition Congress has delegated authority to executive as “fiction”). Most recently, Mark Seidenfeld has articulated the view that it is a doctrine of “self-restraint,” laying out at length the shortcomings of claims that it rests on any actual statutory mandate or congressional intent. 47 Mark Seidenfeld, Chevron’s Foundation, 86 Notre Dame L. Rev. 273, 276–88 (2011) (debunking claim Congress delegated interpretive authority to agencies in any conscious or legally operative way). He has a lot of company. 48 See, e.g., Maureen B. Callahan, Must Federal Courts Defer to Agency Interpretations of Statutes? A New Doctrinal Basis for Chevron U.S.A. v. Natural Resources Defense Council, 1991 Wis. L. Rev. 1275, 1289 (labeling Chevron “principle of self-restraint, related to the various well-established prudential limitations on justiciability in the federal courts”); Frederick Liu, Chevron as a Doctrine of Hard Cases, 66 Admin. L. Rev. 285, 318, 326–29 (2014) (labeling Chevron “a doctrine of judicial self-restraint” that has no basis in congressional command); Note, Justifying the Chevron Doctrine: Insights from the Rule of Lenity, 123 Harv. L. Rev. 2043, 2046 (2010) (“The problem, as many commentators have observed, is that the notion that each statutory ambiguity represents a specific decision by Congress to delegate interpretive authority to the agencies strains credibility.”).

In any event, the theoretical debate is beside the point. Whether self-imposed or derived from an implicit congressional delegation, Chevron’s command to accept any reasonable agency interpretation is in practice a form of self-regulation. The Court made it up and imposed it on itself; it is administrative common law. 49 Gillian E. Metzger, Foreword: Embracing Administrative Common Law, 80 Geo. Wash. L. Rev. 1293, 1300–03 (2012) (describing Chevron and its progeny as “exemplars of administrative common law”). So understood, Chevron is heir to the shortcomings and risks that generally bedevil self-regulation: a lack of transparency, the failure to evaluate or monitor performance, and the absence of meaningful penalties. 50 See Neil Gunningham & Peter Grabosky, Smart Regulation 152–53 (1998) (describing these shortcomings of self-regulation). Judges, like most human beings (especially successful human beings holding prestigious positions, possessed of high self-regard, surrounded by sycophants, and blessed with matchless job security), will only go so far in ceding authority. Of course, this or that individual judge may be too deferential, may overregulate, so to speak. 51 Cf. Jean Roiphe, Aggression, Unconscious Conflict, and the Role of the Lawyer, 16 Cardozo L. Rev. 1205, 1214–18 (1995) (explaining one role of lawyer is to protect client from client’s own superego, which might keep her from being aggressive enough on her own behalf). Teddy Roosevelt famously complained that Oliver Wendell Holmes was such a judge. 52 When, early in his tenure, Holmes dissented in an important government victory in an antitrust case, the President who appointed him fumed, “I could carve out of a banana a Judge with more backbone than that.” Henry J. Abraham, Justices and Presidents: A Political History of Appointments to the Supreme Court 69 (3d ed. 1992). But as an overall tendency, judges generally, federal judges in particular, and Supreme Court Justices most of all, simply are not going to be too constrained. 53 Psychological study of the judiciary has focused on the fact that judges are heirs to all the same decisionmaking shortcomings that plague ordinary mortals. See, e.g., Chris Guthrie, Misjudging, 7 Nev. L.J. 420, 420 (2007) (reviewing literature and extensive evidence showing judges are hampered by cognitive biases and multiple “blinders”). Some of this work may assume a particular judicial tendency to overconfidence, but it does not prove it. Accordingly, the support for the statement in text is impressionistic rather than scientifically established. The impression is widely shared, however. See, e.g., Henry Paul Monaghan, Comment on Professor Van Alstyne’s Paper, 72 Iowa L. Rev. 1309, 1309 (1987) (“Judges constitute an elite group whose self-confidence is seldom matched by genuine learning.”); Louis H. Pollack, “Liberty”: Enumerated Rights? Unenumerated Rights? Penumbral Rights? Other?, 8 U. Pa. J. Const. L. 905, 911 (2006) (“Provided that the judicial miscue [of misconstruing lawyers’ arguments] doesn’t turn a win into a loss, lawyers go back to their offices with their win and get big bucks, and judges go back to their chambers with their big egos inflated some more and get little bucks.”); Jeffrey W. Stempel, Refocusing Away from Rules Reform and Devoting More Attention to the Deciders, 87 Denv. U. L. Rev. 335, 349–60 (2010) (discussing “cult of the judge” and observing “with the pedestal of center stage comes some sense of entitlement and even arrogance”).

Accordingly, the continued reports of Chevron’s demise are not a surprise. Nor is the fact that empiricists have had difficulty determining whether Chevron has actually had an impact in the real world. 54 The literature, and the challenges in attempting to measure the decision’s impact, are summarized in John Manning & Matthew Stephenson, Legislation and Regulation 772–75 (2d ed. 2013). That does not mean it has had no impact, and it seems likely that it has had some. But the fact that it is so difficult to conclusively demonstrate an impact at least establishes that any shift has not been fundamental.

Chevron is an example of what one might call doctrinal self-regulation. It is a doctrine that by its nature will not get out of hand. Self-regulation simply does not lead to overregulation. The risks of self-regulation are one-sided. Like most people, judges will protect their turf. Moreover, even after all these years, strong readings of Chevron remain counterintuitive; judges’ learned intuitions do not generally lead them to defer on questions of law. 55 Gary Lawson and Stephen Kam put it this way:
“To the extent Chevron increases the range of circumstances in which judges defer to agencies on pure legal questions, it seems to reverse the common-sense view of comparative institutional competence in which courts are generally better at determining the law and agencies are generally better at finding facts and making policy. Anyone who subscribes to the legal process approach, in which decisional authority should be allocated where best applied, will find a broad reading of Chevron troublesome at best and absurd at worst. Given the number of judges (and law clerks) trained either at Harvard Law School or by professors who were trained at Harvard Law School, where the legal process approach grew and flourished, it would not be surprising to find serious resistance to the Chevron revolution.”
Gary Lawson & Stephen Kam, Making Law Out of Nothing at All: The Origins of the Chevron Doctrine, 65 Admin. L. Rev. 1, 61 (2013).

The point is not that deference doctrines are meaningless, and this is not a rant about the imperial judiciary or a complaint that the Supreme Court “seems incapable of admitting that some matters—any matters—are none of its business.” 56 Sosa v. Alvarez-Machain, 542 U.S. 692, 750 (2004) (Scalia, J., concurring in part and concurring in the judgment) (emphasis omitted). Indeed, in general, administrative law doctrines align poorly with such objections. 57 See generally Michael Herz, The Rehnquist Court and Administrative Law, 99 Nw. U. L. Rev. 297 (2004) (describing how Rehnquist Court’s administrative law jurisprudence fits poorly with standard academic objections that Court was too sure of itself and inadequately deferential). The point is only that, as with justiciability doctrines, the nature and structure of the judiciary guard against overdoing deference, and the danger will almost always be that courts go too far in the other direction.

A 1992 article by the present author cautioning against too strong a reading of Chevron was entitled Deference Running Riot. 58 Michael Herz, Deference Running Riot: Separating Interpretation and Lawmaking Under Chevron, 6 Admin. L.J. Am. U. 187 (1992) [hereinafter Herz, Deference Running Riot]. In the intervening decades, that title has been shown to be fear mongering. Deference has not run riot, and experience and common sense both suggest that it will not. At the same time, Chevron is far from dead. Properly understood and sensibly applied, that decision is a salutary recognition that Congress delegates broad authority to agencies and courts must respect those delegations. But it leaves courts with an essential role in determining the scope of those delegations. The remainder of this Essay accepts that the revolutionary Chevron, that apotheosis of judicial self-abdication, is indeed dead (if it ever was truly alive) and articulates the nature and scope of the robust and appropriate Chevron that should survive.

II. A Viable Chevron (and Skidmore)

What, then, should a viable Chevron—a Chevron that is sustainable, consistently applied, and aligns with separation-of-powers norms and judicial realities—prescribe? This Part lays out an understanding of Chevron that is both consistent with basic principles of separation of powers and an implementable form of judicial self-regulation. It focuses on the distinctions between Chevron and Skidmore, because each is best understood in comparison to the other.

A. Doctrinal Consequences of the Categorical Model

Chevron and Skidmore are both doctrines of judicial “deference.” It is often said that the first requires “strong deference” and the second “weak deference.” 59 See, e.g., Navarro v. Encino Motorcars, LLC, 780 F.3d 1267, 1268–69 (9th Cir. 2015) (using phrase “strong deference under Chevron” or “strong Chevron deference” in headnotes 4, 9, 10, and 11); Cetto v. LaSalle Bank Nat. Ass’n, 518 F.3d 263, 274 (4th Cir. 2008) (referring to Chevron framework as requiring “strong deference”); Jim Rossi, Respecting Deference: Conceptualizing Skidmore Within the Architecture of Chevron, 42 Wm. & Mary L. Rev. 1105, 1109–10 (2001) (referring to “Skidmore ‘weak deference’” and “Chevron step-two ‘strong deference’”); Charles A. Sullivan, On Vacation, 43 Hous. L. Rev. 1143, 1204 n.287 (2006) (“[T]he Court requires strong deference, usually called Chevron deference, in some situations, and weak deference, usually called Skidmore deference, in others.”). This misconceives the distinction as one of degree when it is in fact a difference in kind. Where Chevron has courts accept any reasonable agency interpretation of an ambiguous statute, Skidmore insists that courts take agency views seriously but cautions that those views are “not controlling upon the courts by reason of their authority,” offer “guidance,” and at most have the “power to persuade.” 60 Skidmore v. Swift & Co., 323 U.S. 134, 140 (1944).

Peter Strauss’s helpful route to understanding the distinction between these two sorts of deference begins with terminology. As the title to his influential article puts it, “deference” is “too confusing”; instead, he would refer to “Chevron space” and “Skidmore weight.” 61 See generally Strauss, Space/Weight, supra note 2. “Weight” was the term used by the Skidmore Court itself. Skidmore, 323 U.S. at 140 (stating “Court has long given considerable and in some cases decisive weight to” agency views and “weight of [the agency’s] judgment in a particular case will depend on” various factors). On this conception, Chevron marks off and insists that courts respect a space within which the agency is in charge. The court must define the boundaries of that space; 62 Strauss, Space/Weight, supra note 2, at 1165 (stating “lines defining an agency’s Chevron space must be judicially determined,” because that determination constitutes “statement of what the law is”). an interpretation that is “unreasonable” exceeds those boundaries. But within Chevron space, the decisionmaker is the agency and the court’s job is to defer rather than second-guess. Chevron space is the area between what a statute “must mean and what it cannot mean.” 63 Id. at 1159. Under Skidmore, in contrast, the court is the ultimate decisionmaker. The agency’s views must be taken seriously; they carry weight. But the court has the ultimate say. 64 Id. at 1156 (“Once a question of statutory interpretation has been put before a court, it is for the court to resolve the question of meaning.”).

Accordingly, there are two fundamental differences between Chevron and Skidmore: the who and the what. First, the ultimate decisionmaker is different in the two settings. Under Skidmore, at the end of the day the decisionmaker is the court; the agency’s views have the power to persuade but they “lack[] power to control” and are “not controlling upon the courts by reason of their authority.” 65 Skidmore, 323 U.S. at 140. Under Chevron, the decisionmaker is the agency; as long as the agency has not gone off the deep end, the court must accept its decision. Thus, in Alaska Department of Environmental Conservation v. EPA, Justice Ginsburg observed that in Chevron, the Court had “accorded dispositive effect to the EPA’s interpretation of an ambiguous [statutory] provision,” whereas the interpretation challenged here “does not qualify for the dispositive force described in Chevron.” 66 540 U.S. 461, 487 (2004) (emphasis added). Instead, it merely “‘warrant[s] respect.’” 67 Id. at 488 (quoting Wash. State Dep’t of Soc. & Health Servs. v. Guardianship Estate of Keffeler, 537 U.S. 371, 385 (2003)). These quotations from Alaska Department of Environmental Conservation are admittedly rather selective; Justice Ginsburg’s wording is not always so precise. The Court agreed with the agency, but it did not defer in the strong sense. 68 For similar Supreme Court decisions upholding agency interpretations under Skidmore because it agreed with them, see Kasten v. Saint-Gobain Performance Plastics Corp., 131 S. Ct. 1325, 1335 (2011) (upholding agency’s interpretation because it was “reasonable” and “consistent with the [statute]”); Fed. Express Corp. v. Holowecki, 552 U.S. 389, 401–02 (2008) (upholding agency view as “consistent with the statutory framework,” whereas respondents’ view was “in considerable tension with [its] structure and purposes”); Estate of Keffeler, 537 U.S. at 385 (applying canons of construction to interpret statutory provision before noting agency guidance document “confirmed” Court’s reading). In other cases, the Court has cited Skidmore but ultimately rejected the agency interpretation—not because it was unreasonable, but because it was wrong. 69 See, e.g., Christopher v. SmithKline Beecham Corp., 132 S. Ct. 2156, 2169 (2012) (rejecting agency’s interpretation of its own regulation); Wyeth v. Levine, 555 U.S. 555, 576–81 (2009) (noting agency’s views were “inherently suspect” and conflicted with “Congress’ purposes”); Gonzales v. Oregon, 546 U.S. 243, 269 (2006) (finding agency’s rule was not persuasive and noting decisionmaker’s “lack of expertise”); Christensen v. Harris County, 529 U.S. 576, 587 (2000) (interpreting statute independently and simply stating agency’s contrary view was “unpersuasive”). Whether agreeing or disagreeing with the agency, the Court was making the decision.

To say that in Skidmore the court is the ultimate decisionmaker whereas in Chevron step two the agency is the ultimate decisionmaker is true but incomplete. Equally important, the court and the agency are making different sorts of decisions. The agency is making a policy decision. By definition, within its Chevron space, the agency is uncon­strained by the statute, which has given out. It must act reasonably, i.e., not arbitrarily and capriciously, and it cannot exceed statutory bound­aries. However, within those limits (which may be expansive), the agency can do what it wants. In contrast, in a Skidmore case, the court is the decisionmaker, but it cannot do whatever it wants. Its role is to do what Congress wants (or, alternatively, what the statute dictates). That descrip­tion of the judicial role may sound naïve or old-fashioned. 70 See generally Sunstein, Beyond Marbury, supra note 29, at 2583 (defending fairly robust version of Chevron in light of “legal realist attack on the autonomy” of legal reasoning and shift of regulatory authority from common law courts to administrative agencies). Interpreters, including judges, are never wholly constrained, must make judgment calls, and will (properly) be influenced by policy concerns. But the constitutional structure, the APA’s distinction between legislative and interpretative rules, centuries of jurisprudence, and most people’s intuitions accept a distinction between interpreting law and making law. The court’s job is to give the statute the best possible reading, not to adopt the best possible policy.

Under Skidmore, the weight due an agency’s interpretation varies according to the circumstances. 71 Skidmore v. Swift & Co., 323 U.S. 134, 140 (1944) (“The weight of [an agency’s] judgment in a particular case will depend upon the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade, if lacking power to control.”). The “Skidmore factors” that determine where the interpretation falls along this sliding scale help illuminate the difference between the judicial task under that decision and the judicial task under Chevron. For example, the most familiar and well-settled factor is that under Skidmore, a long-standing, consistent interpretation merits special weight. 72 Id. (listing factors that give agency decisions persuasive authority). Three justifications might support this principle. First, if the agency has flipped, then at least one of its positions must be wrong, and there is a meaningful chance (though in principle less than fifty–fifty) that the one under review is the wrong one. Therefore, judicial confi­dence in the agency is significantly undercut. Second, if the agency has not flipped, and Congress has left the statute alone, there is implicit congressional approval of the agency view. 73 The same idea underlies the rule that stare decisis is especially robust in statutory cases. See, e.g., Patterson v. MacLean Credit Union, 491 U.S. 164, 172 (1989) (“[T]he burden borne by the party advocating the abandonment of an established precedent is greater where the Court is asked to overrule a point of statutory construction.”). Whether that rule rests in part on an implicit congressional endorsement of the precedent is debated. Compare id. at 175 n.1 (denying congressional failure to overturn precedent is in itself reason for standing by precedent), with id. at 200–01 (Brennan, J., concurring in the judgment) (stating Congress’s failure to pass legislation to override decision is de facto ratification of that decision), and Johnson v. Transp. Agency, 480 U.S. 616, 629–30 n.7 (1987) (making ratification argument). Many things explain congressional inaction besides agreement (e.g., ignorance, distraction, inability to agree on an alternative, lack of time), so the acquiescence argument is thin, but courts still trot it out and it is not absurd. Finally, reliance interests always counsel, to some degree or another, sticking with the status quo; upsetting a long-standing interpretation could be disrup­tive and costly.

The relevance of the first two considerations, though not the third, hinges on the rationale for deferring. Both go to the likelihood that the agency interpretation is “right” in the sense of being an accurate understanding of what the statute means. Thus, the implicit premise is that the statute has a meaning (and a “correct” meaning); the answer lies in the statute; and there is reason to have some (though a varying) degree of confidence in the agency’s understanding of it. Longstanding interpretations get special weight because they are more likely to be correct. In contrast, if the court is deferring because Congress has dele­gated policymaking authority to the agency, then by definition the statute does not have a correct reading. A flip-flop does not undercut the agency decision at all; indeed it may reflect the delegation, expertise, and accountability that are thought to justify deference in the first place. Appropriately, then, the Court has held that this most familiar and central feature of non-Chevron deference—reduced deference to incon­sistent agency positions—is inapplicable under Chevron. 74 See Nat’l Cable & Telecomms. Ass’n v. Brand X Internet Servs., 545 U.S. 967, 981–82 (2005) (explaining agency inconsistency does not preclude Chevron deference because “initial agency interpretation is not instantly carved in stone” and that, “[o]n the contrary, the agency . . . must consider varying interpretations and the wisdom of its policy on a continuing basis” (quoting Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 863–64 (1984))).

This leads to a final point. Within its Chevron space, is the agency really “interpreting”? It is very common to speak of Chevron as involving situations where the agency has been granted “interpretive authority.” 75 See City of Arlington v. FCC, 133 S. Ct. 1863, 1878, 1882–83, 1885–86 (2013) (Roberts, C.J., dissenting) (characterizing Chevron as resting on congressional grants of interpretive authority); United States v. Mead Corp., 533 U.S. 218, 229 (2001) (same); Kenneth A. Bamberger, Provisional Precedent: Protecting Flexibility in Administrative Policymaking, 77 N.Y.U. L. Rev. 1272, 1284–86 (2002) (discussing what it means for Congress to “delegate primary interpretive authority to agencies”); Evan J. Criddle, Chevron’s Consensus, 88 B.U. L. Rev. 1271, 1284–85, 1300, 1303–04, 1313 (2008) (“Arguably the leading rationale for Chevron deference is the presumption that Congress delegates interpretive authority to administrative agencies when it commits regulatory statutes to agency administration.”). Yet to the extent the agency is making a normative, policy-based, prescriptive decision, it is not “interpreting” a statute. It is not explaining what Congress has decided; it is deciding in Congress’s place. Judge Posner’s well-known opinion in Hoctor v. United States Department of Agriculture makes the point. 76 82 F.3d 165 (7th Cir. 1996). An agency memorandum provided that only fences eight feet and higher would satisfy a regulatory requirement that certain wild animals be adequately contained. 77 Id. at 168. The agency argued that the memorandum did not have to go through notice-and-comment since it was an interpretive rule, simply explicating what the regulation already required. 78 Id. at 169. The court disagreed, stating that the shift from the general standard requiring effective containment to the specific rule requiring an eight-foot fence cannot “be derived from the regulation by a process reasonably described as interpretation.” 79 Id. at 170. The same can be said of any step-two case: By definition, interpretation has failed to produce an answer. The statute is silent, so the court goes with the agency. It is a category mistake to view Chevron as being about deference to agency interpretations. 80 See Elizabeth V. Foote, Statutory Interpretation or Public Administration: How Chevron Misconceives the Function of Agencies and Why It Matters, 59 Admin. L. Rev. 673, 684–708 (2007) (making extended argument that treating agency implementation of statutes as being “statutory interpretation” is both incorrect and harmful); Herz, Deference Running Riot, supra note 58, at 196–200 (arguing Chevron deference applies to agency policymaking but not agency “interpretation” in sense of determining what it is Congress has done); Richard J. Pierce, Jr., How Agencies Should Give Meaning to the Statutes They Administer: A Response to Mashaw and Strauss, 59 Admin. L. Rev. 197, 200 (2007) (“Step two . . . does not . . . authorize agencies to ‘interpret’ statutes . . . . [I]t recognizes that institutions may choose among competing constructions of a statutory provision that is within the range of meanings that the statutory language can support . . . . An institution can make that choice only by engaging in a policymaking process.” (footnotes omitted)).

It is more than a little odd that courts and commentators continue to insist that Chevron is about “interpretive authority” because, when courts operate in step two (which is what counts, after all), they do not treat the agency decision under review like an interpretation. They treat it like a policy judgment and review it pursuant to the arbitrary-and-capricious test.

Consider Judulang v. Holder. 81 132 S. Ct. 476 (2011). This was a complicated case in which a unanimous Court set aside a policy of the Board of Immigration Appeals concerning the circumstances in which the government would waive deportation of a deportable alien. The government urged the Court to defer under Chevron, but instead the Court reviewed the policy under the arbitrary-and-capricious test:

The Government urges us instead to analyze this case under the second step of the test we announced in [Chevron] to govern judicial review of an agency’s statutory interpretations. Were we to do so, our analysis would be the same, because under Chevron step two, we ask whether an agency interpretation is “‘arbitrary or capricious in substance.’” But we think the more apt analytic framework in this case is standard “arbitrary [or] capricious” review under the APA. The BIA’s . . . policy . . . is not an interpretation of any statutory language . . . . 82 Id. at 483 n.7 (citations omitted).

So, if the argument is not really about statutory meaning, it is not a Chevron case, but rather an arbitrary-and-capricious case. An example is Fox Television, in which the FCC had found that when Fox aired a fleeting expletive, it had broadcast “indecent” material in violation of the relevant statute. 83 FCC v. Fox Television Stations, Inc., 556 U.S. 502, 510–11 (2009). There was no dispute that fleeting expletives were indecent within the meaning of the statute, but the agency had heretofore not bothered to enforce the prohibition in the case of fleeting expletives. 84 Id. at 507–08 (describing enforcement history). It then changed its policy, and Fox challenged the new policy. 85 Id. at 521–22. That argument sounded in arbitrariness; it was not a Chevron case about the meaning of the term “indecent.” Similarly, State Farm 86 Motor Vehicles Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29 (1983). did not turn on whether the revoked passive-restraint standard met the statutory requirements that it be “practicable,” satisfy “the need for motor vehicle safety,” and be “stated in objective terms.” 87 15 U.S.C. § 1392(a) (1976) (current version at 49 U.S.C. § 30111(a) (2012)). Instead, the Court focused on the adequacy of the reasoning supporting the agency’s judgment calls. 88 See State Farm, 463 U.S. at 47–50 (rejecting agency’s decision because of failure to consider obvious alternative).

So far so good. But note that Justice Kagan explicitly says that were Judulang to be resolved under Chevron, “our analysis would be the same. 89 Judulang v. Holder, 132 S. Ct. 476, 483 n.7 (2011) (emphasis added). So Chevron is the arbitrary and capricious test. 90 On the congruence of step two and the arbitrary-and-capricious test, see Nat’l Mining Ass’n v. Kempthorne, 512 F.3d 702, 710 n.4 (D.C. Cir. 2008) (noting overlap between two standards and concluding that because agency interpretation was reasonable under step two, it also survived arbitrary-and-capricious challenge); Bamberger & Strauss, supra note 2, at 621 & n.39 (citing sources); Levin, supra note 27, at 1254 (arguing step two is identical to arbitrary-and-capricious test); Mark Seidenfeld, A Syncopated Chevron: Emphasizing Reasoned Decisionmaking in Reviewing Agency Interpretations of Statutes, 73 Tex. L. Rev. 83, 128–29 (1994) (arguing for “something akin to hard look review” in step two). Except, she says, that test does not apply to statutory interpretation!

The only way to resolve all this is to view step one as being about statutory interpretation—i.e., figuring out what constraints are actually imposed by the statute—and step two as being about the review of policymaking—i.e., determining whether the agency, free of statutory constraint, engaged in reasoned decisionmaking.

All of this sits comfortably with traditional understandings of the separation of powers. In a Skidmore case, and in Chevron step one, the court and the agency are engaged in the same task, and it is one in which the court has primacy. In Chevron step two, the agency is no longer on the court’s turf, the court is on the agency’s. In the absence of an agency conclusion, the court will do the best it can (just as, in the absence of a directly applicable judicial precedent, an agency will do the best it can in determining statutory meaning). But if the agency has acted, it prevails. 91 See Nat’l Cable & Telecomms. Ass’n v. Brand X Internet Servs., 545 U.S. 967, 984 (2005) (“[A] court’s prior interpretation of a statute [may] override an agency’s interpre­tation only if the relevant court decision held the statute unambiguous.”).

This account is at least part of the response to Justice Thomas’s objection that Chevron is unconstitutional. In Michigan v. EPA, Justice Thomas concurred to challenge the constitutionality of Chevron. 92 Michigan v. EPA, 135 S. Ct. 2699, 2714 (2015) (Thomas, J., concurring) (“We should stop to consider [the Constitution] before blithely giving the force of law to any other agency ‘interpretations’ of federal statutes.”). In his view, it violates either Article I or Article III, which vest the legislative and judicial powers in the courts and Congress, respectively. 93 See id. at 2712–13 (Thomas, J., concurring) (concluding Chevron deference “is in tension with Article III’s Vesting Clause, which vests the judicial power exclusively in Article III courts” while also “run[ning] headlong into the teeth of Article I[]”). Viewed as an allocation of “interpretive authority,” Chevron deference “precludes judges from exercising” the constitutionally required “‘independent judgment in interpreting and expounding the laws.’” 94 Id. at 2712 (quoting Perez v. Mortgage Bankers Ass’n, 135 S. Ct. 1199, 1217 (2015) (Thomas, J., concurring)). For constitutional purposes, and in reality, that just is not happening. Courts retain primacy in interpre­tation; the agency’s views matter but are not dispositive and thus the judicial power has not been ceded to another branch. Where courts must defer is when the agency is making a policy decision within the scope of its delegation, within its Chevron space. That is not a judicial task. (Justice Thomas, of course, would say such deference is still uncon­stitutional, this time because it accepts legislation by the executive branch in violation of Article II. 95 See id. at 2713 (Thomas, J., concurring) (stating Chevron “wrests from Courts the ultimate interpretive authority to ‘say what the law is,’ and hands it over to the Executive” (citation omitted)); see also Dep’t of Transp. v. Ass’n of Am. R.R.s, 135 S. Ct. 1225, 1251–52 (2015) (Thomas, J., concurring in the judgment) (“It may never be possible perfectly to distinguish between legislative and executive power, but that does not mean we may look the other way when the Government asks us to apply a legally binding rule that is not enacted by Congress pursuant to Article I.”). That is a question for another Essay.)

B. Skidmore Within Chevron

Accepting the foregoing, Skidmore turns out to have a role to play within Chevron. These are generally perceived as wholly distinct and alternative regimes. Never the twain shall meet. Yet they do meet. To be precise, they meet in Chevron step one. In determining whether the statute answers the question before it in step one, a court should consider the views of the agency charged with its implementation.

Professor Strauss has said as much, 96 See Strauss, Space/Weight, supra note 2, at 1165 (asserting in determining boundaries of Chevron space, which is “unmistakably [a] judicial determination,” courts should be informed by agency judgments); Strauss, In Search of Skidmore, supra note 2, at 795 (pointing out Skidmore deference is “traditional tool[] of statutory interpretation” that should be employed in step one (internal quotation marks omitted)); Strauss, Overseers or Deciders, supra note 2, at 818 (“Skidmore deference is one of those ‘traditional tools of statutory interpretation’ that bear on a court’s independent conclusion about the extent of agency authority.”). and he is not quite alone. 97 See Herz, Deference Running Riot, supra note 58, at 208–09 (“Within step one, the court seeks to determine the meaning of the statute; for all the traditional reasons the agency’s view may be helpful in doing so.”(footnote omitted)); Michael Herz, Judicial Review of Statutory Issues Outside of the Chevron Doctrine, in A Guide to Judicial and Political Review of Federal Agencies 125, 142–43 (John F. Duffy & Michael Herz eds., 2005) (arguing courts should consider agency views within step one of Chevron); Michael Herz, Purposivism and Institutional Competence in Statutory Interpretation, 2009 Mich. St. L. Rev. 89, 120 (noting one benefit of Skidmore deference within step one is it permits courts to be informed by agencies’ superior understanding of legislative purpose).
Recently, other academics have concurred that Skidmore applies in step one. See, e.g., William N. Eskridge, Expanding Chevron’s Domain: A Comparative Institutional Analysis of the Relative Competence of Courts and Agencies to Interpret Statutes, 2013 Wis. L. Rev. 411, 449–50 (citing Strauss and agreeing “Court should announce that federal judges must give Skidmore deference to an agency’s reading of the statute for purposes of Step One”); Glen Staszewski, Statutory Interpretation as Contestatory Democracy, 55 Wm. & Mary L. Rev. 221, 264 n.208 (2013) (citing Herz and agreeing courts “should give Skidmore deference to an agency’s resolution of questions of law under the first step of Chevron”).
But these have been lonely voices. Apparently no court, ever, has expressly stated that Skidmore applies within step one; some have expressly rejected any sort of deference within step one. 98 See, e.g., Mylan Pharms., Inc. v. FDA, 454 F.3d 270, 274 (4th Cir. 2006) (“At the first step a court focuses purely on statutory construction without according any weight to the agency’s position . . . .”). Overwhelmingly, step one is seen as, to reframe Strauss’s terminology, “judicial space.” The court is on its own. 99 See, e.g., Thomas W. Merrill, Judicial Deference to Executive Precedent, 101 Yale L.J. 969, 998 (1992) (“If the issue is resolved at step one, then the court gives no consider­ation to the views of the executive and decides the matter independently.”).

This is a mistake. What a court is doing in step one is the same thing it is doing in a Skidmore case (or a statutory case where there is no relevant agency interpretation): trying to figure out statutory meaning. If agency views are relevant outside of Chevron, then they are relevant in step one. In examining the statute to see if Congress spoke to the matter in question, a court is performing the traditional judicial task of statutory interpretation, trying to find a legislative answer. A Chevron step-one case, then, is essentially the same as a Skidmore case. They are both settings in which the court itself is determining the meaning of the statute and there is a relevant agency interpretation in the background. Because the enterprise is the same, the tools should be the same.

The Chevron opinion supports this approach in two ways. First, it approaches step one like any other statutory case, examining the standard inputs before abandoning the effort and moving to step two. Second, Justice Stevens directs courts to apply the “traditional tools of statutory construction” in step one. 100 Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843 n.9 (1984). Consideration of agency views is emphatically such a tool. To be sure, Justice Stevens did not give weight to the EPA’s views when searching high and low for clues as to the meaning of “source” under the Clean Air Act. 101 See id. That might suggest he deemed the agency views irrelevant in step one. Alternatively, however, he did not invoke the agency for two other reasons. First, he ultimately accepted the EPA’s views outright; 102 See id. at 865 (holding “EPA’s definition of the term ‘source’ is a permissible construction of the statute”). whether he did so in step one or step two may not have seemed consequential. Second, it is important to remember that step one did not at that point exist as the rigid, distinct step that it has become.

This view may seem counterintuitive. But it is not so different from what the courts already do. Consider the use of substantive canons. After a certain amount of confusion, it is now generally settled that the canons are applicable in step one. 103 See Kenneth A. Bamberger, Normative Canons in the Review of Administrative Policymaking, 118 Yale L.J. 64, 77–80 (2008) [hereinafter Bamberger, Normative Canons] (describing courts’ use of substantive canons to resolve textual ambiguities in step one). That is, in trying to determine whether a statute is sufficiently clear to avoid step two, a court will consider, along with text and, depending on the judge, purpose and history, substantive canons. A statutory provision may be ambiguous unless and until such a canon is applied; the canon can resolve the ambiguity and keep the court in step one. For example, it may be unclear whether a particular provision applies retroactively. Judicial uncertainty disappears, however, in light of the canon that statutes should be construed to apply retroactively only if they explicitly so provide. Read in light of the canon, the statute is not ambiguous. 104 See INS v. St. Cyr, 533 U.S. 289, 320–21 n.45 (2001) (“Because a statute that is ambiguous with respect to retroactive application is construed under our precedent to be unambiguously prospective, there is, for Chevron purposes, no ambiguity in such a statute for an agency to resolve.” (citations omitted)). Agency views can be given weight in step one in just the same way.

C. It’s Not Just “Deference” That Is Too Confusing—A Chevron Lexicon

The terminological problems in the Chevron case law go beyond the difficulty of the term “deference” that Professor Strauss highlights. This section examines additional terms that are also “too confusing.” It begins with three terms—deference, persuasion, construction—that are prob­lematic because they have two different meanings and courts fail to distinguish between them. It then turns to two two-word pairs—interpretation/construction and ambiguity/vagueness—that present the op­posite problem: They mean different things but are employed as synonyms. The goal is partly to get the terminology right and partly to use the terminological discussion to clarify the underlying concepts.

1. Deference. — Professor Strauss would replace “deference” with “Chevron space” and “Skidmore weight.” 105 “Weight” was the term used by the Skidmore Court itself. See Skidmore v. Swift & Co., 323 U.S. 134, 140 (1944) (noting “Court has long given considerable and in some cases decisive weight” to agency views and “he weight of [the agency’s] judgment in a particular case will depend on” various factors). See generally Strauss, Space/Weight, supra note 2. The proposition is that “the d-word” is being used to label two different activities. His point is about judicial usage, but it is worth adding that the roots of the difficulty can be found in the two different dictionary definitions of “deference.” The first is “[s]ubmission or courteous yielding to the opinion, wishes, or judgment of another.” 106 1 The Heritage Illustrated Dictionary of the English Language 346 (1979). The second is “[c]ourteous respect.” 107 Id. The first definition implies subordination; the person deferring “submits” or “yields” to another, evidently superior decisionmaker. The second definition implies authority; the person deferring is the ultimate decisionmaker but must give weight to, take seriously, or respect the views meriting deference. 108 To be sure, different dictionaries give different definitions. The Merriam-Webster Collegiate Dictionary, for example, has only one relevant definition, and it is the Chevron version. See Merriam-Webster’s Collegiate Dictionary 302 (10th ed. 1993) (defining “defer” as “to submit to another’s wishes, opinion, or governance” and giving “yield” as sole synonym). For present purposes, it is enough that the definitions abroad in the land include both the Skidmore version and the Chevron version. “Defer” can mean either, and courts do both.

These two definitions easily map onto the deference case law. The first is Chevron step two; the second is Skidmore. 109 See, e.g., United States v. Mead Corp., 533 U.S. 218, 234 (2001) (“Chevron did nothing to eliminate Skidmore’s holding that an agency’s interpretation may merit some deference whatever its form . . . .”). The phrase “some deference” is only coherent when “deference” is the “courteous respect” kind; the Chevron type is by its nature all or nothing. Thus, if a court “defers” under Chevron, it by definition accepts the agency’s view; under Skidmore, in contrast, a court can give an agency interpretation “deference” while still rejecting it.

Justice Scalia has made half this point. He thinks “defer” has only one meaning, the Chevron meaning. Dissenting in a case where the majority invoked Skidmore and noting that the agency’s views “add force to our conclusion” about the meaning of the statute, 110 Kasten v. Saint-Gobain Performance Plastics Corp., 131 S. Ct. 1325, 1335 (2011). Scalia objected:

In my view this doctrine (if it can be called that) is incoherent, both linguistically and practically. To defer is to subordinate one’s own judgment to another’s. If one has been persuaded by another, so that one’s judgment accords with the other’s, there is no room for deferral—only for agreement. 111 Id. at 1340 n.6 (Scalia, J., dissenting).

He is right, if Chevron is the only deference regime. But if, as a majority of Supreme Court Justices have held, there are two deference regimes, then they involve the two different meanings of “defer” that are found in most dictionaries (though not Justice Scalia’s).

2. Persuasive. — A similar problem arises with regard to “persuasive.” It is also “too confusing” because it has two distinct meanings. “Persuasive” may mean “tending to persuade,” 112 See, e.g., Merriam-Webster’s Collegiate Dictionary, supra note 108, at 868. and it may mean “convincing.” 113 See, e.g., Harbison v. Bell, 556 U.S. 180, 182 (2009) (“[T]he evidence proved persuasive to one Circuit Judge.”); id. at 189 (“Neither argument is persuasive.”); Richlin Sec. Serv. Co. v. Chertoff, 553 U.S. 571, 583 (2008) (“We are not persuaded [by the Government’s statutory arguments].”). That is why a “persuasive advocate” can leave one “unpersuaded.”

Opinions in Skidmore cases are awash with careless use of this term. With dismaying frequency, courts assert that the persuasiveness of the agency’s reasoning is one of the factors affecting the weight to be given the agency’s views under Skidmore. 114 See, e.g., Power v. Barnhart, 292 F.3d 781, 786 (D.C. Cir. 2002) (“Under Skidmore, we grant an agency’s interpretation only so much deference as its persuasiveness warrants.”); Catskill Mountains Chapter of Trout Unlimited, Inc. v. City of New York, 273 F.3d 481, 491 (2d Cir. 2001) (“[T]he agency position should be followed to the extent persuasive.”). This proposition is either a truism or incoherent. Undeniably, a powerfully reasoned agency conclusion is more likely to be upheld than a poorly reasoned one. Courts reward “reasoned decision­making.” However, to accept an interpretation because it is “persuasive” is not to defer (in either of its senses), but rather to agree. 115 In the words of a leading casebook: “We would not say that a court persuaded by the excellent brief of a private litigant to decide in its favor is deferring to the litigant’s lawyers.” Peter L. Strauss et al., Gellhorn & Byse’s Administrative Law: Cases and Comments 975 (11th ed. 2011); see also Kasten, 131 S. Ct. at 1340 n.6 (Scalia, J., dissenting) (“Speaking of ‘Skidmore deference’ to a persuasive agency position does nothing but confuse.”); id. (Scalia, J., dissenting) (“If one has been persuaded by another, so that one’s judgment accords with the other’s, there is no room for deferral—only for agreement.”). The various Skidmore factors (consistency, contemporaneousness, age) all add to the weight of the agency interpretation, making it more “persuasive”; the interpretation’s “persuasiveness” is not itself a Skidmore factor.

The Supreme Court has done much to sow this confusion. For exam­ple, Mead identifies “the persuasiveness of the agency’s position” as one of the Skidmore factors. 116 United States v. Mead Corp., 533 U.S. 218, 228 (2001); see also Vance v. Ball State Univ., 133 S. Ct. 2434, 2443 n.4 (2013) (“[T]o [defer to the EEOC] would be proper only if the EEOC Guidance has the power to persuade . . . . [W]e do not find the EEOC Guidance persuasive.”). But this is not what Skidmore actually says. On Justice Jackson’s formulation, the Skidmore factors are the things that give the agency’s interpretation “the power to persuade”; to say that the “persuasiveness” of the agency’s position is one of the things that give it “the power to persuade” is tautological.

Similarly, in a problematic but much-quoted formulation, the Christensen majority opinion stated that agency interpretations are “‘entitled to respect’ . . . , but only to the extent that those interpretations have the ‘power to persuade.’” 117 Christensen v. Harris County, 529 U.S. 576, 587 (2000); see also Gonzales v. Oregon, 546 U.S. 243, 269 (2006) (observing “under Skidmore, we follow an agency’s rule only to the extent it is persuasive,” and finding Attorney General’s opinion unpersuasive); Pub. Citizen, Inc. v. U.S. Dep’t of Health and Human Servs., 332 F.3d 654, 662–71 (D.C. Cir. 2003) (concluding Skidmore deference due but rejecting agency’s view as “unpersuasive” because statutory text, history, and purpose were all against it); id. at 670 n.25 (invoking views of Department’s Inspector General “not because they legally command our deference, but because we find them logically persuasive”). This is self-contradictory. Everything, including the arguments of the lawyers in the case, is entitled to respect “to the extent [it has] the power to persuade.” 118 Christensen, 529 U.S. at 587 (emphasis added) (internal quotation marks omitted). Under this approach, if the court agrees with the agency it will label the interpretation persuasive and “defer” to it, whereas if it disagrees, it will label the interpretation “unpersuasive” and ignore it. 119 See Rossi, supra note 59, at 1131–34 (describing courts’ tendency to defer under Skidmore only when their independent determination of best interpretation accords with agency’s view); see also Catskill Mountains, 273 F.3d at 491 (noting “agency position should be followed to the extent persuasive” and rejecting agency’s position as unpersuasive because wrong). For example, consider this sentence from a court of appeal’s opinion: “[T]he Director is not entitled to Skidmore respect . . . because his position . . . is simply unpersuasive, notwithstanding its inclusion in the agency’s manual and the Director’s consistent application of [this interpretation] for some time.” 120 Price v. Stevedoring Servs. of Am., Inc., 697 F.3d 820, 832 (9th Cir. 2012); see also id. at 833 (“[T]he Board’s explanations as to the contested issues here are not persuasive and would thus not be entitled to deference in any event.”). This is not deference; the court is reaching an independent conclusion. On this approach, Skidmore means nothing more than “we will defer to the agency if we believe the agency is right.” 121 See Barron & Kagan, supra note 46, at 227 n.98 (wondering aloud whether “Skidmore deference amounts to something more than a court saying ‘we will defer to the agency if we believe the agency is right’”); see also Panel, Agency Preemption: Speak Softly, but Carry a Big Stick?, 11 Chap. L. Rev. 363, 382 (2008) (comments of Ronald A. Cass) (“[T]hat’s why the notion of Skidmore deference is a wonderful concept for courts, which means that they do what the agency says when they would’ve done it without the agency doing it anyway.”). If that’s what it means, it means nothing at all, and Skidmore deference is a fiction.

There are two points here. First, thinking about “persuasiveness” helps clarify the difference between Chevron and Skidmore. The concept is simply irrelevant in step two of Chevron. There, the issue is not whether the agency is right or whether the court agrees with (is persuaded by) the agency’s views. Unless unreasonable (a very different standard), the agency’s views control. It is because the court’s views ultimately control in Skidmore that the question of persuasiveness arises. Second, the relevant meaning of “persuasive” is not “convincing,” but rather “tending to persuade.” Otherwise, the concept is meaningless.

If the Straussian lexicon is adopted, “persuasive” can be abandoned. In the Chevron setting, the term is inapplicable. In the Skidmore setting, where “persuasive” should mean “tending to persuade,” the term “weight” does all the work necessary.

3. Construction. — “Construction,” too, has two meanings that are often confounded. This is because the word is the noun form of two separate verbs with the same etymological root: construe and construct. The first implies interpretation, determining the meaning of the text being construed; the second implies creation, building on the text to construct something new. Which verb is at work in the term “statutory construction”? The answer is both, depending on the context. In Skidmore, the agency, and then the court, are construing the statute, working “to understand or explain the sense or intention of” it. 122 Merriam-Webster’s Collegiate Dictionary, supra note 108, at 248 (defining “construe”). In contrast, under Chevron, the agency has constructed something—not the statute, which by definition was “built” by Congress, but a determination of legal rights or obligations in the space Congress left for the agency to work within—that the court must accept. Courts construe; agencies construct.

To simply abandon the term “construction” would be a mistake, and would not be possible in any event. The following section suggests, however, that clarity would be advanced if “construction” in this setting were limited to being the noun for “construct,” and “interpretation” the noun for “construe.”

4. Construction Versus Interpretation. — Almost without exception, writing about statutory interpretation uses “interpretation” and “con­struction” as synonyms. 123 See, e.g., Bryan A. Garner, A Dictionary of Modern Legal Usage 462 (2d ed. 1995); Gary E. O’Connor, Restatement (First) of Statutory Interpretation, 7 N.Y.U. J. Legis. & Pub. Pol’y 333, 335 n.5 (2004) (“This article treats the terms ‘statutory interpretation’ and ‘statutory construction’ as interchangeable synonyms.”). Indeed, Reed Dickerson defines “construction” with a single word: “interpretation.” 124 Reed Dickerson, The Interpretation and Application of Statutes 283 (1975). In Chevron itself, Justice Stevens uses the two terms interchangeably. 125 See Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 842, 844, 865–66 (1984) (referring interchangeably to EPA’s “interpretation” and EPA’s “construction” of Clean Air Act). Later courts have done the same, alternating between the two mainly, it would seem, to avoid repeating the same word too many times in one sentence. 126 See, e.g., Nat’l Cable & Telecomms. Ass’n v. Brand X Internet Servs., 545 U.S. 967, 986 (2005) (“If the statute is ambiguous on the point, we defer at step two to the agency’s interpretation so long as the construction is [reasonable] . . . .”); id. at 982–83 (noting “statute unambiguously forecloses the agency’s interpretation” and “displaces conflicting agency construction”); Thomas Jefferson Univ. v. Shalala, 512 U.S. 504, 506 (1994) (“The dispositive question is whether the Secretary’s interpretation is a reasonable construction of the regulatory language.”); City of Chicago v. Envtl. Def. Fund, 511 U.S. 328, 339 n.5 (1994) (“In view of our construction of § 3001(i), we need not consider whether an agency interpretation . . . .”); Good Samaritan Hosp. v. Shalala, 508 U.S. 402, 417 (1993) (“[W]here the agency’s interpretation of [its regulation] is at least as plausible as competing ones, there is little, if any, reason not to defer to its construction.”); Vander Boegh v. EnergySolutions, Inc., 772 F.3d 1056, 1062 (6th Cir. 2014) (“Affording the agency’s interpretation Chevron deference, this court held that the agency’s construction was permissible . . . .”); Perry v. Dowling, 95 F.3d 231, 237 (2d Cir. 1996) (“[T]he interpretation is a permissible construction of the statute . . . .”). Commentators also alternate between the two at random. 127 See Beermann, supra note 20, at 744 (“The Chevron doctrine . . . establishes a two-step process for judicial review of agency interpretations of statutory construction.”); Matthew C. Stephenson & Adrian Vermeule, Chevron Has Only One Step, 95 Va. L. Rev. 597, 598 (2009) (“[E]ven if the agency’s interpretation is plausible as a construction of the statutory language, the agency’s choice among plausible interpretations must not be ‘arbitrary [or] capricious.’”).

This was not always so, and it is not so today in some legal fields. This section explores the possible distinction between interpretation and construction as a way of understanding Chevron.

Section II.A concluded that what judges do in Skidmore or step one is interpretation in a standard sense—trying to figure out what the statute means (or, on traditional accounts, what Congress meant or intended). 128 See supra notes 82–97 and accompanying text (explaining “interpretation” is what courts do in Skidmore cases and in Chevron step one, but not what agencies have done in a case decided in step two). In a Skidmore case, where the court has the last word, the underlying premise is Dworkinian: There is a right answer. For Ronald Dworkin, the question “what is the law on this issue?” always has a discoverable right answer, and the judge’s job is to discover it. Even in hard cases, law does not run out and judges are not forced to legislate. 129 See generally Ronald Dworkin, Hard Cases, 88 Harv. L. Rev. 1057 (1975), reprinted in Ronald Dworkin, Taking Rights Seriously 81, 84 (1977) (distinguishing between arguments on principle and arguments on policy and asserting judges’ decisions are and should be based on the former). Now, Dworkin is not asserting that the one right answer is one that Congress put there. To ensure “integrity in legislation,” 130 Ronald Dworkin, Law’s Empire 167, 176–84, 217–28 (1986) [hereinafter Dworkin, Law’s Empire]. courts must interpret a statute, just as they must interpret the Constitution or common law precedents, in light of the underlying principles of the community. In hard cases, the right interpretation is the one that aligns with underlying values and makes the statute the best statute it can be. 131 See Ronald Dworkin, How to Read the Civil Rights Act, in A Matter of Principle 316, 329 (1985) [hereinafter Dworkin, A Matter of Principle] (arguing judges faced with unclear statutory text should apply statute so as to further justification that “is superior as a matter of political morality”); Ronald Dworkin, Is There Really No Right Answer in Hard Cases?, in Dworkin, A Matter of Principle, supra, at 119, 129 (noting view that judge should interpret statute so as to “best advance[] the set of principles and policies that provides the best political justification for the statute at the time it was passed”); Dworkin, Law’s Empire, supra note 130, at 313–54 (characterizing judges as herculean coauthors of statutes in relying on their best judgment when deciding hard cases). That is a tall order, which is why only a judge such as Hercules can be counted upon to get it right. Because herculean judges are few and far between, courts should take help where they can find it. One such source of assistance is the agency charged with administration of the statute.

In contrast, Chevron is positivist. Its very premise is that law some­times runs out. Under Chevron, the question, “What is the law on this issue?” does not always have a right answer (at least if it is understood to mean “what does the statute prescribe?”). Professor Dworkin asserts that because positivists believe the law “runs out,” they must endorse some version of strong discretion whereby judicial decision is unconstrained by legal principles. 132 Ronald Dworkin, The Model of Rules, 35 U. Chi. L. Rev. 14, 35 (1967) (“[W]hen a judge runs out of rules he has discretion, in the sense that he is not bound by any standards from the authority of law . . . .”). Chevron is not inconsistent with the basis of that critique, it just offers a different resolution. Instead of accepting uncon­strained judicial discretion, it requires judicial acceptance of (largely) uncon­strained administrative discretion. In other words, Professor Dworkin does not believe that law ever runs out (so there is no reason not to leave decisionmaking to judges), but Chevron does (which is why, in the absence of law, decisions should be made by agencies).

This is a coherent and even appealing view of the Skidmore/Chevron divide. The question is whether it has any basis in reality. Professor Dworkin came late to the question of statutory interpretation and paid almost no attention to administrative agencies. 133 See, e.g., Patrick McKinley Brennan, Realizing the Rule of Law in the Human Subject, 43 B.C. L. Rev. 227, 323 (2002) (“I know of nothing in Dworkin’s writings that suggests he knows that there is a ‘counter-Marbury, for the administrative state,’ the rule of Chevron.” (footnote omitted)); Cass R. Sunstein, Justice Scalia’s Democratic Formalism, 107 Yale L.J. 529, 566 & n.204 (2007) (reviewing Antonin Scalia, A Matter of Interpretation: Federal Courts and the Law (1997)) (stating writings on statutory interpretation by both Professor Dworkin and Justice Scalia fail to come to terms with centrality of agency interpretation). And the belief that texts have a single, determinate, correct meaning is not a popular one. If all interpretation involves discretion, and any text might have multiple meanings, then the clear distinction between what is happening in Skidmore and what is happening in Chevron starts to collapse.

One possible route out of that difficulty is the distinction that is drawn in some areas of the law between interpretation and construction.

Almost two centuries ago, Francis Lieber, later to author the first and still influential codification of the laws of warfare and to become one of the first professors at the Columbia Law School, wrote a book with the strikingly modern title of Legal and Political Hermeneutics. 134 Francis Lieber, Legal and Political Hermeneutics: Principles of Interpretation and Construction in Law and Politics, With Remarks on Precedents and Authorities (William G. Hammond ed., 3d ed. 1880). This volume, which is not quite as cutting edge as the title makes it sound to modern readers, set out a theory of textual interpretation. Central to Lieber’s hermeneutics was a distinction between interpretation and construction. Interpretation was the narrower task, consisting of “the discovery and representation of the true meaning of any signs used to convey ideas.” 135 Id. at 5; see also id. at 11 (“Interpretation is the art of finding out the true sense of any form of words; that is, the sense which their author intended to convey, and of enabling others to derive from them the same idea which the author intended to convey.”). Every text requires interpretation, for “[t]here is no direct communion between the minds of men.” 136 Id. at 1. In some cases, the interpreter can discover the true meaning based on little besides a careful reading of the text. In others, resort must be had to other materials, but the goal is always to ascertain the speaker’s meaning. 137 Id. at 106–07 (explaining process of interpreting ambiguous terms using context and related sources).

Where interpretation does not “suffice[] . . . we must have recourse to construction.” 138 Id. at 50. “Construction is the drawing of conclusions re­specting subjects, that lie beyond the direct expression of the text, from elements known from and given in the text—conclusions which are in the spirit, though not within the letter of the text.” 139 Id. at 44; see also id. at 46 (“[C]onstruction signifies the discovery of the spirit, principles and rules that ought to guide us according to the text . . . .”). Most often, con­struction is necessary where unanticipated circumstances have arisen or background circumstances have changed so much as to make inter­pretation impossible.

Lieber’s distinction maps tidily onto Chevron, particularly if step one is not especially capacious. Lieber’s definition of interpretation corre­sponds to Chevron’s step one: “First, always, is the question of 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 . . . .” 140 Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 842 (1984). In determining whether Congress has spoken to the issue, the court should employ the “traditional tools of statutory construction.” 141 Id. at 843 n.9. In Chevron itself, the Court used the “traditional tools” of examining language, purpose, and legis­lative history before it gave up and deferred to the agency. The premise here is pure Lieber: Congress had an intent, the statute has one true meaning, and through interpretation (which is not the same as a literal reading and may require more than simply reading the statute and seeing what it says) the court comes to understand Congress’s meaning.

A court moves to Chevron’s step two, upholding any reasonable or permissible agency reading, if “the statute is silent or ambiguous with respect to the specific issue.” 142 Id. at 843. In these circumstances, Congress has not resolved the issue and the court will simply accept any reasonable agency determination. In other words, the court does not try to wring some meaning out of the statute beyond what Congress put there. Or, as Lieber puts it: “Interpretation, seeking but for the true sense, forsakes us when the text is no longer directly applicable; because the utterer, not foreseeing this case, did not mean it, therefore it has no true sense in this particular case.” 143 Lieber, supra note 134, at 111. In these circumstances, the Chevron analysis moves to step two and Lieber moves to construction. For both, the agency is still limited by whatever constraints the statute, the Constitution, and general requirements of reasonableness impose.

Lieber’s distinction has never really taken hold in the realm of statutory interpretation. However, in recent years, several theorists have emphasized the distinction between interpretation and construction as a way of understanding what is going on when courts attempt to under­stand and apply the Constitution. Keith Whittington and Lawrence Solum are leading examples. 144 See, e.g., Keith E. Whittington, Constitutional Construction: Divided Powers and Constitutional Meaning 2, 5 (1999) (arguing “[n]ot everything that courts do is consistent with the ideal of interpretation,” and “[u]nlike jurisprudential interpretation, construction” is more political and “provides for an element of creativity in construing constitutional meaning”); Lawrence B. Solum, The Interpretation-Construction Distinction, 27 Const. Comment. 95, 95 (2010) [hereinafter Solum, Interpretation-Construction Distinction] (arguing “interpretation-construction distinction . . . marks a deep difference in two different stages . . . in the way that legal and political actors process legal texts”); Lawrence B. Solum, Originalism and Constitutional Construction, 82 Fordham L. Rev. 453, 489 (2013) [hereinafter Solum, Originalism] (“The distinction rests on an underlying set of distinctions about communicative content (meaning), legal content (doctrine), and legal effect (decision).”); Keith E. Whittington, Constructing a New American Constitution, 27 Const. Comment. 119, 128 (2010) [hereinafter Whittington, Constructing] (“Constitutional interpretation may put the judiciary on the strongest footing for exercising the power of judicial review and overturning the actions of other government officials, but there are circumstances in which judges like other political actors might engage in constitutional construction.”); see also Jack M. Balkin, Living Originalism 3–6, 69–73 (2011) (distinguishing “interpretation-as-ascertainment” from “interpretation-as-construction”); Randy E. Barnett, Interpretation and Construction, 34 Harv. J.L. & Pub. Pol’y 65, 69 (2011) (describing limits of originalism and emphasizing necessity of judicial interpretation in constitutional construction). They articulate a distinction between understanding the semantic content of the constitutional text (interpretation) and applying that meaning in actual, often unantici­pated, circumstances (construction). The latter necessarily goes beyond the meaning of the document; though guided by it, construction involves some input from, or additions by, the constructer. As Whittington puts it: “The process of constitutional construction is concerned with fleshing out constitutional principles, practices[,] and rules that are not visible on the face of the constitutional text and that are not readily implicit in the terms of the constitution.” 145 Whittington, Constructing, supra note 144, at 120.

The two steps of interpretation and then construction—determining semantic meaning and then applying that meaning to particular circumstances—occur in statutory cases as well, even if the vocabulary is unfamiliar. 146 For example, Reed Dickerson distinguishes between the “ascertainment of meaning” (which he labels “cognitive function”) and the “assignment of meaning” (which he calls “creative function”). Dickerson, supra note 124, at 15. Professor Dickerson mentions that this distinction resembles or overlaps with the interpretation/construction distinction. Id. at 19; see also Robert Martineau, Craft and Technique, Not Canons and Grand Theories: A Neo-Realist View of Statutory Construction, 62 Geo. Wash. L. Rev. 1, 2 n.1 (1993) (quoting Robert Martineau, Drafting Legislation and Rules in Plain English (1991)) (describing statutory construction as involving, first, interpretation to determine meaning and, second, application to facts, and stating often courts do not realize they are engaging in two steps rather than one). In the constitutional sphere, Professor Solum has referred to issues involving construction as opposed to interpretation as arising in “the construction zone.” 147 Solum, Originalism, supra note 144, at 469–72. The term of course resonates with the idea of “Chevron space.” 148 The term also brings to mind the “zone of indeterminacy,” a phrase Peter Strauss may have coined. See Strauss, 150 Cases, supra note 2, at 1124 (stating within “zone of determinacy,” “Supreme Court should tolerate the gradual accretion of circuit interpre­tations of indeterminate statutes, focusing . . . instead on . . . issues on which Congress appears to have ‘directly spoken,’” which “reduces the need for the Court to exercise direct control”); see also Bamberger, Normative Canons, supra note 103, at 106 (referring to “zone of indeterminacy”); David S. Rubenstein, “Relative Checks”: Toward Optimal Control of Administrative Power, 51 Wm. & Mary L. Rev. 2169, 2193 (2010) (same). The parallel would be that interpretation takes place in step one of Chevron and in Skidmore cases, whereas construction, within statutorily defined boundaries, takes place in step two.

Gillian Metzger utilizes the interpretation/construction distinction in writing about administrative constitutionalism, i.e., the various ways in which administrative agencies interpret, implement, and help define the meaning of the U.S. Constitution. 149 See generally Gillian E. Metzger, Administrative Constitutionalism, 91 Tex. L. Rev. 1897 (2013) (analyzing administrative constitutionalism as form of constitutional interpretation). She argues that administrative constitutionalism “stands as a prime example of constitutional construc­tion.” 150 Id. at 1914. Such construction takes place subject to judicial oversight; courts have ultimate but not exclusive (if that is not an oxymoron) authority to determine constitutional meaning. 151 See id. at 1925–27 (discussing relationship between judicial review and admin­istrative constitutionalism). Precisely the same phenomenon exists with regard to “administrative statutism.” Properly read, Chevron is an admonition to respect agency statutory construction—which is different from statutory interpretation.

An illustration is the classic Supreme Court decision of NLRB v. Hearst Publications, Inc. 152 322 U.S. 111 (1944). The National Labor Relations Board (NLRB) had concluded that newspaper sellers were “employees” of the papers they sold within the meaning of the labor laws. 153 Id. at 114. The Court of Appeals rejected this conclusion, relying on the common law of California. 154 Id. at 114–15 (outlining procedural history). The Supreme Court reversed, upholding the Board’s initial decision. It did so in two steps. First, it concluded that the question did not turn on common law standards, which vary from state to state: “Both the terms and the purposes of the statute, as well as the legislative history, show that Congress had in mind no such patchwork plan for securing freedom of employees’ organization and of collective bargaining.” 155 Id. at 123. In this portion of the opinion, the Court made only a passing reference to the Board and did not even mention its views on the legal issue. It then turned to the question of whether, as a matter of federal law, the newsboys were employees. With regard to this issue, the agency loomed large. The Court spent some time on the general purposes of the Act, noted that there will be lots of borderline situations, and stated that the definitional “task has been assigned primarily to the agency created by Congress to administer the Act.” 156 Id. at 130. Accordingly, “the Board’s determination that specified persons are ‘employees’ under this Act is to be accepted if it has ‘warrant in the record’ and a reasonable basis in law,” 157 Id. at 131. which it did.

The two parts of the Court’s opinion might be categorized in many ways. The standard administrative law formulation would be to call the first a “pure question of law” and the second a “mixed question of law and fact.” 158 Justice Stevens, with uncertain support from his colleagues, consistently took the position that Chevron does not apply to “pure questions of statutory interpretation.” See, e.g., Negusie v. Holder, 555 U.S. 511, 531 (2009) (Stevens, J., concurring and dissenting) (stating “pure questions of statutory interpretation . . . remain within the purview of the courts, even when the statute is not entirely clear”); INS v. Cardoza-Fonseca, 480 U.S. 421, 446 (1987) (“The question whether Congress intended the two standards to be identical is a pure question of statutory construction for the courts to decide.”). Those categories correspond to interpretation and construc­tion. When there are no facts, the inquiry is abstract and the answer need not be applied to particular circumstances; it is “pure.” Lawmaking is “pure” in this way as well. But statutory meaning (like the meaning of a common law precedent) is fully developed through application to particular cases. That is construction, and Hearst and Chevron alike teach that agencies have a particular role to play in that activity.

5. Ambiguity Versus Vagueness. — The interpretation-versus-construction distinction rests on one other distinction: that between ambiguity and vagueness. 159 See, e.g., Caleb Nelson, Statutory Interpretation 77–80 (2011) (explicating distinction); Ralf Poscher, Ambiguity and Vagueness in Legal Interpretation, in The Oxford Handbook of Language and Law 128, 128–29 (Peter M. Tiersma & Lawrence M. Solan eds., 2012) (same). A term is ambiguous when it can have two or more meanings; each alone is clear and understandable, but the reader is uncertain as to which is in play. A term is vague when its scope is unclear and there are marginal cases to which it may or may not apply. Professor Solum explains:

A word, phrase, sentence, or clause is ambiguous if it has more than one sense: for example, the word “cool” is ambiguous because it can mean (a) hip, (b) of low temperature, or (c) of even temperament. A word or phrase is vague when it has borderline cases: for example, the word “tall” is vague, because there is no bright line between those individuals who are tall and those who are not. The same word can be both ambiguous and vague in one of its senses: cool is ambiguous and each sense of cool is vague. 160 Lawrence Solum, Legal Theory Lexicon 063: Interpretation and Construction (Apr. 27, 2008), http://lsolum.typepad.com/legal_theory_lexicon/2008/04/legal-theory-le.html [http://perma.cc/Q8LZ-SN9Z]; see also Barnett, supra note 144, at 67 (“[L]anguage is ambiguous when it has more than one sense; it is vague when its meaning admits of borderline cases that cannot definitively be ruled in or out of its meaning.” (emphasis omitted)). See generally Solum, Interpretation-Construction Distinction, supra note 144, at 97–98 (explaining distinction between vagueness and ambiguity).

The distinction maps on to the interpretation/construction distinction. In general, interpretation is the process for resolving ambiguity; construction is the process for resolving vagueness.

If this distinction is indeed applicable to Chevron as suggested above, then step one should involve resolving ambiguity and step two should involve resolving vagueness. And this is indeed (almost always) the case. This is emphatically not what courts say. The Chevron opinion and most subsequent case law and commentators state that under Chevron, deference is triggered by statutory “ambiguity.” 161 See, e.g., Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843 (1984) (“[I]f 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.”); Calix v. Lynch, 784 F.3d 1000, 1007 (5th Cir. 2015) (“When a statute is ambiguous, we defer to an agency’s reasonable interpretation.”). The word “vague” does not appear in Chevron. See generally Chevron, 467 U.S. 837. Sometimes, courts use the term “vague” alongside “ambiguous”; it is not clear that they mean to distinguish the two. 162 See, e.g., Rollins v. Admin. Review Bd., 311 F. App’x 85, 86 (10th Cir. 2008) (noting court moves to step two “if the statute uses a vague or ambiguous term”); Stevens v. Premier Cruises, Inc., 215 F.3d 1237, 1240 (11th Cir. 2000) (noting under Chevron “statute is not vague or ambiguous just because it is broad”). But they are not synonymous, and Chevron is not primarily about ambiguity, it is about vagueness.

When a statute is ambiguous—that is, we are unsure as to which of two (or more) meanings apply—Congress generally has made a decision and the challenge is figuring out what that decision was. This is not to say that Chevron is never applicable in cases of ambiguous statutory language; sometimes, when a term can have either of two meanings, there really is no way of saying which is “right.” But, in general, the challenges of ambiguity are ones of semantic meaning and are resolved by courts. The classic Chevron case involves statutory vagueness. A provision is of inher­ently uncertain scope and no amount of staring at the text or rummaging in legislative history can make the penumbra less penumbral. Are wetlands “waters of the United States”? 163 See United States v. Riverside Bayview Homes, Inc., 474 U.S. 121, 121, 131 (1985) (upholding agency regulation concluding they are under step two). Is an entire facility a “source”? 164 Chevron, 467 U.S. at 840. Are newsboys “employees”? 165 NLRB v. Hearst Publ’ns., 322 U.S. 111, 132 (1944). This is not to say that Chevron is per se inapplicable in cases of ambiguity. If a statute is confoundingly ambiguous, the court must defer, because it cannot find an answer using the traditional tools of statutory construction. However, vague statutes can be inherently uninterpretable; by definition, the right answer not only cannot be found in the statute, it does not exist there. Moreover, to the extent Chevron is understood to rest on an explicit or implicit delegation, it is much more reasonable to discover a delegation in a vague statute than an ambiguous one. This point is made by Ryan D. Doerfler, Mead as (Mostly) Moot: Predictive Interpretation in Administrative Law, 36 Cardozo L. Rev. 499, 500 n.4 (2014) (arguing claim of implicit delegation is more plausible with respect to vague statutes than with respect to ambiguous ones, as matter of principle).

In sum, a different activity is involved in resolving questions of ambiguity and questions of vagueness. This difference is the difference between interpretation and construction.

III. A Case Study: Chevron and Jurisdictional Questions

A much-debated issue of “Chevronology” 166 This neologism seems to have been coined by John Manning, who, at the Columbia conference mentioned in the opening footnote, referred to Peter Strauss as “our leading Chevronologist.” John Manning, Remarks at The Harold Leventhal Memorial Symposium, Celebrating Peter Strauss, Columbia Law School (Apr. 24, 2015). The term is so apt (not least because of hint of a reference to “phrenology” and the quasi-scientific dubiousness of that field) and appealing that it is hard to believe it took three decades to appear. is whether Chevron applies to an agency’s statutory interpretation bearing on its “jurisdiction,” or the scope of its power. Until recently, this was a matter of dispute. 167 See Ernest Gellhorn & Paul Verkuil, Controlling Chevron-Based Delegations, 20 Cardozo L. Rev. 989 (1999) (contending Chevron deference should not extend to jurisdictional issues); Nathan A. Sales & Jonathan H. Adler, The Rest Is Silence: Chevron Deference, Agency Jurisdiction, and Statutory Silences, 2009 U. Ill. L. Rev. 1497, 1518 (describing conflicting lower court decisions and noting “D.C. and Eighth Circuits appear to have resolved the issue both ways”). Compare Miss. Power & Light Co. v. Mississippi ex rel. Moore, 487 U.S. 354, 381–82 (1988) (Scalia, J., concurring in the judgment) (contending “there is no discernible line between an agency’s exceeding its authority [i.e., ‘jurisdiction’] and an agency’s exceeding authorized application of its authority”), with id. at 386–87 (Brennan, J., dissenting) (contending “[a]gencies do not ‘administer’ statutes confining the scope of their jurisdiction, and such statutes are not ‘entrusted’ to agencies”). Many shared an intuition against deference on such issues, fearing that Chevron would enable agency empire-building. In City of Arlington v. FCC, 168 133 S. Ct. 1863 (2013). the Court held that Chevron does apply to jurisdictional questions. For Justice Scalia, this was a case about whether Chevron lived or died:

Make no mistake—the ultimate target [of those arguing for an exception] is Chevron itself. Savvy challengers of agency action would play the “jurisdictional” card in every case . . . . The effect would be to transfer any number of interpretive decisions—archetypal Chevron questions, about how best to construe an ambiguous term in light of competing policy interests—from the agencies that administer the statutes to federal courts. We have cautioned that “judges ought to refrain from substituting their own interstitial lawmaking” for that of an agency. That is precisely what Chevron prevents. 169 Id. at 1873 (citations omitted).

On this account, City of Arlington is the case that saved Chevron. 170 At least for the time being. Some have suggested that King v. Burwell, 135 S. Ct. 2480 (2015), represents the triumph of the Chief Justice’s position in dissent in City of Arlington. See, e.g., Jody Freeman, The Chevron Sidestep: Professor Freeman on King v. Burwell, Blog of the Harv. Law Sch. Envtl. Law Program, http://environment.law.harvard.edu/2015/06/the-chevron-sidestep/ [http://perma.cc/UYB4-LWBT] (last visited Aug. 3, 2015) (stating after City of Arlington, Chief Justice “has clearly been waiting in the weeds” and in King v. Burwell, he “had his way,” “elevat[ing] his Arlington dissent into a majority opinion”). This Part assesses just what version of Chevron survived, using this con­troversy to elucidate the approach to Chevron outlined above, concluding that both the majority and the dissent are consistent with it.

A. Defining “Jurisdictional” Questions

Justice Scalia’s opinion for the Court rejected the idea that a distinct category of jurisdictional questions even existed and admonished that “judges should not waste their time in the mental acrobatics needed to decide whether an agency’s interpretation of a statutory provision is ‘jurisdictional’ or ‘nonjurisdictional.’” 171 City of Arlington, 133 S. Ct. at 1870. Echoing his long-ago opinion in Mississippi Power & Light, 172 See supra note 167 (describing divisions among Justices in that case). Justice Scalia wrote that it just does not matter whether an agency action is characterized as marking an improper assertion of power or an improper application of an accepted power. In either event, if the charge is substantively correct, the agency action is ultra vires. The real question is therefore “always, simply, whether the agency has stayed within the bounds of its statutory authority.” 173 City of Arlington, 133 S. Ct. at 1868. As such, “there is no principled basis for carving out some arbitrary subset of such claims as ‘jurisdictional.’” 174 Id. at 1869.

Despite Justice Scalia’s objections, it is possible to define “jurisdictional questions” coherently. One clear and administrable definition is that they are disputes over an agency’s authority to regulate particular persons, activities, or geographical locations. For example:

  • The Army Corps of Engineers interprets “waters of the United States,” discharges into which require a Clean Water Act permit from the Corps, to include wetlands. 175 E.g., United States v. Riverside Bayview Homes, Inc., 474 U.S. 121, 123–25 (1985).
  • The Food and Drug Administration determines that nicotine in cigarettes is a drug, the cigarette a drug-delivery device, and that the agency can regulate both under the Federal Food, Drug, and Cosmetic Act. 176 E.g., FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 125 (2000).
  • The EPA determines that greenhouse gases are not “air pollutants” and therefore it cannot regulate them under the Clean Air Act. 177 E.g., Massachusetts v. EPA, 549 U.S. 497, 511–13 (2007).
  • The Federal Trade Commission (FTC) determines that lawyers and law firms are “financial institutions” subject to the Gramm-Leach-Bliley Financial Modernization Act. 178 E.g., ABA v. FTC, 430 F.3d 457, 465–67 (D.C. Cir. 2005).
  • The Federal Energy Regulatory Commission determines that a promise to forgo consumption of electricity that would have been purchased in a retail-electricity market is not a “sale of electric energy” (and so subject to state rather than federal regulation) under the Federal Power Act. 179 E.g., Elec. Power Supply Ass’n v. FERC, 753 F.3d 216, 220–21 (D.C. Cir. 2014).
  • The Fish and Wildlife Service concludes that privately owned property can be an “area under Federal jurisdiction” for purposes of the Endangered Species Act. 180 E.g., N. Cal. River Watch v. Wilcox, 620 F.3d 1075, 1078 (9th Cir. 2010).
  • The Commodity Futures Trading Commission asserts authority to adjudicate counterclaims. 181 E.g., Commodity Futures Trading Comm’n v. Schor, 478 U.S. 833, 837 (1986).
  • The Occupational Safety and Health Administration interprets a statute mandating protection for all workers as applying only to the manufacturing sector. 182 E.g., Dole v. United Steelworkers of Am., 494 U.S. 26, 29 (1990).
  • The NLRB determines that newsboys are employees (rather than independent contractors) and therefore within the protection of the National Labor Relations Act. 183 E.g., NLRB v. Hearst Publ’ns, 322 U.S. 111, 114–15 (1944).
  • The Mine Safety and Health Administration determines that a vendor who delivered steel to a mine qualifies as an “operator” of the mine subject to the requirements of the Federal Mine Safety and Health Act. 184 E.g., N. Ill. Steel Supply Co. v. Sec’y of Labor, 294 F.3d 844, 846 (7th Cir. 2002).

Many pre-Arlington commentators seemed to operate from this definition of a jurisdictional question. 185 See, e.g., Quincy M. Crawford, Comment, Chevron Deference to Agency Interpretations that Delimit the Scope of the Agency’s Jurisdiction, 61 U. Chi. L. Rev. 957, 958 (1994) (“[C]ourts should defer to an agency’s interpretation of an ambiguous statute even if the interpretation delimits the scope of the agency’s jurisdiction.”); Gellhorn & Verkuil, supra note 167, at 993 (“[A] requirement of independent judicial evaluation of peripheral jurisdictional issues is consistent with the terms and rationale of Chevron.”); Cass R. Sunstein, Law and Administration After Chevron, 90 Colum. L. Rev. 2071, 2097–101 (1990) (“[C]ourts should probably refuse to defer to agency decisions with respect to issues of jurisdiction . . . .”). It is also, possibly, the view Justice Scalia took in Arlington. Of course, because his bottom line was that jurisdictional questions are indistinguishable from nonjurisdictional questions, he did not have to, and perhaps could not, explain exactly what a jurisdictional question is. Still, his premise seems to be that such questions are of the sort in the list above. Thus, he refers to a brief that “explains, helpfully, that ‘[j]urisdictional questions concern the who, what, where, and when of regulatory power: which subject matters may an agency regulate and under what conditions.’” 186 City of Arlington v. FCC, 133 S. Ct. 1863, 1870 (2013) (first emphasis added). In addition, noting that “[t]he U.S. Reports are shot through with applications of Chevron to agencies’ constructions of the scope of their own jurisdiction,” 187 Id. at 1872. he cites primarily (though not exclusively) cases about whether the agency had authority over particular persons or activities: cigarettes, wetlands, particular toxic pollutants, “pole attachments.” 188 Id. at 1870–73. And he relies in particular on NLRB v. City Disposal Systems, Inc., which explicitly stated that no “exception exists to the normal [deferential] standard of review” for “jurisdictional or legal question[s] concerning the coverage” of an Act. 189 465 U.S. 822, 830 n.7 (1984) (internal quotation marks omitted).

On this definition of a jurisdictional question, the majority was clearly correct that Chevron should apply with full force. Numerous precedents had invoked Chevron in such settings; 190 Each of the cases cited earlier is an example. See supra notes 175–184. Some upheld the agency. See, e.g., United States v. Riverside Bayview Homes, Inc., 474 U.S. 121, 131, 139 (1985) (upholding Corps of Engineers’ reading of “waters of the United States” to include wetlands under Chevron step two). Some did not. See, e.g., Massachusetts v. EPA, 549 U.S. 497, 529 n.26 (2007) (rejecting EPA’s reading of “air pollutant” to exclude greenhouse gases under Chevron step one). But none found Chevron inapplicable. none had said it was inapplicable; agency self-aggrandizement or abuse are not greater threats with regard to jurisdictional issues than with regard to substantive ones; and all the various rationales for Chevron apply equally to jurisdictional and substantive questions. Indeed, not only does Chevron apply to jurisdictional questions so understood, these are the paradigmatic Chevron questions because they are disputes over the boundaries of vague terms. 191 See supra notes 159–165 and accompanying text (describing step two as involving resolution of vague, as opposed to ambiguous, terms).

On the other hand, Chief Justice Roberts’s dissent is also correct. Joined by Justices Kennedy and Alito, the Chief Justice described the question as being whether court or agency determines whether Congress has delegated to the agency authority to act with the force of law. 192 City of Arlington, 133 S. Ct. at 1879–80 (Roberts, C.J., dissenting). That is the threshold requirement for Chevron to apply, and in his view it is necessarily a matter for judicial decision. He began with an admirably succinct summation:

A court should not defer to an agency until the court decides, on its own, that the agency is entitled to deference. Courts defer to an agency’s interpretation of law when and because Congress has conferred on the agency interpretive authority over the question at issue. An agency cannot exercise interpretive authority until it has it; the question whether an agency enjoys that authority must be decided by a court, without deference to the agency. 193 Id. at 1877 (Roberts, C.J., dissenting).

This is exactly right. But it rests on a different definition of a “jurisdictional question” than the majority’s. To have jurisdiction is to have authority to decide. The dissent focuses not on whether an agency has authority to regulate particular places or things, but rather whether it has authority to regulate at all, that is, whether it has authority to act with the force of law. Applied to Chevron, the question is not whether agencies should receive deference regarding their determinations of what persons or activities fall within their authority, but rather whether they should receive deference regarding their determinations of whether they have the authority to act with the force of law, period.

On this conception, applying Chevron to jurisdictional questions is deeply problematic. Chevron applies because the agency has been delegated authority; because the delegation is a condition precedent of deference, it is circular to defer to a determination that the delegation exists. As one influential article put it:

One reason courts should refuse to extend Chevron deference to agencies’ jurisdictional interpretations follows from the nature of Chevron itself: The existence of agency jurisdiction is a precondition of Chevron deference, and Chevron therefore has no bearing on how that threshold question should be resolved. Only after it is determined that Congress has conferred jurisdiction on an agency does Chevron come into play. 194 Sales & Adler, supra note 167, at 1533.

This is Chief Justice Roberts’s point. He rightly insists that courts, not agencies, must determine which questions are within an agency’s Chevron power to resolve with the force of law. 195 See Herz, Deference Running Riot, supra note 58, at 219 (“Chevron does not require a court to accept an agency’s view of the scope of its delegated authority, jurisdictional or substantive. By definition, Congress cannot have left this determination to the agency.”). Strauss, In Search of Skidmore, supra note 2, at 796 (“What are the boundaries of the agency’s authority, conferred by Congress, remains a judicial question, as it must be.”). In fact, Justice Scalia would seem to agree. He asserts essentially the same proposition at the end of his opinion:

The fox-in-the-henhouse syndrome is to be avoided not by establishing an arbitrary and undefinable category of agency decisionmaking that is accorded no deference, but by taking seriously, and applying rigorously, in all cases, statutory limits on agencies’ authority. Where Congress has established a clear line, the agency cannot go beyond it; and where Congress has established an ambiguous line, the agency can go no further than the ambiguity will fairly allow. 196 City of Arlington, 133 S. Ct. at 1874.

The ships-passing-in-the-night aspect of the opinions is highlighted by looking at subsequent cases that rely on City of Arlington. For example, Pharmaceutical Research & Manufacturers of America v. FTC was a challenge to an FTC regulation issued under the Hart-Scott-Rodino Antitrust Improvements Act that required pharmaceutical companies to report certain transfers of exclusive patent rights. 197 44 F. Supp. 3d 95 (D.D.C. 2014). The challengers argued that the statute did not permit the FTC to impose these obligations on only a single industry; it read the statute to authorize only rules of general application that would cover all transfers of exclusive patents. The district court applied Chevron, laying out the two-step test. It cited City of Arlington for the propositions that Chevron applies to jurisdictional determinations and that there is no difference between exceeding the scope of authority and exceeding authorized application of authority. It then concluded the “FTC is Entitled to Deference on Scope of Statutory Authority,” i.e., on the question whether it could issue a rule limited to a particular industry. 198 Id. at 114–15. A thorough review of text, history, and purpose indicated no particular congressional view on industry-specific regulations; thus, since the statute did not address the “precise question,” it was on to step two, which, of course, the regulation survived. 199 Id. at 124–25 (“[O]nce the Court has determined that Congress has not directly addressed the issue, the agency is entitled to Chevron deference of its interpretation of the scope of its authority. Where, as here, the statute is silent on the issue, the agency’s interpretation of its authority is due deference.” (citation omitted)).

So what happened here in City of Arlington terms? The court thought it was doing what the majority in City of Arlington told it to: Apply Chevron to a question about the scope of agency authority. That question was whether the FTC could write a regulation about a single industry even though the relevant statute applied to mergers and the transfer of assets generally. One could describe that as a jurisdictional issue, though it is a slightly odd one, being about whether the greater includes the lesser rather than which industries are regulable at all. So the case could be seen as a perfect illustration of Justice Scalia’s reasoning and application of his conclusion: It is hard to distinguish jurisdictional from non-jurisdictional questions, the key issue is whether the agency has violated the statute, and where the statute is ambiguous, the court should accept the agency’s reading as long as it is permissible.

But the Pharmaceutical Research decision was also completely consis­tent with Chief Justice Roberts’s dissent in City of Arlington. The court did not “defer to [the] agency on whether Congress has granted the agency interpretive authority over the statutory ambiguity at issue.” 200 City of Arlington, 133 S. Ct. at 1879–80 (Roberts, C.J., dissenting). It decided entirely on its own whether this was a case in which deference was mandated. It may not have done a great job at that (it essentially skipped step zero and only inquired whether the statute was ambiguous), but it was only going to defer if it found the statute ambiguous and ultimately made an independent determination that it was. Of course, the FTC had said that the statute was ambiguous, 201 Pharm. Research & Mfrs. of Am., 44 F. Supp. 3d at 115 (“The FTC disputes that Congress has directly addressed the issue in the HSR Act and instead contends, under Chevron Step Two, that the agency’s interpretation of the statute is entitled to deference.”). but the court did not defer on that question. 202 For another post–City of Arlington decision that is consistent with both the majority and the dissent, see Pharm. Research & Mfrs. of Am. v. Dep’t of Health & Human Servs., 43 F. Supp. 3d 28 (D.D.C. 2014) (rejecting agency regulation under step one, with no deference in sight, because Congress had not given agency authority to regulate particular activities reached by rule).

This discussion can be restated in terms of Chevron “step zero.” 203 See, e.g., Cass R. Sunstein, Chevron Step Zero, 92 Va. L. Rev. 187, 191 (2006) (defining “Chevron Step Zero”). That term, which has never been invoked by the Supreme Court itself, refers to the threshold inquiry as to “whether the Chevron framework applies at all.” 204 Id. The leading case is United States v. Mead Corp., which held that Chevron only applies if Congress has delegated authority to write rules with the force of law and the agency has used that authority. 205 United States v. Mead Corp., 533 U.S. 218, 226–27 (2001). Mead coexists somewhat uncomfortably with decisions that interpret statutory ambiguity itself to be a delegation to the agency; the Court’s “cases find in unambig­uous language a clear sign that Congress did not delegate gap-filling authority to an agency; and they find in ambiguous language at least a presumptive indication that Congress did delegate that gap-filling authority.” 206 United States v. Home Concrete & Supply, LLC, 132 S. Ct. 1836, 1843 (2012). Courts never defer in step zero. Thus, another way of explaining the difference between Justice Scalia and the Chief Justice in City of Arlington is that the former understood the case to be about how Chevron applies and the Chief Justice understood it to be about whether Chevron applies. Justice Scalia is completely correct that in deferring (or not) under Chevron, jurisdictional questions should not be treated any differently from other questions of statutory meaning. Chief Justice Roberts is completely correct that courts should not yield to agency under­standings of whether Chevron kicks in or whether a statute is ambiguous.

Finally, to return to “space” and “weight”: In a Chevron step-two case, the court’s role under the statute is to ensure that the agency has stayed within its space, that its interpretation is not unreasonable. It is hard to understand how a court could grant Chevron deference to that determi­nation even if it wanted to. Both Justice Scalia and Chief Justice Roberts in fact endorse that understanding in City of Arlington. Where they go astray, on the Straussian account, is the failure to accord Skidmore deference to questions about statutory meaning. 207 See Strauss, In Search of Skidmore, supra note 2, at 797 (characterizing Skidmore deference as “traditional tool[] of statutory interpretation” (quoting Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843 n.9 (1984))); supra section II.B (arguing for application of Skidmore deference within Chevron step one).

B. “Jurisdictional” Questions and the APA

Ignored in all this is the text of the APA itself. The Act does actually refer to jurisdictional questions. Not surprisingly, it lumps them together with other sorts of statutory questions. Specifically, section 706 provides that a reviewing court “shall . . . hold unlawful and set aside agency action . . . found to be . . . in excess of statutory jurisdiction, authority or limitations, or short of statutory right.” 208 5 U.S.C. § 706(2)(C) (2012) (emphasis added). The APA has a tendency toward the thesauric; often a string of nouns appears where a more parsi­monious drafter might have used just one or two. 209 See Michael Herz, “Data, Views, or Arguments”: A Rumination, 22 Wm. & Mary Bill Rts. J. 351, 356 (2013) (“The APA contains a number of what we might call ‘thesauric’ provisions.”). Perhaps the most striking instance involves licensing. The APA defines “licensing” as an “agency process respecting the grant, renewal, denial, revocation, sus­pension, annulment, withdrawal, limitation, amendment, modification, or conditioning of a license,” 5 U.S.C. § 551(9), and then defines “license” as “the whole or part of an agency permit, certificate, approval, registration, charter, membership, statutory exemp­tion or other form of permission.” Id. § 551(8). That means that “licensing” could in theory be any of eighty-eight different undertakings—the eleven kinds of “licensing” multiplied by the eight kinds of “licenses.” But this reflects care and precision, not sloppiness. The no-surplusage canon admonishes us to take each of those words seriously, as indicating something distinct. And indeed, many APA provisions set out a string of not-quite-synonyms, the differences between which should be taken seriously. Especially with phrases as familiar and well-worn as the key provisions of the APA, it can be helpful to read the text with fresh eyes.

The scope-of-review provisions of the APA are one prominent example of cover-the-waterfront drafting. Not just “arbitrary,” but “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 210 5 U.S.C. § 706(2)(A) (“The reviewing court shall . . . hold unlawful and set aside agency action, findings, and conclusions found to be . . . arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law . . . .”). Not just “unconstitutional,” but “contrary to constitutional right, power, privilege, or immunity.” 211 Id. § 706(2)(B). And not just “contrary to a relevant statute,” but “in excess of statutory jurisdiction, authority, or limitations, or short of statutory right.” 212 Id. § 706(2)(C). If we take that wording seriously, it means a reviewing court has the same authority (or lack thereof) to determine whether an agency has exceeded its statutory jurisdiction that it has with regard to reviewing other statutory limitations or authorizations. That is a textual argument that the great textualist ignored.

On the other hand, the text of this provision undercuts Justice Scalia’s opinion as well. By referring separately to “jurisdiction” and “authority,” 213 Id. (“The reviewing court shall . . . hold unlawful and set aside agency action, findings, and conclusions found to be . . . in excess of statutory jurisdiction, authority, or limitations, or short of statutory right . . . .”). the APA indicates that they are two distinct things. Yet the basis of Justice Scalia’s entire opinion is that there is no distinction between them. In a case arising under the APA, his job was to at least try to figure out how these two separately identified concepts differ, even if the question presented treated them as synonyms. Defining “jurisdiction” as being about the persons, places, and things that an agency can regulate, and “authority” as being about the manner and extent of permissible agency regulation of those things over which it has juris­diction, serves to give each word independent meaning.

Section 706(2)(C) is a Goldilocks provision. Agencies cannot go too far, but they also cannot fail to go far enough. Not too cold and not too hot; neither in excess of maxima nor shy of minima. In a passing comment, Professor Strauss has suggested this formulation may be a statutory peg on which to hang the idea of Chevron space. The court must ensure that the agency’s reading does not exceed statutory boundaries, that it is inconsistent neither with what the statute must mean (“short of statutory right”) nor what it can mean (“in excess of statutory jurisdiction”). 214 See Strauss, Space/Weight, supra note 2, at 1161 (“‘Excess of statutory jurisdiction . . . or short of statutory right’ readily suggests some space between, space within which the agency may exercise policymaking discretion.” (quoting 5 U.S.C. § 706(2)(A))). It would be delightful if the APA really did articulate the concept of Chevron space, but this reading imposes more on the text than it will bear. On the most natural reading of the text, the “in excess of” and “short of” language does not define boundaries of a “geographical” sort. Rather it speaks to the quali­tatively different kinds of limitations a statute can impose: Some provisions impose restrictions, some impose obligations. 215 The legislative history also undercuts this reading. “Short of statutory right” seems to have been intended to reach the situation in which an agency has awarded a prevailing party less than the full relief required by statute; the reports are wholly focused on this question of remedy for an individual party rather than the more abstract question of whether a policy decision exceeds statutory boundaries. See H.R. Rep. No. 79-1980, at 278 (1946) (“‘Short of statutory right’ means that agencies are not authorized to give partial relief where a party demonstrates his right to the whole. Authorized relief must be granted by an agency to the full extent that entitlement is shown.”); Staff of S. Comm. on the Judiciary, 79th Cong., Administrative Procedure 40 (Comm. Print 1945) (“[I]f Congress has prescribed a measure of right or relief, on principle the citizen is entitled to the full measure and there should be no arbitrary power in administrative agencies to grant less in specific cases.”); S. Rep. No. 79-752, at 214 (1945) (“‘Short of statutory right’ means that agencies are not authorized to give partial relief where a party demonstrates his right to the whole.”). That does not undercut the Chevron space conceptualization; it just means that the drafters of the APA did not write it into the statute.

Conclusion

The argument here is for a reading of Chevron that is both weaker and stronger than that often proposed. On the one hand, courts retain an essential and meaningful role in determining the boundaries of Chevron space; they have real work to do. On the other hand, in doing that work, the views of the agency can never be ignored. Accordingly, there is no agency-free space within which courts interpret statutes.

The familiar and oft-told narrative is that Congress retreats to vague generalities in order to achieve agreement and, in an effort to take credit and avoid blame, leaves the hard issues to others to decide. Chevron’s great strength is that it recognizes this narrative is sometimes accurate and admonishes that when it is judges should yield to agencies because of their comparative advantage in policymaking. But this narrative is not always accurate. Sometimes Congress does make a deci­sion. Chevron’s great risk is that it invites judges and agencies to ignore what Congress has decided, discovering uncertainty and discretion where they do not exist. 216 An example—one that is understandable and prompted by “bad facts” but an example nonetheless—is Entergy Corp. v. Riverkeeper, Inc., 556 U.S. 208 (2009), which held that the EPA could determine the “best technology available for minimizing adverse environmental impact” by balancing the costs and benefits of different alternatives rather than just determining which of the feasible alternatives was most effective at reducing environmental harms. Id. at 219–20. Some scholars would essentially write courts out of the picture in statutory cases, concluding that agencies always have a comparative advantage. 217 See, e.g., Adrian Vermeule, Judging Under Uncertainty: An Institutional Theory of Legal Interpretation 205–29 (2006) (arguing for super-strong version of Chevron deference); Lisa Schultz Bressman, Chevron’s Mistake, 58 Duke L.J. 549, 566 (2009) (arguing statutes virtually never have specific meaning and therefore virtually always include broad delegations to implementing agencies). Peter Strauss wisely warns against such an abandonment of the judicial role, while still granting due weight to agency interpretation and, within the congressionally established and judicially policed Chevron space, respecting agency construction. So understood, Chevron is not a revolutionary shift of authority from the judiciary to the executive. That Chevron is dead. Rather, Chevron is an appropriate allocation of decisionmaking responsibility among the three branches, relying on the judiciary to enforce congressional decisions, but protecting agency authority and discretion where Congress has left the decision to the executive. Long may it reign.