George Mason
Law Review

Three Wrong Turns in Agency Adjudication

Aaron L. Nielson
Volume 28
Issue 2


When considering the origins of modern administrative law, many scholars turn their attention to the 1930s.1Compare Gillian E. Metzger, Foreword: 1930s Redux: The Administrative State Under Siege, 131 Harv. L. Rev. 1 (2017) (largely defending the 1930s approach), with Steven G. Calabresi & Gary Lawson, The Depravity of the 1930s and the Modern Administrative State, 94 Notre Dame L. Rev. 821 (2018) (acknowledging the importance of the 1930s but challenging the approach). Other scholars, of course, begin the analysis much earlier. See, e.g., Sophia Z. Lee, Our Administered Constitution: Administrative Constitutionalism from the Founding to the Present, 167 U. Pa. L. Rev. 1699 (2019); Jerry L. Mashaw, Foreward: The American Model of Federal Administrative Law: Remembering the First One Hundred Years, 78 Geo. Wash. L. Rev. 975 (2010). Few disagree, however, that the 1930s were important. The 1940s, however, may be even more important.2See, e.g., Aaron L. Nielson, Confessions of an “Anti-Administrativist, 131 Harv. L. Rev. F. 1, 1 (2017) (“The hard reality is that 1930s administrative law is not a good fit for today. In fact, it was not a good fit for the 1930s — which is why Congress reformed it in the 1940s.”). Many of the key aspects of today’s administrative state originated during that decade. After all, in 1946 Congress enacted the Administrative Procedure Act (“APA”)—“the bill of rights for the new regulatory state.”3George B. Shepherd, Fierce Compromise: The Administrative Procedure Act Emerges from New Deal Politics, 90 Nw. U. L. Rev. 1557, 1558 (1996). Although how the APA is applied today may not always track the 1946 version,4See, e.g., Jeffrey A. Pojanowski, Neoclassical Administrative Law, 133 Harv. L. Rev. 853, 882–912 (2020) (comparing today’s administrative law with administrative law from the time when Congress enacted the APA). by itself the APA’s enactment makes the 1940s the preeminent decade for understanding administrative law.

The APA, however, is not the only significant aspect of today’s administrative law that comes from the 1940s. The Supreme Court’s hands-off approach to nondelegation, for instance, is also in large part a product of the 1940s.5See, e.g., Mistretta v. United States, 488 U.S. 361, 373–74 (1989) (citing five cases from the 1940s that upheld delegations and two 1935 decisions that invalidated them). The Supreme Court, of course, also took a hands-off approach to nondelegation before the 1930s.. Delegation is a defining characteristic of modern government, especially when combined with a broad conception of interstate commerce—which, notably, also largely comes from that same period.6See, e.g., Michael J. Klarman, Constitutional Fetishism and the Clinton Impeachment Debate, 85 Va. L. Rev. 631, 649 (1999) (discussing “revolutionary commerce clause decisions of the early 1940s,” including Wickard v. Filburn, 317 U.S. 111 (1942)). All the while, the bar on post hoc explanations from agencies—a mainstay of administrative law7See, e.g., Kevin M. Stack, The Constitutional Foundations of Chenery, 116 Yale L.J. 952, 955 (2007) (“[One] ‘bedrock’ principle . . . is that a court may uphold an agency’s action only for the reasons the agency expressly relied upon when it acted.”). and arguably a precursor to hard look review8See, e.g., Merrick B. Garland, Deregulation and Judicial Review, 98 Harv. L. Rev. 505, 526 (1985) (tracing “[t]he origins of the hard look doctrine” to the 1940s). and the “ordinary remand rule”9See, e.g., Christopher J. Walker, The Ordinary Remand Rule and the Judicial Toolbox for Agency Dialogue, 82 Geo. Wash. L. Rev. 1553, 1563 (2014) (tracing the origin of the rule that a “court must remand [a decision premised on error] to the agency for reconsideration” to the same period).—began in 1943.10SEC v. Chenery Corp. (Chenery I), 318 U.S. 80, 95 (1943).

The 1940s are particularly important for agency adjudication, and not in positive ways. To be clear, agency adjudication is not inherently problematic. Although it can be abused, agency adjudication—the application of law to facts by the executive branch—is a longstanding feature of administrative law.11See, e.g., City of Arlington v. FCC, 569 U.S. 290, 304–05 n.4 (2013) (“Agencies . . . conduct adjudications (‘This rancher’s grazing permit is revoked for violation of the conditions’) and have done so since the beginning of the Republic.”). Indeed, agency adjudication has been with us from the beginning—“the first Congress provided compensation to disabled veterans” through an executive-controlled process.12See Harold J. Krent, Presidential Control of Adjudication Within the Executive Branch, 65 Case W. Rsrv. L. Rev. 1083, 1089 (2015). Not a bad pedigree.13See, e.g., Harmelin v. Michigan, 501 U.S. 957, 980 (1991) (Scalia, J.) (“The actions of the First Congress . . . are of course persuasive evidence of what the Constitution means . . . .”).

Agency adjudication, however, is also controversial. It is debatable, for instance, whether agencies should be able to adjudicate private, rather than public, rights.14See, e.g., Kent Barnett, Due Process for Article III—Rethinking Murray’s Lessee, 26 Geo. Mason L. Rev. 677 (2019) (discussing the distinction). Notably, Justices William Brennan and Thurgood Marshall believed that sweeping private rights into agency adjudication is a step towards “tyranny,” CFTC v. Schor, 478 U.S. 833, 859–60 (1986) (Brennan, J., dissenting), a view that Chief Justice John Roberts and Justice Clarence Thomas have echoed in recent years. E.g., Wellness Int’l Network, Ltd. v. Sharif, 135 S. Ct. 1932, 1960 (2015) (Roberts, C.J., dissenting) (rejecting that the law “may assign away our hard-won constitutional birthright”); id. at 1963 (Thomas, J., dissenting) (“Disposition of private rights to life, liberty, and property falls within the core of the judicial power, whereas disposition of public rights does not.”). Agency adjudication also carries with it accusations of bias15See, e.g., Daniel B. Listwa & Lydia K. Fuller, Constraint Through Independence, 129 Yale L.J. 548, 556 (2019) (noting accusations of bias); see also Giles D. Beal, IV, Judge, Jury and Executioner: SEC Administrative Law Judges Post-Dodd Frank, 20 N.C. Banking Inst. 413, 417 (2016) (offering a critical assessment of agency adjudication). But see David Zaring, Enforcement Discretion at the SEC, 94 Tex. L. Rev. 1155 (2016) (defending agency adjudication against accusations of bias). and, with notable success, Appointments Clause challenges.16See, e.g., Lucia v. SEC, 138 S. Ct. 2044, 2049 (2018) (holding that “administrative law judges . . . of the Securities and Exchange Commission” are officers under the Appointments Clause); Linda D. Jellum, “You’re Fired!” Why the ALJ Multi-Track Dual Removal Provisions Violate the Constitution and Possible Fixes, 26 Geo. Mason L. Rev. 705, 750 (2019) (“[M]ultiple for-cause removal provisions protecting ALJs from at-will removal are unconstitutional . . . .”). These sorts of criticisms have only escalated in recent years.17See, e.g., Christopher J. Walker, Constitutional Tensions in Agency Adjudication, 104 Iowa L. Rev. 2679, 2680 (2019); see also Jack M. Beermann, The Never-Ending Assault on the Administrative State, 93 Notre Dame L. Rev. 1599, 1637 (2018).

This short Article does not address all that is right and wrong with agency adjudication. Rather, it addresses three particular wrong turns in agency adjudication, each occurring between 1945 and 1947. First, in 1945, the Supreme Court decided Bowles v. Seminole Rock & Sand Co.,18325 U.S. 410 (1945). which, because of sloppy language, created decades of confusion about when and how courts should defer to an agency’s interpretation of its own regulations.19See id. at 414 (setting forth a standard of deference for agency interpretations of their own regulations). Second, in 1946, Congress enacted the APA but mishandled adjudication in at least two key respects: the APA says little about informal adjudication and does not clearly define the trigger for formal adjudication.20See 1 Kristin E. Hickman & Richard J. Pierce, Jr., Administrative Law Treatise §§ 7.1–7.15 (6th ed. 2019) (explaining the APA’s adjudication procedures). And third, in 1947, the Court decided SEC v. Chenery Corp.,21332 U.S. 194 (1947). more commonly known as Chenery II. In Chenery II, the Court took a sound principle—that rulemaking is not always required when an agency announces new policy—too far by downplaying fair-notice concerns.22See id. at 196. The combined consequences of these three wrong turns are confusion (as regulators, regulated parties, and the public alike sometimes do not know what the law means), worse policy (as agencies can make important decisions without the benefits of procedural rigor), and abuse of authority (as agencies can act with unfair retroactivity while leveraging the threat of their power).

Reformers have begun to address these consequences. In 2019, two of the most significant events in administrative law—the Supreme Court’s decision in Kisor v. Wilkie23139 S. Ct. 2400 (2019). and the Trump Administration’s executive orders on agency enforcement and guidance documents—are, in key respects, responses to these wrong turns from the 1940s. Kisor imposed significant limits on the “caricature” of deference that Seminole Rock had become24Id. at 2415.—limits that would not have been necessary if the Court in 1945 had written a better opinion in the first instance. Kisor, notably, also makes it much more difficult for agencies to use adjudication to retroactively create policy,25See id. at 2417–18 (holding that deference is unwarranted if it “creates ‘unfair surprise’ to regulated parties” or “impose[s] retroactive liability on parties for longstanding conduct that the agency had never before addressed” (quoting Long Island Care at Home, Ltd. v. Coke, 551 U.S. 158, 170 (2007))). an effort bolstered by Executive Orders (“EO”) 13891 and 13892.26See Exec. Order No. 13,891, 84 Fed. Reg. 55,235 (Oct. 9, 2019); Exec. Order No. 13,892, 84 Fed. Reg. 55,239 (Oct. 9, 2019). As noted later in this Article, these executive orders have been revoked. See infra note 149. Because they emphasize fair-notice concerns, these reforms also have the potential to tame some of the problematic aspects of Chenery II. By creating some procedures for informal adjudication above and beyond what the APA requires, EO 13982 also partially filled the hole in the APA where procedures for informal rulemaking should go.27See 84 Fed. Reg. at 55,241 (explaining that “before an agency takes any action with respect to a particular person that has legal consequence . . . the agency must afford that person an opportunity to be heard, in person or in writing, regarding the agency’s proposed legal and factual determinations” and then “must respond in writing and articulate the basis for its action”). Finally, by making prospective rulemaking relatively more attractive, these reforms correspondingly reduce the harm caused by inadequate adjudicative procedures.28See, e.g., Aaron L. Nielson, Beyond Seminole Rock, 105 Geo. L.J. 943, 990 (2017).

The reforms of 2019, however, are not cure-alls. And they may have unintended consequences of their own. Studying the wrong turns of 1945 to 1947, however, should help history avoid repeating itself. Accordingly, this Article ends by considering what lessons we can learn from what went wrong.

I.     Three Wrong Turns

Today’s agency adjudication is in many ways a product of the 1940s. This Article examines: (1) Seminole Rock’s sloppiness in 1945; (2) the APA’s incompleteness in 1946; and (3) Chenery II’s aggressiveness in 1947. Each was a serious wrong turn.

A.     Seminole Rock’s Sloppiness

Seminole Rock is a foundational deference case.29Even apart from Seminole Rock, the 1940s was a significant decade for deference. See, e.g., Skidmore v. Swift & Co., 323 U.S. 134 (1944); NLRB v. Hearst Publ’ns, Inc., 322 U.S. 111 (1944). Unfortunately, sloppy language in Seminole Rock paved the way for Auer deference—a much more deferential standard than what the 1945 Court presumably intended.30See Auer v. Robbins, 519 U.S. 452 (1997); Jeffrey A. Pojanowski, Revisiting Seminole Rock, 16 Geo. J.L. & Pub. Pol’y 87, 87–88 (2018) (explaining the relationship between Auerand Seminole Rock); see also Sanne H. Knudsen & Amy J. Wildermuth, Unearthing the Lost History of Seminole Rock, 65 Emory L.J. 47, 77 (2015) (explaining how Auer expands deference beyond what the Seminole Rock Court apparently intended).

The story of Seminole Rock has been told many times before. The case arose out of a price control scheme administered by the Office of Price Administration (“OPA”), a wartime agency that was treated as an outlier even at the time.31See, e.g., Knudsen & Wildermuth, supra note 30, at 55 (“Scholars of the price control era distinguished between OPA and its peer agencies.”) (discussing Helen B. Norem, The “Official Interpretation” of Administrative Regulations, 32 Iowa L. Rev. 697 (1947)). This outlier status continued after Seminole Rock was decided. See id. at 65 (“OPA was almost exclusively the agency asking for and receiving deference under Seminole Rock in the decade following the decision.”). The particular dispute in Seminole Rock concerned the meaning of a regulation that established maximum prices based on going rates in March 1942. The regulation, however, contained an ambiguity—what to do when the sale occurred before March 1942 but the delivery itself occurred in March 1942?32See Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 415 (1945). In an adjudicative order, the OPA concluded that under the agency’s regulation, the delivery date controlled rather than the sale date.33See id. On appeal, the Supreme Court sided with the OPA, with Justice Frank Murphy famously explaining that “the ultimate criterion is the administrative interpretation, which becomes of controlling weight unless it is plainly erroneous or inconsistent with the regulation.”34Id. at 414.

On its face, this test is a muddle. The word “inconsistent” may suggest de novo review,35See, e.g., Webster’s Twentieth-Century Dictionary of The English Language (1937) (defining “inconsistent” as, inter alia, “[i]ncompatible; incongruous; not . . . consistent; contrary”). but the words “plainly erroneous” do not. The problem with the test, however, runs deeper. The Court’s actual analysis in Seminole Rock was much more searching than what a “plainly erroneous” formulation read in isolation suggests. As Professors Sanne Knudsen and Amy Wildermuth have explained, the Court referred to OPA’s interpretation merely “to confirm what [the Court itself] had concluded” independently.36Knudsen & Wildermuth, supra note 30, at 61. Similarly, Seminole Rock arose in an unusual context: the OPA issued its interpretation of the regulation concurrently with the regulation itself, in a bulletin that was signed by OPA’s administrator.37Id. Thus, one could liken the OPA’s interpretation to a “statement of basis and purpose” of the sort that now would be part of a final rule.38Id. These facts, moreover, were no secret to the parties or the Court. The OPA’s briefing emphasized these limitations, arguing that “the Court should give ‘weight’ to the agency’s ‘settledadministrative construction’ and its ‘consistently and repeatedly reaffirmed administrative interpretation,’ which was embodied in a bulletin issued ‘[c]oncurrently with the issuance of the’ regulation” and which was not “a position taken for the first time in this lawsuit.”39Aditya Bamzai, Henry Hart’s Brief, Frank Murphy’s Draft, and the Seminole Rock Opinion, Yale J. on Reg.: Notice & Comment (Sept. 12, 2016), (emphasis added).

Against that backdrop, the actual outcome in Seminole Rock was unsurprising. When interpreting a legal instrument, courts often look to contemporaneous explanations.40See, e.g., United States v. Bd. of Comm’rs, 435 U.S. 110, 131 (1978) (“This contemporaneous administrative construction of the Act is persuasive evidence of the original understanding . . . .”); cf. Myers v. United States, 272 U.S. 52, 175 (1926) (focusing on the “contemporaneous legislative exposition of the Constitution”); Steven G. Calabresi & Saikrishna B. Prakash, The President’s Power to Execute the Laws, 104 Yale L.J. 541, 553 (1994) (similar). Caselaw and commentary of the time also support a narrow understanding of Seminole Rock. Aside from a 1946 case ruling against the government,41See M. Kraus & Bros., Inc. v. United States, 327 U.S. 614, 621–22 (1946) (explaining that a “strict rule of construction” applies to a regulation with criminal implications). the Court did not address Seminole Rock again until 1965.42See Udall v. Tallman, 380 U.S. 1, 16–17 (1965). And within the lower courts, “early cases connected Seminole Rock more closely with the deference framework for an agency’s statutory interpretations under Skidmore [v. Swift & Co.]”43Knudsen & Wildermuth, supra note 30, at 94–95.—a weak form of deference announced the year before Seminole Rock that focuses on the agency’s persuasiveness.44See Skidmore v. Swift & Co., 323 U.S. 134, 140 (1944) (“We consider that the rulings, interpretations and opinions of the Administrator under this Act, while not controlling upon the courts by reason of their authority, do constitute a body of experience and informed judgment to which courts and litigants may properly resort for guidance. The weight of such a 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.”). Indeed, it is debatable whether Skidmore “deference” is actually deference. Compare United States v. Mead Corp., 533 U.S. 218, 250 (2001) (Scalia, J., dissenting) (dismissing Skidmore as “an empty truism”), and Kisor v. Wilkie, 139 S. Ct. 2400, 2424 (2019) (Roberts, C.J., concurring in part) (“[T]here is a difference between holding that a court ought to be persuaded by an agency’s interpretation and holding that it should defer to that interpretation . . . .”), with Kristin E. Hickman & Matthew D. Krueger, In Search of the Modern Skidmore Standard, 107 Colum. L. Rev. 1235, 1271 (2007) (concluding that “Skidmoredeference” is “highly deferential to administrative interpretations as applied”). Even cases “that did not invoke Skidmore also undertook rigorous review before ruling on the validity of the agency’s interpretation—review that is far closer to the less-deferential Skidmore standard.”45Pojanowski, supra note 30, at 96. In other words, lower courts did not recognize Seminole Rock as a game-changing decision. Nor did academic commentators. Professor Kenneth Culp Davis dismissed the language in Seminole Rock “regarding controlling weight for the interpretation ‘[a]s hardly more than dictum.’”46Knudsen & Wildermuth, supra note 30, at 64 (quoting Kenneth Culp Davis, Scope of Review of Federal Administrative Action, 50 Colum. L. Rev. 559, 597 (1950)).

When it was decided, accordingly, Seminole Rock was understood as standing for the unremarkable proposition that a court should respect an agency’s formal and contemporaneous interpretation of truly ambiguous language. Seminole Rockwas not understood to stand for the much bolder notion that courts should defer even if those conditions were absent. But over time, Seminole Rock turned into something very different: Auer. In its 1996 decision in Auer v. Robbins,47519 U.S. 452 (1997). the Supreme Court purported to merely apply Seminole Rock but actually expanded the doctrine to include “informal interpretations announced long after a regulation’s promulgation.”48Aaron L. Nielson, Cf. Auer v. Robbins, 21 Tex. Rev. L. & Pol. 303, 304 (2017). The key contextual limits on when and how Seminole Rock deference applied thus largely fell away.49See id.

How did this happen? It is impossible to say for sure, of course, but it appears that sloppiness played a role. The Court’s opinion in Seminole Rock—a short, quick decision with almost a breezy tone50The case was argued in late April 1945 and decided in early June 1945. It is less than ten pages in the U.S. Reports. All but one of the justices joined the majority opinion without comment. The Court’s decision stated without elaboration that Justice Roberts believed that “the judgment should be affirmed for the reasons given in the opinion of the Circuit Court of Appeals.” Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 410–19 (1945) (citation omitted).—should have been better. The Court’s articulation of the legal rule did not match what the rule meant in context. When the Court dusted the decision off years later in Auer, it repeated the broad formulation but omitted the context in which that formulation was offered.

B.     The APA’s Incompleteness

The next wrong turn occurred in 1946—this time it was Congress’s doing. That year, Congress, acting on concerns about potential abuse,51See, e.g., Wong Yang Sung v. McGrath, 339 U.S. 33, 36–37 (1950) (explaining that the APA followed the “[m]ultiplication of federal administrative agencies and expansion of their functions to include adjudications” and with it “[t]he conviction . . . particularly within the legal profession, that this power was not sufficiently safeguarded”); Aaron L. Nielson, Visualizing Change in Administrative Law, 49 Ga. L. Rev. 757, 773 (2015) (“In short, neither side obtained a complete victory; both reluctantly surrendered important positions.”). enacted the APA. To Congress’s credit, the APA contains entire sections setting forth the procedures for agency adjudication.52See, e.g., 5 U.S.C. § 554(c)(1)–(2) (setting forth requirements for formal adjudication, including a “hearing and decision on notice and in accordance with sections 556 and 557 of this title”); id. §§ 556–57 (detailing those requirements). Unfortunately, the APA is also incomplete in at least two important respects. First, Congress’s attention to procedure only extended so far. Congress addressed formal adjudication but said essentially nothing about informal adjudication.53See Kristin E. Hickman & Aaron L. Nielson, Narrowing Chevron’s Domain, 70 Duke L.J. (forthcoming 2021) (manuscript at 11–12), . In an important forthcoming article, Professor Emily Bremer explains that as originally intended, informal and formal adjudication would proceed In stages. See Emily S. Bremer, The Rediscovered Stages of Agency Adjudication (Feb. 4, 2021) (unpublished manuscript), Assuming that is what Congress intended, however, Congress still could and should have written the statute more clearly. Second, Congress inadequately demarcated the triggers for formal adjudication (which had lots of procedures) and informal adjudication (which essentially had none). The result was that agencies could often avoid the APA’s more strenuous formal procedures.

Under the APA, agency action with legal effect is generally divided into four categories: (1) formal rulemaking; (2) informal rulemaking; (3) formal adjudication; and (4) informal adjudication.54See Hickman & Nielson, supra note 53 (manuscript at 9). Each box carries with it certain procedural obligations. The box for formal adjudication is extensive; it includes provisions for in-person hearings with presiding officials, oaths, subpoenas, burdens of proof, cross-examination, and proposed findings of fact and conclusions of law.55See id. (manuscript at 11–12); see also 5 U.S.C. §§ 554, 556–57. These procedures—which overlap with the procedures for formal rulemaking56Sections 556 and 557 of the APA cover both formal rulemaking and formal adjudication. See 5 U.S.C. §§ 556–57; see also Hickman & Nielson, supra note 53 (manuscript at 11–12).—apply “in every case of adjudication required by statute to be determined on the record after opportunity for an agency hearing.”575 U.S.C. § 554(a). The trigger for formal rulemaking is similar, applying “[w]hen rules are required by statute to be made on the record after opportunity for an agency hearing.”58Id. § 553(c).

The APA, however, has a hole. Unlike formal rulemaking and formal adjudication (with their litany of procedures) and informal rulemaking (with its own procedural requirements),59See id. § 553(b), (d). there are essentially no procedures for informal adjudication—i.e., adjudication not “required by statute to be determined on the record after opportunity for an agency hearing.”60Id. § 554(a); see also Jeffrey S. Lubbers, Paul Verkuil’s Projects for the Administrative Conference of the U.S. 1974-1992, 32 Cardozo L. Rev. 2421, 2432 (2011) (“[I]nformal adjudication was essentially ignored by the APA.”). Informal adjudication, moreover, is not a small box. To the contrary, it may be the largest box of the four, as “informal adjudications are essentially the default form of agency decisionmaking.”612 The Fundamentals of Administrative Law for Natural Resources Practice 1, 1–4 (2019); see also Paul R. Verkuil, A Study of Informal Adjudication Procedures, 43 U. Chi. L. Rev. 739, 744 (1976) (“[A]dministrative decision making labeled here as informal adjudication is largely unaddressed procedurally by the APA, even though those decisions have long been considered ‘truly the life blood of the administrative process.’” (footnote omitted)). The APA’s definition of adjudication—the “process for the formulation of an order”—covers all agency actions that reach a binding decision other than rulemaking, be it “affirmative, negative, injunctive, or declaratory in form.”62See 5 U.S.C. § 551(4)–(7). That capacious language includes obvious agency decisions such as determining whether a company has violated the law.63See Hickman & Nielson, supranote 53 (manuscript at 9–10). But it also covers agency decisions that may not come immediately to mind such as where to place a road.64See Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 417 (1971).

This hole is a significant problem—both as a policy matter (procedural rigor may help create better, more legitimate outcomes65See, e.g., Jessica Mantel, Procedural Safeguards for Agency Guidance: A Source of Legitimacy for the Administrative State, 61 Admin. L. Rev. 343, 346–47 (2009).) and doctrinally (it can be challenging to apply the APA without procedural requirements for informal adjudication).66See Hickman & Pierce, supra note 20, § 15.15 (describing the challenges presented by the lack of procedural requirements in informal adjudication). Consider the doctrinal point first. The APA allows courts to review agency action, including informal adjudication, to determine whether it is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.”675 U.S.C. § 706(2)(A). But if a court does not know why an agency did something and what agency decisionmakers considered while doing so—information relatively easily found in a formal adjudication’s record—then how can a court meaningfully evaluate potential arbitrariness? Courts have long struggled with this tension.68One possible solution is that arbitrary-and-capricious review may have originally been understood as extraordinarily deferential. See Pojanowski, supra note 4, at 892–93. But words like “discretion”—also found in the APA—suggest more than a rubberstamp. See, e.g., Martin v. Franklin Cap. Corp., 546 U.S. 132, 139 (2005) (“We have it on good authority that ‘a motion to [a court’s] discretion is a motion, not to its inclination, but to its judgment; and its judgment is to be guided by sound legal principles.’ Discretion is not whim, and limiting discretion according to legal standards helps promote the basic principle of justice that like cases should be decided alike.” (quoting United States v. Burr, 25 F. Cas. 30, 35 (C.C. Va. 1807) (No. 14,692d) (Marshall, C.J.) (alteration in original))). That said, the meanings of and relationship between arbitrary-and-capricious review and abuse-of-discretion review are beyond this Article’s scope. In Citizens to Preserve Overton Park, Inc. v. Volpe,69401 U.S. 402 (1971). the Supreme Court created de facto procedural requirements for informal adjudication.70See Hickman & Pierce, supra note 20, §§ 11.4, 17.6. The Court held that although courts cannot force agencies to use certain procedures, if the agency record is insufficient to allow for meaningful review, a reviewing court may order discovery to figure out what actually happened and why.71See, e.g., Overton Park, 401 U.S. at 420 (“The court may require the administrative officials who participated in the decision to give testimony explaining their action.”); id. (“[S]ince the bare record may not disclose the factors that were considered or the Secretary’s construction of the evidence it may be necessary for the District Court to require some explanation in order to determine if the Secretary acted within the scope of his authority and if the Secretary’s action was justifiable under the applicable standard.”). The result is that agencies often create a record.72See, e.g., 33 Charles Alan Wright & Arthur R. Miller, Federal Practice & Procedure Judicial Review § 8391 (2d ed. 2020) (“The Court in Overton Park thus created a strong incentive for agencies wishing to avoid later burdensome judicial inquiries to offer contemporaneous explanations and records for informal adjudications even in the absence of any statutory requirement.”). Yet this de facto requirement to create a record is hard to square with the text of the APA. It is also mushy, as Overton Park “requires that an agency adequately explain its decision, even though the court did not explain the meaning of ‘adequate.’”73Gordon G. Young, Judicial Review of Informal Agency Action on the Fiftieth Anniversary of the APA: The Alleged Demise and Actual Status of Overton Park’s Requirement of Judicial Review “On the Record,10 Admin. L.J. Am. U. 179, 210 (1996). To be sure, organic statutes may impose their own procedures for informal adjudication; the APA is only a default. But not all organic statutes do. See, e.g., Overton Park, 401 U.S. at 414–15 (noting no on-point procedural requirements).

The APA also created another problem: it is not settled when formal procedures apply. Based on the APA’s text and history, one might think that formal adjudication is required whenever an organic statute requires a “hearing.”745 U.S.C. § 554(a). That view, however, runs into the Supreme Court’s decision in United States v. Florida East Coast Railway Co.75410 U.S. 224 (1973). that essentially created a “magic words” requirement for formal rulemaking.76See id. at 237–38. Indeed, “since Florida East Coast Railway, no organic rulemaking statute that does not contain the specific words ‘on the record’ has ever been held to require formal rulemaking.”77Jack M. Beermann & Gary Lawson, Reprocessing Vermont Yankee, 75 Geo. Wash. L. Rev. 856, 857 n.9 (2007). This means that with only a few exceptions, formal rulemaking today is essentially nonexistent.78See, e.g., Aaron L. Nielson, In Defense of Formal Rulemaking, 75 Ohio St. L.J. 237 (2014).

Does that same “magic words” requirement apply to formal adjudication? Remarkably, although the APA has been on the books for nearly seventy-five years, the Supreme Court has not decided this issue. The lower courts are also divided. Some cases say there is a presumption favoring formal adjudication whenever a statute requires a hearing, but the majority view is that agencies have discretion—even Chevron-deference-type discretion—over whether formal rulemaking applies.79See Hickman & Nielson, supra note 53 (manuscript at 34–35). This is a problem. Policy-wise, some adjudications that would be well-served by formal procedures do not receive them.80See id. Doctrinally, the fact that courts are still struggling to understand what the APA requires suggests that something has gone wrong.

Both of these problems—the hole for informal adjudication and the imprecise trigger for formal adjudication—flow from the fact that the APA is incomplete. Congress did not finish the job.

C.     Chenery II’s Aggressiveness

The third wrong turn came in 1947. The principle from Chenery II is simply stated: when announcing policy, “an administrative agency must be equipped to act either by general rule or by individual order” (i.e., by rulemaking or adjudication, even if the adjudicatory order may have “a retroactive effect”).81SEC v. Chenery Corp. (Chenery II), 332 U.S. 194, 202–03 (1947). This principle is now black-letter administrative law.82See, e.g., Martin v. Occupational Safety & Health Rev. Comm’n, 499 U.S. 144, 154 (1991) (“[A]djudication operates as an appropriate mechanism not only for factfinding, but also for the exercise of delegated lawmaking powers, including lawmaking by interpretation.”); see alsoNLRB v. Bell Aerospace Co., 416 U.S. 267, 294 (1974) (“[T]he Board is not precluded from announcing new principles in an adjudicative proceeding . . . .”); see generally Russell L. Weaver & Linda D. Jellum, Chenery II and the Development of Federal Administrative Law, 58 Admin L. Rev. 815 (2006) (explaining the rule). But it is also a wrong turn, at least when taken too far.

The facts of Chenery II, also familiar to students of administrative law, illustrate how the principle works. The Federal Water Service Corporation tried to reorganize itself in a way that the Securities and Exchange Commission (“SEC”) disliked, primarily because the agency felt the reorganization would allow existing shareholders to retain too much control over the reorganized entity.83SEC v. Chenery Corp. (Chenery I), 318 U.S. 80, 81 (1943). Accordingly, the SEC concluded that certain “preferred stock” could not “participate in the reorganization on an equal footing with all other preferred stock.”84Id. The relevant stockholders protested that no law barred the company’s preferred reorganization. The SEC disagreed and invoked its authority to block reorganizations “detrimental to the public interest” or not “fair and equitable.”85Id. at 82 n.1. In particular, the SEC concluded that “judge-made rules of equity” condemned the company’s plan.86Id. at 93. That conclusion, however, was “plainly” false—in fact, such traditional rules did “not impose upon officers and directors of a corporation any fiduciary duty to its stockholders which precludes them, merely because they are officers and directors, from buying and selling the corporation’s stock.”87Id. at 88. Accordingly, in Chenery I, the Supreme Court held that the SEC’s order could not stand, and stated that if the SEC wanted to create prohibitions beyond the common law, the agency could “promulgate[] new general standards of conduct” through “the rule-making powers delegated to it.”88Id. at 93; see also id. at 92–93 (“Abuse of corporate position . . . may raise questions so subtle that the law can deal with them effectively only by prohibitions not concerned with the fairness of a particular transaction. But before transactions otherwise legal can be outlawed or denied their usual business consequences, they must fall under the ban of some standards of conduct prescribed by an agency of government authorized to prescribe such standards—either the courts or Congress or an agency to which Congress has delegated its authority.”).

On remand, the SEC opted against rulemaking, perhaps because it could not apply a rule retroactively.89See SEC v. Chenery Corp. (Chenery II), 332 U.S. 194, 196 (1947); see also Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 208 (1988) (explaining the default rule that rulemaking is prospective). Acting through adjudication instead, the SEC reached the same conclusion as before by announcing and enforcing a new prohibition—one that was not found in traditional rules of equity, Congress had not enacted, and the SEC had never previously announced.90See Chenery II, 332 U.S. at 196 (“On remand, the Commission reexamined the problem, recast its rationale and reached the same result.”). Yet when the upset stockholders again sought judicial review, the Supreme Court upheld the SEC’s retroactive application of that new prohibition.91See id. at 209.

Justice Murphy explained that although rulemaking is preferable,92See id. at 202 (“The function of filling in the interstices of the Act should be performed, as much as possible, through this quasi-legislative promulgation of rules to be applied in the future.”). there is no “rigid requirement” mandating rulemaking for the creation of new policy because, among other things, “problems may arise in a case which the administrative agency could not reasonably foresee, problems which must be solved despite the absence of a relevant general rule.”93Id.; see also id. at 202–03 (“[The] agency may not have had sufficient experience with a particular problem to warrant rigidifying its tentative judgment into a hard and fast rule. Or the problem may be so specialized and varying in nature as to be impossible of capture within the boundaries of a general rule.”). He further explained that retroactive application, although not always allowed, is permissible when it serves important interests.94See id. at 203 (“[R]etroactivity must be balanced against the mischief of producing a result which is contrary to a statutory design or to legal and equitable principles. If that mischief is greater than the ill effect of the retroactive application of a new standard, it is not the type of retroactivity which is condemned by law.”). Justice Robert Jackson dissented, accusing the Court of upholding “conscious lawlessness” that “literally” allowed the agency to “take[] valuable property away from its lawful owners for the benefit of other private parties without full compensation.”95Id. at 211, 218 (Jackson, J. dissenting); see also id. at 216 (condemning the decision as “administrative authoritarianism” that allows an agency “to govern the matter without law” based on “reasons . . . locked in its own breast” that it “does not intend to commit . . . to any rule or regulation”).

Chenery II represents another wrong turn in agency adjudication. To be clear, the idea that agencies need not always act by rulemaking is sometimes correct. After all, because the primary concern with retroactivity is whether there was fair warning, it follows that “agencies should be able to announce policy in adjudication when retroactivity concerns are not especially serious.”96Nielson, supra note 28, at 999; see also id. (“Justice Jackson erred by saying that agencies should never be allowed to make policy through adjudication, even though he was correct to worry about retroactivity.”). For instance, if the agency’s reading of the legal text is “the most natural one,” there is little reason to require rulemaking.97See, e.g., NetworkIP, LLC v. FCC, 548 F.3d 116, 123 (D.C. Cir. 2008). In such a case, the text itself provides fair notice. Chenery II, however, took that sound principle and applied it too aggressively.98See, e.g., Qwest Servs. Corp. v. FCC, 509 F.3d 531, 540 (D.C. Cir. 2007) (“[A] mere lack of clarity in the law does not make it manifestly unjust to apply a subsequent clarification of that law to past conduct.”). As Justice Jackson recognized, the company in Chenery II had no way of knowing that the SEC would interpret an open-ended phrase like “fair and equitable” to apply to its conduct.99See Chenery II, 332 U.S. at 216 (Jackson, J., dissenting) (“[T]he Court admits that there was no law prohibiting these purchases when they were made, or at any time thereafter. And, except for this decision, there is none now.”).

Chenery II’s aggressiveness distorts administrative law in at least two ways. First, retroactivity can be unfair to individual parties—as in Chenery II itself. Indeed, unknowable law may be “literally Orwellian.”100NetworkIP, 548 F.3d at 122 n.5 (citing George Orwell, Animal Farm 102–03 (1946)). It is doubtful that Congress intended that “lawless[]” result,101Chenery II, 332 U.S. at 217 (Jackson, J., dissenting). especially given the language from Chenery I, decided just three years before the APA’s enactment.102See SEC v. Chenery Corp. (Chenery I), 318 U.S. 80, 92 (1943) (“Had the Commission, acting upon its experience and peculiar competence, promulgated a general rule of which its order here was a particular application, the problem for our consideration would be very different.”). Second, Chenery II has a dynamic effect. Because regulated parties—operating in the shadow of the law103See generally Robert H. Moonkin & Lewis Kornhauser, Bargaining in the Shadow of the Law: The Case of Divorce, 88 Yale L.J. 950 (1979) (explaining that courts often do not need to resolve issues because the parties themselves anticipate judicial decisions).—know that their regulators can create law retroactively, they must pay close attention to what an agency says informally, such as through guidance documents.104See, e.g., Nielson, supra note 28, at 987 (“At least in part, the reason regulated parties care about guidance documents where the agency has not promulgated a rule is because Chenery II looms in the background. If an agency could not punish regulated parties directly under the statute, but instead could only promulgate rules, regulated parties would have less reason to worry about a guidance document when the agency has not promulgated a rule . . . Hence, Chenery II is the anchor that gives many guidance documents their weight.” (citing David L. Franklin, Two Cheers for Procedural Review of Guidance Documents, 90 Tex. L. Rev. 111, 117 (2012))). If an agency can change the law after the fact, a rational party will often obey all of the agency’s hints about the future. Thus, the power to act retroactively enables agencies to signal where the law is headed, which may in effect create law—at least law in the Holmesian sense of prediction105See, e.g., Oliver Wendell Holmes, Jr., The Path of the Law, 10 Harv. L. Rev. 457, 457 (1897) (explaining that law is a profession precisely because lawyers are trained to predict what the power of the state will do).—without any formal procedures.

II.     Two Responses

The effects of these wrong turns are still felt today. In fact, 2019’s most significant administrative law events can be traced to them. The Supreme Court in Kisor v. Wilkie limited what Seminole Rock had become, while the Trump Administration’s executive orders targeted the use of adjudication and guidance documents to make policy.

A.     Kisor Deference

The most significant event in 2019 for administrative law was the Supreme Court’s decision to cut back on Auerdeference—what Seminole Rock had become106See generally Nielson, supra note 28 (explaining how Auer set aside some of Seminole Rock’s safeguards).—in Kisor v. Wilkie.107See 139 S. Ct. 2400, 2408 (2019). Although purporting to uphold this form of deference, the Court’s majority limited it, while the other justices would have formally overruled both Auer and Seminole Rock.108See, e.g., id. at 2447–48 (Gorsuch, J., concurring in the judgment). Notably, no one on the Court defended the conventional understanding of Auer deference.

The story of the end of Auer deference and the emergence of Kisor deference merits a quick retelling. In 2011, Justice Antonin Scalia began questioning Seminole Rock and Auer.109See Talk Am., Inc. v. Mich. Bell Tel. Co., 564 U.S. 50, 67–68 (2011) (Scalia, J., concurring). Notably, Justice Scalia authored Auer. The attack picked up speed the next year when a majority of the Court refused to defer to an agency interpretation that did not provide “fair warning.”110Christopher v. SmithKline Beecham Corp., 567 U.S. 142, 156 (2012) (quoting Gates & Fox Co. v. Occupational Safety & Health Rev. Comm’n, 790 F.2d 154, 156 (D.C. Cir. 1986) (opinion of Scalia, J.)). Justice Scalia thereafter expressly called for Auerand Seminole Rock to be overruled,111Decker v. Nw. Env’t Def. Ctr., 568 U.S. 597, 617 (2013) (Scalia, J., dissenting in part). a call soon echoed by Justice Thomas.112See Perez v. Mortg. Bankers Ass’n, 575 U.S. 92, 112 (2015) (Thomas, J., concurring in the judgment); see also Garco Constr., Inc. v. Speer, 138 S. Ct. 1052, 1053 (2018) (Thomas, J., dissenting from the denial of certiorari); United Student Aid Funds, Inc. v. Bible, 136 S. Ct. 1607, 1608 (2016) (Thomas, J., dissenting from the denial of certiorari). Eventually, the Court agreed to consider the issue in Kisor.

James Kisor served as a soldier during the Vietnam War. Alleging post-traumatic stress disorder from his participation in Operation Harvest Moon,113This was a troubled mission that “led to heavy losses on the allied side.” Nicholas J. Schlosser, In Persistent Battle: U.S. Marines in Operation Harvest Moon, 8 December to 20 December 1965 1 (2017). Kisor unsuccessfully sought disability payments in the early 1980s from the Department of Veterans Affairs (“VA”).114Kisor v. Wilkie, 139 S. Ct. 2400, 2409 (2019). Over two decades later, he returned to the VA, urging the agency to reopen the matter.115Id. See 38 C.F.R. § 3.156(c)(1). This time, the VA concluded that Kisor was eligible—but only prospectively.116Kisor, 139 S. Ct. at 2409. A relevant regulation stated that the VA may grant benefits retroactively only if the agency had not previously considered “relevant official service department records.”11738 C.F.R. § 3.156(c)(1). But what does “relevant” mean? After all, Kisor “had come up with two new service records,” both confirming his participation in Operation Harvest Moon, but the VA concluded “those records were not ‘relevant’ because they did not go to the reason for the denial—that Kisor did not have PTSD.”118Kisor, 139 S. Ct. at 2409. The matter eventually ended up before the Court of Appeals for the Federal Circuit, which deferred to the VA under Auer.119See Kisor v. Shulkin, 869 F.3d 1360, 1367 (Fed. Cir. 2017) (“We hold that § 3.156(c)(1) is ambiguous as to the meaning of the term ‘relevant.’ In our view, the regulation is vague as to the scope of the word, and canons of construction do not reveal its meaning.”).

Justice Elena Kagan—writing parts of her opinion for a majority of the Court and other parts for a plurality—declined to formally overrule anything, but her analysis takes a good chunk out of Auer. In particular, in a portion of her opinion that commanded a majority, Justice Kagan significantly limited Auer in a number of ways. She stressed, for instance, that before deference is allowed, the regulation must be “genuinely ambiguous, even after a court has resorted to all the standard tools of interpretation.”120Kisor, 139 S. Ct. at 2414. To be “reasonable,” the agency’s reading “must come within the zone of ambiguity the court has identified after employing all its interpretive tools.”121Id. at 2415–16. And flatly rejecting the view then held by many lower courts, Justice Kagan stressed that Auer is not more deferential than Chevron,122See Chevron, U.S.A., Inc. v. Nat. Res. Def. Council, Inc., 467 U.S 837, 842–43 (1984) (outlining a two-step inquiry for determining whether or not an agency’s interpretation of an authorizing statute merits judicial deference). notwithstanding the language from Seminole Rock about “plainly erroneous” interpretations or “controlling weight.”123Compare Kisor, 139 S. Ct. at 2416, with William Yeatman, Note, An Empirical Defense of Auer Step Zero, 106 Geo. L.J. 515, 530 (2018) (collecting citations showing that “U.S. Courts of Appeals for the First, Third, Fifth, Sixth, District of Columbia, and Federal Circuits ha[d] claimed that Auer is more deferential than Chevron” (footnotes omitted)).

Justice Kagan, for the majority, further held that deference is not required even for “all reasonable agency constructions” of truly ambiguous regulations.124Kisor, 139 S. Ct. at 2414. For instance, no deference is appropriate where the “interpretation does not reflect an agency’s authoritative, expertise-based, fair, or considered judgment.”125Id. (alterations omitted). In other words, the interpretation cannot be an “ad hoc statement,” but rather “must at the least emanate from those actors, using those vehicles, understood to make authoritative policy in the relevant context.”126Id. at 2416. Further, the agency’s interpretation must reflect real “expertise” with the policy at issue so that the relevant question does not “fall more naturally into a judge’s bailiwick.”127See id. at 2417. Finally, a court should not defer when an agency is trying to fend off litigation or where the agency’s interpretation “creates ‘unfair surprise’ to regulated parties,” for instance “when an agency substitutes one view of a rule for another” or otherwise “impose[s] retroactive liability on parties for longstanding conduct that the agency had never before addressed.”128Id. at 2417–18. Chief Justice Roberts did not join all of Justice Kagan’s opinion, refusing, pointedly, to join her first-principles defense of deference. But he did endorse her new limits on Auer. See id. at 2424 (Roberts, C.J., concurring in part). And for his part, Justice Neil Gorsuch argued that the Court should have said “goodbye to Auer” and observed that Auer now has “so many new and nebulous qualifications and limitations” that it is not the same doctrine. See id. at 2425–26 (Gorsuch, J., concurring in the judgment); see also Aaron L. Nielson, Kisor Deference, Yale J. on Reg.: Notice & Comment (June 26, 2019), (“[T]he version of ‘Auer’ that the Court created today is so far removed from how Auerwas understood yesterday that isn’t really accurate to call it Auer anymore.”). These are important safeguards.

Especially relevant here is Justice Kagan’s assessment of Seminole Rock as a work of judicial craftsmanship. She did not give it high marks, observing that Seminole Rock’s language “may suggest a caricature of the doctrine” and acknowledging that “Kisor has a bit of grist for his claim that Auer ‘bestows on agencies expansive, unreviewable’ authority.”129Kisor, 139 S. Ct. at 2415 (quoting Brief for Petitioner at 25, Kisor, 139 S. Ct. at 2405 (No. 18-15)). She also observed that “[s]ome courts” may have been confused by “Seminole Rock’s ‘plainly erroneous’ formulation” into believing that this form of deference is particularly strong, even though it “is not so.”130Id. at 2416. Whose fault was that confusion? The truth is that if the Seminole Rock Court had written a better opinion in the first place, all the confusion and unfairness cataloged by Justice Kagan could have been avoided.

Even after Kisor, it is doubtful that administrative law has returned to what the Supreme Court in Seminole Rock likely intended: an application of—or at least a close cousin to—Skidmore. But the law is at least closer to that standard.131See id. at 2424–25 (Roberts, C.J., concurring in part) (explaining that after Kisor, “cases in which Auer deference is warranted largely overlap with the cases in which it would be unreasonable for a court not to be persuaded by an agency’s interpretation of its own regulation”). No matter how you slice it, Kisor is a sharp rebuke of Seminole Rock’s sloppiness.

B.     The 2019 Executive Orders

The second-most significant event in 2019 for administrative law132No doubt some will disagree with this assessment. The Supreme Court’s decision in Gundy v. United States, 139 S. Ct. 2116 (2019), which rejected a nondelegation challenge, may prove to be more significant—not because of what it did, but for what it may portend. See, e.g., Gary Lawson, “I’m Leavin’ It (All) Up to You”: Gundy and the (Sort-of) Resurrection of the Subdelegation Doctrine, 2018-2019 Cato Sup. Ct. Rev. 31, 61, 68 (speculating what Gundy may “suggest about future cases and the administrative state more broadly”). But see Aditya Bamzai, Delegation and Interpretive Discretion: Gundy, Kisor, and the Formation and Future of Administrative Law, 133 Harv. L. Rev. 164, 174 (2019) (suggesting that Gundy itself is significant because it “breathes new life into the constitutional avoidance approach, with uncertain implications for the future”). was the Trump Administration’s decision to issue two executive orders that address agency adjudication. The first EO on guidance documents only indirectly addressed adjudication. The second EO, however, was all about “[t]ransparency and [f]airness in [c]ivil [a]dministrative [e]nforcement and [a]djudication.”133Exec. Order No. 13,892, 84 Fed. Reg. 55,239 (Oct. 9, 2019). The two executive orders were issued the same day and targeted overlapping issues.

Beginning with the second EO (EO 13892), “[r]egulated parties” should “know in advance the rules by which the Federal Government will judge their actions,” and “[n]o person should be subjected to a civil administrative enforcement action or adjudication absent prior public notice of both the enforcing agency’s jurisdiction over particular conduct and the legal standards applicable to that conduct.”134Id. Relevant here, EO 13892 ordered that when an agency conducts adjudication, “it may apply only standards of conduct that have been publicly stated in a manner that would not cause unfair surprise,” with “unfair surprise” being defined as “a lack of reasonable certainty or fair warning of what a legal standard . . . requires.”135Id. at 55,240–41. EO 13892 further stated that, with a few exceptions, “before an agency takes any action with respect to a particular person that has legal consequence for that person . . . the agency must afford that person an opportunity to be heard, in person or in writing, regarding the agency’s proposed legal and factual determinations.”136Id. at 55,241. Notably, several months later, the Office of Management and Budget issued a call for comments “on regulatory reforms that will better safeguard due process in the regulatory enforcement and adjudication settings (i.e., non- Article III adjudications),” including whether adjudications “coerce Americans into resolutions/settlements.” Off. of Mgmt. & Budget, Exec. Off. of the President, Improving and Informing Regulatory Enforcement and Adjudication, 85 Fed. Reg. 5,843, 5,483–84 (Jan. 30, 2020). That provision—seemingly imposing new procedures for informal adjudication—did not apply in situations without time for a written response “because of a serious threat to health, safety, or other emergency.”13784 Fed. Reg. at 55,241. Even in those circumstances, however, EO 13892 required the agency to provide “an opportunity to be heard, in person or in writing, regarding the agency’s legal determinations and [to] respond in writing as soon as practicable.”138Id. Those procedures went beyond what the APA requires.

EO 13892 also addressed how guidance documents can be used in adjudication. In particular, it said that an agency “may not” use a guidance document “to impose new standards of conduct on persons outside the executive branch except as expressly authorized by law or as expressly incorporated into a contract.”139Id. at 55,240. Thus, a guidance document could “do no more, with respect to prohibition of conduct, than articulate the agency’s understanding of how a statute or regulation applies to particular circumstances,” and even then, before the agency could cite the document for that limited purpose, it must have previously notified the public.140Id.

For its part, EO 13891’s main policy was that agencies should “impose legally binding requirements on the public” through “notice-and-comment rulemaking.”141Exec. Order No. 13,891, 84 Fed. Reg. 55,235, 55,235 (Oct. 9, 2019). Although agencies “may clarify existing regulations through” guidance, they should not “regulate the public without following the rulemaking procedures of the APA.”142Id. Nor should agencies exercise de facto regulatory power by issuing guidance documents “carry[ing] the implicit threat of enforcement action if the regulated public does not comply.”143Id. As relevant here, EO 13891 further required agencies to make guidance documents easily accessible to the public and to clearly state that guidance documents are not binding.144See id. The Office of Management and Budget thereafter issued a memorandum expanding on EO 13891’s themes. That memorandum stated that “a guidance document should never be used to establish new positions that the agency treats as binding.”145Memorandum from Dominic J. Mancini, Acting Adm’r, Off. of Info. & Regul. Affs. to Regul. Pol’y Officers at Exec. Dep’ts & Agencies & Managing & Exec. Dirs. of Certain Agencies & Comm’ns 3 (Oct. 31, 2019), The memorandum further stated that regardless of the label, agencies should not use guidance documents to “effectively . . . coerce private-party conduct, for instance by suggesting that a standard in a guidance document is the only acceptable means of complying with statutory requirements, or by threatening enforcement action against all parties that decline to follow the guidance.”146Id.

The intended effect of these executive orders on the problems for agency adjudication from 1945 to 1947 was apparent. Not only would they reduce Seminole Rock and Chenery II’s effects by incorporating stronger fair-notice requirements, but they also addressed, at least partially, the procedural hole in the APA for informal adjudication.

But these executive orders were not perfect. Even the bolstered procedures for informal adjudication were not as formal as those for formal adjudication. This is not always a problem, of course; not everything needs trial-like procedures.147See, e.g., Mila Sohoni, Agency Adjudication and Judicial Nondelegation: An Article III Canon, 107 Nw. U. L. Rev. 1569, 1606 (2013). Lack of formality, however, can be a problem. Both orders  also stated that they did not create any judicially enforceable rights.148See Exec. Order No. 13,892, 84 Fed. Reg. 55,239, 55,242 (Oct. 9, 2019) (“This order is not intended to, and does not, create any right or benefit, substantive or procedural, enforceable at law or in equity by any party against the United States, its departments, agencies, or entities, its officers, employees, or agents, or any other person.”). Any enforcement of them thus would had to have come from within the executive branch itself.

Even more significantly, executive orders do not bind future administrations, and President Joseph Biden decided not to retain these orders.149See Exec. Order No. 13,992, 86 Fed. Reg. 7,049 (Jan. 20, 2021).  Yet some of the ideas behind them may have staying power. For instance, the Administrative Conference of the United States, a nonpartisan agency that studies the regulatory process, has endorsed a number of these principles, including that agencies should not use interpretative rules, policy statements, and guidance documents to create policy.150See Administrative Conference of the United States: Adoption of Recommendations, 84 Fed. Reg. 38,927, 38,928 (Aug. 8, 2019); Administrative Conference of the United States: Adoption of Recommendations, 82 Fed. Reg. 61,728, 61,736 (Dec. 29, 2017) (documenting Administrative Conference Recommendation 2017-5: Agency Guidance Through Policy Statements); see also Nicholas R. Parrillo, Admin. Conf. of the U.S., Federal Agency Guidance: An Institutional Perspective 26 (Sept. 18, 2017). Such principles should cut across party lines.

III.     Lessons

What went wrong between 1945 and 1947? This is an important question, all the more so because, like the 1940s, ours is also a day of reform. As cases like Kisor reflect, today’s Supreme Court overflows with administrative law talent151See, e.g., Aaron L. Nielson, D.C. Circuit Review – Reviewed: The Supreme (Administrative Law) Court, Yale J. on Reg.: Notice & Comment (Sept. 8, 2018) (noting that the Supreme Court has “deep expertise in administrative law”), and interest.152See, e.g., Metzger, supra note 1, at 22, 27–28. Thus, change is likely—especially because both Congress and the White House are also considering reform.153See, e.g., Christopher J. Walker, Modernizing the Administrative Procedure Act, 69 Admin. L. Rev. 629, 631 (2017). The history of administrative law provides lessons for these reform efforts. Three principles in particular stand out. When stating legal rules, it is important to: (1) anticipate how actors are likely to respond; (2) write carefully to address that likely response; and (3) monitor regularly to ensure that the rules are working. Incorporating these principles into reform should prevent future wrong turns, or at least minimize their wrongness.

A.     Anticipate Prudently

Successful reform requires an understanding of human nature and keen awareness that “[t]he law tends to snowball”154AKM LLC v. Sec’y of Labor, 675 F.3d 752, 764 (D.C. Cir. 2012) (Brown, J., concurring) (“The law tends to snowball. A statement becomes a holding, a holding becomes a precedent, a precedent becomes a doctrine, and soon enough we’re bowled over at the foot of a mountain, on our backs and covered in snow. So it is with our deference doctrines.”). and something “that is amply justified in principle may go terribly wrong in practice.”155Cass R. Sunstein, Political Equality and Unintended Consequences, 94 Colum. L. Rev. 1390, 1390 (1994). Even if well-intentioned, reform that gives short shrift to human nature is unlikely to succeed.156See, e.g., Aaron L. Nielson, Reflections on the End of the Federal Law Clerk Hiring Plan, 112 Mich. L. Rev. First Impressions 22, 27–28 (2013).

When it comes to administrative law, it is safe to suppose that those who work for agencies are generally predisposed to the use of regulatory power.157See, e.g., David E. Bernstein, Antidiscrimination Laws and the Administrative State: A Skeptic’s Look at Administrative Constitutionalism, 94 Notre Dame L. Rev. 1381, 1403 (2019) (“Numerous scholars have concluded that federal agencies tend to attract employees who are committed to the agency’s regulatory mission.” (collecting citations)). It is also no stretch that such individuals may prioritize substantive issues over procedural ones, especially procedural issues that increase the burden on the agency. At the margins, abstract concerns about legitimacy and procedural rigor may seem less important than immediate concerns about substantive problems: “after all, how could anyone think that the team across the hallway is illegitimate?”158Nielson, supra note 78, at 288 (emphasis omitted). Accordingly, although the point should not be overstated, there is reason to worry that regulators sometimes may be tempted to take shortcuts.159See, e.g., William F. Pedersen, Jr., Formal Records and Informal Rulemaking, 85 Yale L.J. 38, 59–60 (1975) (explaining how some within agencies try to avoid scrutiny). This dynamic has been identified in a related context.160See, e.g., Connor Raso, Agency Avoidance of Rulemaking Procedures, 67 Admin. L. Rev. 65, 68 (2015) (“This Article’s original empirical analysis shows that agencies seize upon . . . ambiguity [about when certain procedures are required] to avoid rulemaking procedures more frequently as the threat of a successful lawsuit challenging that avoidance declines. But even when litigation ensues, courts do not consistently require agencies to comply with rulemaking procedures. The result is frequent agency avoidance of rulemaking procedures.”).

A preference for substance over procedure puts pressure on the law. Over the long run, we should not be surprised to see agencies push the line to make it easier for them to regulate. The story of Seminole Rock fits the pattern. Agencies at first may have been reluctant to press too hard for deference; the Seminole Rock Court spoke broadly but also understood the context of its statement and would have presumably rejected an aggressive interpretation of its decision.161Compare, e.g., Knudsen & Wildermuth, supra note 30, at 77 (explaining how Seminole Rock was originally understood), with San Diego Gas & Elec. Co. v. FERC, 913 F.3d 127, 145 & n.4 (D.C. Cir. 2019) (Randolph, J. dissenting) (“The majority . . . grants this deference . . . without taking the necessary first step of identifying an ambiguity in the regulation . . . This case [thus] demonstrates the perils of deferring to an agency’s wayward interpretation of its own regulation.”). Eventually, however, agencies found that they could move the line as the judiciary turned over and new judges were left with only the Supreme Court’s sloppy language, rather than institutional memory, to guide them.

The shift away from formal adjudication also fits this pattern. For a long while, courts expected agencies to use formal procedures.162See, e.g., Hickman & Pierce, supra note 20, § 6.2. But the APA’s language is not crisp, and agencies gradually persuaded courts to grant them greater discretion over the issue.163See id. There is little reason to think that Congress intended judges to defer to agencies about whether formal or informal adjudication applies but that is where we have ended up in many courts.164See, e.g., id. (noting legal evolution well after the APA’s enactment); see alsoHickman & Nielson, supra note 53 (manuscript at 7–8). Had Congress spoken more clearly, this uncertainty could have been avoided.

Chenery II fits this pattern too, at least in a sense. The Chenery II Court started in an extreme place; it is hard to see how the regulated party in the Chenery cases had fair notice, for all the reasons that Justice Jackson identified in dissent. But why did the majority uphold the agency decision? Surely because it underestimated the risk of abuse of power.165SeeSEC v. Chenery Corp. (Chenery II), 332 U.S. 194, 207 (1947) (deferring to the SEC’s “informed, expert judgment on the problem”). As others have noted, Chenery II sounds in the notion that agencies are technocratic experts.166See, e.g.,Ronald J. Krotoszynski, Jr., Why Deference?: Implied Delegations, Agency Expertise, and the Misplaced Legacy of Skidmore, 54 Admin. L. Rev. 735, 736 (2002) (explaining Chenery II as a product of faith in expertise). That notion, however, can be taken too far. As then-Professor Elena Kagan explained, in reality, agency officials face complex incentives.167See, e.g., Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2261 (2001) (rejecting as “almost quaint” the notion that “‘expertness’ imposed its own guideposts, effectively solving the problem of administrative discretion,” and that professional administrators “could ascertain and implement an objective public interest”); see also Mach Mining, LLC v. EEOC, 575 U.S. 480, 487–89 (2015) (explaining that agencies may abuse their authority if they are not subject to judicial review). The Chenery II Court’s failure to give teeth to fair-notice concerns forgot an older wisdom: “If men were angels, no government would be necessary.”168The Federalist No. 51 (James Madison).

The two 2019 executive orders deserve credit for recognizing the relationship between adjudication and guidance documents.169See generally Hickman & Pierce, supra note 20, § 15.15 (discussing agency guidance). If adjudication becomes more difficult, agencies will be tempted to shift to other policy-making tools that may be easier like guidance documents. Accordingly, anticipating that reaction, the executive orders also prudently targeted guidance documents.170See Exec. Order No. 13,892, 84 Fed. Reg. 55,239, 55,240 (Oct. 9, 2019).

B.     Write Carefully

The second lesson is an easy one—write carefully. Seminole Rock deference overgrew its fence because that fence was too low. It would have been much harder for judges to read Seminole Rock broadly if the Court hadn’t used open-ended language like “plainly erroneous” and “controlling weight.”171Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 414 (1945). Would the Court, decades later, have said that it “need not tarry” over a regulation’s text172United States v. Larionoff, 431 U.S. 864, 872 (1977). but for that unfortunate language from Seminole Rock? Hopefully not. And here’s the thing: it is doubtful that the Court in Seminole Rock intended such a grandiose idea,173See, e.g., Knudsen & Wildermuth, supra note 30, at 52. yet its sloppiness opened the door for abuse. The old axiom about an ounce of prevention being worth a pound of cure comes to mind. The fact that no one wrote a full separate opinion in Seminole Rock may have been the problem, illustrating that courts should be especially careful with unanimous opinions.174See, e.g., Jennifer Senior, In Conversation: Antonin Scalia, New York (Oct. 4, 2013), (“[T]he worse opinions in my court have been unanimous. Because there’s nobody on the other side pointing out all the flaws.”).

The APA’s adjudication provisions are also poorly drafted. To be sure, some of that sloppiness is intentional. Congress made a deliberate choice to not spell out the procedures for informal adjudication.175See, e.g., Michael Asimow, The Spreading Umbrella: Extending the APA’s Adjudication Provisions to All Evidentiary Hearings Required by Statute, 56 Admin. L. Rev. 1003, 1006 n.15 (2004) (“Earlier drafts of the ABA adjudication recommendations included provisions calling for a set of barebones procedures for informal adjudication. They would have required an agency—when it would be practicable to do so—to give appropriate notice to parties, allow parties to submit comments or objections, and furnish reasons for its order . . . This proposal encountered serious opposition and was dropped.”). Regardless, though, that procedural hole is recipe for confusion, particularly because Congress did not explain how informal adjudication interacts with the APA’s provisions for judicial review. Accordingly, courts have been forced to muddle through because Congress never resolved that key question. Congress’s failure to clearly identify the trigger for formal adjudication is lamentable for similar reasons.

C.     Monitor Regularly

Finally, it is important to monitor regularly. The Supreme Court should have never let Seminole Rock grow untended for so long and Congress should regularly amend the APA. Administrative law requires regular monitoring because the external and internal pressures on the system are not constant. No one should be surprised that a statute enacted in 1946 as part of a massive compromise is often a poor fit for the different world of 2021.

Granted, it is easier to say that regular monitoring is essential than to actually do the monitoring. No one has time to monitor everything. Professor Peter Strauss’s observation that the Supreme Court can only hear 150 cases per year also cannot be brushed aside.176Peter 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, 1093 (1987). But the Court now hears less than half that number; there is significant unused capacity. Congress, of course, can also dedicate more time to this issue. Oversight hearings, for example, are important and should address issues such as how agencies engage in administrative adjudication.177See generally Investigations and Oversight, U.S. House of Representatives Off. of the Historian, (“Hearings are most commonly held . . . to investigate and oversee federal programs.”).

Regular monitoring will be especially important going forward. One concern is that revisiting Seminole Rock deference (as the Court has now done in Kisor) without also revisiting Chenery II (which did not happen in Kisor) may have unintended consequences.178See generally Nielson, supra note 28 (addressing this concern at length). In particular, to the extent that an agency values flexibility—either consciously, as some of the more aggressive criticisms of Seminole Rock contend or, perhaps more likely, because the agency does not want to expend the resources to identify and then resolve potential ambiguity—the agency may forego rulemaking altogether and instead create law under the statute itself via adjudication. A vigorous fair-notice doctrine would mitigate that concern. Thinking dynamically, reformers should carefully observe whether substitution is occurring. For any complex system, it is important to monitor how a change in one part of the system affects other parts.


This Article does not attack agency adjudication. Indeed, a great deal of agency adjudication should not be controversial. Yet because of three mistakes in the 1940s, agency adjudication is much more controversial than it needs to be. Focusing on what went wrong is the first step to fixing the problem. Unfortunately, even after seventy-five years, we still have work to do.

Share this article

Share on twitter
Share on facebook
Share on linkedin

Share it

Share on twitter
Share on facebook
Share on linkedin