Why the Court Should Reexamine Administrative Law’s Chenery II Doctrine

Executive Summary

The Chenery II “Doctrine” has been taught in administrative law courses for decades. It is widely interpreted to be foundational. The deciding Supreme Court case, SEC v. Chenery Corp. 1947, or Chenery II, is frequently cited as the case that gives federal agencies broad power to make law and policy either through rulemaking or adjudication, unless explicitly stated otherwise in their governing statutes. 

In this paper, CGO contributing scholars Gary Lawson and Joseph Postell argue that the conventional interpretation of Chenery II is a serious overreading of the Court’s decision. 

From the start, the Chenery II opinion has rested on questionable foundations. Only four justices signed the Court’s opinion, one justice concurred in the result but not the opinion, two justices vigorously dissented, and two justices did not participate in the decision.

The authors call into question the constitutionality of the Chenery II “doctrine” that gives federal agencies the power to create law through adjudication. Congress, which writes the governing statutes creating federal agencies, has no judicial power. Therefore Congress cannot delegate that power to agencies. As the authors argue, this should bring up concerns about separation of powers. It’s constitutionally problematic enough to merit a closer examination of the “doctrine,” as well as a presumption against recognizing agency power to make laws via adjudication—as opposed to the current presumption in favor of it. 

Lawson and Postell go on to examine the Chenery II decision itself. In it, the Court did actually limit the power it was handing to agencies. Some of those limits were written explicitly in the opinion and others were implied but later became more apparent in subsequent rulings. The authors analyze later cases as well as correspondences among the justices that further clarify the implied limits of Chenery II

Finally, the authors discuss how administrative law has changed since the 1947 decision. Other doctrines are now in place that conflict with the expansive interpretation of Chenery II, including doctrines governing agency legal interpretation, arbitrariness, and unfair surprise.

The authors encourage the Court to either abandon Chenery II as a principle or confine the decision by focusing on the already-present doctrinal limits and enforcing those limits consistently. The change would most dramatically affect the agencies principally using adjudication to create policy—the National Labor Relations Board (NLRB), the Federal Trade Commission, and the Drug Enforcement Administration (DEA). 

Much of the administrative state would be largely untouched, but these three agencies would be significantly changed, and in a way that would create greater predictability and greater protection for individual liberty.

The Supreme Court’s 1943 and 1947 decisions in SEC v. Chenery Corp.1318 U.S. 80 (1943) (hereinafter Chenery I); 332 U.S. 194 (1947) (hereinafter Chenery II).—colloquially referred to as Chenery I and Chenery II—are typically taught in administrative law courses as establishing foundational administrative law principles or doctrines. Specifically, Chenery II2Chenery I, for its part, says that agencies can only defend their decisions on grounds actually relied upon by the agencies when those decisions were made. See Gary Lawson, Federal Administrative Law 542-50 (9th ed. 2022). We do not question that doctrine here. See infra n.98 and accompanying text. is frequently understood to establish the principle that agencies may make law or policy through either rulemaking or adjudication, as long as their governing statutes do not explicitly state otherwise.3See, e.g., Cary Coglianese, Gabriel Scheffler & Daniel E. Walters, Unrules, 73 Stan. L. Rev. 885, 959 n. 309 (2021); Aaron L. Nielson, Beyond Seminole Rock, 105 Geo. L.J. 943, 948 (2017); Todd Phillips. A Change of Policy: Promoting Agency Policymaking by Adjudication, 73 Admin L. Rev. 495, 501-02 (2021). Organic statutes sometimes give agencies no choice in the matter; for example, the Environmental Protection Agency can only set pollution standards by rule,4See 42 U.S.C. § 7409(a) (2018). and the Occupational Safety and Health Review Commission can only impose sanctions by issuing orders in adjudications.5See 29 U.S.C. § 661(j) (2018). The OSHRC can issue procedural rules, see id. at § 661(g), but any binding rulings must be an “order.” Id. at § 661(j). The strong form of the Chenery II doctrine kicks in when, as frequently happens, organic statutes can be read to authorize agencies to act by either mode.

In our view, this conventional account is a serious overreading of the Court’s decision in Chenery II. If it was an accurate reading of what the Court said in Chenery II, that decision would be so constitutionally problematic that it must be overturned or at least cabined. Agency lawmaking through adjudication raises constitutional questions serious enough at least to warrant a strong presumption against such authority rather than, as the conventional understanding of Chenery II prescribes, a presumption in its favor.

Even apart from these constitutional concerns, the Chenery II opinion rested on relatively flimsy foundations. Only four justices signed the Court’s opinion, one justice concurred in the result without opinion, two justices vigorously dissented, and two justices did not participate in the decision. Moreover, the Court in Chenery II recognized limits on the ability of federal agencies to choose to make law through either rulemaking or adjudication, as we explain in Part II of this article.

These limits, we argue, are inherent in the nature of agency power and the due process of law. Accordingly, courts should rigorously enforce the limits to Chenery II, as well as the other background rules that protect individual liberty from ad hoc administrative decision-making.

Part I of this article begins by discussing some fundamental constitutional principles that were raised, sometimes implicitly and indirectly, in the Chenery cases. Those principles point to limits on administrative adjudication that go well beyond those recognized in current doctrine. We do not here seek to push those principles as far as they can go, though we offer no resistance to anyone who wants to trod that path. Instead, we identify and raise those principles to help understand the scope and limits of actual doctrine. Our modest claims here are that constitutional concerns about at least some classes of agency lawmaking in adjudication (1) are serious enough to warrant a close look at unqualified articulations of a Chenery II “doctrine” and (2) warrant at least a presumption against recognizing agency power to choose adjudication as a form of lawmaking. In other words, they form a lens through which one can take a fresh look at a now-canonical case.

Part II then discusses the progress of the litigation and decisions in both Chenery I and Chenery II. As the correspondence among the justices and other circumstances of the cases reveal, the Court did not conclude in Chenery II that, as an absolute rule, “the choice made between proceeding by general rule or by individual, ad hoc litigation is one that lies primarily in the informed discretion of the administrative agency.”6Chenery II, 332 U.S. at 203. While that language comes from the Court’s decision in Chenery II, the broader context of the case indicates that the Court qualified the scope and domain of this principle.

Part III discusses the development of the law following the Chenery II case. The limits implicit in the Court’s 1947 decision have largely been lost in the ensuing three-quarters of a century. Nonetheless, in Part IV, we suggest that these later developments can be interpreted in two ways, both of which question the notion of a limitless Chenery II “doctrine.” Either agencies are interpreting and applying their governing statutes when issuing orders that are not pursuant to general rules, or they are establishing “embedded rules” that are contained in orders.7See Matthew C. Stephenson, Embedded Rules, 39 Yale J. Reg. Bull. 59 (2021). If the former, then various doctrines governing agency legal interpretation, arbitrariness, and unfair surprise apply. If the latter, then doctrines addressing embedded rules should apply. Both paths suggest that it is incorrect simply to think of a Chenery II doctrine that enables agencies to act via rulemaking or adjudication at their discretion. Thus, we argue that the Court should overturn, clarify, or simply ignore unqualified recitations of a broad Chenery II principle in future cases, relying instead on alternative principles to address agencies’ choice between rulemaking and adjudication.

Part V briefly suggests a few potential applications of this new approach to explain how the law might change in a post-Chenery II world. While we believe that a post-Chenery II legal regime would afford better protection for individual rights and due process of law, the roots of much of what we recommend can already be found in various administrative law sources and doctrines, none of which have proven to be fundamentally disruptive to the administrative state.8To be clear, neither of us is especially reluctant to disrupt the administrative state. But this project is primarily doctrinal rather than normative, and we are just being realistic about the modest consequences of recognizing limits to the modern Chenery II doctrine.

Part I: The Constitution and Agency Lawmaking Through Adjudication

The most basic problem with reading the Chenery II decision to give agencies free rein (absent direct statutory limits) to make law through adjudication is that, over a large range of cases, it would quite possibly be unconstitutional for an agency to proceed in that vein by adjudication rather than rulemaking, regardless of whether Congress purported to authorize the agency to proceed by adjudication and regardless of how formal or robust are the procedures that the agency employs for its action. Specifically, when the agency action deprives someone of life, liberty, or property, agency action by adjudication rather than rulemaking violates core principles of both due process of law and subdelegation. The former conclusion may seem paradoxical, because agency adjudications are often more procedurally robust than agency rulemakings. But in this context, due process of law is about substance rather than procedure, as we will shortly explain.

Before we defend these conclusions, some preliminary qualifications are necessary.

First, in setting the constitutional context for understanding Chenery II, we are describing the Constitution’s original meaning,9We do not here enter into the multifaceted disputes about what “original meaning” actually involves. See, e.g., Gary Lawson, Equivocal Originalism, ___ Tex. Rev. L. & Politics  ­_­­__ (2023). For the topics that we discuss in this article, those disputes probably do not matter; all plausible versions of originalism will converge on more or less the same results. not the state of current doctrine. As we will explain below, current doctrine regarding due process of law has strayed far from the Constitution’s actual meaning—so far that a direct constitutional argument along our lines by an actual modern litigant would be quixotic at best and perhaps even borderline frivolous at worst. Even nearly a century ago, the Chenerys did not advance precisely the argument that we formulate here,10See infra Part I.a.. though, as we shall see, one can find hints of it in the arguments of counsel, the Chenery I decision, and Justice Robert H. Jackson’s dissenting opinion in Chenery II. Nonetheless, for anyone who considers the meaning of the Constitution either relevant or interesting, it is still worthwhile, even today, to know that meaning and to reflect on how it might bear on construing statutory authorizations to agencies.

Second, we are not remotely claiming that all agency adjudication is unconstitutional as a matter of original meaning. To the contrary, the vast majority of agency adjudications are indisputably constitutional, often on multiple grounds. Most of those adjudicatory proceedings involve claims for government benefits, and as a matter of original meaning such proceedings do not implicate principles of due process of law or subdelegation of legislative authority because they do not involve deprivations of life, liberty, or property.11For a comprehensive account of the difference between adjudications of rights and benefits (or privileges), see Caleb Nelson, Adjudication in the Political Branches, 107 Colum. L. Rev. 559 (2007). In 1947, the development of modern doctrine extending due process of law protections to benefits claims was still a few years away.12See, e.g., Bailey v. Richardson, 182 F.2d 46, 57 (D.C. Cir. 1950), aff’d by an equally divided Court, 341 U.S. 918 (1951) (“It has been held repeatedly and consistently that Government employ is not ‘property’”). That modern extension perhaps has some justification as a second-best response to other modern developments,13The scope of governmental activity has grown beyond anything plausibly within the contemplation of the Constitution of 1788. Does it make sense to allow that growth without also growing, perhaps without textual justification, the constraints on government action? We pose the question without answering it. For a thoughtful exploration of “second-best” problems in adjudication, see Peter B. McCutchen, Mistakes, Precedent, and the Rise of the Administrative State: Toward a Constitutional Theory of the Second-Best, 80 Cornell L. Rev. 1 (1994). but it has no plausible foundation in original constitutional meaning. Our analysis here is confined to the modest but important subset of agency adjudications that result in deprivations of life, liberty, or property, as the Constitution uses those terms.

Third, we refer often to principles of due process of law rather than to the text of the Fifth Amendment’s due process of law clause.14See U.S. Const. amend. V (“No person shall . . . be deprived of life, liberty, or property, without due process of law”). We are discussing only federal agencies, so we do not address the Fourteenth Amendment’s due process of law clause and whether it might have a different meaning from the similar provision in the Fifth Amendment. See French v. Barber Asphalt Paving Co., 181 U.S. 324, 328-29 (1901). As one of us has explained at great length elsewhere,15See Gary Lawson, Take the Fifth . . . Please!: The Original Insignificance of the Fifth Amendment’s Due Process of Law Clause, 2017 B.Y.U. L. Rev. 611. those principles of due process of law were part of the Constitution of 1788. The due process of law clause, ratified more than three years after the Constitution took effect, at most confirmed and clarified those principles. Indeed, it is quite possible that the precise phrase “due process of law” as used in the Fifth Amendment refers solely to the appropriate process for initiating a cause of action. That was the phrase’s meaning 800 years ago16See Keith Jurow, Untimely Thoughts: A Reconsideration of the Origins of Due Process of Law, 19 Am. J. Legal Hist. 265 (1975). and quite possibly was its meaning in 1791 as well.17See Max Crema & Lawrence B. Solum, The Original Meaning of “Due Process of Law” in the Fifth Amendment, 108 Va. L. Rev. 447 (2022). There is evidence that the different phrase “due course of law” described the broader principles of legality that require government to act in a certain fashion throughout a legal action before imposing burdens on citizens.18See id. at 502-04. We do not engage that question here because we are not offering an interpretation of the words of the Fifth Amendment. As explained below, we are describing fundamental structural features of the Constitution that were in place well before December 15, 1791, when the Fifth Amendment was ratified.

Fourth, we take no position on whether agency rulemaking as well as adjudication would be constitutionally suspect in the circumstances that we discuss. We take as given that Congress has power to subdelegate a sufficient measure of legislative authority to allow the SEC, if it had chosen to do so in the 1930s or 1940s, to promulgate rules governing stock acquisitions during reorganization negotiations. As a matter of first principles, we are not at all confident that Congress has such power, but discussion of that point would require a separate paper (or perhaps several separate papers).19For some thoughts on these broader questions, which may surprise people expecting to hear about a categorical ban on all legislative subdelegation (which is not in fact the result that one gets from applying original meaning), see Gary Lawson, A Private-Law Framework for Subdelegation, in The Administrative State before the Supreme Court: Perspectives on the Nondelegation Doctrine 125 (Peter J. Wallison & John Yoo eds., 2022). We are here claiming only that whatever power of subdelegation Congress might possess does not include the power to authorize agency adjudications that deprive people like the Chenerys of their property by creating new law. We start with that proposition about subdelegation and then connect it up to principles of due process of law.

a. Congress Cannot Subdelegate the Power to Make Law Through Adjudication

As we explain in more detail in Part II,20See infra Part II. the (now-repealed) Public Utility Holding Company Act (PUHCA) of 193521Pub. L. No. 74-333, ch. 687, 49 Stat. 803 (1935). For an overview of the Act’s drafting and passage, see A.C. Pritchard and Robert B. Thompson, Securities Law and the New Deal Justices, 95 Va. L. Rev. 841, 862-68 (2009). gave broad powers to the newly created Securities and Exchange Commission to reorganize utility holding companies in a fashion that was “fair and equitable to the persons affected.”22Id. § 11(e). Reorganizations could take place “[i]n accordance with such rules and regulations or order as the Commission may deem necessary or appropriate in the public interest or for the protection of investors.”23Id. (emphasis added). On its face, the statute gave equal adjudicative and rulemaking power to the SEC, and no one has ever argued otherwise. The agency used that adjudicative power to approve a reorganization plan for a utility holding company primarily owned by the Chenerys; the approved plan denied the Chenerys an ownership stake in the reorganized company by refusing to allow stock purchased by the Chenerys during reorganization negotiations to participate in the newly formed company on an equal footing with other stock of the same class.24The details of the case were, of course, much more involved. See infra Part II. The agency initially claimed that it was simply applying settled equity case law, but in Chenery I the Supreme Court roundly disagreed.25See Chenery I, at 87-89. For whatever it is worth, one of us has read all the equity cases relied upon by the SEC and agrees with the Court that those cases did not remotely support the agency’s broad proposition about the law regarding management stock transactions during reorganizations circa 1935. Not even close. The agency then re-entered its order, this time relying on its expertise and ability to craft new standards of management conduct that go beyond what common law, equity, regulations, or statutory law had previously imposed.26See In re Federal Water Service Corp., 18 S.E.C. 231 (1945). The Court in Chenery II upheld that exercise of authority,27See Chenery II, 332 U.S. at 201-03. thus allowing the agency to create new standards of conduct through adjudication.

There is very little in this scheme that is atypical of post-New Deal agency authorizations. The essentials could be replicated across many agencies.

Our central constitutional claim is that Congress cannot authorize agencies in such rights-denying adjudications to create new standards of conduct, which is precisely what the Chenery II case permitted, rather than to apply preexisting legal norms. The reason is simple: Congress itself has no power to engage in such activity and thus cannot subdelegate that power to agencies even if Congress may generally subdelegate some portion of its legislative powers. Nor do agencies have any inherent power to create law—through any procedural format. Even if Congress can subdelegate some measure of its lawmaking power to agencies, that would only authorize agency rulemaking within the scope of that subdelegation. It would not validate agency lawmaking through adjudication, and it would not support agency power to supplant the role of the courts in effectuating deprivations of life, liberty, or property. These conclusions are obviously counter to much of current practice and doctrine, including the holding in Chenery II, but they follow naturally from basic features of the U.S. Constitution. Bear with us as we lay out some fundamentals that are necessary for understanding this point.

Start with the simple civics-book model of constitutional structure. The Constitution’s three vesting clauses, at the beginning of each of the first three articles, identify three specific kinds of power possessed by various federal actors: (1) legislative power, some aspects of which are vested in a Congress consisting of a House and Senate28See U.S. Const. art. I, § 1 (“All legislative Powers herein granted shall be vested in a Congress of the United States, which shall consist of a Senate and House of Representatives.”). and some aspects of which are not granted to any federal institution or actor;29One potential source of that lacuna in legislative powers is the language in Article I, Section 1 referring to “legislative Powers herein granted,” U.S. Const. art. I, § 1, cl. 1 (emphasis added) rather than to the entire conceptual category of legislative powers. It is not clear that this language alone is sufficient to establish that Congress lacks general legislative powers. See Richard Primus, Herein of “Herein Granted”: Why Article I’s Vesting Clause Does Not Support the Doctrine of Enumerated Powers, 35 Const. Commentary 301 (2020). A complete argument would look also to the overall structure of the Constitution, its character as a fiduciary instrument, and the Tenth Amendment. See U.S. Const. amend. X (referring to “powers not delegated to the United States”). Taking into account all of these sources, it is not surprising that Chief Justice John Marshall could say without controversy that the federal government “is acknowledged by all, to be one of enumerated powers,” McCulloch v. Maryland, 17 U.S. (4 Wheat.) 316, 405 (1819). (2) executive power, which is vested in the person of the president;30U.S. Const. art. II, § 1, cl. 1 (“The executive Power shall be vested in a President of the United States”). By saying that this power is vested in the person of the president, we are endorsing some form of the so-called “unitary executive” thesis. The form that locates all executive power in the president is mandated by the text of the Article II vesting clause, as well as by a wide range of other textual and structural clues. See Steven G. Calabresi, The Vesting Clauses as Power Grants, 88 Nw. U.L. Rev. 1377 (1994); Gary Lawson & Guy Seidman, The Jeffersonian Treaty Clause, 2006 U. Ill. L. Rev. 1, 22-43. It is another matter entirely, however, how that unitary presidential power gets applied to control exercises of power by subordinate executive officials, the extent to which the power is internally delegable by the president, and substantively how far the “executive Power” extends. We have no occasion here to comment on any of those long-debated controversies. For some preliminary ruminations, see Gary Lawson, Command and Control: Operationalizing the Unitary Executive, ___ Fordham L. Rev. ___ (2023) (forthcoming). and (3) judicial power, which is vested in federal judges31See U.S. Const. art. III, § 1 (“The judicial Power of the United States, shall be vested in one supreme Court, and in such inferior Courts as the Congress may from time to time ordain and establish.”). who, by constitutional command, have tenure during good behavior and guarantees against diminishment in salary while in office.32See id. (“The Judges, both of the supreme and inferior Courts, shall hold their Offices during good Behaviour, and shall, at stated Times, receive for their Services, a Compensation, which shall not be diminished during their Continuance in Office.”). The Constitution’s basic structure describes not just the powers of the federal government but also which specific institutions and actors within that government are capable of exercising those powers. In addition, each designated institution or actor can act only through the forms and procedures prescribed by the Constitution. Thus, every federal power must be exercised by specific actors through specific mechanisms and forms. This seemingly small point turns out to have very large consequences.

It is true that the Constitution does not contain an express separation-of-powers clause akin to those found in early state constitutions, such as those of Virginia,33See Va. Const. of 1776, § 3 (“The legislative, executive, and judicial departments, shall be distinct, so that neither exercise the powers properly belonging to the other”). Georgia,34See Ga. Const. of 1777, art. I (containing the same language found in the Virginia Constitution of 1776). and (to this day) Massachusetts, which boldly proclaim things like:

[T]he legislative department shall never exercise the executive and judicial powers, or either of them: The executive shall never exercise the legislative and judicial powers, or either of them: The judicial shall never exercise the legislative and executive powers, or either of them: To the end it may be a government of laws and not of men.35Mass. Const, pt. I, art. 30.

But the U.S. Constitution achieves the same effect as those express clauses through the enumeration of powers, and the enumerations of forms and mechanisms for action, that are allocated to particular institutions. Thus,

Congress cannot exercise executive power because Congress is not granted executive power by the Constitution; that power goes to the President in Article II. The President and the courts cannot exercise legislative power because “[a]ll legislative Powers herein granted” are vested in Congress. Congress and the President cannot exercise judicial power because “[t]he judicial Power of the United States” is vested in federal judges with tenure during good behavior and guarantees against diminishment of salary while in office and is not vested in Congress or the President. The Constitution did not need an express “separation of powers” clause because the scheme of enumerated institutional power secures that separation by giving to each institution and actor only a certain subset of the total mass of potential governmental powers. That is why when James Madison tried to introduce an express separation-of-powers clause as part of the Bill of Rights, see 1 Annals of Cong. 453 (1789), it was rejected as “altogether unnecessary, inasmuch as the constitution assigned the business of each branch of the Government to a separate department.” Id. at 789 (statement of Rep. Sherman).36Steven Gow Calabresi & Gary Lawson, The U.S. Constitution: Creation, Reconstruction, the Progressives, and the Modern Era 144-45 (2020).

This kind of scheme—whether instantiated through an enumeration of powers, an express separation-of-powers clause, or both—only works if terms such as “legislative power,” “executive power,” and “judicial power” have ascertainable meanings. The Constitution does not define those terms, nor did any of the founding-era state constitutions, including those with express separation-of-powers clauses. The fuzziness of these classifications was well understood in the founding era, as reflected in Madison’s famous observation that “no skill in the science of government has yet been able to discriminate and define, with sufficient certainty, its three great provinces: the legislative, executive, and judiciary. . . . Questions daily occur in the course of practice which prove the obscurity which reigns in these subjects, and which puzzle the greatest adepts in political science.”37The Federalist No. 37 (James Madison). Nonetheless, 18th-century constitutional drafters uniformly chose to employ those terms as the central concepts in constitutional design, on the assumption that, at least over a significant range of cases, the terms had sufficiently clear and distinctive meanings to allow the machinery of government to move forward. In 1825, Chief Justice John Marshall could say that “[t]he difference between the departments undoubtedly is, that the legislature makes, the executive executes, and the judiciary construes the law,”38Wayman v. Southard, 23 U.S. (10 Wheat.) 1, 42-43 (1825). See also, e.g., 4 The Works of John Adams 579 (Charles Francis Adams ed., 1851) (“three branches of power have an unalterable foundation in nature; . . . the legislative and executive authorities are naturally distinct”). although fixing those boundaries is often “a delicate and difficult inquiry.”3923 U.S. (10 Wheat.) at 43. And nearly a century later, in holding that a state commission identified in its state constitution as a judicial body nonetheless was exercising legislative rather than judicial power when it set railroad rates, the Court said:

But we think it equally plain that the proceedings drawn in question here are legislative in their nature, and none the less so that they have taken place with a body which, at another moment, or in its principal or dominant aspect, is a court. . . . A judicial inquiry investigates, declares, and enforces liabilities as they stand on present or past facts and under laws supposed already to exist. That is its purpose and end. Legislation, on the other hand, looks to the future and changes existing conditions by making a new rule, to be applied thereafter to all or some part of those subject to its power. The establishment of a rate is the making of a rule for the future, and therefore is an act legislative, not judicial, in kind.40Prentis v. Atlantic Coast Line Co., 211 U.S. 210, 226 (1908).

The categories of governmental power may puzzle adepts (and modern law professors and political scientists), but they are foundational to the entire American scheme of government. If one is not trying to give those categories some meaning, one is abandoning the enterprise of constitutional interpretation.

A few profound things follow from these seemingly banal observations about basic constitutional design. First, and most importantly, the “executive Power” is, as the name suggests, a power to execute or implement existing laws.41See Saikrishna Prakash, The Essential Meaning of Executive Power, 2003 U. Ill. L. Rev. 701. It is not a power to make new law. There are, to be sure, limited contexts in which the president can function as an actual lawmaker, but they involve wartime exigencies and extend, in accordance with norms of international law, only to occupied territory, not to the United States proper.42See Gary Lawson & Guy Seidman, The Constitution of Empire: Territorial Expansion & American Legal History 151-52 (2004). This means that executive agents have no baseline or inherent authority to make law—even if they divide themselves into bicameral bodies and present their proposals to the president and even if they dress up in robes and follow most of the procedures used by courts. The Supreme Court got at least that one right in Youngstown Sheet & Tube Co. v. Sawyer.43343 U.S. 579 (1952). The Securities and Exchange Commission, being neither Congress nor a federal court, is by default an executive institution and thus has no inherent lawmaking power.44Whether federal courts have anything that can be called inherent lawmaking power depends on how one understands the “judicial Power.” Fortunately, we do not need to go down that rabbit hole here, as agencies are not courts. It has executive power (subject to presidential control45As noted above, see supra note 30, the precise forms of presidential control—removal, veto, or direct decision-making—are the subjects of long and inconclusive debates, which we avoid here. The language of the Article II vesting clause mandates some form of presidential control over agency action, but we need not worry here about what form that might be.) and nothing more.

Second, executive agents also do not have, and thus cannot exercise, the “judicial Power.” The quintessential use of “judicial Power” involves, as James Wilson put it, “applying, according to the principles of right and justice, the constitution and laws to facts and transactions in cases, in which the manner or principles of this application are disputed by the parties interested in them.”461 The Works of James Wilson 296 (Robert Green McCloskey ed, 1967). Of course, when put so abstractly, Wilson’s account describes a good portion of executive activity as well, including everything from prosecutorial decisions to adjudication of benefits claims, all of which involve applying law to facts. This close connection between judicial and executive activity is not surprising, as for most of English legal history the courts were formally an arm of the executive.47See Lawson, supra note 15, at 619-20. The identification of “judicial Power” as something distinct from “executive Power” is really an 18th-century innovation. Nonetheless, a constitution that separates those powers obviously has something in mind as a distinctively judicial function; surely, for instance, the attorney general cannot preside over the trial of a criminal defendant.

The line between executing and judging gets crossed when executive agents try to apply law to facts in order to deprive someone of life, liberty, or property. This limitation on the scope of executive authority is an idea that has roots in Magna Carta,48See id. at 618-21. and it is the driving force behind the whole concept of due process (or due course) of law: Executive agents cannot unilaterally deprive people of life, liberty, or property. They can only seek such deprivations in accordance with law, and the final adjudications of those deprivations must run through the courts. That is why principles of due process of law, as we have been capaciously using that term, are intimately connected with the separation of powers.49See Nathan S. Chapman & Michael W. McConnell, Due Process as Separation of Powers, 121 Yale J.J. 1672 (2012). Principles of due process of law are first and foremost constraints on what kinds of actions executive agents can take.

Third, Congress, which wields “[a]ll [federal] legislative Powers herein granted,” can add to or subtract from the substantive law that authorizes executive agents to pursue deprivations and authorizes courts to finalize them. The legislative power is just the power to prescribe rules of conduct, within the constraints of substance and form that define the powers of the legislature. Those federal “legislative Powers” operate against a backdrop of general law, the law of nations, natural law, and other sources of law, but they are capable of creating rules of law beyond those other sources. For example, in the mid-1930s, there was no principle of general law that categorically forbade corporate officials from trading in the stock of their companies during reorganization negotiations. Nor was there any other source of law that categorically forbade such transactions. But if Congress had passed a statute specifically declaring that corporate officials in reorganization negotiations were forbidden from trading stock in their companies while those negotiations were ongoing, whatever constitutional provisions authorized Congress to enact PUHCA would presumably have authorized as well such a provision dealing with stock purchases.50We do not address here whether there is any actual constitutional authorization for the Public Utility Holding Company Act. A statute to that effect would create, out of thin air, a standard of conduct that did not previously exist. Legislatures can do that sort of thing even when executive agents cannot.

Fourth, Congress can only create new standards of conduct for people by enacting legislation pursuant to the Constitution’s prescribed lawmaking process in Article I, Section 7. Those procedures of bicameralism and presentment define what it takes to make something “a Law;”51U.S. Const. art. I, § 7, cl. 2. Congress cannot make a law through other means. It can impose legal consequences of sorts through adjudication in cases of impeachment,52See id. art. I, § 2, cl. 5; id. art. I, § 3, cl. 6. and while established practice allows Congress to issue subpoenas through procedures other than Article I, Section 7 lawmaking53At least one of us thinks this practice is unconstitutional. See Gary Lawson, Burning Down the House (and Senate): A Presentment Requirement for Legislative Subpoenas under the Orders, Resolutions, and Votes Clause, 83 Tex. L. Rev. 1373 (2005). and venerable (if mistaken) precedent allows it to hold people in contempt to maintain the integrity of its proceedings,54See Anderson v. Dunn, 19 U.S. (6 Wheat.) 204 (1821). these are limited and specific exceptions to the general rule about how Congress must exercise its powers.

Fifth, and perhaps most significantly, when Congress exercises its formal legislative power, there are constitutional limits on the ways in which it can impose obligations. The bill of attainder and ex post facto clause55U.S. Const. art. I, § 9, cl. 3 (“No Bill of Attainder or ex post facto Law shall be passed.”). For discussion of this clause, see Calabresi & Lawson, supra note 36, at 308-28. imposes some level of generality on legislation that imposes burdens.56Private bills are different because they confer benefits rather than impose burdens or obligations. Laws cannot be targeted at specific individuals or (at least in the criminal context) behavior that has already occurred.57We do not address here the long-lived debate about whether the ex post facto clause applies only to criminal laws. A full discussion of the Constitution’s rules on generality of lawmaking would require a separate article or book58One of us has postponed that project for more than two decades now. See Gary Lawson, Controlling Precedent: Congressional Regulation of Judicial Decision-Making, 18 Const. Commentary 191, 208 (2001).; for the present, it is enough to note that some such set of rules exists.

The combined effect of these foregoing principles is that Congress cannot authorize the SEC, or any other executive agency, to make new law through adjudication.

Let us assume that Congress has some power to subdelegate legislative power to administrative agencies. The only power that it can subdelegate is the power that it has, which is legislative power. Congress, outside of the narrow exceptions noted above, has no adjudicatory power. It cannot subdelegate what it does not have. If any kind of subdelegation is permissible, it can only be subdelegation of lawmaking power—the power to promulgate the kinds of norms that Congress would promulgate through legislation. The adjudicatory power of agencies comes from the fact that the “executive Power” includes powers of adjudication, but that executive power of adjudication does not extend to lawmaking that results in deprivations of life, liberty, or property. Agencies cannot receive power to make adjudicatory law from Congress, and agencies do not possess adjudicatory power to make law from their executive power.

To be sure, the case that we are laying out for this claim is not a deductively airtight argument. It cannot be such an argument, because no matter what procedures agencies employ to make rules, they will not be complying with Article I, Section 7, and thus will not literally be producing what the Constitution considers “a Law.” Congress, when subdelegating, conveys to agencies the power to promulgate binding norms of conduct, but it cannot convey to agencies the power to do so using the procedures set out by the Constitution for valid congressional action. The question is whether the act of subdelegation requires that Congress tie the conveyance of power to modes of proceeding that most closely resemble, given the realities of agency governance, the constitutionally prescribed lawmaking process, and whether exercises of that power by agencies must conform to the norms of generality required by the Constitution for legislation. If so, and if all agency action with legal consequences falls within either rulemaking or adjudication, agencies can only constitutionally make law through promulgation of general norms (i.e., through rulemaking), which is precisely the proposition that the strong language in Chenery II rejected.

b. Rulemaking, Adjudication, and the Nature of Agency Action

Thus, the question is whether all of the foregoing premises are true. Does all agency action with legal consequences fall under either rulemaking or adjudication, and does the logic of subdelegation require that agencies receiving congressional lawmaking power exercise it through the former?

Neither premise is self-evidently true, but we think a soft case can be made for both. Start with the rulemaking/adjudication distinction. This distinction is not articulated in the Constitution because the Constitution does not address the forms of executive action. It grants “[t]he executive Power” to the president but says nothing specific about the appropriate means for exercising that power, beyond the injunction to “take Care that the Laws be faithfully executed.”59U.S. Const. art. II, § 3. Nor does it say anything about how subordinate executive institutions must do their jobs. Congress can presumably provide some procedural structure through its power to make “all Laws which shall be necessary and proper for carrying into Execution”60Id. art. I, § 8, cl. 18. executive power, but that is all that the Constitution has to say on the subject.

But, of course, the constitutional structure was not created in a vacuum. Many of the terms used in the Constitution had well-developed histories that flesh out their content. For example, the Constitution provides no structure of note on how to exercise the “judicial Power.” Does that mean that courts in 1788, 1789, 1790, and 1791 (remember that there was no Bill of Rights until December 15, 1791) could hold secret ex parte trials without notice? Of course not. There were understood background forms for exercising the “judicial Power”—as Congress well recognized when it passed the Judiciary Act of 1789 that made multiple references to those background understandings about how courts functioned.61See Lawson, supra note 15, at 630 (describing the founding-era statutory references to background understandings about judicial power). “In the founding era, there was no need to specify in detail precisely how federal courts were to carry out their constitutionally vested function. Everyone knew what a judicial process looked like.”62Id.

There was less background structure to the concept of executive power, but the rulemaking/adjudication distinction was implicit in the operations of government. All executive action can be mapped along several dimensions, such as its level of generality and its prospective or retrospective operation. These are not neatly defined concepts, and they both operate along continuums rather than on/off switches. But some actions are highly specific and retrospective, while others are highly general and forward-looking. Granting benefits to a Revolutionary War veteran under a statute defining eligibility for such benefits is both particular and retrospective, as it gives effect to past events. Specifying forms for applications for those benefits is both general and prospective; it applies to all applicants going forward. Granting a particular applicant a license to trade with American Indian tribes is particularized but forward-looking. Nothing in the nature of things dictates that these various actions be labelled “rulemaking” or “adjudication.” But when 20th-century scholars tried to articulate that distinction for the modern administrative state,63See, e.g., John Dickinson, Administrative Justice and the Supremacy of Law in the United States 16-21 (1927); Ralph F. Fuchs, Procedure in Administrative Rule-Making, 52 Harv. L. Rev. 259, 260-65 (1938). they could look back on past practices to see the continuum of agency action and how certain activities looked more like what legislatures did and others looked more like what courts did. Relationships were there for centuries even if people at the time did not name them. When Congress got around to legislating formal definitions of rulemaking and adjudication in the Administrative Procedure Act in 1946,645 U.S.C. §§ 551(4)-(9) (2018). it had a long history to draw upon (and at least partially ignore65See Ronald M. Levin, The Case for (Finally) Fixing the APA’s Definition of “Rule,” 56 Admin. L. Rev. 1077 (2004).).

The more difficult question is whether the logic of subdelegation requires that agency lawmaking take a general and prospective form. It is not true that all legislation is general and prospective. Much congressional legislation takes (and took in the founding era) the form of private bills.66See Note, Private Bills in Congress, 79 Harv. L. Rev. 1684 (1966). Sometimes legislation has retroactive effect, though such effect is disfavored.67See Landgraf v. USI Film Products, 511 U.S. 244 (1994). So it is not true that Congress only has power to enact general and prospective norms. The case against agency power to make law through adjudications is subtler than that. It draws on all of the considerations presented thus far.

Agencies, as executive actors, have no inherent power to make law; if they have such power, it must stem from congressional subdelegation. And Congress can only subdelegate what it has. In ascertaining what and how Congress can subdelegate to agencies, one key consideration is the difference between congressional action that provides benefits and congressional action that imposes burdens that deprive people of life, liberty, or property. The former can be particularized and retrospective (private bills that operate on past events), while the latter must be generalized and prospective (enforced at least in part through the explicit prohibitions on ex post facto laws and bills of attainder and in part through immanent principles of due process or course of law). The nature of subdelegation in the context of the Constitution requires agencies to adhere to similar constraints.

Matters could have been otherwise. The Constitution, as was typical of fiduciary instruments in the 18th century,68On the use of preambles in 18th-century fiduciary instruments, see Gary Lawson & Guy Seidman, “A Great Power of Attorney”: Understanding the Fiduciary Constitution 20-23 (2017). identifies the purposes behind its delegations of power to various agents. Those purposes are identified in the Preamble as “to form a more perfect Union, establish Justice, insure domestic Tranquility, provide for the common defense, promote the general Welfare, and secure the Blessings of Liberty to ourselves and our Posterity.”69U.S. Const. Preamble. Suppose that the Constitution then said something like, “To these ends, the Congress may exercise the following powers,” and then listed various powers, with nothing more, including no specification of the mode by which Congress must make law. Presumably, Congress would have free choice of means and forms to carry those purposes into effect. To the extent that Congress was permitted to subdelegate some portion of that authority, there would be no obvious limitation on the means and forms that subagents could use pursuant to those subdelegations.

The Constitution, however, does not simply identify purposes and powers. It also identifies specific means through which those purposes can be pursued and the powers executed: The Article I, Section 7 process for legislation, along with the requirements of prospectivity and generality built into the instrument through the bill of attainder and ex post facto clause. The document specifies how Congress must act. The Supreme Court got at least this much right in INS v. Chadha.70462 U.S. 919 (1983).

Accordingly, if Congress subdelegates power, that subdelegation takes place not just in the shadow of the purposes and powers granted to Congress but also of the means and forms through which those purposes and powers are effectuated. To the extent that Congress is limited to enacting general and prospective rules that deprive people of life, liberty, or property, Congress’s subagents are subject to the same requirements. Surely Congress cannot grant to agencies more power than Congress itself possesses. Agencies, of course, cannot follow precisely the forms and means of lawmaking employed by Congress because they cannot precisely duplicate the Article I, Section 7 process. But they can emulate the substance of those forms and procedures by confining lawmaking to general and prospective rules. The inner morality of constitutional subdelegation demands nothing less.

Strands of this argument, as we discuss in the subsequent Section, can be found in the Chenery proceedings, though the doctrine of the time made a direct argument along these lines impossible. Again, we do not argue here that this amounts to a knock-down claim that agency lawmaking through adjudication is categorically unconstitutional. Rather, we suggest only that there are sufficient questions about the constitutionality of the practice to warrant caution before concluding that agencies have a free hand to choose how to make law.

c. Principles of Due Process of Law Do Not Counsel Against Restricting Agency Lawmaking to Rulemaking

The idea that agencies can only make law through rulemaking has an odd ring to modern ears, especially in view of recent suggestions that agency rulemaking is precisely the problem with subdelegation.71See Dep’t of Transportation v. Ass’n of Am. R.Rs., 575 U.S. 43, 66 (2015) (Thomas, J., concurring in the judgment). Nicholas Parrillo wrote a lengthy article rebutting the suggestion that all rulemaking is unconstitutional. See Nicholas R. Parrillo, A Critical Assessment of the Originalist Case Against Administrative Regulatory Power: New Evidence from the Federal Tax on Private Real Estate in the 1790s, 130 Yale L.J. 1288 (2021). Rulemakings typically involve fairly modest procedures of public notice and comment.72See, e.g., 5 U.S.C. § 553 (2018). Numerous organic statutes contain comparable provisions for public notice and comment. Affected parties in rulemakings generally have no right to individualized presentations, discovery, or cross-examination.73They have such rights only in so-called formal rulemakings—which in the modern world is close to a null set. See Lawson, supra note 2, at 382-83. Many adjudications are even less procedurally robust, but adjudications that deprive people of rights are usually highly formalized, approaching in most respects the kind of procedures that one would see in court (minus an Article III judge, a jury, and perhaps stylized rules of evidence). It is standard doctrine that the Fifth and Fourteenth Amendment due process of law clauses require no specific procedures in agency rulemaking but do require some measure of procedures for adjudications that deprive people of life, liberty, or property.74The doctrine is typically traced to Londoner v. City and County of Denver, 210 U.S. 373 (1908), and Bi-Metallic Investment Co. v. State Board of Equalization of Colorado, 239 U.S. 441 (1915). Isn’t agency lawmaking through adjudication better, from a due process of law perspective, than agency lawmaking through rulemaking?

If one is looking only at the kinds of procedures that agencies are required to employ, then yes, the class of adjudications that deprive people of life, liberty, or property is likely to be, on the whole, more procedurally robust for individually affected parties than rulemakings. For purposes of understanding the constitutional framework of Chenery II, however, none of this is relevant. As an original matter, due process of law has nothing—or at least almost nothing—to do with agency procedures.

The focus in due process of law doctrine on agency procedures is a post-founding development, resulting from the Fourteenth Amendment’s extension of “due process of law” to state and local governments and the resulting desire of courts to have a unified body of doctrine that applies equally to state and federal actors.75For the long version of the story, see Lawson, supra note 15. None of that was part of the original meaning of due process of law as a concept in 1788 or “due process of law” as a term in the Fifth Amendment in 1791. Due process of law, as originally understood, was about substance, not about procedure. “It concerns what the ‘executive Power’ can do, not how or by what procedures it can do it.”76Id. at 626.

This means that when we speak of agency “rulemaking,” we are not talking about a procedural mode of the sort defined in the Administrative Procedure Act. We are speaking of a substantive action that involves promulgation of general and prospective norms, and adjudication in this sense means application of existing norms to a specific set of facts. Essentially by definition, adjudication does not involve lawmaking; it involves law application. As a functional matter, interpretation of a set of preexisting norms can sometimes yield an outcome that was not foreordained, and in that respect adjudication can involve an element of “lawmaking,” but it is interstitial and incremental lawmaking. By the time of Chenery II, no one thought that the SEC decision regarding the Chenerys was of this incremental character. Everyone understood that the agency was creating a new norm of conduct out of thin air.

Under principles of due process of law, the legislature has power to create those new norms. If some measure of subdelegation is permissible, some of that creative lawmaking power can be exercised by executive agents. How much can be exercised is determined by the contours of the subdelegation doctrine, not by principles of due process of law.

We are only saying here that principles of due process of law do not work counter to those subdelegation principles by pushing agency lawmaking into adjudication. If the structure of subdelegation indicates that agency lawmaking must take the form of general and prospective norms, nothing in due process of law says otherwise, because Congress will have overcome the inherent limitations on the executive power by authorizing a limited exercise of legislative power. Due process of law has done its job once it has prescribed those inherent limits on the executive power. The law of subdelegation, not due process of law, then determines to what extent Congress can authorize agencies to go beyond those inherent limits.

With these considerations in mind, we can now re-examine the Chenery cases. When studied closely, they depart from the baseline constitutional framework less than modern convention suggests. Or, at the very least, they can be so read.

Part II: “One of the Worst Ever” Decisions

Congress enacted PUHCA in 1935 to regulate the structure and activities of public utility holding companies. These companies were enormously profitable during the first decades of the 20th century as electric power generation emerged as a significant portion of the American economy.77See Roy A. Schotland, A Sporting Proposition—SEC v. Chenery, in Administrative Law Stories 169-70 (Peter L. Strauss ed., 2005). Simplifying these companies’ complicated activities and structures was a primary goal of PUHCA. Consequently, PUHCA contained a “death sentence” provision that required all public utility holding companies to reorganize on terms that the SEC deemed to “fairly and equitably” distribute the voting power of stockholders.78See id. at 173.

It was this reorganization provision that led to the Chenery cases. Christopher Chenery, founder and President of the Federal Water Service Company, filed the company’s reorganization plan with the SEC in November 1937. The SEC rejected Chenery’s reorganization plan, as it did with many other plans, on the grounds that it did not fairly and equitably distribute shareholders’ voting power.79See id. at 176. Chenery submitted amended reorganization plans to the SEC several times, each one rejected on the same grounds. This process occurred over several years.

While submitting his various reorganization plans to the SEC, Chenery also added to his holdings of Federal’s preferred stock. The SEC had made it clear to Chenery that it would only approve plans in which preferred stockholders, who were owed massive back dividends, received essentially all ownership of the new company. Chenery made his stock purchases on the open market and in compliance with the very modest legal regulations of insider trading in effect at the time. As required by law, he reported all of his stock purchases to the SEC and did not sell any stock. When the insider trading division of the SEC notified the PUHCA division of Chenery’s purchases, the Commission arranged a meeting with Chenery.80See id. at 176-77. Chenery explained that he was merely trying to protect his controlling interest in the company—another party had been buying shares of Federal’s preferred stock, which threatened Chenery’s control of the company.

While the SEC accepted Chenery’s legitimate reasons for purchasing preferred stock, the agency explained that it could not allow insiders to purchase stock during the PUHCA reorganizations. Other parties similarly situated may be taking advantage of their insider knowledge by buying during reorganization, even if Chenery was not doing so.81See id. at 177. On the advice of his attorney, Chenery rejected an SEC proposal to sell his shares while retaining control over them. The SEC rejected Chenery’s final reorganization plan in 1940, and stated that it would refuse to approve any plan in which “the preferred stock acquired by [Chenery] would be permitted to share on a parity with other preferred stock.”82Federal Water Service Corp., 8 S.E.C. 893 (1941). In support of its decision, SEC claimed that Chenery was a fiduciary “under a ‘duty of fair dealing’ not to trade in the securities of the corporation while plans for its reorganization were before the Commission.”83Chenery I, 318 U.S. at 85. Chenery challenged the SEC-amended plan in the D.C. Circuit Court of Appeals.

a. Chenery I

In the first Chenery decision in the court of appeals, the D.C. Circuit (with future Chief Justice Fred Vinson then a member of the D.C. Circuit) made clear that the agency could act to deprive the Chenerys of their property interest in preferred stock only if there was preexisting law to support the action:

[I]f we are brought to conclude that there was at the time no regulation of the Commission, no provision of the statute, and no rule of common law or equity prohibiting the purchase of stock by an officer or director of a corporation during the pendency of the proceedings we are concerned with . . . , it would seem logically to follow that the action of the Commission in applying the rule it did was an erroneous assumption of power and an invasion of the legislative field.84Chenery Corp. v. SEC, 128 F.2d 303, 306-07 (D.C. Cir. 1942).

After finding no support for the SEC’s actions in either statute, case law, or regulation, the court concluded:

[I]f the Commission’s objective is to be attained, it should be only after the pros and cons have been carefully weighed in their relation, respectively, to the dangers and the benefits, and the scales should be controlled by Congress and not by the Commission. In short, all that we hold is that this vital question of policy is one for the Congress and not for the Commission. Until Congress acts to change the standard it has expressly set up in the Act, action by the Commission to expand or enlarge its terms, and to make such expansion or enlargement apply to transactions three years old, is we think, with great deference to the Commission, neither more nor less than retrospective legislation.85Id. at 311. Judge Miller, in dissent, thought the majority was unduly interfering with the Commission’s broad discretion. See id. at 315-16 (Miller, J., dissenting).

The case did not address the Commission’s power to enact a rule with prospective effect. It said only that the agency, in an adjudication, could only deprive the Chenerys of property in accordance with preexisting law of some kind.

The bulk of the government’s brief to the Supreme Court defended the agency’s decision on the basis of existing law, contending that the action was “in accord with established principles”86Brief for the Securities and Exchange Commission, SEC v. Chenery Corp., No. 254, at 37. that have “long been settled.”87Id. at 35. At the end of the brief, the SEC argued that it had great discretion in the application of the PUHCA,88See id. at 39-45. but it stopped short of saying that it could create new standards of conduct in proceedings under Section 7 of the statute. Instead, it said that its “conclusion that the principles relating to the powers and duties of reorganization managers were applicable in the situation before it was peculiarly within the Commission’s special administrative competence.”89Id. at 39. That is not quite saying that the Commission can make law in its adjudicatory proceedings.

The Chenerys, for their part, openly argued that only general and prospective action could limit their rights: “Only a legislature can adequately determine, after hearings on the subject, whether . . . , in the absence of fraud or inequity, officers and directors should ever be precluded from purchasing stock, and if so when—whether during reorganization, pending reorganization, or always. . . . Such legislative action obviously would be prospective in its operation.”90Brief for the Respondents Chenery Corporation, H. M. Erskine, R. H. Neilson, et al., SEC v. Chenery Corp., No 254, at 11. They did not claim that such action would be unconstitutional, but only that “[f]or an administrative body to formulate a new principle and then to apply it retroactively to transactions which had been reported to that body and which had not been objected to is a harsh procedure”91Id. at 30. and that Congress had fully dealt with issues of corporate insider trading elsewhere in the Public Utility Holding Company Act.92See id. at 27-30.

The first Chenery decision, typically referred to as Chenery I, was handed down in February 1943. Only seven of the justices participated. Justice William Douglas led the Commission as chairperson during the Chenery proceedings, and thus he did not participate in either decision. Justice Wiley Rutledge was confirmed later in February 1943 to replace Justice James Byrnes, but this left the Court with only seven votes to decide the first Chenery case.

The Court’s decision was 4-3, with Justice Felix Frankfurter writing for the majority. Frankfurter’s opinion was unclear as to exactly why the SEC’s order was unlawful. “To ascertain the precise basis of [SEC’s] determination, we must look to the Commission’s opinion,” he reasoned. The Commission’s decision rested on the conclusion that it was “merely applying ‘the broad equitable principles enunciated in the cases’”93Chenery I, 318 U.S. at 87. cited in the SEC’s order.

According to the SEC, in other words, it was merely applying existing and established principles of equity “derived from judicial decisions.”94Id. at 88. “If,” Frankfurter wrote, however, “the rule applied by the Commission is to be judged solely on the basis of its adherence to the principles of equity derived from judicial decisions, its order plainly cannot stand.”95Id. That is because, Frankfurter demonstrated, the judicial decisions that the SEC cited “do not establish principles of law and equity which in themselves are sufficient to sustain its order.”96Id. at 89.

Frankfurter left open the possibility that, perhaps, the Commission could find another argument to sustain its order against Chenery. However, Frankfurter concluded, “[s]ince the decision of the Commission was explicitly based upon the applicability of principles of equity announced by courts, its validity must likewise be judged on that basis.”97Id. at 87. See also id, at 89: “Since the Commission professed to decide the case before it according to settled judicial doctrines, its action must be judged by the standards which the Commission itself invoked.” In other words, Frankfurter signaled that he would be willing to consider the possibility that the SEC’s order could be sustained on other grounds, but the Court was required to consider only the grounds that the agency offered. This, of course, is what administrative lawyers know today as the Chenery I principle: “an administrative order cannot be upheld unless the grounds upon which the agency acted in exercising its powers were those upon which its action can be sustained.”98Id. at 95.

That was surely enough to decide the case, but the Court went on to give broad assurance that the SEC had ample power to create new law regarding the duties of corporate insiders: “In evolving standards of fairness and equity, the Commission is not bound by settled judicial precedents. Congress certainly did not mean to preclude the formulation by the Commission of standards expressing a more sensitive regard for what is right and what is wrong than those prevalent at the time the Public Utility Holding Company Act of 1935 became law.”99Id. at 89. The only catch is that it needed to do so through rulemaking rather than adjudication:

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. . . . 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.100Id. at 92-93.

The Court did not indicate whether this seeming requirement of rulemaking came from the Constitution, the statute, administrative common law, or some combination thereof. Three justices specifically denied that there was any such requirement: “The intimation is that the Commission can act only through general formulae rigidly adhered to. . . . But . . . the Act gives the Commission wide powers to evolve policy standards, and this may well be done case by case. . . .”101Id. at 99-100 (Black, J., dissenting).

In other words, Justice Frankfurter seemed to lay out two options for the SEC. The Commission could have simply applied the statute, which authorizes it to decide “under § 7(d)(6) of [PUHCA], whether the proposal was ‘detrimental to the public interest or the interest of investors or consumers,’ and, under § 11(e), whether it was ‘fair and equitable.’” These broad legal provisions, Frankfurter emphasized, already “confer upon the Commission broad powers for the protection of the public.”102Id. at 90. The SEC could, in short, have simply cited the statute’s broad provisions and rested its order on that basis. Its decision could simply have been an application and interpretation of the governing statute, subject to review as such.: “if the action is based upon a determination of law . . . the reviewing authority of the courts does come into play, an order may not stand if the agency has misconceived the law.”103Id. at 94.

Alternatively, Frankfurter suggested, the agency could have “promulgated a general rule of which its order here was a particular application.” In that instance, he granted, “the problem for our consideration would be very different,”104Id. at 92. as the Commission would be issuing orders that enforce a clear legal prescription. As A.C. Pritchard and Robert B. Thompson note, an earlier draft of Frankfurter’s opinion was more explicit about proceeding only after the promulgation of a general rule, perhaps at the urging of Chief Justice Harlan Stone.105See Pritchard & Thompson, supra note 21, at 897.

Justice Frankfurter was joined by Chief Justice Stone and Justices Owen Roberts and Jackson in the majority. Justice Jackson wrote to Frankfurter in January 1943 signaling that he may “decide to ride a separate horse” and write a concurring opinion.106Letter from Justice Robert Jackson to Justice Felix Frankfurter (Jan. 27, 1943). Three days later, he wrote Frankfurter again concluding that “a separate opinion by me would add nothing but words.”107Letter from Justice Robert Jackson to Justice Felix Frankfurter (Jan. 30, 1943). He did, however, offer a paragraph that might be added to Frankfurter’s majority opinion, most of which Frankfurter incorporated. The thrust of that paragraph was that if anyone had been wronged, “it would be against the stockholders from which they purchased and who thereby parted with their stock at less than its book value.”108Chenery I, 318 U.S. at 93. By ordering that Chenery surrender his shares, but not to return them to or compensate the previous shareholders, the Commission was implicitly acknowledging that nobody was harmed by Chenery’s purchases. In fact, Justice Jackson wrote to Frankfurter, if the previous shareholders “were selling under compulsion, the bids of these directors may well have sustained their market, and they may well have benefited therefrom as against the terms they must have accepted in the absence of such bids.”109Letter from Jackson, supra note 107.

A few days later, Justice Frankfurter made the same point to Justice Stanley Reed, who was voting with the dissent. As Pritchard and Thompson have explained, “Frankfurter had been courting Reed since his appointment to the Court,”110Pritchard & Thompson, supra note 21, at 898. and he and Jackson separately urged Reed to consider voting with the majority. Jackson explained that if the dissent were correct that Chenery’s group had opportunities to purchase stock that other stockholders did not have as a result of their insider status, “I should be prepared to reverse [the D.C. Circuit] in a paragraph. If the Commission anywhere in its opinion had stated anything that could be tortured into a finding of fact that Chenery & Co. utilized ‘their peculiar information’ in purchasing the preferred stock in controversy, there would be an end of the matter for me. The whole point is, as far as I am concerned, that the Commission does not say so, did not so find.”111Letter from Justice Robert Jackson to Justice Stanley Reed (Feb. 1, 1943). Frankfurter indicated he would be prepared to uphold the SEC’s order if it had been based on a finding of fact that Chenery’s stock purchases harmed other shareholders, thus rendering his reorganization plan unfair and inequitable under the terms of PUHCA. Without such a finding, the SEC’s decision appeared lawless.

Justice Reed’s response suggested a fair amount of agreement. As Reed explained, the SEC’s order was simply “[a]n application of the Commission’s idea as to ‘fair and equitable.’” If the Commission’s interpretation of the statute “does not go too far (arbitrary, capricious, and unreasonable), it should be upheld.”112Letter from Justice Stanley Reed to Justice Felix Frankfurter (Jan. 29, 1943). This is, of course, precisely what the Commission did not say, which was the primary reason the majority voted to overturn the order.

Jackson’s concurring opinion, which he wrote but never published, was more explicit about the need to rely on a preexisting rule rather than the bare terms of PUHCA to sustain the Commission’s order. Jackson looked at the statute and concluded that there was “no support for this rule of law in the Public Utility Holding Company Act. So far as it touches the subject [of stock buying while under reorganization] it seems to slant the other way. . . . No prohibition of such a purchase can be spelled out of anything Congress has provided.”113Robert Jackson, Chenery I concurring opinion draft, No. 254, October Term, 1942, at 3. If the SEC were to claim that its order was merely an interpretation of the underlying statute, Justice Jackson would likely still have voted against it.

However, even Justice Jackson accepted that if the SEC had “promulgated legislatively a rule that during reorganization proceedings directors and officers were forbidden to acquire stock . . . I should not as presently advised see doubt as to their validity.”114Id. at 4. The problem, of course, was that the Commission had not done so: “The rule the Commission proposes may well be a good rule, but no one can deny that it is an innovation. . . . Surprise law is sometimes inevitable, but it seems almost bromidic to say that citizens are entitled to have some way of learning the general principles that they will suffer in person or property for transgressing.”115Id. at 5-6. Jackson ultimately chose to withhold his opinion and sign on to Frankfurter’s, and Reed remained with the dissent.

b. Chenery II

Three years later the issue returned to the Court. By this time, the Court’s membership had changed significantly, and the case went the other way, for the Commission and against Chenery.

On remand, the agency reached precisely the same result as before without promulgating any new rule (which would presumably have only prospective effect and therefore would not apply to the Chenerys).116See In re Federal Water Service Corp., 18 S.E.C. 231 (1945). The D.C. Circuit again, this time rather nastily, cut down the agency, insisting that it could outlaw transactions such as the Chenerys’s stock purchases only through promulgation of a general rule.117See Chenery Corp. v. SEC, 154 F.2d 6, 12 (D.C. Cir. 1946): In nothing we have said do we wish to be understood as expressing any opinion as to the right of the Commission under its broad powers to promulgate a rule of general application forbidding officers and directors of a corporation in process of reorganization from buying—and perhaps also from selling—securities of the corporation during the pendency of proceedings before the Commission. That question is not present in this case. What we do say is that, without such a rule, of which notice is given so that all may know of its existence, transactions in themselves fair and just and honest and in accord with traditional business practices, and which ‘Congress itself did not proscribe,’ and which ‘judicial doctrines do not condemn,’ may not properly be ‘outlawed or denied’ their ordinary effect. The court located this requirement solely in the Supreme Court’s prior opinion,118See id. at 8-9. which, as noted above, did not identify the requirement’s legal source.

The Chenery II case thus squarely posed the question whether and how the SEC could promulgate and apply a new principle of law. The intervenors made this point succinctly: “If the [agency’s] policy is desirable, it can only be made effective through the exercise of the legislative power of the Commission.119Brief for Federal Water & Gas Corp., SEC v. Federal Water & Gas Corp., No. 82, at 7. The Chenerys similarly said that “any power in the Commission to resolve problems of policy is legislative in its nature and should be exercised by the promulgation of regulations, prospective in their operation, which will serve as a fair warning to such persons as may come within their provisions.”120Brief for the Respondents Chenery Corporation et al., SEC v. Chenery Corp., No. 81, at 19. While this might have seemed like the central question facing the Court in Chenery II, the SEC’s nearly 60-page-long brief devoted only a few paragraphs, spanning about four pages, to the question. The agency’s response was simple: While there might be circumstances where proceeding by prospective rule is a good idea,121See Brief for the Securities and Exchange Commission, SEC v. Federal Water & Gas Corp., Nos. 81 & 82, at 57 n.25. requiring agencies to make law only through rulemaking “would, we submit, substantially impair and, indeed, hamstring the work of every administrative agency with power to make general rules as well as specific orders. . . . Such a requirement for the adoption of general rules in advance of every step forward in the agency’s effectuation of statutory policies would go far to defeat the intention of Congress of promoting flexible administrative machinery for the very purpose of allowing the agencies to use varied facilities to cope with the specialized problems before them.”122Id. at 53. In response to the charge that adjudicatory lawmaking leads to improper retroactivity, the agency answered that retroactivity is fine when it is “merely implementation of the statutory command”123Id. at 54. to ensure that reorganization plans are “fair and equitable.”

The Chenerys, at long last, brought due process into the case in their brief, arguing that “The Commission’s Order Amounts to Retroactive Legislation Depriving the Respondents of Property Without Due Process of Law.”124Brief for the Respondents Chenery Corporation et al., supra note 120, at 29. The argument did not draw on fundamentals of separation of powers but focused narrowly on the retroactive effect of the agency’s action.125See id. at 29-34. The closest the Chenerys came to grounding their claims in basic constitutional structure was one passage:

The Commission, as the body created by Congress to administer the Act, has a further power not granted to courts, the power to promulgate regulations of general application. . . .

But such delegated power to resolve problems of policy is legislative. Where a new standard is to be created by the Commission, ordinary justice as well as a reasonable interpretation of the statute requires that such a standard be put into effect by a regulation prospective in its operation, so that it will serve as a fair warning to such persons as may come within its provisions.126Id. at 31.

The argument was not developed further.

The Court released its decision June 23, 1947, late in the term. Justice Frank Murphy, who had dissented in Chenery I, wrote the majority opinion, joined by Justices Hugo Black and Reed, who had also dissented in Chenery I, along with Justice Rutledge. Justice Harold Burton concurred in the result but did not join Justice Murphy’s opinion.127Chenery II, 332 U.S. at 209. Justices Frankfurter and Jackson dissented, and the remaining two justices (Douglas and Chief Justice Vinson) did not participate. Thus, the Court’s decision was 5-2 with two abstentions, and four of the justices signed on to the majority opinion.

In writing for the Court, Justice Murphy indicated that the SEC’s “latest order . . . definitely avoids the fatal error of relying on judicial precedents which do not sustain it. This time . . . the Commission has concluded that the proposed transaction is inconsistent with the standards of §§ 7 and 11 of the Act.”128Id. at 199. In other words, the SEC was now resting on the statute’s prohibition of “unfair” and “inequitable” distribution of shareholder voting power to ground its decision. Its decision would be reviewable as an interpretation and application of statutory law.

Of course, there remained the obvious question from the Court’s opinion in the previous Chenery case: Didn’t the Court say that the Commission could not reject Chenery’s reorganization plan unless it relied on a preexisting legal requirement, and that the Commission should therefore promulgate a rule against buying stock during a reorganization in order to provide that legal basis for its decision? Justice Murphy explained that such a view “grows out of a misapprehension of our prior decision and of the Commission’s statutory duties. . . . The absence of a general rule or regulation governing management trading during reorganization did not affect the Commission’s duties in relation to the particular proposal before it.”129Id. at 200-01. While a preexisting rule would have clarified the nature of the Commission’s statutory duties, Murphy wrote, “we did not mean to imply that the failure of the Commission to anticipate this problem and to promulgate a general rule withdrew all power from the agency to perform its statutory duty in this case.”130Id. at 201-02. When a statute imposes a responsibility on an agency, such as ensuring that all reorganization plans are fair and equitable, the agency’s order is merely carrying out that statutory duty. Thus, as Murphy concluded, “the choice between proceeding by general rule or by individual, ad hoc litigation is one that lies primarily in the informed discretion of the administrative agency.”131Id. at 203. Such use of adjudication, Murphy hinted, was simply “the case-by-case evolution of statutory standards,” again intimating that the legal basis for such adjudications was the statute itself.132Id.

However, Justice Murphy also emphasized the desirability of specifying general legal provisions through prospective rules. He wrote that “[t]he 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. But any rigid requirement to that effect would make the administrative process inflexible and incapable of dealing with many of the specialized problems which arise.”133Id. at 202. There would not be a rigid requirement for rulemaking to precede adjudication, but the majority signaled a clear preference for rules to precede adjudication where possible. In cases where it is not possible, the majority seemed to indicate, the legal authority must come from the underlying statute, and the reviewing court’s primary question is whether the agency’s adjudication is consistent with the statute. As Murphy explained, because the SEC’s determination “is a judgment that can justifiably be reached in terms of fairness and equitableness,”134Id. at 209. terms that are found in PUHCA itself, it should be sustained by the courts.

Justice Jackson was astonished. In December 1946 Jackson wrote to Justice Black that he had reviewed the case “[w]ith every impulse to sustain the Commission.” Nevertheless, he wrote, “I cannot escape the conviction that the Commission has decided this case ad hoc without any reference to considerations that would govern it in the same case tomorrow.”135Letter from Justice Robert Jackson to Justice Hugo Black (Dec. 23, 1946). When the Chenery II opinion was released in June, Justices Frankfurter and Jackson dissented, but claimed that “there is not now opportunity for a response adequate to the issues raised by the Court’s opinion.”136Chenery II, 332 U.S. at 209 (Frankfurter and Jackson, JJs., dissenting). Justice Jackson drafted a dissenting opinion that he sent to Frankfurter in July, which gave “the grounds of objection that I have to the decision and the opinion, which is one of the worst ever.”137Letter from Justice Robert Jackson to Justice Felix Frankfurter (July 25, 1946). That draft opinion, which was filed in revised form at the beginning of the Court’s next term, opened with famous lines that were subsequently moved to later parts of the opinion:

The unprecedented reasoning which the Court employs to reverse the Court of Appeals and sustain the order of the Securities and Exchange Commission warrants analysis in a dissenting opinion. While the opinion, because of disqualifications, does not have the adherence of a majority of the full Court, it is none the less an ominous one, and defines a fundamental cleavage in the Court. It appears to strip judicial review of administrative proceedings, even where directed by Congress, of all substance and meaning. Moreover it seems to open ways of government without law, which we had supposed were foreclosed by our form of government.138Robert Jackson, Chenery II dissenting opinion draft, Nos. 81 and 82, October Term, 1946, July, 1947, at 1.

Justice Frankfurter praised Jackson’s draft effusively. As he wrote to Jackson, “[y]our Chenery dissent is a rip-snorter, or a sockdolager . . . or both, a rip & a sock! Leave it be, don’t . . . subtract any of the mother wit.”139Letter from Justice Felix Frankfurter to Justice Robert Jackson (July 23, 1946). He hastened to add a suggestion for the last paragraph of the opinion: “The first Chenery opinion was also by a Court of four,” because Justice Rutledge had not yet joined the Court. In order to “guard yourself against a tu quoque,” he suggested that Jackson expressly note that the Court was not questioning the first decision but was in fact claiming to follow it.140Id. Frankfurter could not close the letter without another dig at Justice Reed: “What does ‘a great friend of the administrative process’ like Stanley think he is doing in joining so suicidal an opinion?”141Id.

Examining the correspondence among the justices and the Chenery cases as a whole suggests some important qualifications of what has come to be known as the Chenery II rule that agencies may proceed by rule or by order. First, and most obviously, the opinion of the Court in Chenery II was only signed by four of the justices. Second, the Court’s opinion in Chenery II emphasized that the SEC was not acting in the absence of law. It was, instead, carrying out its statutory duties in issuing the order. It follows from this, however, that the Commission’s order was based on an interpretation of law, an interpretation which would be subject to judicial review. The Commission’s action would need to be justified as an application of preexisting legal standards rather than as creation of new law. Third, and related, the clearest statement of the holding in Chenery II announced that “the choice made between proceeding by general rule or by individual, ad hoc litigation is one that lies primarily in the informed discretion of the administrative agency.”142Chenery II, 332 U.S. at 203. “Primarily” does not mean “exclusively,” and thus the Court left an opening for courts that might impose requirements that agencies proceed by rulemaking in some instances. The Court would consider some of these possibilities in later cases.

Part III: The Post-Chenery World

In the years and decades after Chenery II was decided, courts continued—sometimes—to impose requirements that agencies use rules to establish generally applicable policies rather than proceed through ad hoc adjudication. We admit that these developments did not follow a clear or coherent path. Nevertheless, the law post-Chenery II did not seem to take for granted that agencies could use either rulemaking or adjudication to decide any issue whatsoever. Therefore, the post-Chenery II world either suggests that Chenery II was not universally understood to be a core principle of administrative law after the case was handed down, or it suggests that courts have consistently limited its applicability in important ways.143That assessment of Chenery II comports with some scholarly assessments. See, e.g., Merton C. Bernstein, The NLRB’s Adjudication-Rule Making Dilemma Under the Administrative Procedure Act, 79 Yale L. J. 571, 607 (1970) (arguing that in Chenery II “the Court did not endorse any and all rule making in adjudication. Read with due recognition of the facts before the court, Chenery II is not carte blanche to administrative agencies to use adjudication and rule making as the spirit moves them.”).

a. The Lower Courts

Courts have sporadically appeared to require agencies, before issuing orders, to use rulemaking to specify how broad statutory language would be interpreted and applied. In rendering its decision in Amalgamated Meat Cutters v. Connally,144373 F. Supp. 737 (D.D.C. 1971). for instance, the district court for the District of Columbia, with Judge Harold Leventhal from the D.C. Circuit writing, explained that “any action taken by the Executive under the law . . . must be in accordance with further standards as developed by the Executive.”145Id. at 758. This requirement was not expressly rooted in the law but was “inherent in the Rule of Law and implicit in the Act” creating the wage and price controls under consideration in the case.146Id.

To be fair, this argument differs slightly from the contention in Chenery II. The court was saying that an agency, once it has issued standards implementing the law, must follow those standards. It was not saying that agencies had to issue standards before applying the law to specific parties. But the notion that the rule of law requires agencies to specify legal standards and follow them in later cases bears a close resemblance. As Cass Sunstein and Adrian Vermeule note, “Leventhal’s basic approach played a central role in several important decisions by the D.C. Circuit,” requiring agencies to issue rules specifying the meaning of broad statutory provisions.147Cass R. Sunstein & Adrian Vermeule, Law and Leviathan 49 (2020).

In Whitman v. American Trucking Ass’n,148531 U.S. 457 (2001). the Supreme Court unceremoniously put an end to this requirement, largely because it focused on the wrong issue: nondelegation. Correctly, the Court stated that it was “internally contradictory” to think that “an agency can cure an unconstitutionally standardless delegation of power by declining to exercise some of that power.”149Id. at 473. The Court did not address the deeper question, however, of whether agencies should be required to specify the meaning of a statute through rules before issuing orders.

On several occasions in the early 1980s, the Ninth Circuit found circumstances in which new law could be created only through rulemaking.150See Patel v. INS, 638 F.2d 1199 (9th Cir. 1980); Ford Motor Co. v. FTC, 673 F.2d. 1008 (9th Cir. 1981). Those cases acknowledged the language from Chenery II declaring that agencies have discretion to make law through adjudication where their organic statutes do not forbid it, but the court found in each case that the agency had abused its discretion by using adjudication rather than rulemaking.151See 638 F.2d at 1203; 673 F.2d at 1009. Frankly, it is hard to see why these cases are different from the run-of-the-mill post-Chenery cases in which agencies make law through adjudication; there did not seem to be any extraordinary circumstances beyond the generality of the norms created by the agencies. As it happens, however, these cases appear to be something of a historical anomaly; they have not developed into a body of jurisprudence in the ensuing four decades.

The Ninth Circuit cases relied on a statement from the Supreme Court in 1974 that “there may be situations where the reliance on adjudication would amount to an abuse of discretion.”152NLRB v. Bell Aerospace Co., 416 U.S. 267, 294 (1974). Does that mean that the Court has backed away from a strong reading of Chenery II? The answer is actually unclear.

b. The Supreme Court Post-Chenery

The Supreme Court’s first major case after Chenery II involving an agency’s choice of procedural form came in 1969 in NLRB v. Wyman-Gordon Co.153394 U.S. 759 (1969). In a 1966 adjudication, the National Labor Relations Board announced that, henceforth, employers would be required to give unions a list of people eligible to vote in unionization elections, but that no such requirement would be applied in the case in which that new norm was announced.154Excelsior Underwear, Inc., 156 N.L.R.B. 1236 (1966). Shortly thereafter, the Board applied that norm in another adjudication involving Wyman-Gordon. The company protested that the norm was actually a rule that had not been promulgated in accordance with the Administrative Procedure Act’s rulemaking requirements.155See 5 U.S.C. § 553 (2018). The District Court enforced the agency’s order on the ground that the norm had been created in an adjudication rather than a rulemaking and that Chenery II allowed the agency to choose which procedural form to use.156See NLRB v. Wyman-Gordon Co., 270 F. Supp. 280 (D. Mass. 1967). The First Circuit reversed, finding that the agency’s action, involving promulgation of a general norm that was not used to decide the case before it, was rulemaking in disguise and without required procedures.157See Wyman-Gordon Co. v. NLRB, 397 F.2d 394 (1st Cir. 1968). In essence, the court held that Chenery II allows adjudicating agencies to create law only as an incidental aspect of their case-deciding function. Note that such a holding was consistent with Chenery II because the agency there was applying its new standard to the Chenerys. Thus, this case did not directly challenge Chenery II but instead involved a possible extension of it to allow agencies to make law using adjudicatory rather than rulemaking procedures even when the new norm was not actually employed in the adjudication.

The Supreme Court took the case and split three ways, with no majority opinion.158See NLRB v. Wyman-Gordon Co., 394 U.S. 759 (1969). A four-justice plurality of the Court seemed to agree with Wyman-Gordon and the First Circuit that the agency in 1966 had circumvented the APA’s rulemaking requirements, declaring that the NLRB’s use of adjudication to craft rules “does not comply with statutory command” and “falls short of the substance of the requirements of the Administrative Procedure Act.”159Id. at 764 (plurality opinion). While the Court accepted that “[a]djudicated cases may and do, of course, serve as vehicles for the formulation of agency policies, which are applied and announced therein,” those policies cannot be “‘rules’ in the sense that they must, without more, be obeyed by the affected public.”160Id. at 765-66. The plurality nonetheless ruled that Wyman-Gordon must comply with the NLRB order to produce a voting list because the order to do so was issued in an adjudication:

In the present case, however, the respondent itself was specifically directed by the Board to submit a list of the names and addresses of its employees for use by the unions in connection with the election. This direction, which was part of the order directing that an election be held, is unquestionably valid. Even though the direction to furnish the list was followed by citation to ‘Excelsior Underwear Inc., 156 NLRB No. 111,’ it is an order in the present case that the respondent was required to obey. Absent this direction by the Board, the respondent was under no compulsion to furnish the list because no statute and no validly adopted rule required it to do so.

Because the Board in an adjudicatory proceeding directed the respondent itself to furnish the list, the decision of the Court of Appeals for the First Circuit must be reversed.161Id. at 766.

We confess to some measure of bafflement about what this meant. Was the Court saying that Excelsior Underwear had failed to create a principle of law but that the agency had then successfully created such a principle in the subsequent Wyman-Gordon case? That would be an affirmation of a strong Chenery II principle—though perhaps at the cost of Chenery I, if the basis for the agency’s decision regarding Wyman-Gordon was that it had already created a valid norm in a prior adjudication.162Justice Harlan made this point in his dissenting opinion. See id. at 781-82 (Harlan, J., dissenting). The plurality, however, made no reference to the Chenery cases.

Justice Black, joined by Justices William Brennan and Marshall, concurred in the result but did so specifically citing Chenery II in support of their position. The concurring justices believed that the agency had full power in the Excelsior proceeding to adopt a broad prospective policy because the policy arose out of a procedurally proper adjudication.163See id. at 772-74 (Black, J., concurring in the result). This is a full-blown affirmation of the strong reading of Chenery II: The NLRB could create new law, and even apply it prospectively only, in the course of adjudication. Justice Black argued that although the Chenery II decision “did not involve the Labor Board or the Administrative Procedure Act, [it] is nonetheless equally applicable here. As we explained in that case, ‘the choice made between proceeding by general rule or by individual, ad hoc litigation is one that lies primarily in the informed discretion of the administrative agency.’”164Id. at 772 (Black, J., concurring in the result).

The dissent in Wyman-Gordon was written by Justice Douglas and joined by Justice John Marshall Harlan II. Justice Douglas, of course, was on the Court when Chenery II was decided and did not participate in the decision because he was SEC Chairman during the negotiations with Chenery. Justice Douglas was also the only member of the Court circa 1969 who could plausibly claim expertise in administrative law. Justice Douglas admitted that “if the Board decided to treat each case on its special facts and perform its adjudicatory function in the conventional way, we should have no difficulty in affirming its action.”165Id. at 775-76 (Douglas, J., dissenting) The problem was that it made a rule in the Excelsior decision and then applied it prospectively to future cases. Thus, Douglas concluded, “it should be bound to follow the procedures prescribed in the [Administrative Procedure] Act.”166Id. at 776 (Douglas, J., dissenting). “A rule like the one in Excelsior is designed to fit all cases at all times. It is not particularized as to special facts. . . . It should therefore have been put down for public hearing prescribed by the Act.”167Id. at 777 (Douglas, J., dissenting). While this is nominally a discussion about the appropriate procedures employed by the agency, it amounts in practice to a limitation on the agency’s capacity to make broad policy pronouncements in adjudications rather than rulemakings.

Justice Harlan also dissented, essentially for the same reasons as Justice Douglas, plus his belief that the Chenery I decision required the Court to assume that the agency in Wyman-Gordon had relied solely on its (invalid) policy from Excelsior.168See id. at 780 (Harlan, J., dissenting).

Thus, while it is difficult to draw strong conclusions from the splintered, and sometimes puzzling, reasoning in Wyman-Gordon, the Court left considerable space for agencies to make “policies” through adjudication.

In another case involving the NLRB several years later, the Court strongly reaffirmed Chenery II by clarifying that agencies do generally have the ability to make policy through adjudication,169See NLRB v. Bell Aerospace Co., 416 U.S. 267, 291-94 (1974). although “there may be situations where the Board’s reliance on adjudication would amount to an abuse of discretion or a violation of the Act.”170Id. at 294. Consistent with the Court’s opinion in Chenery II, the Court in Bell Aerospace indicated that an agency’s ability to render decisions through adjudication in the absence of rulemaking was limited both by the fact that such decisions are applications and interpretations of the governing statute and general requirements not to abuse discretion through things like unfair surprise.

This wiggle room in the Bell Aerospace decision, as Sunstein and Vermeule note, has enabled lower courts, in some cases, to strike agency decisions for circumventing the APA’s rulemaking provisions and to impose requirements that agencies only adjudicate after promulgating rules.171See Sunstein & Vermeule at 55 & nn. 72-73.

Thus, while the law since Chenery II has been generous toward agencies’ ability to make law through adjudications, there is no conclusive presumption that agencies can simply make policy through adjudication without engaging in rulemaking first. Decisions applying Chenery II indicate that the APA’s rulemaking provisions cannot be circumvented by establishing general rules in adjudications, and adjudicatory policies are subject to general rule-of-law requirements against unfair surprise, as well as requirements that agencies follow their own standards in future cases. These requirements at least somewhat limit the scope of the Chenery II principle. If truly taken seriously, they could undermine it altogether. If taken even moderately seriously, they would require courts to be a bit more vigorous than they have been thus far about policing agencies’ choices to make law through adjudication.

Part IV: Limiting or Ignoring Chenery II

As a recent article on the subject explains, “[a]lthough agencies may generally articulate new policies by adjudication, courts have enacted guardrails to guarantee private parties in agency adjudications the same rule-of-law protections as are provided in Article III cases, ensuring that agencies do not violate due process rights while engaging in policymaking.”172Phillips, supra note 2, at 502-03. Courts have not, however, been clear or systematic about the limits on agencies’ discretion to use rulemaking or adjudication to make law. Nor, in our view, have these “guardrails” been sufficiently protective of individual rights and due process of law. Thus, if Chenery II is not overturned, courts should be clearer, more systematic, and more aggressive in checking abuses that arise from ad hoc decision-making.

One of those guardrails stems from the Chenery II decision itself. As the Court explained in Chenery II, the SEC’s use of adjudication to enforce a rule against management stock purchases during reorganizations was acceptable only because it was an order that implemented the statute. Under this view of the Chenery II principle, the legal basis for any adjudication, if there is no rule upon which it is based, must be the text of the statute itself.

Today, the prospect of court review of agency statutory interpretation conjures up the shadow—or perhaps the ghost—of Chevron U.S.A. v. Natural Resources Defense Council, Inc.,173467 U.S. 837 (1984). under which agencies receive considerable deference for their interpretations of the statutes they administer. In 1947, there was no such doctrine. Courts had just begun to construct a doctrine of deference to agency interpretations made in the course of law application,174See Lawson, supra note 2, at 601-06. but there was no categorical rule of deference in all cases.175For the definitive account of deference to agency legal interpretations in the pre-Chenery era, see Aditya Bamzai, The Origins of Judicial Deference to Executive Interpretation, 126 Yale L.J. 908 (2017). In the same year as Chenery II was decided, the Court famously declined to give deference to the NLRB in a major case of law application.176See Packard Motor Co. v. NLRB, 330 U.S. 485 (1947). Given the constitutional concerns regarding agency lawmaking in adjudication, deference to agency interpretations that clearly create new law in this setting seems dubious at the very least.

Rule-of-law concerns suggest similar limits on the application of Chevron in this context. With respect to agency constructions of their own regulations, the Court has made clear that no deference is due when those interpretations generate “unfair surprise” by crafting new legal norms from rules that are so vague that regulated parties could not know what is prohibited or required in advance of an agency determination.177See Christopher v. SmithKline Beecham Corp., 567 U.S. 142 155-59. In the same way, when the governing statutes are so broad and vague that no regulated party could reasonably know what is required before the adjudication, due process principles prohibit an unfair surprise in which the agency has not notified parties in advance of its interpretation of the law. If it offers that interpretation through guidance, it would not receive any deference from reviewing courts under Chevron.178This proposal comports with a recent suggestion by Kristin Hickman and Aaron Nielson in Narrowing Chevron’s Domain, 70 Duke L. J. 931, 940 (2021).

b. “Embedded Rules” and the APA

There is a second possible way to understand and limit Chenery II. Perhaps agencies can openly acknowledge that their orders contain rules that they will apply in future cases. The Court in Wyman-Gordon acknowledged that agencies could set policies through adjudication, but noted that the APA foreclosed issuing rules without following the requirements of the APA’s rulemaking provisions. The APA does, however, allow for rules to be promulgated without complying with these provisions in limited cases. It specifies that the rulemaking provisions do not apply “when the agency for good cause finds (and incorporates the finding and a brief statement of reasons therefor in the rules issued) that notice and public procedure thereon are impracticable, unnecessary, or contrary to the public interest.”1795 U.S.C. § 553(b)(3)(B) (2018).

As Matthew Stephenson writes, agencies frequently “embed” rules within orders, most of which can be understood as “interpretive rules (agency statements that explain what some existing statute or regulation means) or policy statements . . . rather than legislative rules (which create new rights, duties, or prohibitions).”180Stephenson, supra note 7, at 59. He argues that “certain doctrinal anomalies and uncertainties might be easier to resolve if we recognize the agency orders often contain embedded rules.”181Id. at 60. These rules are legislative rules if they create new rights or obligations rather than clarifying legal rights or obligations in the underlying statute or regulation.182See, e.g., Catholic Health Initiatives v. Sebelius, 617 F.3d 490, 494 (D.C. Cir. 2010).

The issue in Chenery I, of course, was that the SEC was imposing a new obligation that was not grounded in the preexisting legal standards the Commission cited. Thus, the SEC’s order contained a new obligation rather than one clarifying an existing legal obligation. It was establishing, in other words, a legislative rule.

This is precisely what was disputed in Chenery II. The dissent continued to claim that the agency was establishing a new legal obligation without an underlying rule of law upon which it was relying. The majority simply responded that the statute itself contained the legal requirement, and the order was interpreting the law in the course of administering the requirement that reorganizations be “fair and equitable.” The difficulty with the majority’s position in Chenery II is that it was impossible to show how Chenery could have known that buying stock during reorganization was rendered unlawful by PUHCA. This is why the dissent also considered the case to present the problem of unfair surprise.

To address the obvious rule of law concerns raised by embedded legislative rules, Stephenson proposes that agencies rely on the APA’s good cause exception when it would be “contrary to the public interest” to promulgate a rule in advance of an adjudication.183Stephenson, supra note 7, at 65. This is a sensible way of resolving the problem of agencies using adjudication to establish legislative rules, as long as it is also paired with a stronger rule against retroactivity.

Part V: The Stakes

If the Chenery II doctrine were to be ignored or significantly limited, what would be the result? What are the stakes? Since the well-known “flight away from individualized, adjudicatory proceedings to generalized disposition through rulemaking”184Antonin Scalia, The APA, the D.C. Circuit, and the Supreme Court, 1978 Sup. Ct. Rev. 345, 376. took place across most of the administrative state in the 1960s and 1970s, the number of agencies that proceed principally through adjudication rather than rulemaking has decreased dramatically. Much of the administrative state, therefore, would be largely untouched. Three areas, however, could be significantly changed, and in a way that would create greater predictability and greater protection for individual liberty.

a. The National Labor Relations Board

First, as is well-known, the National Labor Relations Board (NLRB) continues to use adjudication rather than rulemaking to establish general legal principles and requirements.185See Phillips, supra note 2, at 516. This feature of the NLRB has concerned scholars for decades.186The literature calling upon the NLRB to exercise rulemaking authority is voluminous and began decades ago. See, e.g., Cornelius J. Peck, The Atrophied Rulemaking Powers of the National Labor Relations Board, 89 Yale L. J. 982 (1961); David L. Shapiro, The Choice of Rulemaking or Adjudication in the Development of Administrative Policy, 78 Harv. L. Rev. 921 (1965). For a recent summary see Jeffrey Lubbers, The Potential of Rulemaking by the NLRB, 5 FIU L. Rev. 411, 414 n. 20 (2010). The NLRB treats its adjudications as quasi-rulemaking procedures in various ways. For instance, it invites amicus briefs when deciding important cases, akin to inviting public comments as required by the notice-and-comment rulemaking procedures under the APA. The Board has established a procedural rule enshrining this practice, and its website features a list of ongoing invitations for interested parties to file amicus briefs.187Invitations to File Briefs, NLRB.gov, https://www.nlrb.gov/cases-decisions/filing/invitations-to-file-briefs (last updated Dec. 15, 2021). The NLRB also publishes a bench book that contains Board precedents and authorities to be consulted in future cases.188NLRB Division of Judges Releases 2022 Bench Book, NLRB.gov, https://www.nlrb.gov/news-outreach/news-story/nlrb-division-of-judges-releases-2022-bench-book (last updated Feb. 8, 2022). The NLRB has engaged in a few high-profile rulemakings in recent years, and overturning or limiting Chenery II would encourage the Board to increase the use of rules rather than case-by-case adjudication to make general policies.

b. The Federal Trade Commission

The Federal Trade Commission (FTC) also “largely acts through adjudications and enforcement actions” rather than through rulemaking.189See Phillips, supra note 2, at 516. When applying the statutory prohibition against “unfair or deceptive acts or practices”19015 U.S.C. § 45(a) (2018). (UDAPs) contained in Section 5 of the FTC Act, FTC does not define the nature of unfair and deceptive acts and practices in advance through rulemaking. Instead, it issues a complaint and adjudicates before an administrative law judge. If the Commission finds that there was a violation of the prohibition against unfair or deceptive acts or practices, it issues an order requiring the party to cease and desist from the conduct.

This more limited adjudicatory authority to issue cease and desist orders under Section 5 of the Act was the FTC’s sole enforcement authority for the first half-century of its history. In the 1970s, however, Congress granted new enforcement powers to the Commission. In 1973 the Commission was granted authority to go to federal court to seek civil penalties against those who violate cease and desist orders (under Section 13), and two years later the FTC obtained power (under Section 19) to seek relief to redress injury to consumers, including monetary penalties. In other words, in the 1970s, the FTC was given authority to go straight to court rather than use its own adjudicatory proceedings, but also to seek civil penalties rather than merely issue cease and desist orders.

The best explanation for why Congress originally granted the FTC such broad powers over UDAPs is the limited extent of its enforcement authority. All that the FTC could do, originally, is issue a cease and desist order that must be enforced by an independent court. The Commission would not have to issue rules defining what conduct is prohibited in advance of issuing such orders because they do not violate rights protected by constitutional due process.191Indeed, the FTC did not have statutory rulemaking power at all until 1975. See Federal Trade Commission Improvement Act of 1975, Pub. L. No. 93-637, § 202, 88 Stat. 2183. Section 5 enforcement orders could serve as the basis for definition of unfair or deceptive practices in future decisions, but civil penalties would not attach to these orders.

The effect of this sequence is to authorize the FTC to seek civil penalties against parties who violate the Commission’s orders against unfair or deceptive practices, in the absence of rules governing UDAPs. Scholars have recommended that the FTC use rulemaking authority to provide greater consistency and predictability to the Commission’s enforcement.192See, e.g., Rohit Chopra & Lina M. Khan, The Case for “Unfair Methods of Competition” Rulemaking, 87 U. Chi. L. Rev. 357 (2020). Limiting Chenery II would encourage this process.

c. The Drug Enforcement Administration

A third, less-well known area of law relates to pharmacy licensing by the Drug Enforcement Administration (DEA). The DEA uses enforcement orders to implement a registration (i.e., a de facto licensing) scheme for pharmacies. It uses the threat of revoking registration through enforcement orders as the chief means to impose regulatory requirements on pharmacies.

The DEA’s own regulations require prescriptions to be issued for a “legitimate medical purpose.”19321 C.F.R. § 1306.04. This language essentially parrots the language of the DEA’s governing statute, the Controlled Substances Act, which in Section 830(b) requires that drugs be issued for a “legitimate medical purpose.” The DEA has used the requirement that drugs be issued for a “legitimate medical purpose” to revoke pharmacy registrations for violating its “red flags” policy. That policy triggers when the DEA determines that a pharmacy has filled prescriptions under conditions that suggest drug diversion or abuse.

The DEA has never specified which conditions, or combination of conditions, create red flags that will lead to registration revocation, and it declines to publish a list of red flags either through rulemaking or guidance. The policy is developed entirely through enforcement orders and pharmacies have no way of knowing in advance how to avoid losing their registration for violating the red flags policy. “The red flags are constantly changing,” writes the George Washington University Regulatory Studies Center, “making it difficult for pharmacists to keep track of when they will run afoul of DEA regulations.”194Laura Stanley, Geo. Wash. Univ. Reg. Stud. Ctr., Policymaking through Adjudication: DEA’s Red Flags 2 (Aug. 12, 2022).The Fifth Circuit Court of Appeals has determined that the red flags policy is not an administrative rule under the APA, which means that it is an example of proceeding through adjudication.195Id.

This instance of policymaking through adjudication in the absence of a rule, according to the Regulatory Studies Center, “has major implications for the opioid crisis. It makes pharmacies reluctant to fill prescriptions for drugs used to treat opioid use disorder.”196Id. at 3. Pharmacies, including Walmart, have presented legal challenges to the DEA’s enforcement regime, but these challenges have not raised the specific legal challenge that we highlight in this article.

Conclusion

The conventional account of the Supreme Court’s decision in Chenery II is a serious overreading of the Court’s opinion. Four justices voted to sustain the SEC in that decision, and only three of them joined Justice Murphy’s opinion. Even that opinion, which is often misunderstood to state that agencies may make law or policy through rulemaking or adjudication, acknowledged the limits on an agency’s choice in this regard. These limits have appeared, in various forms, throughout cases subsequent to the Chenery II decision.

Going forward, the Court should either abandon Chenery II as a principle, or it should cabin that decision by developing the already-present doctrinal limits, and enforcing those limits consistently.

CGO scholars and fellows frequently comment on a variety of topics for the popular press. The views expressed therein are those of the authors and do not necessarily reflect the views of the Center for Growth and Opportunity or the views of Utah State University.