Administrative Sabotage

Government can sabotage itself. From the president’s choice of agency heads to agency budgets, regulations, and litigating positions, presidents and their appointees have undermined the very programs they administer. But why would an agency try to put itself out of business? And how can agencies that are subject to an array of political and legal checks sabotage statutory programs?

This Article offers an account of the “what, why, and how” of administrative sabotage that answers those questions. It contends that sabotage reflects a distinct mode of agency action that is more permanent, more destructive, and more democratically illegitimate than more-studied forms of maladministration. In contrast to an agency that shirks its statutory duties or drifts away from Congress’s policy goals, one engaged in sabotage aims deliberately to kill or nullify a program it administers. Agencies sabotage because presidents ask them to. Facing pressure to dismantle statutory programs in an environment where securing legislation from Congress is difficult and politically costly, presidents pursue retrenchment through the administrative state.

Building on this positive theory of administrative sabotage, this Article considers legal responses. The best response, this Article contends, is not reforms to the cross-cutting body of administrative law that structures most agency action. Rather, the risk of sabotage is better managed through changes to how statutory programs are designed. Congress’s choices about agency leadership, the concentration or dispersal of authority to implement statutory programs, the breadth of statutory delegations, and other matters influence the likelihood that sabotage will succeed or fail. When lawmakers create or modify federal programs, they should design them to be less vulnerable to sabotage by the very agencies that administer them.


In 2010, Congress enacted the Dodd-Frank Act in an effort to address the market failures that triggered the 2008 financial crisis.1Dodd-Frank Wall Street Reform and Consumer Protection Act, Pub. L. No. 111-203, 124 Stat. 1376 (2010) [hereinafter Dodd-Frank Act] (codified as amended in scattered sections of the United States Code). Following the recommendations of an article coauthored by then-Professor Elizabeth Warren,2Oren Bar-Gill & Elizabeth Warren, Making Credit Safer, 157 U. Pa. L. Rev. 1 (2008). the Act sought to regulate “unsafe” financial products that Congress believed imposed a variety of negative externalities on society.3See. e.g., Dodd-Frank Act §§ 501–533 (regulating insurance markets); id. §§ 601–628 (regulating depository institutions); id. §§ 1400–1498 (regulating mortgages). Title X of Dodd-Frank, known as the Consumer Financial Protection Act of 2010, created the Consumer Financial Protection Bureau (CFPB) and charged it with enforcing eighteen consumer financial laws.4Id. § 1011 (establishing CFPB); id. §§ 1021(a), (c) (authorizing CFPB to enforce federal consumer financial laws); id. § 1022(12) (enumerating consumer financial laws the Bureau administers).

In its early years under Director Richard Cordray, the CFPB pursued its statutory mission with zeal. The Bureau aggressively investigated predatory lenders, mortgage companies, and credit-card companies.5See Christopher L. Peterson, Consumer Financial Protection Bureau Law Enforcement: An Empirical Review, 90 Tul. L. Rev. 1057, 1076–77 (2016). Information from supervision and enforcement drove rulemaking in areas such as payday lending6See Press Release, CFPB, Consumer Financial Protection Bureau Proposes Rule to End Payday Debt Traps (June 2, 2016),‌_Proposes_Rule_End_Payday_Debt_Traps.pdf []. and debt collection.7See Richard Cordray, Dir., CFPB, Remarks at the Consumer Advisory Board Meeting (June 8, 2017), []. Within five years, the Bureau was returning an average of $43 million per week from financial services companies to consumers.8Christopher L. Peterson, Consumer Fed’n of Am., Dormant: The Consumer Financial Protection Bureau’s Law Enforcement Program in Decline 15 tbl.1 (2019), [].

But in December 2017, Cordray resigned to run for governor of Ohio,9Sylvan Lane, Former Consumer Bureau Director Cordray Announces Run for Ohio Governor, Hill (Dec. 5, 2017, 1:44 PM), []. and President Trump used the Federal Vacancies Reform Act105 U.S.C. §§ 3345–3349d. to replace him with Mick Mulvaney, the director of the Office of Management and Budget (OMB).11See Renae Merle, The CFPB Now Has Two Acting Directors. And Nobody Knows Which One Should Lead the Federal Agency, Wash. Post (Nov. 24, 2017), https://www.washingtonpost‌.com/news/business/wp/2017/11/24/the-cfpb-now-has-two-acting-directors-and-nobody-knows-which-one-should-lead-the-federal-agency []. In one of the stranger episodes of the Trump administration, Cordray appointed Leandra English as the CFPB’s deputy director before he left office; she and Mulvaney both claimed the right to lead the agency. A district court rejected English’s motion for a preliminary injunction recognizing her right to lead the agency, and English abandoned the case before it proceeded further. English v. Trump, 279 F. Supp. 3d 307 (D.D.C. 2018); see also Sylvan Lane, Biden Consumer Bureau Pick Could Take Over Agency on Inauguration Day, Hill (Jan. 19, 2021, 4:47 PM), []. A long-time ally of the payday lending industry,12See George Zornick, Why Loan Sharks, Car Salesmen, and Payday Lenders Love Mick Mulvaney, Nation (Sept. 13, 2018), []. Mulvaney once sponsored legislation to abolish the CFPB and stated at a House hearing that he didn’t “like the fact that CFPB exists.”13Merle, supra note 11.

At the helm of an agency he “detest[ed],”14Devin Leonard & Elizabeth Dexheimer, Mick Mulvaney Is Having a Blast Running the Agency He Detests, Bloomberg Businessweek (May 25, 2018, 5:28 PM), [] (cleaned up). Mulvaney moved to cripple it. Mulvaney declined to request money to fund the Bureau’s operations;15Kate Berry, Mulvaney Requests ‘Zero’ Money for CFPB, Am. Banker (Jan. 18, 2018, 10:11 AM), []. installed “policy associate directors”16Nicholas Confessore, Mick Mulvaney’s Master Class in Destroying a Bureaucracy from Within, N.Y. Times Mag. (Apr. 16, 2019), []. to shadow Bureau chiefs protected by the civil service laws;17Kate Berry, Meet Mulvaney’s ‘Politicos’: Six Senior Staff Remaking the CFPB, Am. Banker (May 7, 2018, 5:11 PM), []. rescinded, stayed, or delayed major rules on payday lending, overdraft fees, and student loan servicing;18See Ganesh Sitaraman, The Supreme Court, 2019 Term—Comment: The Political Economy of the Removal Power, 134 Harv. L. Rev. 352, 372 (2020). and lent the Bureau’s support to a constitutional challenge to the Bureau’s structure, in which the petitioner asked the Supreme Court to “invalidate Title X.”19See CFPB, Semi-annual Report of the Bureau of Consumer Financial Protection 1–2 (2018), []; Reply Brief for the Petitioner at 19, Seila L. LLC v. CFPB, 140 S. Ct. 2183 (2020) (No. 19-7). According to one commentator, these actions left the CFPB in a “vegetative state.”20Sitaraman, supra note 18, at 373 (quoting Bess Levin, Mick Mulvaney Won’t Rest Until He’s Exorcised Elizabeth Warren from the C.F.P.B., Vanity Fair (May 25, 2018), []).

Mulvaney’s attack on the CFPB and the Bureau’s attack on Title X under him highlight an important gap in our understanding of the U.S. administrative state. From the Affordable Care Act to laws protecting consumers, financial markets, and the environment, major statutory programs are administered by executive departments and administrative agencies.21In federal statutory law, executive agencies, executive departments, and independent regulatory commissions are distinct kinds of entities. The differences among these kinds of institutions are for the most part unimportant to my goals in this Article. Thus, I use “agencies” as a shorthand to refer to all three. I use “statutory program” to refer to legislation that is administered, in whole or part, by an agency or executive department. Scholars have long appreciated that Congress’s delegation of authority to agencies creates risks of slacking, drift, and capture: agencies might perform their functions lethargically,22See, e.g., Sean Farhang, The Litigation State: Public Regulation and Private Lawsuits in the U.S. 70 (2010) (“[A]n enacting legislative coalition, mindful that the stickiness of the status quo will make legislative override of bureaucrats difficult, will want to safeguard in advance against bureaucrats deviating from the coalition’s preferences based upon bureaucrats’ own policy preferences, careerism, propensity to shirk, or acquiescence to capture.”); Matthew C. Stephenson, Public Regulation of Private Enforcement: The Case for Expanding the Role of Administrative Agencies, 91 Va. L. Rev. 93, 110 (2005) (noting that private enforcement regimes can correct “the tendency of government regulators to underenforce certain statutory requirements because of political pressure, lobbying by regulated entities, or the laziness or self-interest of the regulators themselves” (footnotes omitted)). depart from Congress’s preferences in administering a program,23See, e.g., David Epstein & Sharyn O’Halloran, Delegating Powers: A Transaction Cost Politics Approach to Policy Making Under Separate Powers 48 (1999) (“[G]iven that delegation implies surrendering at least some residual rights of control over policy, legislators will be loath to relinquish authority in politically sensitive policy areas where they cannot be assured that the executive will carry out their intent.”); Matthew D. McCubbins, Roger G. Noll & Barry R. Weingast, Structure and Process, Politics and Policy: Administrative Arrangements and the Political Control of Agencies, 75 Va. L. Rev. 431, 438–39 (1989) (“Once a policy is enacted, the agency must implement it. In so doing, the agency may shift the policy outcome away from the legislative intent . . . .”). or consistently advance the interests of a regulated industry.24See Daniel Carpenter & David A. Moss, Introduction, in Preventing Regulatory Capture: Special Interest Influence and How to Limit It 1, 13 (Daniel Carpenter & David A. Moss eds., 2014) (“Regulatory capture is the process by which regulation, in law or application, is consistently or repeatedly directed away from the public interest and toward the interests of the regulated industry, by the intent and action of the industry itself.” (emphasis omitted)). But with a handful of recent exceptions,25See, e.g., Abbe R. Gluck & Thomas Scott-Railton, Affordable Care Act Entrenchment, 108 Geo. L.J. 495, 548 (2020); Madeline June Kass, Presidentially Appointed Environmental Agency Saboteurs, 87 UMKC L. Rev. 697 (2019); Patricia A. McCoy, Inside Job: The Assault on the Structure of the Consumer Financial Protection Bureau, 103 Minn. L. Rev. 2543 (2019). Two recent works offer broader perspectives on issues related to administrative sabotage. In a cross-national study, Michael Bauer and Stefan Becker identify sabotage as one of four strategies that populists who are elected to office use to “transform public administration according to their anti-pluralist ideology.” Michael W. Bauer & Stefan Becker, Democratic Backsliding, Populism, and Public Administration, 3 Persps. on Pub. Mgmt. & Governance 19 (2020). Focusing on the Trump administration, Jody Freeman and Sharon Jacobs examine structural deregulation, a set of strategies that track what this Article terms “systemic” sabotage. Jody Freeman & Sharon Jacobs, Structural Deregulation, 135 Harv. L. Rev. 585 (2021). scholars have not focused on the possibility that agencies might affirmatively attack programs they administer—a phenomenon this Article terms “administrative sabotage.”

Administrative sabotage raises thorny questions. An influential strand of public choice scholarship views agencies as “budget maximizers” that continuously undersupply policy outcomes desired by Congress while maximizing their own budgets.26The classic work is William A. Niskanen, Jr., Bureaucracy and Representative Government (1971). Why, contrary to this image, would an agency aim to put itself out of business? Is sabotage different than agency slacking and drift or merely an extreme form of those phenomena? Given the political and legal checks on agencies, how can they successfully sabotage a statutory program?

This Article offers a theoretical and legal account of administrative sabotage that answers those questions. The core claim is that presidents use agencies to pursue statutory retrenchment that is costly, if not impossible, to obtain directly from Congress. This affects our understanding of what agencies are. Agencies not only enforce, elaborate, and implement statutory policy but can undermine and dismantle the programs they administer.

I begin by making the case that administrative sabotage reflects a distinct mode of agency action. Part I offers additional examples of administrative sabotage, distinguishes sabotage from other kinds of maladministration, and explains why sabotage is normatively objectionable.

Part I also explains that sabotage exists in something of a legal grey area. The use of agency power to attack statutory programs encroaches on Congress’s legislative authority, is in tension with the executive’s constitutional duty to “take Care that the Laws be faithfully executed,”27U.S. Const. art. II, § 3. and violates statutory provisions that contemplate good-faith policy implementation. But concerns about intruding on a coordinate branch of government frequently prevent courts from effectively checking administrative sabotage.

Having defined administrative sabotage, Part II turns to its origins. Administrative sabotage is a tool for retrenching federal statutory programs. Demand for it originates in larger political and ideological opposition to an activist federal government. Yet sabotage would not occur were it not for other long-term trends in U.S. law and politics.

Those trends weaken presidents’ ability to secure new legislation while strengthening their power over programs that have already been enacted into law. On one hand, the complexity of the legislative process and the distinctive politics of statutory retrenchment make it difficult for presidents to secure legislation rolling back enacted programs. On the other hand, Congress has delegated significant authority to agencies, presidents have expanded the White House’s ability to control agencies through “presidential administration,” congressional oversight has been weakened by the emergence of polarized political parties, and courts are increasingly open to reconfiguring or dismantling statutory programs based on creative legal arguments.

Constrained on the one hand and empowered on the other, presidents pursue statutory retrenchment that Congress will not give them through the administrative state. In Part III, I catalogue the tools available to a motivated administration for sabotaging statutory programs and analyze the likelihood that the tools will be checked by Congress or the courts. Overall, checks on administrative sabotage are inconsistent, and their effectiveness depends on contingent historical conditions such as partisan control of Congress. Sabotage, then, is not just a distinct mode of agency action but likely represents a new normal.28Accord Nicholas Bagley, Executive Power and the ACA, in The Trillion Dollar Revolution: How the Affordable Care Act Transformed Politics, Law, and Health Care in America 192, 198 (Ezekiel J. Emanuel & Abbe R. Gluck eds., 2020); cf. James A. Morone, Partisanship, Dysfunction, and Racial Fears: The New Normal in Health Care Policy?, 41 J. Health Pol. Pol’y & L. 827 (2016). Particularly during conservative administrations, agencies are likely to use their delegated authority to attack disfavored programs.

Administrative sabotage complicates traditional accounts of the bureaucracy. After briefly highlighting these theoretical implications, Part IV considers legal responses. Although existing checks on sabotage are weak, changes to procedural requirements, statutory and institutional design, and the law of judicial review could potentially counteract it. A crucial question about such reforms is whether sabotage is best addressed through cross-cutting reforms that apply to agencies in general or through changes to specific agencies and programs.

I argue that cross-cutting reforms are a poor response to the risk of sabotage. Subjecting agencies to new procedural and analytical requirements and expanding the scope of judicial review would make sabotage more difficult at the margin, but these reforms would also impede legitimate efforts to implement statutory policy. Instead, I argue that the risk of sabotage is better addressed through statutory and institutional design. A host of choices affect statutory programs’ resilience against sabotage. While the risk of sabotage cannot be eliminated, it complicates and in some cases reorients long-running debates about the design of statutory programs.

I.        Fundamentals

Consider the following executive actions:

  • The president appoints an EPA administrator who opposed federal environmental regulation as a state lawmaker. In her first speech to the agency, she announces, “We’re going to do more with less and we’re going to do it with fewer of you.”

Each of these cases involves actions that are the ordinary stuff of the administrative process: appointments, office reorganizations, budgeting, rulemaking, and public communications. But the actions also differ from the textbook image of agency administration, insofar as they seem designed to undermine programs that the executive is charged with administering. This tension raises the question whether it is coherent—and useful—to view administrative sabotage as a distinctive mode of agency action.

This Part makes the case that it is. Administrative sabotage is defined by intent: an agency aims to kill or nullify a statutory program. Section I.A develops this definition. Section I.B considers sabotage’s relationship to agency slacking, drift, and capture. Section I.C explores why administrative sabotage is normatively objectionable and rebuts the argument that the president’s popular mandate makes sabotage democratically legitimate. Finally, Section I.D explores the reasons that courts, for the most part, do not directly police sabotage.

A.     Defining Administrative Sabotage

An agency can sabotage many things: economic conditions, a presidential administration, a business or industry. This Article focuses on the sabotage of statutory programs by agencies that administer them.

In defining administrative sabotage, I draw on work in law and public administration on related forms of sabotage. John Brehm and Scott Gates study bureaucrats in state and federal bureaucracies and classify their actions as “working, shirking, and sabotage.”29John Brehm & Scott Gates, Working, Shirking, and Sabotage: Bureaucratic Response to a Democratic Public 21 (1997) (cleaned up). They define sabotage as bureaucratic action that “deliberately undermin[es] the policy objectives of their superiors.”30Id. at 21. Marissa Martina Golden studies executive agencies’ responsiveness to the president’s policy agenda and approaches “sabotage” as bureaucratic action that undermines the president.31See Marissa Martino Golden, What Motivates Bureaucrats? Politics and Administration During the Reagan Years 17 (2000) (finding little evidence that bureaucratic sabotage occurred during the Reagan administration). In a formal model of legislative delegation to an imperfectly controlled, expert bureaucrat, Sean Gailmard defines “subversion” as any action outside the discretionary policy bounds set by the legislature.32Sean Gailmard, Expertise, Subversion, and Bureaucratic Discretion, 18 J.L. Econ. & Org. 536, 537 (2002).

The first point suggested by this literature is that administrative sabotage takes place within a principal-agent relationship.33See Brehm & Gates, supra note 42, at 25. The principal in this relationship is Congress. The agent is the administrative agency that administers the program.34This view of agencies follows from the black-letter definition of an agency: “[A]n agency literally has no power to act . . . unless and until Congress confers power upon it.” La. Pub. Serv. Comm’n v. FCC, 476 U.S. 355, 374 (1986). It also comports with the standard view of agencies in positive political theory. See Sean Gailmard & John W. Patty, Formal Models of Bureaucracy, 15 Ann. Rev. Pol. Sci. 353 (2012). It bears noting, however, that agencies are subject to some degree of control by the president, the boundaries of which are deeply contested. See Peter L. Strauss, Foreword, Overseer, or “The Decider”? The President in Administrative Law, 75 Geo. Wash. L. Rev. 696 (2007).

Second, administrative sabotage involves a specific stance on the part of the agency toward the program it administers. The agency does not over- or underenforce a policy relative to a hypothesized baseline,35See Gailmard & Patty, supra note 47, at 365–66. or shift statutory policy within the “space” a statute creates,36See Peter L. Strauss, Essay, “Deference” Is Too Confusing—Let’s Call Them “Chevron Space” and “Skidmore Weight, 112 Colum. L. Rev. 1143, 1146–47 (2012) (explaining that when Chevron deference applies to an agency’s statutory interpretation, “Congress has created the agency’s freedom to act within its space anticipating presidential oversight”). but seeks to eliminate a program it administers. Identifying sabotage thus requires analysis of agency decisionmakers’ intent. The intent to kill or nullify a statutory program distinguishes sabotage as a mode of agency action.37Cf. Freeman & Jacobs, supra note 25, at 627 (“The real question is whether . . . agency reforms that have potential to reduce functional capacity appear to have been taken in good faith with a legitimate purpose other than weakening the agency’s ability to perform its legislative tasks.”); Randy E. Barnett, The President’s Duty of Good Faith Performance, Reason: Volokh Conspiracy (Jan. 12, 2015), [] (arguing that determining whether an enforcement policy complies with the Take Care Clause requires analysis of whether nonenforcement is motivated by disagreement with the law not being enforced).

Third, administrative sabotage can target an entire statutory program, such as the ACA, or a discrete aspect of a program, such as the ACA’s minimum essential coverage provisions. Every presidential administration likely engages in some small-bore forms of sabotage. As the scope of sabotage expands, however, so do the normative concerns it raises. More on those concerns in a moment.

Fourth, administrative sabotage can succeed, fail, or produce a range of outcomes in between. Given the stickiness of enacted law, it is not surprising that many attempts at sabotage fail. But the line between successful and unsuccessful sabotage is uncomfortably thin. Consider the Trump Justice Department’s support for Texas v. United States, the most recent constitutional challenge to the ACA.38See infra notes 113–118 and accompanying text. There, a stay pending appeal was all that prevented a district court from invalidating the Act in its entirety.39Texas v. United States, 352 F. Supp. 3d 665, 690 (N.D. Tex. 2018), aff’d in part, 945 F.3d 355 (5th Cir. 2020), rev’d sub nom. California v. Texas, 141 S. Ct. 2104 (2021).

Based on the foregoing, this Article defines administrative sabotage as follows: an agency engages in administrative sabotage when it deliberately seeks to kill or nullify a statutory program, in whole or part, that Congress has charged the agency with administering.40I use “kill” to refer to agency action that deprives a statutory program of the force of law and “nullify” to refer to action that leaves a program in place but deprives it of any practical effect.

B.     Sabotage, Slacking, Drift, and Capture

Administrative sabotage is related to, but different from, other forms of agency action that scholars have explored at length. We can sharpen our understanding of sabotage by comparing it to these phenomena.

First, administrative sabotage is not the same as the undersupply of policy outcomes Congress desires, variously known as slacking,41E.g., Stephenson, supra note 22, at 110. shirking,42E.g., Mathew D. McCubbins, Roger G. Noll & Barry R. Weingast, Administrative Procedures as Instruments of Political Control, 3 J.L. Econ. & Org. 243, 247 (1987). or nonenforcement.43See, e.g., Daniel T. Deacon, Note, Deregulation Through Nonenforcement, 85 N.Y.U. L. Rev. 795 (2010). For example, Congress might direct the Internal Revenue Service to audit ten thousand high-income taxpayers per year and provide the requisite authority and appropriations.44Cf. H.R. 1200, 117th Cong. § 3(a) (2021) (directing IRS to audit 50 percent of individual tax returns with an income over million a year). If the agency simply refuses to conduct the audits, it is slacking. If it refuses to perform the audits and fires all of its personnel who are capable of performing the audits, it is engaged in slacking and sabotage.

As this suggests, what distinguishes sabotage from mere slacking is the agency’s intent to kill or nullify the program. The agency does not simply take its foot off the pedal but totals the vehicle, making it impossible for later agency heads to restart. Where sabotage succeeds, its effects are more lasting than mere slacking.45Analyzing the impact of Secretary Purdue’s office moves at the Department of Agriculture, see supra text accompanying note 35, the House Appropriations Committee concluded that the two affected offices “are shells of their former selves” and that “the loss of institutional knowledge each agency has suffered will take years to overcome.” Ryan McCrimmon, Farm Spending Bill Set for House Markup, POLITICO (July 9, 2020, 10:00 AM), [], quoted in Skowronek et al., supra note 33, at 116. Seen in terms of the agency’s fealty to its statutory mandate, this is an important difference of kind. The agency does not simply fail to enforce but uses nonenforcement as a means of eliminating or disabling the sabotaged program.

Sabotage also differs from agency “drift,” or the “risk that implemented policy will differ from what the enacting legislative coalition would prefer.”46Jacob E. Gersen, Designing Agencies, in Research Handbook on Public Choice and Public Law 333, 338 (Daniel A. Farber & Anne Joseph O’Connell eds., 2010). If we model statutory policy along a simple liberal-to-conservative spectrum, a conservative agency might implement a liberal statute more conservatively than the enacting Congress expected, and vice versa.

Most analyses of agency drift assume that it occurs within the space created by the statute.47See, e.g., id. at 337; J.R. DeShazo & Jody Freeman, The Congressional Competition to Control Delegated Power, 81 Tex. L. Rev. 1443, 1445 (2003); McCubbins et al., supra note 23, at 438–39. For example, the Federal Trade Commission might adopt an understanding of anticompetitive conduct that is more or less restrictive than Congress contemplated.48See John J. Flynn, The Reagan Administration’s Antitrust Policy, “Original Intent” and the Legislative History of the Sherman Act, 33 Antitrust Bull. 259 (1988). Sabotage is different. The agency does not implement a statute in a manner that departs from Congress’s preferences but aims to kill the program itself.

Finally, administrative sabotage is different than agency or regulatory capture, the “process by which regulation, in law or application, is consistently or repeatedly directed away from the public interest and toward the interests of the regulated industry, by the intent and action of the industry itself.”49Carpenter & Moss, supra note 24. As this definition suggests, capture is a process through which public authority is used for private ends. A regulated industry might capture an agency and direct it to engage in sabotage. But the industry might also have the agency protect it from regulation, protect it from competition, or even overregulate.50See Sidney A. Shapiro, The Complexity of Regulatory Capture: Diagnosis, Causality, and Remediation, 17 Roger Williams U. L. Rev. 221, 223–25 (2012). Conceptually, then, whether an agency has been captured is distinct from whether the agency is engaged in sabotage.

C.     Normative Objections to Administrative Sabotage

The basic objection to administrative sabotage is that it is incompatible with democratic government under the Constitution. Even fervent critics of the administrative state acknowledge Congress’s authority to define national policy through law, create executive departments and agencies, and charge them with enforcing and elaborating statutory policies.51See Free Enter. Fund v. Pub. Co. Acct. Oversight Bd., 561 U.S. 477, 499 (2010) (“No one doubts Congress’s power to create a vast and varied federal bureaucracy.”).

Administrative sabotage subverts that model. Rather than use delegated authority to enforce and elaborate statutory policy, an agency engaged in sabotage uses that authority to undermine the program it administers. In structural constitutional terms, this use of delegated authority is at odds with the principle of legislative supremacy.52See Kendall v. United States ex rel. Stokes, 37 U.S. (12 Pet.) 524, 613 (1838) (rejecting argument that the Take Care Clause empowers the president to disregard enacted law on the grounds that such a power would “cloth[e] the President with a power entirely to control the legislation of congress, and paralyze the administration of justice”). To the extent that it is done with the president’s blessing, sabotage also breaches the executive’s duty to “take Care that the Laws be faithfully executed.”53U.S. Const. art. II, § 3. Although many aspects of the Take Care Clause’s meaning are unresolved,54Jack Goldsmith & John F. Manning, The Protean Take Care Clause, 164 U. Pa. L. Rev. 1835 (2016). virtually everyone agrees that it requires good-faith execution of duly enacted laws.55See id. at 1858; Andrew Kent, Ethan J. Leib & Jed Handelsman Shugerman, Faithful Execution and Article II, 132 Harv. L. Rev. 2111, 2118 (2019); Josh Blackman, The Constitutionality of DAPA Part II: Faithfully Executing the Law, 19 Tex. Rev. L. & Pol. 213, 219 (2015); Zachary S. Price, Enforcement Discretion and Executive Duty, 67 Vand. L. Rev. 671, 698 (2014). Administrative sabotage is bad-faith, not good-faith, administration.56As David Pozen has explored, the notion of bad faith is used to describe many kinds of objectionable conduct in constitutional law and politics. Administrative sabotage involves a “classic” form of bad faith: “the use of deception to conceal or obscure a material fact, a malicious purpose, or an improper motive or belief, including the belief that one’s own conduct is unlawful.” David E. Pozen, Constitutional Bad Faith, 129 Harv. L. Rev. 885, 892 (2016).

The maladministration of the laws inherent to sabotage affects the beneficiaries of sabotaged programs. Programs protecting health, the environment, and markets seek to overcome collective action problems to address pressing social challenges. When those programs are sabotaged, their ability to provide the goods that lawmakers enacted them to provide is compromised.

But there is more to dislike about sabotage. The programs that sabotage undermines are enacted through a legislative process that is designed to accommodate competing interests and provide opponents of those programs the opportunity to voice their objections. Scholars have advanced powerful arguments that Congress’s design entrenches minority rule, hampers responses to urgent collective action problems, and perpetuates white supremacy.57See, e.g., Adam Jentleson, Kill Switch: The Rise of the Modern Senate and the Crippling of American Democracy (2021); Sanford Levinson, Our Undemocratic Constitution (2006); Robert A. Dahl, How Democratic Is the American Constitution? (2d ed. 2003). Yet for all of Congress’s flaws, enacted legislation has a strong claim to democratic legitimacy. Enacted laws must navigate an obstacle course of veto points and draw support from a wide range of representatives and senators.58See infra text accompanying note 150. Presidents are subject to none of these constraints. Insofar as sabotage privileges the president’s views about the desirability of maintaining statutory programs over congressional judgments expressed in law, sabotage is undemocratic.

Defending their attacks on the statutory programs they administer, agency heads have argued that the president’s popular mandate legitimates administrative sabotage.59For example, in remarks to a bank association, Mulvaney boasted that the CFPB was “a different bureau” under him than under his predecessor Richard Cordray. That was simply “the nature of the business” because “elections do have consequences.” Steve Cocheo, “Elections Have Consequences,” Mulvaney Tells Bankers, Banking Exch. (Apr. 25, 2018, 6:58 PM), []. In doing so, they draw on a body of doctrine and scholarship that celebrates presidential control of agency action.60See, e.g., Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245 (2001); Steven G. Calabresi, Some Normative Arguments for the Unitary Executive, 48 Ark. L. Rev. 23 (1995); Lawrence Lessig & Cass R. Sunstein, The President and the Administration, 94 Colum. L. Rev. 1, 105–06 (1994). This strand of legal scholarship elaborates on Theodore J. Lowi’s analysis of the “plebiscitary president,” which he argued had displaced the original constitutional conception of the presidency. Theodore J. Lowi, The Personal President 97–133 (1985); see also Jide Nzelibe, The Fable of the Nationalist President and the Parochial Congress, 53 UCLA L. Rev. 1217, 1226 & nn.28–33 (2006). Proponents of this “presidentialist” view argue that “because the President has a national constituency, he is likely to consider, in setting the direction of administrative policy on an ongoing basis, the preferences of the general public, rather than merely parochial interests.”61Kagan, supra note 73, at 2335. Thus, the argument goes, an agency that engages in sabotage at the president’s behest does not act undemocratically, because it merely channels the president’s views.

An initial problem with this defense of sabotage is that it misconstrues the presidentialist arguments it invokes. Presidentialists argue that presidential control of administrative agencies is necessary to maintain coherence among programs, allocate resources efficiently, and strengthen the connection between the actions of unelected bureaucrats and electoral politics.62See id. at 2331–45. Thus, presidentialists defend presidential control over agencies because it promotes the faithful execution of the laws.63See, e.g., Seila L. LLC v. CFPB, 140 S. Ct. 2183, 2191 (2020) (“[T]he Constitution gives the President ‘the authority to remove those who assist him in carrying out his duties. Without such power, the President could not be held fully accountable for discharging his own responsibilities; the buck would stop somewhere else.’” (citation omitted) (quoting Free Enter. Fund v. Pub. Co. Acct. Oversight Bd., 561 U.S. 477, 513–14 (2010))); Calabresi, supra note 73, at 58 (“It is to [the President and officers directly accountable to him] that the text of Article II of the Constitution entrusts the power of law execution, and upon them that the same text imposes the duty that the laws be ‘faithfully executed.’”). By contrast, the presidentialist defense of administrative sabotage assumes that the president’s mandate makes it democratically legitimate for him to take any action that they believe advances the national interest, no matter the lawfulness of the agency’s actions. No serious presidentialist argues this.

The more fundamental problem with the defense is its assumption that the president’s gestalt views about the continued desirability of a program are a better reflection of democratic preferences than enacted law. That assumption invokes the president’s national constituency: because the president is elected by voters nationwide, he ostensibly speaks for voters nationwide. But there are a number of problems with the idea that the president’s preferences should serve as a switch that can turn off programs enacted through regular lawmaking processes.

To begin with, when considering the extent to which they represent voters, the proper comparison is not between the president and individual members of Congress, but between the president and Congress as a whole.64Nzelibe, supra note 73, at 1249. Congress as a collective body represents practically the same number of voters as the president.65The qualifier is necessary because residents of the District of Columbia vote for president but are not fully represented in Congress. Thus, in terms of raw numbers of voters they represent, there is no reason to view the president’s views as more democratically legitimate than Congress’s.

One might think that, because the president must appeal to a national constituency to win office, the president’s views are more likely to represent the median national voter. But the idea that the president represents a national constituency is questionable. As Jide Nzelibe shows, the rules governing presidential elections allow a candidate to win with the support of as little as 25 percent of the national electorate.66Nzelibe, supra note 73, at 1234–37. This creates incentives for presidents to govern on behalf of narrow parochial interests rather than the entire nation.

Moreover, the argument that the president’s mandate legitimates sabotage ignores civic republican values that Congress is designed to foster. When an agency engages in sabotage, its actions reflect the will of a single individual whose actions are not guided by law or by predictable, rational norms. The arbitrariness of this exercise of authority is in striking contrast to the kind of deliberative, inclusive decisional process that proponents of civic republicanism celebrate.67E.g., Glen Staszewski, Political Reasons, Deliberative Democracy, and Administrative Law, 97 Iowa L. Rev. 849 (2012); Mark Seidenfeld, A Civic Republican Justification for the Bureaucratic State, 105 Harv. L. Rev. 1511 (1992).

Of course, recognizing that judgments reflected in legislation take priority over the president’s views results in a form of democratic lag. Major legislation is heavily negotiated and takes years if not decades to enact. Accordingly, legislation cannot capture popular preferences as quickly as executive action. But major programs are designed with this lag in mind and delegate authority precisely because of the difficulty of amending enacted law.68See Epstein & O’Halloran, supra note 23, at 29. Moreover, that conventional constitutional arrangements freeze Congress’s judgments in the face of changing circumstances is usually thought to be a feature, not a bug, of the constitutional scheme. Organizing social life around legislation gives individuals a voice in the norms that become law and allows them to organize their affairs around a (comparatively) stable system of legal regulation.69See Jeremy Waldron, Legislation and the Rule of Law, 1 Legisprudence 91, 107 (2007). A lag between popular preferences and enacted law is inherent in the system.

If sabotage is objectionable on constitutional, welfarist, and democratic grounds, however, this is not to say that it can never be normatively justified. Suppose, for example, that President Lincoln had used control of the executive branch to undermine the 1850 Fugitive Slave Act in the decade prior to the Civil War.70See generally Stanley W. Campbell, The Slave Catchers: Enforcement of the Fugitive Slave Law, 1850–1860 (1970). An intentional effort to subvert congressional regulation of slavery would have qualified as sabotage. Yet, seen through modern eyes, no one would say that it would have been normatively objectionable. To the contrary, administrative sabotage of the Fugitive Slave Act would have vindicated the Constitution’s highest ideals. Arguments in favor of the Justice Department’s refusal to defend the Defense of Marriage Act rely on the same logic.71E.g., Joseph Landau, Presidential Constitutionalism and Civil Rights, 55 Wm. & Mary L. Rev. 1719 (2014).

But we should be clear about the logic of these arguments. The claim is not that sabotage is compatible with the constitutional scheme. It is that, in extraordinary circumstances, sabotage may be justified despite its tension with that scheme.

D.    Sabotage’s Legal Position

Although sabotage is objectionable on both constitutional and democratic grounds, it does not follow that courts will stop agencies from engaging in it. In this Section, I argue that sabotage exists in a kind of legal grey area—formally incompatible with the Constitution and agencies’ authorizing legislation, but mostly insulated from judicial scrutiny. This immunity reflects courts’ hesitance to engage the question whether agency action is sabotage. Because of that hesitance, challenges to administrative sabotage cannot rely solely on the fact that it is sabotage.

I first explain the evidentiary and separation of powers problems that prevent courts from inquiring into whether agency action aims to kill or nullify a statutory program. I then trace how those problems inform doctrine under the Take Care Clause and § 706 of the Administrative Procedure Act (APA), the two bodies of law that provide the most obvious avenues for challenging administrative sabotage in court. Finally, I discuss three recent cases in which litigants unsuccessfully challenged agency actions on the ground that they were sabotage.

1.       Evidentiary and Separation of Powers Obstacles to Judicial Review of Agency Sabotage

The understanding of administrative sabotage developed in this Part implies that agency officials act in a particular kind of bad faith when they engage in sabotage: officials charged with administering a statutory program instead use authority created by law to attack it.

As I argued above, the bad faith inherent in administrative sabotage is at odds with the Constitution’s Take Care Clause.72U.S. Const. art. II, § 3. It also conflicts with specific delegations of statutory authority to agencies. When Congress charges an agency with ensuring “that markets for consumer financial products and services are fair, transparent, and competitive”7312 U.S.C. § 5511(a). or with setting air pollution limits “requisite to protect the public health,”7442 U.S.C. § 7409(b)(1). it assumes that the agency will act in good faith to advance policy objectives defined by law. Administrative sabotage is not that. The entire point is to dismantle a disfavored program.

But for two reasons, courts are unlikely to invalidate agency action based on an agency’s failure to carry out a statutory mandate in good faith. The first obstacle is evidentiary. Officials sometimes acknowledge they are engaged in sabotage,75See Robert Pear & Reed Abelson, Foiled in Congress, Trump Moves on His Own to Undermine Obamacare, N.Y. Times (Oct. 11, 2017),‌/politics/trump-obamacare-executive-order.html []. but it is simple to invent technocratic explanations for agency actions designed to undermine a statutory program, and government lawyers routinely offer such explanations in court.76See, e.g., Dep’t of Com. v. New York, 139 S. Ct. 2551, 2557 (2019) (noting the Trump administration’s contention that it sought to add a citizenship question to the 2020 census form “to use in enforcing the Voting Rights Act”); Trump v. Hawaii, 138 S. Ct. 2392, 2404, 2417 (2018) (crediting the Trump administration’s contention that a proclamation implementing Trump’s campaign promise to halt immigration from Muslim countries “sought to improve vetting procedures by identifying ongoing deficiencies in the information needed to assess whether nationals of particular countries present ‘public safety threats’”). This creates a difficult evidentiary problem for courts, because many tools of sabotage are also the stuff of good-faith policy implementation.

Courts can order discovery into agency decisionmakers’ subjective motivations, but this highlights the second obstacle. If assessing decisionmakers’ subjective reasons would not violate the separation of powers, it nevertheless requires courts to probe far into the administrative process. What were the relevant decisionmakers’ thinking at the time of the agency’s action? Do they take the oath of office seriously? Are they telling the truth?

Courts traditionally have avoided such inquiries and instead evaluated agencies’ explanations for their actions in light of the contemporaneous administrative record.77Dep’t of Com., 139 S. Ct. at 2573. As the Supreme Court explained in Department of Commerce v. New York, this approach to judicial review “reflects the recognition that further judicial inquiry into ‘executive motivation’ represents ‘a substantial intrusion’ into the workings of another branch of Government and should normally be avoided.”78Id. (quoting Village of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 268 n.18 (1977)). To respect the executive branch’s prerogatives, courts afford agency actions a “presumption of regularity.”79Note, The Presumption of Regularity in Judicial Review of the Executive Branch, 131 Harv. L. Rev. 2431, 2432 (2018).

The influence of these problems is apparent in doctrine interpreting the two bodies of law that provide the most obvious avenues for challenging administrative sabotage: the Take Care Clause and the APA.

2.       The Take Care Clause

As already noted, the Take Care Clause obligates the president to “take Care that the Laws be faithfully executed.”80U.S. Const. art. II, § 3. Precedent interpreting the Take Care Clause does not answer many questions it raises.81See Goldsmith & Manning, supra note 67, at 1836–38. However, courts have traditionally held that the Clause provides an affirmative basis for relief only when an executive-branch actor is subject to a specific, nondiscretionary duty.82See, e.g., Detroit Int’l Bridge Co. v. Gov’t of Can., 189 F. Supp. 3d 85, 99 (D.D.C. 2016). This distinction traces to Marbury v. Madison, where Chief Justice Marshall distinguished between “cases in which the executive possesses a constitutional or legal discretion” and those in which “a specific duty is assigned by law, and individual rights depend upon the performance of that duty.” 5 U.S. (1 Cranch) 137, 166 (1803). When a court had jurisdiction, Marshall held, a writ of mandamus would issue only to correct violations of the latter category of duties. Id. at 169; see also Mississippi v. Johnson, 71 U.S. (4 Wall.) 475, 499 (1867) (reasoning that for a court to police the president’s performance of discretionary duties under the Take Care Clause would be “an absurd and excessive extravagance” (quoting Marbury, 5 U.S. at 170)). Absent such a duty, allegations that an executive-branch actor failed to execute the law in good faith do not state a claim upon which relief may be granted.83See Detroit Int’l Bridge Co., 189 F. Supp. 3d at 99. Consistent with this view, claims that an executive-branch actor violated the Clause by implementing statutory policy in bad faith appear to be a nonstarter. My research did not locate a single case in which a federal court held that the president or an agency acting under his control violated the Take Care Clause by failing to execute, administer, or enforce a duly enacted law in good faith.84Perhaps the closest that a court has come to affirmatively enforcing the Take Care Clause is National Treasury Employees Union v. Nixon, 492 F.2d 587 (D.C. Cir. 1974). That case involved President Nixon’s failure to implement a pay raise for federal employees that had been approved by Congress. Employees petitioned for a writ of mandamus compelling Nixon to implement the raise, as well as a declaratory judgment that Nixon had failed to comply with the law mandating the raise. Id. at 591. The court concluded that Nixon had unlawfully failed to implement the pay raise but felt compelled “to show the utmost respect to the office of the Presidency and to avoid, if at all possible, direct involvement by the Courts in the President’s constitutional duty faithfully to execute the laws.” Id. at 616. Accordingly, the court limited itself to issuing a declaratory judgment. Id.

3.       Administrative Law

The same evidentiary and separation of powers concerns are reflected in doctrine interpreting the APA. The APA contains two chapters: one establishing procedures for “each authority of the Government of the United States,”855 U.S.C. § 551(1). and another that provides for review of agency action.86Id. §§ 701–706. The centerpiece of the chapter on judicial review, § 706, directs courts to hold unlawful and set aside agency action that is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.”87Id. § 706(2)(A).

By its terms, § 706 seems to apply to administrative sabotage: what greater “abuse of discretion” could there be than sabotaging a statutory program? But the black letter formulation of the arbitrary-and-capricious standard carefully avoids inquiry into executive motive—the crucial question in identifying sabotage. To withstand arbitrary-and-capricious review, an “agency must examine the relevant data and articulate a satisfactory explanation for its action including a ‘rational connection between the facts found and the choice made.’”88Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (quoting Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168 (1962)). Moreover, judicial review ordinarily proceeds on the basis of a record compiled by the agency and submitted to the court.892 Kristen E. Hickman & Richard J. Pierce, Jr., Administrative Law Treatise § 10.5, at 1145–48 (6th ed. 2019). Only when a party makes a “strong showing of bad faith or improper behavior” may a court order discovery into “the mental processes of administrative decisionmakers.”90Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 420 (1971).

Within this framework, a litigant challenging alleged sabotage cannot rest on the fact that the agency acted in bad faith but must also show that the challenged action is substantively unreasonable. The agency will defend its action as a permissible exercise of statutory discretion. Ordinarily, the litigant will not be allowed to go beyond the administrative record to rebut that defense.

The Supreme Court’s recent decision in Department of Commerce v. New York potentially changes this equilibrium in recognizing “pretext” as a basis for finding agency action arbitrary and capricious.91See Dep’t of Com. v. New York, 139 S. Ct. 2551, 2576 (2019). At issue was the Commerce Department’s decision to include a citizenship question on the short-form 2020 census form.92Id. at 2561–62. Writing for the Court, Chief Justice Roberts held that the addition of the citizenship question was not “substantively invalid”—that is, that the addition of the citizenship question was within the Commerce Department’s statutory authority and satisfied the ordinary arbitrary-and-capricious standard.93Id. at 2571 (concluding that the decision to reinstate the citizenship question “was reasonable and reasonably explained, particularly in light of the long history of the citizenship question on the census”). But Roberts went on to hold that the citizenship question was unlawful “because it rested on a pretextual basis.”94See id. at 2573, 2576.

To reach this conclusion, the Court relied on extrarecord discovery that the district court ordered after finding that the Justice Department’s explanation of the Commerce Department’s actions involved bad faith or improper behavior.95Id. at 2573–74; see also New York v. U.S. Dep’t of Com., 339 F. Supp. 3d 144, 147 (S.D.N.Y. 2018), aff’d in part, rev’d in part sub nom. Dep’t of Com. v. New York, 139 S. Ct. 2551. That discovery revealed that Commerce went to “great lengths” to elicit a request to include the citizenship question from the Justice Department and that Commerce and Justice’s rationale for including the citizenship question—that citizenship data would “better enforce” the Voting Rights Act—was “contrived.”96Dep’t of Com., 139 S. Ct. at 2575. The Court remanded the matter to the Commerce Department because its “sole stated reason” for the addition of the citizenship question was pretextual.97Id.

Although a few examples of sabotage might be vulnerable to legal challenge on this ground, Department of Commerce will not have a major effect on judicial review of administrative sabotage. The Court’s conclusion depended on extrarecord discovery that the Supreme Court held was ordered prematurely.98See id. at 2574 (agreeing “with the Government that the District Court should not have ordered extra-record discovery when it did” but concluding that “the new material that the parties stipulated should have been part of the administrative record . . . largely justified such extra-record discovery as occurred”). More importantly, the Court held that “a court may not reject an agency’s stated reasons for acting simply because the agency might also have had other unstated reasons.”99Id. at 2573. Thus, while invalidating the particular abuse of power at issue in that case, Department of Commerce gave a roadmap for using agency power to attack statutory programs. In the future, an agency need only articulate a lawful statutory objective and support it with evidence to avoid the fate of the citizenship question.

4.       Administrative Sabotage in the Courts

When we consider cases where litigants challenged agency action on the ground that it was sabotage, courts have not applied the Take Care Clause or § 706 to examine whether agencies were in fact engaged in sabotage. Instead, they have applied standard administrative law and jurisdictional doctrines to evaluate the lawfulness of agency action or find cases nonjusticiable.

In Texas v. United States, a coalition of states alleged that President Obama and agencies under him violated the Take Care Clause, the APA, and the Immigration and Nationality Act in promulgating the DAPA program, which temporarily exempted certain undocumented immigrants with American children from deportation.100Texas v. United States, 86 F. Supp. 3d 591, 614 (S.D. Tex.), aff’d, 809 F.3d 134 (5th Cir. 2015), aff’d by an equally divided court, 136 S. Ct. 2271 (2016). Quoting statements that Obama made during negotiations on a failed comprehensive immigration reform bill, the complaint alleged that the president “took an action to change the law.”101Complaint for Declaratory & Injunctive Relief at 3, Texas, 86 F. Supp. 3d 591 (No. 14-cv-00254) (emphasis omitted). The theory, Josh Blackman explains, was that DAPA amounted to “a deliberate effort to undermine the laws of Congress, not to act in good faith.”102Blackman, supra note 68, at 219.

The district court preliminarily enjoined DAPA on the ground that it should have been promulgated using notice-and-comment procedures.103Texas, 86 F. Supp. 3d at 677. A divided panel of the Fifth Circuit affirmed the district court’s ruling and added that DAPA was invalid because it conflicted with the Immigration and Nationality Act.104Texas v. United States, 809 F.3d 134, 178, 186 (5th Cir. 2015), aff’d by an equally divided court, 136 S. Ct. 2271 (2016). Both courts declined to address plaintiffs’ argument that “DAPA was not consciously and expressly adopted as a means to enforce the laws of Congress or to conserve limited resources.”105Blackman, supra note 68, at 216. The district court “specifically” noted that it was not addressing “Plaintiffs’ likelihood of success on their . . . constitutional claims under the Take Care Clause/separation of powers doctrine.” Texas, 86 F. Supp. 3d at 677. The Fifth Circuit found it “unnecessary, at this early stage of the proceedings, to address or decide the challenge based on the Take Care Clause.” Texas, 809 F.3d at 146 n.3.

City of Columbus v. Trump, decided four years later, followed the same pattern. A group of cities contended that President Trump and various agencies charged with administering the ACA had violated the APA and the Take Care Clause through “a relentless campaign to sabotage and, ultimately, to nullify the law.”106Complaint for Declaratory & Injunctive Relief at 1, City of Columbus v. Trump, 453 F. Supp. 3d 770 (D. Md. 2020) (No. 18-cv-02364). The complaint alleged that the Trump administration aimed “to pressure Congress to repeal the Act or, if that fails, to achieve de facto repeal through executive action alone.”107Id. It cited a variety of administrative and presidential actions that seemed to present clear-cut examples of sabotage.108See id. at 2 (detailing various agency actions that had “promot[ed] insurance that does not comply with the ACA’s requirements,” “slash[ed] funding for outreach strategies that have been proven to encourage individuals, and healthy individuals in particular, to sign up for coverage,” “misus[ed] federal funds for advertising campaigns aimed at smearing the ACA and its exchanges,” and “impos[ed] unnecessary and onerous documentation requirements”).

The district court dismissed plaintiffs’ Take Care claim because the complaint did not plausibly allege the violation of a nondiscretionary duty.109Columbus, 453 F. Supp. 3d at 803. The court allowed plaintiffs’ APA claims to proceed past a motion to dismiss.110Id. at 795–96. However, the court made clear that it would approach those claims as ordinary requests for judicial review of agency action.111Id. at 796 (reasoning that “dismissal based solely on [the complaint’s] contents would be premature here because a review of the administrative record is necessary to a determination of whether the Secretary’s methodology was arbitrary and capricious”). The court did not suggest that administrative sabotage, if proved, would violate the arbitrary-and-capricious standard.

In Maryland v. United States, the State of Maryland alleged that “the President and his administration have waged, and continue to wage, a relentless campaign to sabotage and nullify the ACA, through legislative and executive action.”112360 F. Supp. 3d 288, 312 (D. Md. 2019). Among other remedies, Maryland sought a declaration that would have precluded the administration from supporting a lawsuit designed to nullify the ACA.113Second Amended Complaint at 35, Maryland, 360 F. Supp. 3d 288 (No. 18-cv-2849). Citing the Take Care Clause, the district court acknowledged that the president’s authority to execute the laws “does not entail the authority to disregard a federal statute.”114Maryland, 360 F. Supp. 3d at 317. But the court ruled the case was not justiciable.115Id.

Under the law as it now stands, administrative sabotage is unlawful but unlikely to be remedied by the courts. Insofar as law regulates administrative sabotage, prohibitions of it have the status of an “underenforced” norm.116For a classic treatment of this concept, see Lawrence Gene Sager, Fair Measure: The Legal Status of Underenforced Constitutional Norms, 91 Harv. L. Rev. 1212 (1978). From the standpoint of constitutional and administrative law doctrine, an allegation that an agency is engaged in sabotage is sauce on the main dish. The allegation might affect how a court applies standard review doctrines. In itself, it is unlikely to provide a basis for judicial relief.

II.      The Political Economy of Administrative Sabotage

Case studies such as the CFPB’s attack on the Consumer Financial Protection Act under Mick Mulvaney and the Trump’s administration’s assault on the ACA provide empirical evidence that administrative sabotage is an actual phenomenon. But why administrative sabotage happens is still something of a theoretical puzzle.

Consider William Niskanen’s “budget maximizing” theory of the bureaucracy.117See Niskanen, supra note 26. Niskanen posited that agency employees want the same material goods as most people.118See id. at 40 (pointing to greater opportunities for promotion and greater job security as reasons for agency employee interest in larger budgets). While an agency does not generate profits that it can pass on to owners and employees, it possesses private information about the cost of regulating that it can use to dupe Congress into increasing its budget.119See id. at 29. Bigger budgets, Niskanen posited, filter down to agency employees in the form of higher wages, better work, and nicer parking spots.120See id. at 40. So Niskanen predicted that bureaucrats would “maximize the total budget of their bureau during their tenure.”121Id. at 42.

For a Niskanenian agency, administrative sabotage is irrational. Why would a budget-maximizing agency attack the very programs it administers—up to requesting nothing to fund its operations?

To answer that question, it helps to move beyond simplified theories of the bureaucracy and situate sabotage within larger trends in American politics, legislation, and administrative policymaking. In this Part, I sketch a theory of sabotage’s origins in three such trends: ideological opposition to an activist federal government, the “stickiness” or entrenchment of programs that are enacted into law, and the rise of an administrative state that is highly responsive to the president’s preferences.

In brief, the argument is that sabotage is the product of the incentives, constraints, and powers of the modern presidency. Presidents face intense pressure to retrench statutory programs, but the textbook lawmaking process is difficult and politically costly to navigate.122For the distinction between “textbook” and “unorthodox” lawmaking processes, see Barbara Sinclair, Unorthodox Lawmaking (5th ed. 2017). Facing this dilemma, the administrative state offers presidents an alternative way of attacking statutory programs that imposes fewer barriers to action than, and offers many of the same rewards as, formal statutory retrenchment.

Sections II.A and II.B review the dilemma for presidents created by political demands for statutory retrenchment and a legislative process that makes satisfying those demands difficult. Section II.C shows how presidents, caught on the horns of this dilemma, use control of agencies to pursue retrenchment that they are unwilling or unable to get from Congress.

A.     Demand for Statutory Retrenchment

Administrative sabotage is a mechanism for deconstructing, dismantling, rolling back, or minimizing federal statutory programs. As such, it must be understood within the context of broader opposition to federal state-building in the United States. Theoretically, such opposition might come from either the political left or the right. But for complicated historical reasons, opposition to federal legislation and regulation comes predominantly from the right. Thus, while left-wing politics occasionally generates demands to dismantle statutory programs,123See, e.g., A. Naomi Paik, Abolitionist Futures and the US Sanctuary Movement, Race & Class, Oct.–Dec. 2017, at 3. sabotage is an “asymmetric” phenomenon that is most likely to occur under conservative presidents.124Cf. Joseph Fishkin & David E. Pozen, Essay, Asymmetric Constitutional Hardball, 118 Colum. L. Rev. 915 (2018) (positing a similar asymmetry in use of “constitutional hardball”); Matt Grossmann & David A. Hopkins, Asymmetric Politics: Ideological Republicans and Group Interest Democrats 14 (2016) (showing that the parties are asymmetric in their composition, approach to politics, and policymaking stances).

A few key moments in modern political history underscore the parties’ asymmetric incentives to engage in administrative sabotage. In the decades after World War II, the Republican Party was captured by the modern conservative movement.125See generally Grossmann & Hopkins, supra note 137; Heather Cox Richardson, To Make Men Free: A History of the Republican Party (2014). Among the movement’s defining positions is a commitment to freedom, individualism, and economic liberty.126See generally Elizabeth A. Fones-Wolf, Selling Free Enterprise: The Business Assault on Labor and Liberalism, 1945–60 (1994). While this free-market, antiregulatory philosophy has been central to conservative politics since the New Deal, Republican officeholders “have increasingly embraced a highly confrontational approach that eschews inside strategies premised on the pursuit of compromise in favor of maximizing partisan conflict, emphasizing symbolic acts of ideological differentiation, and engaging in near-automatic obstruction of initiatives proposed by the opposition.”127Grossmann & Hopkins, supra note 137, at 12. After the Tea Party helped Republicans retake control of the House in 2010, members’ votes increasingly aligned with Tea Party demands on both substantive and procedural issues.128Rachel M. Blum, How the Tea Party Captured the GOP 87–89 (2020); Fishkin & Pozen, supra note 137, at 933; Theda Skocpol & Vanessa Williamson, The Tea Party and the Remaking of Republican Conservatism 4 (2012).

As illustrated by a Tea Partier’s demand to “keep the government’s hands off my Medicare,” conservative opposition to federal regulation is both selective and informed by race, class, and notions of deservingness.129See Skocpol & Williamson, supra note 141, at 60, 79–80. For a somewhat sympathetic account of these attitudes, which hesitates to ascribe them entirely to racial prejudice, see Katherine J. Cramer, The Politics of Resentment: Rural Consciousness in Wisconsin and the Rise of Scott Walker (2016). Thus, conservative demands to retrench federal regulation have not translated into demands to, say, eliminate interior immigration enforcement or defund the police.130In fact, the size of the federal government has grown during periods of Republican control. See, e.g., Lisa Rein, The Federal Government Puts Out a ‘Help Wanted’ Notice as Biden Seeks to Undo Trump Cuts, Wash. Post (May 21, 2021, 6:00 AM),‌/politics/2021/05/21/biden-trump-government-rebuilding [] (noting that size of federal bureaucracy grew 3.4 percent during Trump administration). Even so, opposition to federal regulation remains a defining feature of conservative politics.

This stance was prominent in the early days of the Trump administration. On the heels of Trump’s electoral college victory, Steve Bannon vowed the new administration would fight for the “deconstruction of the administrative state.”131Max Fisher, Stephen K. Bannon’s CPAC Comments, Annotated and Explained, N.Y. Times (Feb. 24, 2017), []. “The way the progressive left runs,” Bannon argued, “is if they can’t get it passed, they’re just going to put in some sort of regulation in an agency.”132Id. Bannon promised: “That’s all going to be deconstructed and I think that’s why this regulatory thing is so important.”133Id.

B.     Sticky Legislation

Politics may generate intense demand for presidents to dismantle statutory programs, but how to satisfy that demand is easier asked than answered. For reasons familiar to scholars of deregulation, repealing enacted programs is both difficult and politically treacherous.134See, e.g., Farhang, supra note 22, at 38; Paul Pierson, Dismantling the Welfare State? Reagan, Thatcher, and the Politics of Retrenchment 18 (1994); Herbert Kaufman, Brookings Inst., Are Government Organizations Immortal? (1976). Enacted programs thus tend to become “sticky” or entrenched.135See Bryan D. Jones, Tracy Sulkin & Heather A. Larsen, Policy Punctuations in American Political Institutions, 97 Am. Pol. Sci. Rev. 151, 151–52 (2003).

This stickiness has both structural and psychological explanations.136See Pierson, supra note 147, at 17–19; Jones et al., supra note 148. As for structure, the many “veto points” in the federal legislative process insulate enacted programs from repeal.137See generally George Tsebelis, Veto Players: How Political Institutions Work (2002) (showing how changing the legislative status quo requires that a certain number of “veto players” in the political system agree to the change). Suppose Congress enacts a new federal program. Following an election, party control of Congress switches, but the prior majority party still controls a veto point such as the Senate filibuster. Control of that veto point allows lawmakers now in the minority to protect the new program.

The psychological explanation involves the way that “loss aversion” interacts with interest group dynamics.138See Pierson, supra note 147, at 18. Organizing collective action is a major obstacle to passing any legislation.139The classic work developing this point is Mancur Olson, The Logic of Collective Action (1965). As Olson explains, someone seeking to persuade Congress to enact legislation must coordinate the actions of thousands of people and pay for collective goods such as legislative drafting and lobbying that no individual would rationally subsidize. Cutbacks to statutory programs, however, “generally impose immediate pain on specific groups, usually in return for diffuse, long-term, and uncertain benefits.”140Pierson, supra note 147, at 18. A credible threat to repeal a statutory program—at least one that generates benefits for statutory beneficiaries—can accordingly be expected to mobilize intense interest-group opposition.141Id. at 19–24.

The unsuccessful effort to “repeal and replace” the ACA illustrates these dynamics.142See Jacob S. Hacker & Paul Pierson, The Dog That Almost Barked: What the ACA Repeal Fight Says About the Resilience of the American Welfare State, 43 J. Health Pol. Pol’y & L. 551 (2018). In the 2016 election cycle, virtually every Republican candidate ran on a platform of repealing “Obamacare.”143David A. Fahrenthold & Jenna Johnson, Republicans’ Obamacare ‘Repeal and Replace’ Dilemma Joins Presidential Contest, Wash. Post (Aug. 18, 2015),‌08/18/b620ee94-45ce-11e5-846d-02792f854297_story.html []. President Trump’s victory gave Republicans control of the House, Senate, and presidency. But the House’s passage of legislation repealing the ACA triggered a massive mobilization to defend the law.144Christopher Ingraham, Here’s a List of Medical Groups Opposing the Cassidy-Graham Health-Care Bill, Wash. Post (Sept. 22, 2017),‌/wp/2017/09/22/heres-a-list-of-medical-groups-opposing-the-cassidy-graham-health-care-bill []; Hacker & Pierson, supra note 155, at 552. With Republicans divided between far-right factions that favored using the ACA repeal to roll back Medicare and Medicaid and centrists who favored more modest reforms, the repeal effort culminated in a dramatic early morning vote on July 28, 2017.145The “Skinny Repeal” attempted to bridge differences among congressional Republicans by repealing central parts of the ACA while putting off the design of replacement legislation. See H.R. 1628, 115th Cong. (2017); see also Mary Ellen McIntire, McConnell Reveals ‘Skinny’ Bill Text as Midnight Vote Looms, Roll Call (July 27, 2017, 10:51 PM),‌/2017/07/27/mcconnell-reveals-skinny-bill-text-as-midnight-vote-looms []. With a simple thumbs down, Senator John McCain killed the “Skinny Repeal,” effectively ending congressional efforts to repeal the ACA.146Peter W. Stevenson, How John McCain’s ‘No’ Vote on Health Care Played Out on the Senate Floor, Wash. Post (July 28, 2017), [].

This is not to say that enacted programs are set in stone. Scholars have found that, on some measures, restructuring or termination of statutory programs is commonplace.147Christopher R. Berry, Barry C. Burden & William G. Howell, After Enactment: The Lives and Deaths of Federal Programs, 54 Am. J. Pol. Sci. 1 (2010). But the failed ACA repeal highlights the barriers a president must overcome to successfully unwind an enacted program. In 2016 and 2017, Republicans ran on an anti-ACA platform, unified around the necessity of repealing the law, and controlled Congress, the presidency, and the Supreme Court. Even with these advantages, repeal proved impossible.

C.     The Administrative State as a Tool of Statutory Retrenchment

Imagine you are a president—probably a conservative one—elected with a mandate to roll back a statutory program. Opposition to the program is central to your political appeal, and voters expect you to use the presidency to deliver promised statutory retrenchment. But a credible threat to repeal the program will mobilize intense interest-group opposition. Even if your party controls Congress, the repeal effort is likely to fail. All of this imposes political costs, diverting time and attention from other priorities and contributing to a perception of fecklessness.

The basic logic of administrative sabotage is to use administrative agencies to do what is difficult or costly to accomplish through Congress. Through control of administrative agencies, the president can kill or nullify a statutory program or lay the groundwork for its repeal or nullification by Congress or the courts. Seen from this perspective, the fact that presidents engage in sabotage reflects the comparative costs and benefits of different modes of retrenching programs they don’t like. All else being equal, presidents might prefer to retrench statutory programs through formal legislation. When that is difficult or politically costly, administrative sabotage offers a close substitute.

Part III surveys tools available to a motivated administration to sabotage a statutory program. This Section works with a wider lens and considers the structural conditions that allow presidents to engage in administrative sabotage. Five features of the contemporary administrative state—none of which is sufficient in itself—facilitate sabotage.

1.       Delegation

To begin, administrative sabotage takes advantage of Congress’s creation of agencies and departments and its broad delegations of statutory authority to them. Since the New Deal, Congress has created scores of agencies and executive departments and granted them power to regulate particular domains, at times through open-ended delegations.148See, e.g., U.S. Gov’t Manual (Nov. 2020 ed. 2020), [] (listing some four hundred federal administrative agencies). Contrary to critics, the story of congressional delegation is not one of ever-increasing agency power.149In the leading empirical study of congressional delegation, David Epstein and Sharyn O’Halloran found that major statutes’ “delegation ratio” (a measure of the discretion that Congress allowed agencies to exercise) declined slightly in the second half of the twentieth century. Epstein & O’Halloran, supra note 23, at 114–16. Later studies are consistent with Epstein and O’Halloran’s finding that Congress adjusts the amount of authority it delegates to agencies in response to information problems, the area in which Congress is regulating, and changes in party control of Congress and the executive branch. See, e.g., Jordan Carr Peterson, All Their Eggs in One Basket? Ideological Congruence in Congress and the Bicameral Origins of Concentrated Delegation to the Bureaucracy, 7 Laws art. 19,; Kathleen M. Doherty & Jennifer L. Selin, Does Congress Differentiate? Administrative Procedures, Information Gathering, and Political Control (Mar. 27, 2015) (unpublished manuscript) (on file with the Michigan Law Review). But there are enough statutory delegations—and they are sufficiently flexible—to allow agencies to sabotage important statutory programs.

2.       Presidential Administration

If agencies were led by individuals who were committed to implementing statutory mandates in good faith, Congress’s delegations of authority would not create a risk of sabotage. But rather than empower independent administrators, public law in recent decades has expanded the president’s control over them. When presidents face pressure to dismantle statutory programs, presidential administration is a transmission belt for sabotage.

As then-Professor Elena Kagan wrote in 2001, presidents beginning with Reagan asserted authority to direct the actions of administrative agencies and portrayed agencies as an extension of the presidency.150See Kagan, supra note 73, at 2247. Formal tools of “presidential administration” include the president’s power to nominate and appoint agency heads, executive orders, formal review of agency policymaking via OMB’s Office of Information and Regulatory Affairs, regulatory czars and interagency working groups, and control of agency budgets and proposed legislation.151See id. at 2270–80, 2284–303. More important than these formal tools is the White House’s informal power to direct agency action. Appointees’ commitment to toeing the party line, back-channel communications, and the “shadow cabinet” allow the White House to control what agencies do while “leaving no tracks.”152See, e.g., Jo Becker & Barton Gellman, Leaving No Tracks, Wash. Post (June 27, 2007), [] (describing Vice President Dick Cheney’s behind-the-scenes interventions to undercut environmental rules).

In the years since Kagan’s article appeared, presidential administration has grown more powerful. At the level of constitutional doctrine, the Roberts Court has broadly embraced the “unitary executive” theory, restricting Congress’s authority to protect agency heads from being removed by the president and suggesting in dicta that any substantial limitation on the president’s authority to direct how the law is executed is incompatible with Article II.153See Blake Emerson, Liberty and Democracy Through the Administrative State: A Critique of the Roberts Court’s Political Theory, 73 Hastings L.J. (forthcoming 2022), []; Seila L. LLC v. CFPB, 140 S. Ct. 2183, 2197, 2205 (2020); Collins v. Yellen, 141 S. Ct. 1761, 1784 (2021). Informal norms have shifted even further in the president’s favor.154See Emerson, supra note 166 (manuscript at 11–18); Jerry L. Mashaw & David Berke, Presidential Administration in a Regime of Separated Powers: An Analysis of Recent American Experience, 35 Yale J. on Regul. 549 (2018).

3.       Partisanship and Polarization

In classic accounts of the U.S. separation of powers system, faithful execution of the law is ensured by Congress and the judiciary protecting their own interests: ambition counteracts ambition.155The Federalist No. 51, at 322 (James Madison) (Clinton Rossiter ed., 1961). But as Daryl Levinson and Richard Pildes observe, “the rise of partisan politics worked a revolution in the American system of separation of powers, radically realigning the incentives of politicians and officeholders.”156Daryl J. Levinson & Richard H. Pildes, Separation of Parties, Not Powers, 119 Harv. L. Rev. 2311, 2321 (2006).

Today, party identification is a far better predictor of how government officials will exercise their authority than the formal position they occupy in Congress or the executive branch.157Id. at 2326. As elected officials and voters have come to identify more strongly with their party, the parties have also become ideologically polarized.158See Matthew Levendusky, The Partisan Sort: How Liberals Became Democrats and Conservatives Became Republicans (2009); Grossmann & Hopkins, supra note 137, at 75–79. According to a standard measure of lawmaker ideology, the ideological difference between the average Republican member of Congress and the average Democratic member has increased nearly twofold since the 1970s, with Republicans moving further to the extreme than Democrats.159See Nolan McCarty, Polarization: What Everyone Needs to Know 30–31 (2019).

Because of partisanship and polarization, one cannot assume that Congress will necessarily check administrative sabotage out of institutional self-interest. To the contrary, “the degree and kind of competition between the legislative and executive branches vary significantly, and may all but disappear, depending on whether the House, Senate, and presidency are divided or unified by political party.”160Levinson & Pildes, supra note 169, at 2315. When, say, the Senate and the presidency are under unified party control, the Senate is unlikely to reject antiregulatory agency heads out of an abstract commitment to faithful execution of the law.161See, e.g., Coral Davenport, Senate Confirms Scott Pruitt as E.P.A. Head, N.Y. Times (Feb. 17, 2017), [] (noting Republicans’ support for the confirmation of Scott Pruitt as EPA administrator despite Pruitt having previously called for the dissolution of agency).

Indeed, a president’s co-partisans in Congress and the judiciary are more likely to support administrative sabotage than oppose it. The attack on the ACA provides an example. In 2018, health-insurance company Anthem withdrew from Wisconsin and Indiana in response to what it artfully described as “continual changes and uncertainty in federal operations, rules and guidance.”162Dan Mangan & Bertha Coombs, Anthem Pulls Out of Obamacare Markets in Wisconsin and Indiana for 2018, CNBC (Jun. 21, 2017, 3:22 PM), []. Speaker Paul Ryan did not decry the unstable regulatory environment that caused Anthem’s withdrawal from the two states. Instead, he cited the move as evidence that “Obamacare is clearly collapsing” and called for replacing the ACA “before more families get hurt.”163Barnini Chakraborty, Anthem Pulls Out of ObamaCare Exchanges in Midwest, Fueling GOP Repeal Push, Fox News (June 21, 2017), [].

4.       Legal Culture

Changes in legal culture have also facilitated sabotage. Courts are increasingly willing to strike or reconfigure statutory programs based on creative legal arguments. This effectively gives opponents of statutory programs a second opportunity to dismantle them after failing to prevail in the political process.164See, e.g., Sitaraman, supra note 18, at 364; Gluck & Scott-Railton, supra note 25, at 517.

These legal challenges draw on well-developed networks of institutional support. As Amanda Hollis Brusky, Steven Teles, and others have documented, the founders of the Federalist Society set out in the early 1980s “to reorient the legal culture” against perceived liberal orthodoxy and lend intellectual legitimacy to the view that “the state exists to preserve freedom.”165See Amanda Hollis-Brusky, Ideas with Consequences: The Federalist Society and the Conservative Counterrevolution 5, 22 (2015); Steven M. Teles, The Rise of the Conservative Legal Movement (2008). From modest beginnings at Yale and the University of Chicago, the society grew into a network of more than forty thousand lawyers, with members or associates in the legal academy, think tanks, the executive branch, and the courts.166Hollis-Brusky, supra note 178, at 1–2.

The success of the Federalist Society and the broader conservative legal movement creates a number of conditions that facilitate administrative sabotage. The movement generates an endless stream of arguments designed to undermine disfavored programs that can be adopted by conservative administrations.167The decade-long campaign to kill the ACA through litigation provides an illustration. See Abbe R. Gluck, Mark Regan & Erica Turret, The Affordable Care Act’s Litigation Decade, 108 Geo. L.J. 1471 (2020); Hollis-Brusky, supra note 178, at 21–22. The Federalist Society’s publications, conferences, meetings, and informal gatherings help to legitimize those arguments. Lawyers, academics, and experts aligned with the movement can be expected to support those arguments, “carrying the ideas of the Federalist Society into their roles as legal professionals, including judges, academics, executive branch officials, litigators, or friends of the court.”168Hollis-Brusky, supra note 178, at 25. Perhaps most importantly, the society now serves as the de facto credentialing authority for conservative judges and even law clerks.169See Lawrence Baum, Hiring Supreme Court Law Clerks: Probing the Ideological Linkage Between Judges and Justices, 98 Marq. L. Rev. 333, 348, 356 (2014). Training and socialization provided by the organization increase the odds that courts will adopt creative antiregulatory arguments that agencies advance.170The transformation of the federal judiciary under President Trump suggests the importance of this development. Trump appointed three Supreme Court justices, 30 % of active appeals-court judges, and 27 % of active district-court judges. About 84 % of Trump’s lower-court appointees were white and 76 % were men. Moreover, Trump’s circuit court appointees “are more uniformly conservative than any of his recent Republican predecessors.” Elena Mejía & Amelia Thomson-DeVeaux, It Will Be Tough for Biden to Reverse Trump’s Legacy of a Whiter, More Conservative Judiciary, FiveThirtyEight (Jan. 21, 2021), []; see also Stacy Hawkins, Trump’s Dangerous Judicial Legacy, 67 UCLA L. Rev. Discourse 20 (2019).

5.       The Presidential Power of Unilateral Action

A final feature of the administrative state that facilitates administrative sabotage is what Terry Moe and William Howell term the “presidential power of unilateral action.”171See Terry M. Moe & William G. Howell, The Presidential Power of Unilateral Action, 15 J.L. Econ. & Org. 132 (1999) (cleaned up). Under the Constitution, the executive branch, Congress, and the courts operate in a perpetual power competition, in which the actual power of each branch depends on whether and how it is checked by the others.172See id. at 135.

But the competition does not proceed on a level playing field. Congressional action requires a proponent of legislation to overcome major collective action problems.173See Olson, supra note 152. Courts only act in response to a proper case or controversy,174See, e.g., Spokeo, Inc. v. Robins, 136 S. Ct. 1540, 1547 (2016). and action by an appellate court generally requires a lower court to act first.175But cf. Stephen I. Vladeck, The Supreme Court, 2018 Term—Essay: The Solicitor General and the Shadow Docket, 133 Harv. L. Rev. 123 (2019) (“As the Justices have interpreted it, the Constitution allows the Court to exercise broad and sweeping appellate jurisdiction that is not limited to review of final rulings by lower Article III and state supreme courts.”). By contrast, when presidents wish to challenge the status quo, “they can simply do it—quickly, forcefully, and (if they like) with no advance notice.”176Moe & Howell, supra note 184, at 138. This gives presidents two advantages in disputes over power allocation. Barriers to presidential action are lower than the barriers to action by Congress or the courts. And presidents’ power to act unilaterally allows them to shift facts on the ground and force other branches to respond to their actions.

This does not mean that presidents are free to do whatever they want. As Moe and Howell argue, however, the big picture is one of “presidents who move strategically and moderately to promote their imperialistic designs—and do so successfully over time, gradually shifting the balance of power in their favor.”177Id. at 153–54.

This creates a welcoming environment for administrative sabotage. Knowing that Congress and the judiciary must organize collective action to respond, presidents are likely to conclude there are few downsides to using agencies to attack statutory programs. In the best case, administrative sabotage will go unchecked because its opponents cannot organize a response or the president’s allies in Congress and the courts will prevent those institutions from responding. At worst, Congress or the courts will check particular acts of sabotage, but as the other branches scramble to respond, agencies’ action will have immediate and perhaps irreversible effects.

*     *     *

The answer to why an agency would try to put itself out of business, then, is that the budget-maximizing model of the bureaucracy does not account for the actual incentives that presidents and agency leaders face in the current political environment. In an environment defined by ideologically opposed parties, partisanship, and intense opposition to federal legislation and regulation, it is not inevitable that agencies will seek to expand their budgets. Often, they will be forces of creative destruction.

III.     How to Sabotage a Federal Program

I have argued so far that administrative sabotage originates in a dilemma for presidents seeking to retrench statutory programs. Facing political and ideological pressure to roll back programs, presidents can pursue retrenchment in Congress. But direct attacks on statutory programs are unlikely to succeed, leaving presidents worse off than they would be had they done nothing at all. Confronted with this dilemma, presidents turn to agencies. Through their power to direct agency action, presidents can kill or nullify statutory programs—or encourage Congress or the courts to do so—while incurring fewer costs than they would if they engaged Congress.

But this presents still another puzzle. From the Take Care Clause to statutes and regulations governing agencies, federal law is full of checks that aim to ensure faithful execution of the law. Given these checks, how can agencies successfully sabotage a statutory program?

To gain traction on that question, this Part surveys the legal and institutional tools that a motivated administration can use to attack statutory programs. Drawing on case studies introduced in Part I and literatures in administrative law, constitutional law, and positive political theory, I consider the extent to which those tools of sabotage are checked by judicial review and congressional oversight.178Of course, congressional oversight and judicial review are not the only checks on agencies. Scholars have explored checks such as public opinion, the “internal” separation of powers, and professional norms and culture. See generally Christopher J. Walker, Constraining Bureaucracy Beyond Judicial Review, Daedalus, Summer 2021, at 155. Although I briefly consider some of these checks, I focus on the external constraints that are the focus of traditional separation of powers theory, leaving analysis of other checks for future work.

This survey shows, first, that presidents have access to an array of tools for sabotaging statutory programs. In addition, the survey shows that many tools of sabotage are effectively insulated from review by courts and Congress. Rather than inevitably provoking an interbranch response as classical separation of powers theory predicts, many acts of sabotage will not provoke any response from Congress or the judiciary. When sabotage is checked by the other branches, the effectiveness of the check depends on partisan dynamics and other contingent historical factors. The overall picture is one of sabotage being checked inconsistently, if not randomly. Given presidents’ strong incentives to engage in sabotage, we should expect to see it continue as a regular feature of agency policymaking, especially but not exclusively during conservative presidential administrations.

Section III.A introduces some basic distinctions among tools of administrative sabotage. Sections III.B and III.C survey the tools of administrative sabotage, beginning with systemic tools that attack institutions and procedures through which statutory programs are administered, then turning to programmatic tools that agencies use to attack statutory programs directly.

A.     Mapping Tools of Sabotage

In mapping the tools of sabotage, it is helpful to begin with some basic distinctions. The most important is between systemic and programmatic forms of sabotage, a distinction suggested by Paul Pierson’s classic work on the politics of welfare reform.179See Pierson, supra note 147.

Pierson argued that politicians seeking to scale down entitlement programs face a choice between direct and indirect strategies. “Systemic” retrenchment strategies “alter the broader political economy” of welfare politics—for example, by lowering tax revenues or weakening the institutions that administer welfare programs.180Id. at 15–17. “Programmatic” strategies, by contrast, involve direct cuts to benefits or restrictions on program eligibility.181Id. at 15. Direct cuts and benefits restrictions impose immediate costs on program beneficiaries and are likely to mobilize intense political opposition.182Id. at 21. Pierson argued that successful efforts to retrench welfare programs would accordingly use systemic strategies.183See id. at 19. Professors Freeman and Jacobs highlight the same difference between direct and indirect strategies for attacking statutory programs in their analysis of “structural deregulation.” Freeman and Jacobs define structural deregulation as presidential undermining of administrative capacity that “targets an agency’s core capacities,” deprives it of essential resources, and diminishes it in the eyes of key stakeholders, making it more difficult for the agency to accomplish its delegated tasks. Freeman & Jacobs, supra note 25, at 587. They distinguish structural deregulation from “substantive” deregulation, which “aims to weaken or rescind particular agency rules or policies.” Id. at 588.

A president using agencies to undermine statutory programs faces a similar choice between indirect and direct strategies. Systemic tools of sabotage do not directly attack statutory programs but involve legal and institutional changes that undermine the government’s ability to administer a program. Programmatic tools, by contrast, seek to kill a program directly.

In addition to their systemic or programmatic nature, we can distinguish among tools of sabotage based on the actors who wield them. Tools such as appointments and executive orders are wielded by the president directly. Other tools are wielded by agency heads or lower-level agency officials.

Finally, we can distinguish among forms of sabotage based on the likelihood that they will be checked through congressional oversight or judicial review, as well as the form that check will take.

Table 1 summarizes major tools of administrative sabotage. A check denotes that use of the tool will likely produce a strong counter-reaction from Congress or the courts. A dash indicates the response will vary with conditions such as partisan control of Congress. An X mark means Congress and the courts are unlikely to check the tool.

B.     Systemic Sabotage

1.       Appointments

The most powerful tool of systemic sabotage is created by the Constitution. Under the Appointments Clause, the president appoints “principal” officers of the United States.184U.S. Const. art. II, § 2. This empowers the president to appoint agency heads who oppose the programs they administer.

For example, as the Introduction noted, President Trump’s choice to lead the CFPB tried unsuccessfully to abolish the agency as a congressman, once describing it as a “sick, sad joke.”185Glenn Thrush, Mulvaney Defends Role Running Agency He Wants to Cripple, N.Y. Times (Apr. 11, 2018), []. Trump’s first appointee to head the Department of Health and Human Services was a former member of the Tea Party Caucus who played a leading role in Republican opposition to the ACA.186Amy Goldstein & Philip Rucker, Trump Names Rep. Tom Price as Next HHS Secretary, Wash. Post (Nov. 29, 2016),‌/11/28/trump-to-name-rep-tom-price-as-next-hhs-secretary [].

Under the Appointments Clause, the appointment of a “principal” officer requires the Senate’s advice and consent.187See Edmond v. United States, 520 U.S. 651, 659 (1997). In theory, the advice-and-consent requirement might check the appointment of agency heads who oppose the programs they administer. But the effectiveness of this check depends on partisan control of the Senate. When the presidency and the Senate are under Republican control, the Senate is likely to accede to the appointment of antiregulatory agency heads, if not celebrate their appointment.188See, e.g., How Senators Voted on Scott Pruitt for E.P.A. Administrator, N.Y. Times (Feb. 17, 2017), []; How Senators Voted on Tom Price for Health and Human Services Secretary, N.Y. Times (Feb. 10, 2017), []. With the elimination of the filibuster for executive-branch nominees, nominees can now be confirmed with a bare majority of senators.189Jeremy W. Peters, In Landmark Vote, Senate Limits Use of the Filibuster, N.Y. Times (Nov. 21, 2013), [].

The Federal Vacancies Reform Act (FVRA) gives the president further power to appoint saboteurs by providing for the appointment of “acting” agency heads without the consent of the Senate.190See 5 U.S.C. § 3345(a). The FVRA limits the tenure of acting agency heads to 210 days.191Id. § 3346(a). But the statute’s complicated timing provisions allow this period to be extended for years when one or more permanent nominations are sent to the Senate.192See Anne Joseph O’Connell, Actings, 120 Colum. L. Rev. 613, 630–31 (2020). In a study of appointments from 1981 to 2020, Anne Joseph O’Connell found that “[a]ll modern Presidents have relied heavily on acting officials” and that nearly half of all cabinet secretaries in her sample were neither confirmed by the Senate nor appointed during a recess.193Id. at 641–43.

Other checks on the appointment of agency saboteurs are weak or depend on contingent historical circumstances. My research did not locate a single case in which a court considered a claim that opposition to a statutory program disqualified an official from serving as an agency’s head. Reporting on administrative sabotage and ethical scandals can slow the pace of sabotage and prompt congressional investigations.194See, e.g., Peter Baker, Glenn Thrush & Maggie Haberman, Health Secretary Tom Price Resigns After Drawing Ire for Chartered Flights, N.Y. Times (Sept. 29, 2017), []. Yet there is no guarantee that agency saboteurs will trigger scandals. And presidents and their appointees can be expected to learn from past scandals in which the revelation of corruption slowed sabotage.

Once in office, agency heads control an array of tools, detailed below, that can be deployed to attack statutory programs. While staff buy-in is necessary for some of these tools, agency heads can use other tools on their own. Agency heads, moreover, can overrule their technical staff195See, e.g., Dep’t of Com. v. New York, 139 S. Ct. 2551, 2571 (2019). and outsource work that requires technical and legal expertise from recalcitrant bureaucrats to more pliant contractors.196See, e.g., Jon D. Michaels, Privatization’s Pretensions, 77 U. Chi. L. Rev. 717, 749 n.120 (2010) (noting complaints that “[d]ozens of critical EPA activities have been turned over to contractors,” who were assigned to “draft rules, review public comments, prepare the final drafts . . . [,] and provide interpretive guidance to the public once those regulations are published”). In authorizing agencies to obtain services from other agencies, the Economy Act, 31 U.S.C. § 1535, provides another mechanism through which agency heads can engage in sabotage without buy-in from lower-level staff. See generally Jason Marisam, The Interagency Marketplace, 96 Minn. L. Rev. 886 (2012).

2.       Nonappointments

The Appointments Clause also empowers presidents to sabotage programs by failing to appoint agency heads—what might be termed “nonappointments.” Many agency functions can only be performed by the agency head,197See, e.g., 12 U.S.C. § 5493(a) (authorizing CFPB director to fix the number of the Bureau’s employees and appoint and direct all employees of the Bureau). a quorum of a multimember commission,198See, e.g., New Process Steel, L.P. v. NLRB, 560 U.S. 674, 679–80 (2010) (holding that the NLRB could not act without a statutory quorum by subdelegating its powers to Board members). or a lower-level official designated by statute or regulation.199See, e.g., Jennifer Nou, Essay, Subdelegating Powers, 117 Colum. L. Rev. 473, 477, 501 (2017). Presidents can stop an agency’s work simply by failing to make necessary appointments. For example, after the Merit Systems Protection Board lost a quorum in 2017, President Trump did not put forward a nominee for fourteen months.200Compare Katz, supra note 37 (indicating that the Merit Systems Protection board lacked a quorum starting in January 2017), with Press Release, White House, Eleven Nominations Sent to the Senate Today (Mar. 12, 2018), [] (indicating that Dennis Kirk and Andrew Maunz were nominated to serve on the Merit Systems Protection Board on March 12, 2018), and Press Release, White House, Seven Nominations Sent to the Senate Today (June 20, 2018), [] (indicating that Julia Clark was nominated to serve on the Merit Systems Protection Board on June 20, 2018). By failing to make a nomination, he effectively halted the Board’s enforcement of the civil service laws.201Systematic analysis of appointment patterns confirms that presidents use nonappointments strategically. Christina M. Kinane found that in the period prior to the 1998 enactment of the FVRA, presidents were less likely to fill positions when they prioritize contracting policies an agency administers. Conversely, presidents were more likely to fill positions with interim appointees when they prioritized a policy area. Christina M. Kinane, Control Without Confirmation: The Politics of Vacancies in Presidential Appointments, 115 Am. Pol. Sci. Rev. 599, 599–600, 612 (2021).

As Professors Freeman and Jacobs observe, agency heads can use the same strategy to hollow out lower-level staff and replace civil servants with political appointees.202See Freeman & Jacobs, supra note 25, at 595–600. During Mulvaney’s thirteen-month tenure as the CFPB’s acting director, the Bureau lost at least 129 of its 1,600 employees.203Robert O’Harrow Jr., Shawn Boburg & Renae Merle, How Trump Appointees Curbed a Consumer Protection Agency Loathed by the GOP, Wash. Post (Dec. 4, 2018), https://www‌ []. In 2020, the Occupational Health and Safety Administration (OSHA) employed only 761 inspectors—an “all-time low” for the agency.204Freeman & Jacobs, supra note 25, at 596; cf. Adam M. Finkel, A Healthy Public Cannot Abide Unhealthy and Unsafe Workplaces, 108 Am. J. Pub. Health 312 (2018) (indicating that OSHA had about two thousand inspectors in 2018).

Apart from calling attention to nonappointments, there is little that actors outside the White House can do to counter them. The Senate’s power of advice and consent creates a natural inertia against the confirmation of agency heads. In the courts, separation of powers doctrine provides nonstatutory remedies for parties who suffer a legal injury at the hands of an “unconstitutionally structured” agency.205David Zaring, Toward Separation of Powers Realism, 37 Yale J. on Regul. 708, 728 (2020); see Elizabeth Earle Beske, Litigating the Separation of Powers, 73 Ala. L. Rev. (forthcoming 2022), []. But in a striking example of administrative law’s bias toward protecting negative liberty, no such remedies are available for parties who are injured by the failure to constitute a lawfully created agency in the first place. Under Article III standing doctrine, courts would likely view such a claim as a “‘generalized grievance’ shared in substantially equal measure by all or a large class of citizens” that does not warrant the exercise of jurisdiction.206Warth v. Seldin, 422 U.S. 490, 499 (1975).

3.       Removal

Presidents’ powers of appointment and nonappointment work hand in hand with their power to remove agency officials.207See Seila L. LLC v. CFPB, 140 S. Ct. 2183, 2191 (2020). When an agency is led by an official who is faithfully executing the agency’s statutory mandate, the president can dismiss her or force her from office.208See, e.g., Ashley Parker & Josh Dawsey, Postmaster General Who Was Target of Trump’s Ire Announces Retirement, Wash. Post (Oct. 16, 2019),‌/608e6286-c51d-11e9-850e-c0eef81a5224_story.html []; Rosalind S. Helderman & Matea Gold, Federal Ethics Chief Who Clashed with White House Announces He Will Step Down, Wash. Post (July 6, 2017), []. Having done so, the president is free to install an agency head who opposes the agency’s mission or to leave the agency headless.

Although the president’s removal power is undoubtedly broad, its precise scope is contested. Congress has attempted to protect certain agency heads by limiting the grounds upon which they may be removed, but conservative justices are engaged in a long-term project to dismantle those protections.

A major recent development in this project came in Seila Law v. CFPB, which involved a provision of the Consumer Financial Protection Act. The provision specified that the CFPB director could be removed only for “inefficiency, neglect of duty, or malfeasance in office.”20912 U.S.C. § 5491(c)(3). In an opinion for five justices, Chief Justice Roberts framed the question as whether “[o]ur precedents” allowed Congress to require that the director of a single-headed agency (that is, one led by a single executive rather than a board) only be removed for cause.210Seila L., 140 S. Ct. at 2192. Removal protections are commonplace in “independent” regulatory commissions,211See Kirti Datla & Richard L. Revesz, Deconstructing Independent Agencies (and Executive Agencies), 98 Cornell L. Rev. 769, 786 (2013). and since 1996, the Social Security Act has provided that the commissioner of the Social Security Administration may be removed only “pursuant to a finding by the President of neglect of duty or malfeasance in office.”21242 U.S.C. § 902(a)(3). However, the Court read Myers v. United States213272 U.S. 52 (1926). A paean to executive power written by Chief Justice (and former President) Taft, Myers held that the Congress could not prevent the president from removing a postmaster. In a lengthy historical analysis, Taft purported to demonstrate that the Constitution impliedly authorizes the president to remove all “executive officers of the United States whom he has appointed by and with the advice and consent of the Senate.” Myers, 272 U.S. at 106. Myerss historiography is disputed by many historians. See John F. Manning, Separation of Powers as Ordinary Interpretation, 124 Harv. L. Rev. 1939, 1964 n.135 (2011). to establish that Article II vests the president with “authority to remove those who assist him in carrying out his duties.”214Seila L., 140 S. Ct. at 2198 (quoting Free Enter. Fund v. Pub. Co. Acct. Oversight Bd., 561 U.S. 477, 513–14 (2010)). Because the CFPB was led by a single director, not a multimember board, “our precedents” did not allow Congress to protect the director from removal. Accordingly, the Court declared the offending provision of the Consumer Financial Protection Act void.215Id. at 2211. In Collins v. Yellen, 141 S. Ct. 1761 (2021), the Court extended Seila Law to agencies that are not engaged primarily in law enforcement, holding that the president could remove the director of the Federal Housing Finance Agency (FHFA) notwithstanding a provision of the Economic Recovery Act of 2008 that provided the director could only be removed for cause. Pub. L. No. 110-289, sec. 1101, § 1312(b)(2), 122 Stat. 2654, 2662 (codified at 12 U.S.C. § 4512(b)(2)).

Some commentators suggest that Seila Law’s overall effect is neutral: proregulatory presidents will use their removal power to fire antiregulatory agency heads, and antiregulatory presidents will use theirs to fire proregulatory agency heads.216See, e.g., Sitaraman, supra note 18, at 355; Richard Cordray, Opinion, Why the CFPB’s Loss at the Supreme Court Is Really a Win, Wash. Post (June 29, 2020), []. Thus, the decision gives with one hand while it takes with the other.

But this analysis ignores the asymmetrical nature of statutory removal protections. Under the typical statutory language, mere policy disagreement is not “cause” for removing an agency head, but faithless execution of the law is.217For an exhaustive historical analysis, see Jane Manners & Lev Menand, The Three Permissions: Presidential Removal and the Statutory Limits of Agency Independence, 121 Colum. L. Rev. 1, 67 (2021). Administrative sabotage is the paradigm case of faithless execution of the law. Thus, the long-term effect of invalidating statutory removal protections is to favor administrative sabotage. When removal protections are enforceable, they protect agency heads who are engaged in good-faith policy implementation while allowing saboteurs to be fired for cause. Seila Law upends this arrangement.

Apart from judicial enforcement of statutory removal protections, checks on the removal power are weak. While presidents are prolific users of the removal power,218See, e.g., David E. Sanger, Biden Team Rushes to Take Over Government, and Oust Trump Loyalists, N.Y. Times (Jan. 29, 2021), []; Jan Diehm, Sam Petulla & Zachary B. Wolf, Who Has Left Trump’s Administration and Orbit?, CNN (Oct. 21, 2019), https://www‌ []. it is rare for firings to lead to congressional investigations.219Among recent examples, Congress has investigated President Trump’s firing of FBI director James Comey, President Obama’s 2009 firing of AmeriCorps’s inspector general, and George W. Bush’s midterm firing of seven United States attorneys. When it comes to more powerful checks, Congress is missing in action. The last president to face censure or impeachment over a removal was Andrew Johnson.220See Josh Chafetz, Congress’s Constitution: Legislative Authority and the Separation of Powers 112–13 (2017).

4.       Executive Orders

While changes to agency leadership might put the gears of sabotage in motion, agencies may need a nudge to overcome inertia or coordinate their work with other agencies. Presidents have discovered that executive orders are a powerful tool for addressing these problems.

In his first official act as president, Trump ordered “the heads of all other executive departments and agencies” to “exercise all authority and discretion available to them to waive, defer, grant exemptions from, or delay the implementation” of the ACA.221Exec. Order No. 13,765, 3 C.F.R. 260 (2017). Executive Order 13,772 directed agencies to “restore public accountability within Federal financial regulatory agencies and rationalize the Federal financial regulatory framework”—a reference to long-standing conservative complaints about Dodd-Frank.222Exec. Order No. 13,772, 3 C.F.R. 286 (2017); see Sitaraman, supra note 18, at 365–66 (describing efforts by Republican senators and business allies to portray the CFPB as an authoritarian Soviet-style agency). And Executive Order 13,771 imposed a government-wide “regulatory budget” and purported to require agencies to rescind two regulations before issuing a new one.223Exec. Order No. 13,771, 3 C.F.R. 284 (2017).

Presidents’ position at the head of the executive branch allows them to issue executive orders quickly, easily, and without advance notice.224As a legal matter, the president may issue an executive order simply by publishing it in the Federal Register. 44 U.S.C. § 1505(a)(1). The process of issuing an executive order is formally governed by a Kennedy-era order that calls for extensive interagency collaboration and lawyering before an executive order issues. Exec. Order No. 11,030, 3 C.F.R. 610 (1959–1963). The Trump administration simply ignored this process in some of its most important executive actions. See, e.g., Kim Soffen & Darla Cameron, How Trump’s Travel Ban Broke from the Normal Executive Order Process, Wash. Post (Feb. 9, 2017),‌/politics/trump-travel-ban-process []. The volume of executive orders, as well as partisan dynamics in Congress, makes congressional scrutiny of any given executive order unlikely.225See Moe & Howell, supra note 184, at 165 (finding, based on a list of all congressional responses to executive orders during the period 1973–1997, that “Congress rarely even attempts to respond to executive orders”). In a solo project, Howell catalogued all the bills introduced in Congress between 1945 and 1998 that “sought to amend, extend, overturn, or codify in law a particular executive order.” William G. Howell, Power Without Persuasion: The Politics of Direct Presidential Action 113 (2003). In the period after 1972, “less than 3 percent of the executive orders issued by presidents received any measure of critical attention by Congress.” Id. at 113, 116. In Franklin v. Massachusetts, the Supreme Court held that the president is not an “agency” under the APA.226505 U.S. 788, 800–01 (1992). Thus, when an executive order merely charges agencies with carrying out a presidential policy, judicial review is generally unavailable until the order is implemented in final agency action, preventing judicial review from being a major check on executive policymaking.2275 U.S.C. § 704 (“Agency action made reviewable by statute and final agency action for which there is no other adequate remedy in a court are subject to judicial review.”); see also Kathryn E. Kovacs, Constraining the Statutory President, 98 Wash. U. L. Rev. 63, 79–82 (2020) (reviewing D.C. Circuit caselaw applying Franklin); Howell, supra note 238, at 154 (finding in a database of eighty-three cases challenging executive orders that courts affirmed the president’s order 83 percent of the time).

5.       Budgeting and Spending

The White House’s role in agency budgeting and spending gives presidents further opportunities to attack statutory programs. The president’s power over the budget traces to the Budget and Accounting Act, which charges OMB with proposing an annual budget to Congress and overseeing the implementation of appropriations legislation.228See Budget and Accounting Act, 1921, ch. 18, § 201(a), 42 Stat. 20, 20 (current version at 31 U.S.C. § 1105(a)). As Eloise Pasachoff describes, OMB’s resource-management offices are pervasively involved in agency budgeting decisions.229See Eloise Pasachoff, The President’s Budget as a Source of Agency Policy Control, 125 Yale L.J. 2182, 2199–2200 (2016). This allows the White House to both bolster and undermine statutory programs.

For example, in 2017 OMB proposed “deep cuts to the Environmental Protection Agency’s budget that would reduce the agency’s staff by one-fifth in the first year and eliminate dozens of programs.”230Juliet Eilperin & Brady Dennis, White House Eyes Plan to Cut EPA Staff by One-Fifth, Eliminating Key Programs, Wash. Post (Mar. 1, 2017),‌/news/energy-environment/wp/2017/03/01/white-house-proposes-cutting-epa-staff-by-one-fifth-eliminating-key-programs []. Congress essentially ignored the administration’s budget proposal, enacting a budget that increased the agency’s funding by approximately 9 percent.231See EPA’s Budget and Spending, EPA, []. But the administration used the appropriation to fund buyouts that hollowed out the agency’s scientific staff. In September 2018, the Washington Post reported that generous retirement packages contributed to an exodus of nearly 1,600 employees from the agency, among them the agencies’ most senior scientific personnel.232Brady Dennis, Juliet Eilperin & Andrew Ba Tran, With a Shrinking EPA, Trump Delivers on His Promise to Cut Government, Wash. Post (Sept. 8, 2018), [].

The president can also shift money from disfavored programs to programs he supports. In 2018, the Department of Homeland Security diverted nearly $10 million from the Federal Emergency Management Authority to Immigration and Customs Enforcement to pay for detention programs.233Isaac Stanley-Becker, Trump Administration Diverted Nearly Million from FEMA to ICE Detention Program, According to DHS Document, Wash. Post (Sept. 12, 2018), []. The small portion of Trump’s border wall that was completed during his presidency was funded from unrelated programs in the Department of Defense’s budget.234Trump’s Border Wall Hits a Wall as Pentagon Cancels Parts Funded from Its Budget, Guardian (Apr. 30, 2021, 5:32 PM), []. Although one court of appeals ruled that the transfers violated the Appropriations Clause and another hinted that it agreed, the Supreme Court stayed injunctions stopping the transfers because of doubts over private parties’ authority to sue.235Sierra Club v. Trump, 929 F.3d 670, 676 (9th Cir.) (concluding that the reallocation of Department of Defense funds for border-wall construction is a violation of the Appropriations Clause), rev’d on other grounds, 140 S. Ct. 1 (2019) (holding that plaintiffs lacked a cause of action to challenge reallocation of funds); U.S. House of Representatives v. Mnuchin, 976 F.3d 1, 14 (D.C. Cir. 2020).

Agency heads with independent budget authority can use it to attack statutory programs.236“Independent” budget authority refers to various mechanisms through which agencies are funded outside of the OMB-managed budget process. See Ctr. on Budget & Pol’y Priorities, Introduction to the Federal Budget Process (2020),‌/sites/default/files/atoms/files/3-7-03bud.pdf []. In January 2018, Mulvaney requested nothing from the Federal Reserve to fund the CFPB’s operations.237Letter from Mick Mulvaney, Acting Dir., CFPB, to Janet L. Yellen, Chair, Bd. of Governors of the Fed. Rsrv. Sys. (Jan. 17, 2018), []. In July 2020, Mulvaney and a protégé urged Kathy Kraninger, Mulvaney’s replacement as CFPB director, to stop drawing funds from the Federal Reserve until Congress altered the Bureau’s structure—effectively holding the Bureau hostage to force changes to its authorizing legislation.238Mick Mulvaney & Eric Blankenstein, Opinion, The Justices’ Wakeup Call to Congress, Wall St. J. (July 14, 2020, 6:57 PM), [].

Under the Constitution, “[n]o Money shall be drawn from the Treasury” except as authorized by Congress.239U.S. Const. art. I, § 9, cl 7. See generally Chafetz, supra note 233, at 45–77. To some extent, Congress’s power of the purse and authority to set overall funding for the government through appropriations bills operates as a check on budgetary sabotage. But Congress’s control over agency budgets diminishes once funds have been appropriated. An example is provided by the 2019 Ukraine scandal. Congress appropriated funds to support Ukraine’s defense against Russian aggression, but OMB withheld the funds as part of a White House–directed pressure campaign to force the Ukrainian government to announce an investigation of Joe Biden.240Office of Management and Budget—Withholding of Ukraine Security Assistance, B-331564, 2020 WL 241373, at 1 (Comp. Gen. Jan. 16, 2020). A Government Accountability Office report concluded that OMB’s actions violated the 2019 Department of Defense Appropriations Act and the Impoundment Control Act, a Nixon-era measure that seeks to prevent the president from failing to spend appropriated money.241Id.; see also Allen Schick, The Federal Budget: Politics, Policy, Process 18–19 (3d ed. 2007). Yet OMB’s failure to spend the funds only was revealed by the filing of an extraordinary whistleblower complaint that the Trump administration unsuccessfully attempted to suppress.242See Zachary B. Wolf & Curt Merrill, The Whistleblower Complaint, Annotated, CNN (Sept. 26, 2019), []; Ellen Nakashima, Whistleblower Sought Informal Guidance from Schiff’s Committee Before Filing Complaint Against Trump, Wash. Post (Oct. 2, 2019), [] (describing the Office of Legal Counsel’s conclusion that the complaint was not an “urgent concern” under the Intelligence Community Whistleblower Protection Act, 5 U.S.C. app. § 8H(i)(1)). Were it not for that complaint and the ensuing political scandal, OMB’s failure to transfer the funds to Ukraine would likely have escaped all congressional scrutiny.

Nor are agency spending decisions subject to regular judicial review. In Lincoln v. Vigil, the Supreme Court ruled that the allocation of a lump-sum appropriation is a matter “committed to agency discretion” under the APA.243508 U.S. 182, 193 (1993) (quoting 5 U.S.C. § 701(a)(2)).

6.       Agency Organization, Office Management, and Agency Procedure

Legislation “may stipulate some basic details of agency structure, such as specific divisions or departments and requirements for key agency personnel.”244Gillian E. Metzger & Kevin M. Stack, Internal Administrative Law, 115 Mich. L. Rev. 1239, 1253 (2017) (footnote omitted). But agency heads resolve many other matters, from the organization of particular offices to who reports to whom.245See id. These choices provide further opportunities for administrative sabotage.

For example, Mulvaney eliminated CFPB enforcement attorneys’ power to issue civil investigative demands246See Jean Veta, Jason Grimes & Eitan Levisohn, Potential Revamp of CFPB Civil Investigative Demand Process, Law360 (Jan. 31, 2018, 1:09 PM),‌/1007257/potential-revamp-of-cfpb-civil-investigative-demand-process []. and moved the Office of Fair Lending and Equal Opportunity out of the CFPB’s enforcement division.247Kate Berry, CFPB’s Mulvaney Strips His Fair-Lending Office of Enforcement Powers, Am. Banker (Feb. 1, 2018, 6:43 PM), []. Wearing his hat as OMB director, Mulvaney later boasted to a group of donors that he had forced a wave of retirements by moving a Department of Agriculture office from Washington, D.C., to Kansas City.248Eric Katz, Mulvaney: Relocating Offices Is a ‘Wonderful Way’ to Shed Federal Employees, Gov. Exec. (Aug. 5, 2019), []; see also Freeman & Jacobs, supra note 25, at 598–99 (“A similar relocation effort by the Department of the Interior’s Bureau of Land Management prompted more than eighty-seven percent of employees asked to move from Washington, D.C., to Grand Junction, Colorado to resign or retire . . . .”). Like other tools of sabotage, office moves can be used for legitimate purposes. Compare David Fontana, Federal Decentralization, 104 Va. L. Rev. 727 (2018) (arguing for federal decentralization as a means of diffusing power), with David Fontana, Destructive Federal Decentralization, 29 Wm. & Mary Bill Rts. J. 619 (2021) (examining the Trump administration’s relocation of federal officials outside of Washington, D.C., to reduce the federal government’s capacity).

Given the size and complexity of the administrative state, Congress cannot hope to monitor all the decisions agency heads make concerning internal agency organization and procedure.249Brian D. Feinstein, Avoiding Oversight: Legislator Preferences and Congressional Monitoring of the Administrative State, 8 J.L. Econ. & Pol’y 23, 25–26 (2011). Nor are those decisions likely to be checked through judicial review. The APA exempts “rules of agency organization, procedure, or practice” from the ordinary notice-and-comment process.2505 U.S.C. § 553(b)(3)(A). In addition, matters “related solely to the internal personnel rules or practices of an agency” are “exempted from disclosure” under the Freedom of Information Act. 5 U.S.C. § 552(b)(2). Apart from this bar, there are serious obstacles to challenging organizational and procedural changes in court. While such changes occasionally give rise to justiciable injuries,251See Lodge 1858, Am. Fed’n of Gov’t Emps. v. Paine, 436 F.2d 882, 892 (D.C. Cir. 1970) (concluding that NASA’s authorizing legislation intended to protect “425 of the many thousands of NASA jobs from” private-sector competition and that employees had standing to challenge demotions and discharges connected to a reduction in force). many cases fail.252See Air Courier Conf. of Am. v. Am. Postal Workers Union, 498 U.S. 517 (1991) (holding that injury to Postal Service employees’ employment prospects caused by liberalization of international courier rules did not give employees standing to challenge changes to the rules); Nat’l Fed’n of Fed. Emps. v. Cheney, 883 F.2d 1038, 1039 (D.C. Cir. 1989) (holding that Army’s decision to “contract out” services formerly provided by government employees of the Directorate of Logistics was not subject to judicial review).

7.       Metarules

One form of agency procedure—what might be termed “metarules”—deserves special mention. These rules leverage the “Accardi principle,” which requires agencies to follow their own regulations until they are properly modified or rescinded.253See United States ex rel. Accardi v. Shaughnessy, 347 U.S. 260, 266–68 (1954); Thomas W. Merrill, The Accardi Principle, 74 Geo. Wash. L. Rev. 569, 569 (2006). By establishing rules for the agency’s own rulemaking process, metarules may weaken or eliminate an agency’s ability to carry out statutory mandates.

The EPA’s “Science Transparency” rule, for example, would have prohibited the agency from relying on scientific studies when the underlying scientific data was not made available to the public.254Strengthening Transparency in Pivotal Science Underlying Significant Regulatory Actions and Influential Scientific Information, 86 Fed. Reg. 469 (Jan. 6, 2021) (to be codified at 40 C.F.R. pt. 30). The agency presented the rule as a commonsense effort to improve the reliability of air-quality regulations.255See Press Release, EPA, EPA Finalizes Rule Strengthening Transparency in Pivotal Science Underlying Significant Regulatory Actions and Influential Scientific Information (Jan. 5, 2021), []. But as the editors of the nation’s leading scientific journals explained, it would have prohibited the agency from relying on gold-standard epidemiological studies that measured the effects of air pollution using private medical records that could not lawfully be disclosed to the public.256See H. Holden Thorp et al., Joint Statement on EPA Proposed Rule and Public Availability of Data, Science (Dec. 6, 2019),

Another metarule proposed by the EPA would have required cost-benefit analyses prepared by the agency to focus on direct, monetizable benefits of regulation, limiting the agency’s ability to consider indirect benefits that justify air-quality regulations.257See Increasing Consistency and Transparency in Considering Benefits and Costs in the Clean Air Act Rulemaking Process, 85 Fed. Reg. 84,130 (Dec. 23, 2020) (to be codified at 40 C.F.R. pt. 83). Meanwhile, the Department of Energy’s Process Rule would have prohibited the department from promulgating energy-efficiency standards for furnaces, water heaters, and boilers.258Lauren Urbanek, New DOE Policy Kneecaps Energy Efficiency Standards, Nat. Res. Def. Council (Jan. 17, 2020), [].

Metarules represent “final” agency action and are subject to review under the APA. Because of legal and analytical errors, courts vacated many Trump-era metarules. For instance, following the change in administrations, in February 2021 a district court vacated the Science Transparency rule because the agency failed to follow notice-and-comment procedures.259Env’t Def. Fund v. EPA, 515 F. Supp. 3d 1135, 1148–51 (D. Mont. 2021). In April 2021, the D.C. Circuit vacated the EPA’s Significant Contribution rule based on a concession by the agency’s new leadership that the rule was arbitrary.260California v. EPA, No. 21-1035, 2021 U.S. App. LEXIS 10426 (D.C. Cir. Apr. 5, 2021); Melissa Horne, Mack McGuffy & Emily Guillaume, Rule Limiting EPA Regulation of GHG Emissions Vacated by D.C. Circuit, JD Supra (Apr. 14, 2021), [].

Congress, however, is unlikely to respond to agencies’ promulgation of metarules for much the same reasons that it is unlikely to respond to changes in agency organization and procedure. The rules merely change the way agencies carry out their functions and thus do not provide a focal point that galvanizes congressional action. And the volume of potential metarules gives agencies the advantage over Congress in battles over the way that agencies do business. Notably, the 117th Congress did not repeal a single Trump-era metarule through the Congressional Review Act, despite Democrats’ unified control of Congress and the presidency and the availability of an expedited, filibuster-proof legislative process under the Act.261Congressional Review Act Tracker, Geo. Wash. Univ. Regul. Stud. Ctr., []; Daniel R. Pérez, Geo. Wash. Univ. Regul. Stud. Ctr., Congressional Review Act Fact Sheet1–2 (2019), [‌/DC45-ZJK6].

8.       Stealth Science

The difficulty of basing legislation on timely, high-quality scientific information and agencies’ comparatively greater access to such information are classic rationales for Congress to delegate regulatory authority to agencies.262See Epstein & O’Halloran, supra note 23, at 48. Administrative law seeks to ensure that agencies use science responsibly, for example by requiring that agencies disclose certain data and analyses.263See Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983); Portland Cement Ass’n v. Ruckelshaus, 486 F.2d 375, 393 (D.C. Cir. 1973). But agencies retain significant discretion over how scientific information is created and interpreted—a power that can be used to attack statutory programs.

Thomas McGarity and Wendy Wagner document ten specific techniques that federal environmental agencies have used to undermine the programs they administer.264Thomas O. McGarity & Wendy E. Wagner, Deregulation Using Stealth “Science” Strategies, 68 Duke L.J. 1719 (2019). They include censoring agency scientists,265Id. at 1724–28 (discussing “policy reviews” of climate-change studies by the U.S. Geological Service and the EPA’s delay of a report finding that formaldehyde caused leukemia). limiting opportunities for scientific input,266Id. at 1727–29 (discussing efforts by the George W. Bush administration to limit advice of independent scientific bodies on climate change). tweaking model inputs and assumptions,267Id. at 1728–33 (discussing an EPA proposal to allow private parties to submit risk-assessment models that the agency was required to evaluate). rewriting scientific reports that might prompt calls for rulemaking,268Id. at 1740–42 (discussing an EPA report edited by political appointees to state that fracking did not have “widespread systemic” impacts on drinking water). and manipulating the rules for scientific deliberation by disbanding or reconstituting advisory panels.269Id. at 1758 (discussing the EPA’s disbanding of two advisory committees that the agency had used for thirty years to assist in the formulation of particulate-matter regulations).

These strategies overlap with other forms of sabotage, and their susceptibility to judicial review depends on the specific form of policymaking that an agency uses to manipulate the scientific process. But stealth science is unlikely to be checked through congressional oversight. After all, Congress’s perceived inability to base regulation on scientific and technical information is often the reason it delegates in the first place.

C.     Programmatic Sabotage

Tools of systemic sabotage change institutional and legal conditions in ways that undermine statutory programs. The indirect nature of these tools obscures their effects on statutory policy, making them less likely to provoke a response from Congress, the courts, and the public. Thus, as Pierson’s work suggests, systemic sabotage is more likely to succeed in retrenching statutory programs.270See Pierson, supra note 147, at 15–17. Activists seeking to undermine federal legislation and regulation share the sentiment. Lisa Rein & Andrew Ba Tran, How the Trump Era Is Changing the Federal Bureaucracy, Wash. Post (Dec. 30, 2017),‌/politics/how-the-trump-era-is-changing-the-federal-bureaucracy/2017/12/30/8d5149c6-daa7-11e7-b859-fb0995360725_story.html [] (quoting a statement by Grover Norquist that “slow and steady” approaches are “how you downsize government without engendering blowback”).

Yet the same indirectness that insulates those tools from checks means that they take time to work. Presidents who are in a hurry to dismantle a statutory program can use another set of programmatic tools. Some of these tools can also be used for systemic sabotage. But in contrast to the tools of systemic sabotage, they aim to directly kill or nullify a statutory program.

1.       Rulemaking and Adjudication

The most basic tools of programmatic sabotage are the staples of agency regulation under the APA: rulemaking and adjudication. In hundreds of statutes, Congress has authorized agencies to develop subsidiary statutory policy that carries the force of law.271See Epstein & O’Halloran, supra note 23, at 114–16. The broad language of these delegations allows agencies to make policy decisions that kill or nullify statutory programs.

For example, an August 2018 regulation issued in part by the Department of the Treasury eliminated restrictions on the sale of short-term health-insurance plans, paving the way for the sale of “junk” insurance plans that the ACA sought to exclude from the insurance marketplace.272Short-Term, Limited-Duration Insurance, 83 Fed. Reg. 38,212 (Aug. 3, 2018) (to be codified at 26 C.F.R. pt. 54); see also Bagley & Gluck, supra note 40. In July 2020, the CFPB issued a revision of its payday lending rule that gave the Bureau’s imprimatur to high-interest loans that force borrowers into “debt traps”—one of the problems that prompted Congress to enact the Consumer Financial Protection Act in 2010.273See Payday, Vehicle Title, and Certain High-Cost Installment Loans, 85 Fed. Reg. 44,382 (July 22, 2020) (to be codified at 12 C.F.R. pt. 1041); Susanna Montezemolo, Ctr. for Responsible Lending, Payday Lending Abuses and Predatory Practices 18 (2013), [‌/J9E9-EV5V].

Of all the tools of administrative sabotage, rulemaking and adjudication are subject to the strongest external checks. Scholars have found that members of Congress keep tabs on major rulemakings and even participate in them.274See Kenneth Lowande & Rachel Augustine Potter, Congressional Oversight Revisited: Politics and Procedure in Agency Rulemaking, 83 J. Pol. 401 (2021) (documenting ways that members of Congress participate in notice-and-comment rulemaking). Adjudication and rulemaking produce final agency action that is subject to judicial review under the APA,275See Norton v. S. Utah Wilderness All., 542 U.S. 55, 61–62 (2004). and empirical analyses of regulatory rollback suggests that such review is a substantial check on agency deregulation.276See Bethany A. Davis Noll, “Tired of Winning”: Judicial Review of Regulatory Policy in the Trump Era, 73 Admin. L. Rev. 353, 366–68 (2021). So why would agencies use rulemaking and adjudication for sabotage when more insulated tools are available?

First, it is not inevitable that all uses of rulemaking and adjudication for sabotage will provoke a reaction from Congress and the courts. Interest groups, nongovernmental organizations, and others regularly seek judicial review of agencies’ actions.277See Roundup: Trump-Era Agency Policy in the Courts, N.Y.U. Sch. of L.: Inst. for Pol’y Integrity (Apr. 1, 2021), [‌/3ZNT-YHGH]. But to do so, a challenger must monitor the agency’s regulatory docket, participate in agency proceedings, locate a suitable petitioner or plaintiff (or demonstrate organizational standing), and file and brief a case.278See Bethany Davis Noll, Trump’s Regulatory ‘Whack-a-Mole, POLITICO (Apr. 10, 2019, 5:13 AM), []. In Steve Bannon’s phrase, an agency can “flood the zone with shit” in the hope that some regulations will avoid review.279See Michael Lewis, Opinion, Has Anyone Seen the President?, Bloomberg (Feb. 21, 2018, 10:31 PM), [] (“The real opposition is the media. And the way to deal with them is to flood the zone with shit.”).

The second reason involves the effect of successful rulemaking and adjudication. Features of administrative law from the arbitrary-and-capricious standard280See FCC v. Fox Television Stations, Inc., 556 U.S. 502 (2009); Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29 (1983). to the Accardi rule281See United States ex rel. Accardi v. Shaughnessy, 347 U.S. 260, 266–68 (1954). make it difficult for agencies to change policy that is established through relatively formal administrative procedures. Thus, rulemaking (and to a lesser extent adjudication) allow agencies to establish “sticky regulations” that later administrations can change only if they are willing to incur substantial costs.282See Aaron L. Nielson, Sticky Regulations, 85 U. Chi. L. Rev. 85, 96–100 (2018); Roundup: Trump-Era Agency Policy in the Courts, supra note 290 (finding that challengers prevailed in approximately 80 percent of challenges to the Trump administration’s efforts to change administrative policies through agency action); cf. David Zaring, Reasonable Agencies, 96 Va. L. Rev. 135, 169 (2010) (finding that “agencies win between 60 and 70% of their appeals with few exceptions”). As Melissa Wasserman argues, these findings are biased by “deference asymmetries” that favor regulated entities in judicial review of agency action. See Melissa F. Wasserman, Deference Asymmetries: Distortions in the Evolution of Regulatory Law, 93 Tex. L. Rev. 625 (2015).

Using rulemaking and adjudication for sabotage is thus a gamble for agencies. Compared to other tools of sabotage, Congress and the courts are more likely to check rulemaking and adjudication. But those tools also offer presidents the prospect of lasting damage to statutory programs.

2.       Big Waiver

A closely related tool is what Judge David Barron and Todd Rakoff term “big waiver.”283See David J. Barron & Todd D. Rakoff, In Defense of Big Waiver, 113 Colum. L. Rev. 265 (2013). Congress increasingly has “give[n] agencies the broad, discretionary power to determine whether the rule or rules that Congress has established should be dispensed with.”284Id. at 267 (emphasis omitted). As Jed Steiglitz has noted, such big-waiver provisions give rise to information problems that allow states to undermine federal safety-net programs.285Edward H. Stiglitz, Forces of Federalism, Safety Nets, and Waivers, 18 Theoretical Inquiries L. 125 (2017). They also invite direct attacks on programs by the federal agencies that exercise waiver authority.286See Elizabeth Y. McCuskey, Big Waiver Under Statutory Sabotage, 45 Ohio N.U. L. Rev. 213 (2019).

Consider how the Department of Health and Human Services (HHS) used waivers to promote Medicaid work requirements. Medicaid’s “primary purpose” is “to enable states to provide medical services to those whose ‘income and resources are insufficient to meet the costs of necessary medical services.’”287Gresham v. Azar, 950 F.3d 93, 100 (D.C. Cir.) (quoting Pharm. Rsch. & Mfrs. of Am. v. Concannon, 249 F.3d 66, 75 (1st Cir. 2001)), cert. granted, 141 S. Ct. 890 (2020). During the Trump administration, HHS nonetheless encouraged states to apply for waivers authorizing them to limit Medicaid eligibility to individuals who were employed or actively seeking work.288See Letter from Thomas E. Price, Sec’y, Dep’t of Health & Hum. Servs., & Seema Verma, Adm’r, Ctrs. for Medicaid & Medicare Servs., to U.S. Governors (Mar. 14, 2017) [hereinafter Price Letter], []. Such work requirements would transform Medicaid from a program that aims to expand health access into a form of government-subsidized health insurance for the working poor.289See Nicole Huberfeld, Can Work Be Required in the Medicaid Program?, 378 New Eng. J. Med. 788 (2018). The proposed change was explicitly premised on disagreement with Congress’s decision to expand Medicaid eligibility in the ACA. In a letter inviting states to apply for waivers, the heads of HHS and the Centers for Medicare and Medicaid Services (CMS) described the ACA’s Medicaid expansion as “a clear departure from the core, historical mission of the program.”290Price Letter, supra note 301, at 1.

Because of collective action and coordination problems, Congress is unlikely to check the use of big-wavier authority for sabotage. Indeed, the structure of big-waiver provisions exacerbates Congress’s institutional disadvantage vis-à-vis the executive by lowering the barriers to executive nullification of statutory requirements.

Recognizing this risk, many statutes require agencies to follow specified procedures when granting waivers and subject waiver decisions to judicial review.291See, e.g., Stiglitz, supra note 298, at 147 (discussing the Medicaid Act’s review provisions). In Gresham v. Azar, the D.C. Circuit concluded that HHS’s approval of Arkansas’s work requirement was arbitrary and capricious because the department failed to account for the waiver’s negative effect on health coverage.292Gresham, 950 F.3d at 102. But just as some regulations and orders will escape judicial review, so will some waivers.293See supra text accompanying note 291. And the standards for assessing the lawfulness of a waiver may allow for waivers that kill or nullify statutory programs. The challenge to Arkansas’s waiver succeeded only because the Medicaid Act requires HHS to find that a waiver was “likely to assist in promoting the objectives” of Medicaid—a requirement that an access-restricting work requirement could not satisfy.29442 U.S.C. § 1315(a). If HHS operated under a waiver provision that was worded differently, Medicaid work requirements might well have withstood judicial review.295Related to big waiver is the phenomenon of “unrules,” or “decisions that regulators make to lift or limit the scope of a regulatory obligation through, for instance, waivers, exemptions, or exceptions.” Cary Coglianese, Gabriel Scheffler & Daniel E. Walters, Unrules, 73 Stan. L. Rev. 885, 885 (2021). Based on a text analysis of the Federal Register, Coglianese, Scheffler, and Walters find that unrules are “pervasive.” Id. at 893.

3.       Friendly Litigation

When agencies do not proceed through rulemaking and adjudication, they still have many tools of programmatic sabotage available. One might be termed friendly litigation. Rather than defend statutory programs from legal attacks, agencies might concede error—or affirmatively support the challengers.

A striking recent example occurred in Texas v. United States.296945 F.3d 355 (5th Cir. 2019), rev’d sub nom. California v. Texas, 141 S. Ct. 2104 (2021). In February 2018, a coalition of nineteen states and Governor Paul LePage of Maine filed suit against the federal government in the Northern District of Texas, seeking a declaration that the ACA was invalid in its entirety.297Texas v. United States, 340 F. Supp. 3d 579, 591 (N.D. Tex. 2018), aff’d in part, 945 F.3d 355 (5th Cir. 2019), rev’d sub nom. California v. Texas, 141 S. Ct. 2104 (2021). The states’ argument exploited the 2017 tax bill’s elimination of the tax penalty for failing to comply with the ACA’s individual insurance mandate.298See Tax Cuts and Jobs Act of 2017, Pub. L. No. 115-97, § 11081(a), 131 Stat. 2054, 2092 (codified at I.R.C. § 5000A(c)(3)(A)) (eliminating the ACA’s shared-responsibility payment). In NFIB v. Sebelius, the Supreme Court concluded that the ACA’s individual insurance mandate was a valid exercise of Congress’s tax power.299567 U.S. 519, 547–57 (2012) (opinion of Roberts, C.J.). Plaintiffs contended that because the tax bill zeroed out the penalty for violating the mandate, the mandate could no longer be upheld as an exercise of the tax power.300Texas, 340 F. Supp. 3d at 591. But plaintiffs didn’t stop there. Adopting the reasoning of the NFIB dissent, they contended that the individual mandate could not be severed from the remainder of the ACA and asked the court to declare the entire law void.301Id.

In the district court, the government declined to defend the individual mandate but argued that all but two provisions of the ACA could be severed from it.302Fed. Defendants’ Memorandum in Response to Plaintiffs’ Application for Preliminary Injunction at 16, Texas, 340 F. Supp. 3d 579 (No. 18-cv-167). The district court adopted the plaintiffs’ argument and invalidated the entire ACA but stayed its ruling pending appeal.303Texas v. United States, 352 F. Supp. 3d 665, 665–66 (N.D. Tex. 2018), aff’d in part, 945 F.3d 355 (5th Cir. 2019), rev’d sub nom. California v. Texas, 141 S. Ct. 2104 (2021). At that point, the government changed positions and urged the Fifth Circuit to declare that the ACA could not be enforced in the eighteen plaintiffs’ states.304See Brief for the Fed. Defendants at 49, Texas v. United States, 945 F.3d 355 (5th Cir. 2019) (No. 19-10011). The Fifth Circuit largely followed the district court’s analysis but remanded for further analysis of the appropriate remedy, following which the Supreme Court granted certiorari.305See Texas, 945 F.3d at 403, rev’d sub nom. California v. Texas, 141 S. Ct. 2104 (2021). In the Supreme Court, the government continued to argue that the entire ACA was invalid.306Brief for the Fed. Respondents at 36, California v. Texas, 141 S. Ct. 2104 (2021) (No. 19-840) (“The individual mandate is not severable from the rest of the Act.”). Only after President Biden took office did the government return to its original view that the individual mandate could be severed from the remainder of the ACA.307Letter from Edwin S. Kneedler, Deputy Solic. Gen, to Scott S. Harris, Clerk, Sup. Ct. of the U.S. (Feb. 10, 2021),‌/20210210151147983_19-840%2019-1019%20CA%20v%20TX.pdf []. The Supreme Court found that the case did not present a justiciable controversy, reversed the Fifth Circuit’s judgment with respect to standing, and directed that the case be dismissed.308California, 141 S. Ct. at 2113, 2120. As commentators noted, the government’s litigating positions in Texas transparently sought the very repeal of the ACA that the GOP tried and failed to secure in the 115th Congress.309See, e.g., Nicholas Bagley, A Case That Should Have Been Laughed Out of Court May Kill Obamacare, Atlantic (Dec. 19, 2019),‌/affordable-care-acts-unconstitutional-flaw/603871 []; Maggie Haberman & Robert Pear, Trump Sided with Mulvaney in Push to Nullify Health Law, N.Y. Times (Mar. 27, 2019), [].

Texas underscores the seriousness of the threat to statutory programs that friendly litigation poses. The success of the conservative legal movement ensures an endless stream of creative arguments against federal legislation and regulation and a bench that is receptive to those arguments. Venue rules allow plaintiffs to handpick the court—and in some cases, the judge—that will hear constitutional and statutory challenges to agency-administered programs.310See Adam Liptak, Texas’ One-Stop Shopping for Judge in Health Care Case, N.Y. Times (Dec. 24, 2018), []. For an excellent analysis of how venue provisions enable plaintiffs to pick a judge through their choice of where to file, see Alex Botoman, Note, Divisional Judge-Shopping, Colum. Hum. Rts. L. Rev., Winter 2018, at 297, 315–20. And the broad scope of federal courts’ remedial authority effectively allows district judges to dictate whether statutory programs continue in effect.311See, e.g., Dep’t of Homeland Sec. v. New York, 140 S. Ct. 599 (2020) (Gorsuch, J., concurring) (“[A] single judge in New York enjoined the government from applying [a new regulatory requirement] to anyone, without regard to geography or participation in this or any other lawsuit.”).

Congress’s main response so far has been to authorize houses of Congress to defend programs that the Justice Department refuses to defend.312See Tami Luhby, New House Rules to Allow Democrats to Defend Obamacare, CNN (Jan. 3, 2019, 8:07 PM), []. But this leaves much to be desired. Congressional defenses of statutory programs depend on divided government. And the House and Senate lack the Justice Department’s resources, expertise, and prestige.

4.       Enforcement Policy

Traditionally, criminal prosecutors and administrative agencies were thought to possess virtually unchecked discretion in choosing which cases to pursue. The classic statement of the rationale for this position appears in Heckler v. Chaney.313470 U.S. 821 (1985). Declining to review the FDA’s decision not to investigate the use of pharmaceuticals for lethal injections, the Supreme Court observed that “[a]n agency generally cannot act against each technical violation of the statute it is charged with enforcing.”314Heckler, 470 U.S. at 823, 831, 837–38. The choice of cases to pursue reflects the agency’s balancing of factors that typically are not governed by law.315Id. at 831 (“[T]he agency must not only assess whether a violation has occurred, but whether agency resources are best spent on this violation or another, whether the agency is likely to succeed if it acts, whether the particular enforcement action requested best fits the agency’s overall policies, and, indeed, whether the agency has enough resources to undertake the action at all.”). Thus, the Court reasoned that when Congress does not legislate enforcement criteria, courts lack standards for judging enforcement decisions.316Id. at 830. In the language of the APA, those decisions are “committed to agency discretion.”3175 U.S.C § 701(a)(2).

When an agency adopts a nonenforcement policy as part of an effort to undermine a statutory program, the power to set enforcement policy recognized in Heckler can be an effective tool of sabotage. In practical terms, a statute that is not enforced may be indistinguishable from one that has been formally repealed. Yet not every nonenforcement policy is sabotage. In considering whether nonenforcement crosses the line, close attention must be paid to the agency’s intentions and the permanence or impermanence of its policy.

The debate over the Obama administration’s immigration enforcement initiatives illustrates how difficult these judgments can be to make, at least for individuals who are not directly involved in the agency decisionmaking process. The administration promulgated two enforcement policies, DACA and DAPA, through memoranda that identified specific types of cases agencies would decline to prosecute.318See Texas v. United States, 809 F.3d 134 (5th Cir. 2015) (describing the Deferred Action for Parents of Americans and Lawful Permanent Residents (DAPA) and Deferred Action for Childhood Arrivals (DACA) programs), aff’d by an equally divided court, 136 S. Ct. 2271 (2016). In contrast to the enforcement policy at issue in Heckler, the Obama policies were made available to the public in advance of specific prosecution decisions. They also set out seemingly rule-like criteria that frontline officers were directed to apply in making enforcement decisions. Going further still, the policies invited immigrants who lacked lawful status to apply for “deferred action,” a long-established form of executive relief from deportation.319Anil Kalhan, Deferred Action, Supervised Enforcement Discretion, and the Rule of Law Basis for Executive Action on Immigration, 63 UCLA L. Rev. Discourse 58 (2015). Under similarly long-standing regulations, successful deferred-action applicants became eligible for a range of benefits—most notably, temporary authorization to work in the United States.3208 C.F.R. § 274a.12(c)(14) (2021).

Implicitly equating categorical nonenforcement with administrative sabotage, some commentators argued the publicity and categorical nature of these policies distinguished them from ordinary exercises of enforcement discretion that Heckler recognizes are within agencies’ authority.321See, e.g., Zachary S. Price, Enforcement Discretion and Executive Duty, 67 Vand. L. Rev. 671, 759–61 (2014); Peter Margulies, President Obama’s Immigration Plan: Rewriting the Law, Lawfare (Nov. 23, 2014, 4:15 PM), []. Because the policies effectively set rules governing primary behavior, they amounted to an attempt to legislate that both intruded on Congress’s legislative power and violated the president’s duty to faithfully execute the law. In a compelling response, Adam Cox and Christina Rodríguez argued that the Constitution and federal immigration laws implicitly delegate authority to the president to set broad-scale enforcement priorities.322Adam B. Cox & Cristina M. Rodríguez, The President and Immigration Law Redux, 125 Yale L.J. 104 (2015). Cox and Rodríguez argued that, rather than reflecting an impermissible attempt to legislate, the categorical, public nature of the Obama initiatives served the rule of law by addressing low-level bureaucrats’ resistance to prior presidential directives, ensuring that like cases were treated alike, establishing transparent enforcement policies, and exposing the president to political accountability for his oversight of the immigration system.323See id. at 112. Given these rule-of-law benefits, the administration’s recognition of the limited legal benefits deferred-action status confers, and the administration’s willingness to be held politically accountable for its enforcement decisions, Cox and Rodríguez argued that the initiatives were a permissible exercise of enforcement discretion.324See id. at 222–24.

Cox and Rodríguez have the better of the debate over the Obama immigration initiatives, but it is not difficult to see why critics view the initiatives as examples of administrative sabotage. If the administration had set out to undermine the Immigration and Nationality Act (and its lawyers were gaslighting the public when they intoned that only Congress could confer lawful status on DACA and DAPA recipients), the initiatives would be examples of sabotage. The Supreme Court’s holding that the recission of DACA was subject to judicial review, which had the effect of entrenching the policy, provides further support for this view.325See Dep’t of Homeland Sec. v. Regents of the Univ. of Cal., 140 S. Ct. 1891, 1911–14 (2020).

Much the same can be said about the federal government’s marijuana enforcement policies. Under the federal Controlled Substances Act, marijuana is a Schedule I drug, which may not be cultivated or distributed except as allowed by federal law.32621 U.S.C. § 841(b)(1)(A). Since 1996, however, thirty-seven states have authorized the use of marijuana for medical purposes and eighteen have decriminalized the use of marijuana for adults over the age of 21.327Jeremy Berke, Shayanne Gal & Yeji Jesse Lee, Marijuana Legalization Is Sweeping the US. See Every State Where Cannabis Is Legal, Insider (July 9, 2021, 9:20 AM), []. A large marijuana industry has developed in these states, subject to “regulation that ranges from robust to haphazard.”328Sam Kamin, Prosecutorial Discretion in the Context of Immigration and Marijuana Law Reform: The Search for a Limiting Principle, 14 Ohio St. J. Crim. L. 183, 187 (2016) (footnote omitted).

In 2009, the Justice Department published a memorandum signed by Deputy Attorney General David Ogden that provided guidance to United States attorneys on enforcement of the Controlled Substances Act in states with medical marijuana laws. The memorandum stated that prosecuting “significant traffickers of illegal drugs” and “the disruption of illegal drug manufacturing and trafficking networks” were “core” priorities of the department.329Memorandum from Deputy Att’y Gen. David W. Ogden to Selected U.S. Att’ys 1 (Oct. 19, 2009), []. It then observed that “[a]s a general matter, pursuit of these priorities should not focus federal resources in your States on individuals whose actions are in clear and unambiguous compliance with existing state laws providing for the medical use of marijuana.”330Id. at 1–2.

The Ogden memorandum “was taken by many as a green light to the development of large-scale, commercial marijuana facilities in the states.”331Kamin, supra note 341, at 189. When DOJ later waffled on its stance toward prosecuting medical-marijuana facilities, Congress began to enact appropriations riders that prohibited the use of funds “to prevent such States from implementing their own State laws that authorize the use, distribution, possession, or cultivation of medical marijuana.”332See, e.g., Commerce, Justice, Science, and Related Agencies Appropriations Act, 2015, Pub. L. No. 113-235, § 538, 128 Stat. 2130, 2217. DOJ does not appear to have set out to undermine the Controlled Substances Act,333Ogden, supra note 342, at 2 (noting that “no State can authorize violations of federal law” and that “the list of factors above is not intended to describe exhaustively when a federal prosecution may be warranted”). but it is not difficult to imagine a similar enforcement policy that was motivated by hostility to a statutory regime. Had DOJ set out to undermine the Act and Congress codified its policy, the Ogden memo would be a clear-cut example of administrative sabotage.

Other enforcement tools can be used for sabotage. For example, settlement agreements,334Authority of the United States to Enter Settlements Limiting the Future Exercise of Executive Branch Discretion, 23 Op. O.L.C. 126, 127 (1999). deferred prosecution agreements,335See Brandon L. Garrett, Structural Reform Prosecution, 93 Va. L. Rev. 853, 864–69 (2007). As Garrett notes, however, deferred prosecution typically gives the government sole and unreviewable discretion to determine whether an agreement has been breached. Id. and industry-wide consent agreements can bind agencies through leadership changes and presidential administrations.336See Deacon, supra note 56, at 813–15. In some agencies, enforcement drives rulemaking, so eliminating enforcement affects the agency’s rulemaking agenda.337See Confessore, supra note 16 (describing how changes in CFPB supervision and enforcement undermined the Bureau’s rulemaking). Still other enforcement policies give rise to legally protected reliance interests, requiring later administrations to go through lengthy administrative processes to change enforcement policies.338See Dep’t of Homeland Sec. v. Regents of the Univ. of Cal., 140 S. Ct. 1891, 1911–14 (2020) (holding that the Department of Homeland Security was required to consider a forbearance-only policy and to account for reliance interests before rescinding the DACA program).

For the most part, these tools are not subject to robust checks. As the marijuana example illustrates, Congress can influence enforcement policy through appropriations and oversight. Heckler’s “presumption of unreviewability” can be overcome when, say, a statute lays down specific criteria for an agency to follow when considering whether to bring cases.339Heckler v. Chaney, 470 U.S. 821, 834–35 (1985); Todd Garvey, Cong. Rsch. Serv., R43708, The Take Care Clause and Executive Discretion in the Enforcement of Law (2014). In addition, the Supreme Court recently reiterated that agency policies that go beyond mere nonenforcement and create substantive benefits are subject to judicial review.340See Regents of the Univ. of. Cal., 140 S. Ct. at 1911–12. Yet even with these qualifications, many enforcement policies fall outside the scope of review.341See, e.g., NLRB v. United Food & Com. Workers Union, Local 23, 484 U.S. 112, 133 (1987) (holding that decision to settle an unfair labor practice charge was not subject to judicial review); Allen v. Wright, 468 U.S. 737, 753 (1984) (holding that private plaintiffs lacked standing to challenge IRS’s failure to enforce statute that required agency to deny tax-exempt status to racially discriminatory private schools); Elec. Priv. Info. Ctr. v. Presidential Advisory Comm’n on Election Integrity, 878 F.3d 371, 375 (D.C. Cir. 2017) (holding that plaintiff lacked standing to challenge defendant agencies’ failure to complete privacy impact assessment before collecting voter data).

5.       Contracting and Grantmaking

Section III.B.5 described how the White House can use agency budgets as a tool of systemic sabotage. Money decisions can also be used to attack statutory programs directly. Consider HHS’s decision to reduce or eliminate funding for ACA “Navigators” under the Trump administration.

Navigators are private-sector groups that help individuals enroll in insurance plans offered through ACA exchanges.342See generally Robert Vargas, How Health Navigators Legitimize the Affordable Care Act to the Uninsured Poor, 165 Soc. Sci. & Med. 263 (2016). In August 2017, the Centers for Medicare and Medicaid Services (CMS) announced it would reduce funding for marketplace outreach by at least 90 percent and for enrollment assistance by about 40 percent.343Sabotage Watch: Tracking Efforts to Undermine the ACA, Ctr. on Budget & Pol’y Priorities (Feb. 2, 2021), []. The following year, CMS cut funding for consumer enrollment assistance and outreach through the Navigator program to $10 million for the thirty-four states whose Affordable Care Act marketplaces are managed by the federal government.344Id. This brought Navigator funding to less than 20 percent of its 2016 level.345Id. In April 2019, CMS finalized a rule that reclassified Navigators’ consumer-assistance functions as “optional.”346Id.

These funding decisions supported Trump’s stated goal of showing that the ACA was “failing.”347See Trump on North Korea: ‘Nobody’s Safe, Fox News (May 24, 2017), []. But the cost of litigation and barriers to congressional oversight meant the decisions were effectively final when they were announced. After one group’s funding was cut from $1.2 million to $129,899, its director told the Washington Post that it would likely shut down.348Juliet Eilperin & Amy Goldstein, HHS Slashes Funding to Groups Helping ACA Consumers Enroll by Up to 92 Percent, Wash. Post (Sept. 14, 2017),‌/national/health-science/hhs-slashes-funding-to-some-aca-navigator-groups-by-more-than-60-percent/2017/09/14/729c394c-9957-11e7-b569-3360011663b4_story.html [].

6.       Propaganda and Disinformation

Finally, executive-branch actors can spread falsehoods, lies, and propaganda about programs they administer. An assortment of statements highlighted by the complaint in City of Columbus v. Trump illustrates the White House’s use of this strategy. Trump claimed that “[t]he Democrats [sic] ObamaCare is imploding,” that “ObamaCare is a broken mess,” and that “Obamacare is virtually dead. At best, you could say it’s in its final legs.”349Complaint for Declaratory & Injunctive Relief at 25–26, City of Columbus v. Trump, 453 F. Supp. 3d 770 (D. Md. 2020) (No. 18-cv-2364). His former chief strategist boasted the administration was undertaking executive action to “blow [the ACA] up” and “blow those [insurance] exchanges up.”350Id. at 25.

Nor is the ACA the only statutory program that executive-branch actors have targeted with propaganda and disinformation. An EPA scientist stated, “Since Trump was elected, the palpable sense is just that they don’t like what we do.”351Freeman & Jacobs, supra note 25, at 618 (quoting Rachel M. Cohen, “I Fully Intend to Outlast These People”: 18 Federal Workers on What It’s Really Like to Work for the Trump Administration, Washingtonian (Apr. 7, 2019), []). During the COVID-19 pandemic, Trump and White House officials disparaged the Centers for Disease Control and reportedly altered agency reports and guidance to conform with the president’s conspiratorial views on the pandemic.352See Dan Diamond, Trump Officials Interfered with CDC Reports on Covid-19, POLITICO (Sept. 12, 2020, 11:11 AM), [].

Propaganda and disinformation can undermine statutory programs through a number of mechanisms. Most immediately, officials’ statements can lead to a program’s failure to accomplish its policy goals—the apparent objective of Trump’s statements concerning ACA exchanges. They can sap agency morale, leading to a hollowing out of the agency’s staff and expertise. And they can diminish an agency’s reputation, increasing the odds that Congress or the courts will dismantle a program.

Of all the tools of sabotage, propaganda and disinformation are the most difficult for Congress and the courts to check. No law regulates high-level officials’ speech, and any effort to do so would face insurmountable First Amendment and separation of powers hurdles. Nor is counter-speech likely to be an effective response to executive branch propaganda and misinformation. Agencies such as the FDA and CDC have a well-deserved reputation for scientific expertise that saboteurs can exploit. And the president is covered obsessively by the press. This gives presidents the advantage in battles to shape popular opinion,353See Katherine Shaw, Beyond the Bully Pulpit: Presidential Speech in the Courts, 96 Tex. L. Rev. 71, 80 (2017). which they are free to use for sabotage.

*     *     *

Other tools of systemic and programmatic sabotage are doubtless available to presidents and their agency heads. But the tools collected in this survey illustrate the basic points: Congress’s delegation of statutory authority gives agencies an array of tools for attacking statutory programs. While agencies are subject to an array of political and legal checks, there is no guarantee that Congress or the courts will check particular instances of sabotage. Some tools of sabotage are virtually immune from external checks. Others will be checked only under contingent political conditions.

These findings, together with Part II’s account of administrative sabotage’s origins, shed light on the unusual amount of sabotage that occurred during the Trump administration. It is tempting to see the Trump-era sabotage as an extension of Trump himself—the natural product of entrusting the federal government to an unaccomplished manager whose populist political brand centered contempt for inherited laws, institutions, and norms. But while presidents vary in their ethics, fealty to law, and commitment to inherited statutory programs, the president’s personal characteristics are not central to the account of sabotage offered here. Sabotage is the product of entrenched political and legal dynamics that operate independently of any particular president. Accordingly, the experience of sabotage in the Trump administration is more likely a harbinger of things to come than a historical aberration limited to his presidency.

IV.     Legal Responses

The primary goals of this Article are to highlight the phenomenon of administrative sabotage and develop a theory of its causes and mechanisms. As the prior Parts argued, administrative sabotage reflects a distinct mode of agency action, which emerges from presidents’ incentives to retrench statutory programs in an environment where formal retrenchment imposes high political costs. For an administration motivated to engage in sabotage, the contemporary administrative state provides a cornucopia of tools.

We have already seen the challenge that sabotage poses to Niskanen’s “budget maximizing” theory of the bureaucracy. That theory views the institutions of the federal government as vehicles for advancing the preferences of agency heads and employees. It does not account for presidents’ incentives to undermine statutory programs. Nor does it consider the tools of sabotage available to presidents and their appointees. This leads the theory to predictions that are at odds with the reality of the contemporary administrative state.

Other well-known theories of the bureaucracy similarly fail to account for the dynamics of sabotage. For example, James Q. Wilson seeks to explain the supposed inefficiency of government agencies compared to their private-sector counterparts.354See James Q. Wilson, Bureaucracy: What Government Agencies Do and Why They Do It (1989). Wilson argues that government agencies are subject to constraints that do not apply to private firms.355See id. at 115, 315–16. This leads agencies to emphasize process and equity over effective task performance.356See id. at 132. Like Niskanen, Wilson does not grapple with agency heads’ ideological incentives to attack the programs they administer or the incentives created by the difficulty of formally amending statutory programs. Even formal theorists of the bureaucracy who recognize the risk of sabotage potentially underestimate it. Sean Gailmard, for example, develops a formal model of delegation in which an agency faces increasing exogenous costs for stepping outside a policy window defined by the legislature.357See Gailmard, supra note 45. That assumption does not hold when members of the president’s party support sabotage and control the levers of power in Congress.

This Article does not aim to develop a theory of the bureaucracy, so I take no position on how sabotage should be integrated into those theories. However, it seems clear that action to dismantle or disable statutory regimes is an important part of what federal administrative agencies now do. As such, theorists should incorporate it into their models of the administrative state.

Beyond these theoretical implications, sabotage raises more immediate practical concerns—most importantly, how the law might prevent and constrain it. Responses to sabotage might come from a variety of actors and institutions: Congress, agency officials, courts, the media, interest groups, and more.358On the prospects for bureaucratic resistance, see, for example, Rosemary O’Leary, The Ethics of Dissent: Managing Guerrilla Government (3d ed. 2019); Gary E. Hollibaugh, Jr., Matthew R. Miles & Chad B. Newswander, Why Public Employees Rebel: Guerrilla Government in the Public Sector, 80 Pub. Admin. Rev. 64 (2020); Jennifer Nou, Civil Servant Disobedience, 94 Chi.-Kent L. Rev. 349 (2019); and Jon D. Michaels, The American Deep State, 93 Notre Dame L. Rev. 1653 (2018). On the role of the media and open-government laws, see, for example, Sellers et al., supra note 29; David E. Pozen, Freedom of Information Beyond the Freedom of Information Act, 165 U. Pa. L. Rev. 1097 (2017); and Jonathan Lash, Katherine Gillman & David Sheridan, A Season of Spoils: The Story of the Reagan Administration’s Attack on the Environment (1984). On the role of organizational culture and professional standards, see Brehm & Gates, supra note 42, and William G. Resh, Rethinking the Administrative Presidency (2015). Similarly, responses might take many different forms, from changes to agencies’ authorizing legislation to actions by low-level agency officials and outside monitors.

As this Article focuses on the law of administrative sabotage, I focus in this Part on the legislation governing agencies and to a lesser extent the law of judicial review. In the federal system, this body of law is a mix of trans-substantive legislation such as the APA, program-specific legislation, and judicial precedent. I contend that while law should address the risk of administrative sabotage, this is better done through changes to the design of particular statutory programs than through cross-cutting reforms.

Section IV.A explains the case against cross-cutting reforms: although those reforms would make it more difficult for agencies to attack statutory programs, they would also interfere with legitimate efforts to implement statutory policy. On balance, their costs to legitimate policy implementation exceed their benefits in checking sabotage. Section IV.B develops the case for program-specific reforms.

A.     Against Cross-Cutting Reforms

Federal administrative agencies are governed by overlapping bodies of law. Congress defines agencies’ authorities and mandates in agency-specific legislation and may require agencies to follow specific procedures. Many of the requirements governing agencies, however, are defined through government-wide legislation such as the APA, which sets out requirements for “each authority of the Government of the United States.”3595 U.S.C. § 551(1). Finally, courts have elaborated a complex body of precedent interpreting both bodies of legislation that some scholars consider an “administrative common law.”360E.g., Gillian E. Metzger, Foreword, Embracing Administrative Common Law, 80 Geo. Wash. L. Rev. 1293, 1293 (2012). In addition to these sources of external law, Professors Metzger and Stack suggest that agencies are governed by “internal administrative law.” See Metzger & Stack, supra note 257, at 1250.

A basic question about legal responses to administrative sabotage is where they should be located within this framework: Is sabotage better addressed through cross-cutting reforms or changes to the design of specific agencies and programs?

Building on the observation that administrative law is biased in favor of protecting negative liberty, cross-cutting reforms would subject agencies to new procedures and forms of judicial review that encourage faithful implementation of statutory mandates. For instance, Congress might amend the APA (or the courts might reinterpret it) to impose stricter procedural controls on executive orders, agency budget decisions, and internal agency reorganizations.361See, e.g., Shalev Roisman, Presidential Law, 105 Minn. L. Rev. 1269, 1320–22 (2021) (proposing that courts require the president to establish that she deliberated before exercising statutory powers); Kovacs, supra note 240, at 98–112 (proposing to subject the Executive Office of the President to the procedural requirements and forms of judicial review that govern other forms of agency action). Congress or the courts might expand the scope of judicial review over enforcement policies and contracting and grantmaking decisions.362E.g., Daniel E. Walters, The Judicial Role in Constraining Presidential Nonenforcement Discretion: The Virtues of an APA Approach, 164 U. Pa. L. Rev. 1911, 1915 (2016) (arguing that challenges to the exercise of nonenforcement discretion be evaluated as ordinary administrative law problems); Lisa Schultz Bressman, Judicial Review of Agency Inaction: An Arbitrariness Approach, 79 N.Y.U. L. Rev. 1657, 1686 (2004) (arguing that courts generally should reject prohibitions on judicial review of agency inaction in favor of requiring agencies to give reasons explaining their decisions and supplying standards disciplining their discretion). They might recognize new mechanisms to force agencies to act when they fail to implement statutory policy.363E.g., Bressman, supra note 375, at 1693; Michael D. Sant’Ambrogio, Agency Delays: How a Principal-Agent Approach Can Inform Judicial and Executive Branch Review of Agency Foot-Dragging, 79 Geo. Wash. L. Rev. 1381, 1434–35 (2011). And reforms might modify standards of judicial review to clarify that administrative sabotage is an abuse of discretion that violates APA § 706.364See supra notes 100–103 and accompanying text (discussing courts’ reluctance to hold that sabotage is an abuse of discretion under the APA).

These reforms, however, would limit agencies’ ability to engage not only in sabotage but also in good-faith policy implementation. Suppose Congress required agencies to follow notice-and-comment procedures before making a material change to their enforcement priorities and made the agencies’ response to comments reviewable under the arbitrary-and-capricious standard. These changes would impede the use of enforcement policy as a tool of sabotage. But they would also frustrate good-faith attempts to refocus enforcement to reflect changed circumstances or the views of a new presidential administration.365Thus, the reasons for resisting cross-cutting administrative law reforms parallel the reasons for resisting what Professor Nicholas Bagley terms “the procedure fetish.” See Nicholas Bagley, The Procedure Fetish, 118 Mich. L. Rev. 345 (2019) (cleaned up). While acknowledging that agency procedure has a role to play in preserving the legitimacy of administrative action and countering agency capture, he contends that strict procedures can exacerbate the very problems they aim to solve: “New procedures should be greeted with suspicion; old procedures should be revisited, with an eye to cutting them back or eliminating them altogether.” Id. at 400–01.

New action-forcing mechanisms are an especially dangerous response to sabotage. For one thing, proposals to expose inaction to judicial review are in tension with the Supreme Court’s understanding of Article III.366In Lujan v. Defenders of Wildlife, the Court described the argument that the violation of a statutory procedure gives interested persons an injury that can be redressed by the courts as “remarkable.” 504 U.S. 555, 572–74 (1992). In another leading precedent, the Court stated that plaintiffs seeking “programmatic improvements” in the administration of statutory programs belonged in “the offices of the [agency] or the halls of Congress.” Norton v. S. Utah Wilderness All., 542 U.S. 55, 64 (2004) (quoting Lujan v. Nat’l Wildlife Fed’n, 497 U.S. 871, 891 (1990)). Against this backdrop, expanding the scope of judicial review to include more forms of agency inaction would inevitably provoke Article III objections. More importantly, the tools could be used by a hostile administration to carry out sabotage.

Suppose, for example, that courts required agencies “to supply explanations for particular nonenforcement decisions” and “require[d] agencies to promulgate standards governing all such decisions.”367Bressman, supra note 375, at 1693. Following a switch from an antiregulatory to a proregulatory administration, an agency seeking to adopt new enforcement policies would have to develop an administrative record that explained the changes to the satisfaction of reviewing courts. An ostensibly action-forcing mechanism could thus be used to lock in an enforcement policy that undermined enforcement of a statute.

One might respond that the burden of justifying changes in enforcement policy is minimal and that courts are unlikely to resist changes that redirect enforcement policies for legitimate purposes. But neither response is compelling. Many observers believe that “reasoned decisionmaking” requirements that courts have imposed on agencies have contributed to the “ossification” of agency regulations.368See Richard J. Pierce, Jr., Rulemaking Ossification Is Real: A Response to Testing the Ossification Thesis, 80 Geo. Wash. L. Rev. 1493, 1493–94 (2012). And as explained above, courts will often be willing participants in administrative sabotage.369For an example of how this is likely to play out, consider the Biden administration’s efforts to temporarily halt deportations at the beginning of the new presidential administration. On January 20, 2021, the acting secretary of homeland security issued a memorandum that ordered a comprehensive review of enforcement policies and directed “an immediate pause on removals of any noncitizen with a final order of removal [with certain exception] for 100 days” while the review was conducted. Memorandum from David Pekoske, Acting Sec’y, U.S. Dep’t of Homeland Sec., to Troy Miller, Senior Off., U.S. Customs & Border Prot., Tae Johnson, Acting Dir., U.S. Immigr. & Customs Enf’t & Tracey Renaud, Senior Off., U.S. Citizenship & Immigr. Servs. (Jan. 20, 2021) (footnote omitted), []. In response to a suit filed by Texas attorney general Ken Paxton, a district judge appointed by President Trump entered a preliminary injunction barring implementation of the hundred-day pause, reasoning that the freeze would injure Texas’s pecuniary interests. Texas v. United States, 524 F. Supp. 3d 598, 619–21 (S.D. Tex. 2021).

B.     Anticipating Sabotage in Statutory Design

If addressing sabotage through cross-cutting reforms is misguided, what is the alternative? In this Section, I argue that sabotage is better addressed through changes to the design of specific programs and agencies and, more broadly, that sabotage requires rethinking conventional wisdom about the design of federal statutory programs.

At the outset, it is important to acknowledge the limitations of this Section’s focus on statutory design. The choices described here are largely resolved through legislation. In arguing that administrative sabotage should be addressed through program-specific legislation, I assume that Congress is capable of revisiting agencies’ authorizing legislation. I do not address the feasibility of amending specific agencies’ authorizing legislation within a particular time frame and recognize the complex political economy of that project.370That said, recent scholarship shows that Congress continues to make a substantial amount of statutory policy, although it is packaged into fewer laws. See James M. Curry & Frances E. Lee, The Limits of Party: Congress and Lawmaking in a Polarized Era (2020); Sean Farhang, Legislative Capacity and Administrative Power Under Divided Polarization, Daedalus, Summer 2021, at 49. The contribution is not to suggest that any particular reform is feasible in the short term but to highlight the importance of addressing sabotage when Congress does act.

In simplest form, the argument is that statutory programs should be designed for sabotage. That is, when designing statutory programs, policymakers should not assume that programs will be administered in good faith by officials who are committed to a program’s objectives. To some extent, this is old news: many choices about institutional and statutory design, agency procedure, and judicial review are already shaped by the recognition that agencies are imperfect agents for enforcing and elaborating statutory policy. Yet the risk of sabotage is largely missing from debates over the design of particular programs. In a seventeen-volume legislative history of the ACA, I found a single reference to “sabotage.”371See Health Care Reform in the 21st Century: A Conversation with Health and Human Services Secretary Kathleen Sebelius: Hearing Before the H. Comm. on Ways & Means, 111th Cong. 65 (2009), reprinted in Health Care Reform: A Legislative History of the Patient Protection and Affordable Care Act, Public Law No. 111-148 (2010) (Bernard D. Reams, Jr. & Michael P. Forrest eds., Supp. VIII 2011), HeinOnline. In a fifteen-volume legislative history of the Dodd-Frank Act, I found none.372See 1–15 Dodd-Frank Wall Street Reform and Consumer Protection Act: A Legislative History (William H. Manz ed., 2010), HeinOnline.

Given the complexity of statutory and agency design, I cannot address all the ways that legislation might be designed to resist sabotage. Instead, I consider sabotage’s implications for three questions that are the subject of long-running debates in public law scholarship: the legal position of agency heads, the scope of statutory delegations to agencies, and the choice to divide or concentrate authority to implement a statutory scheme. In each area, the risk of sabotage complicates long-standing debates; at times, it reorients conventional wisdom.

1.       Constraining the Appointment of Agency Saboteurs

Since the Supreme Court’s decisions in Myers v. United States373272 U.S. 52 (1926). and Humphrey’s Executor v. United States,374295 U.S. 602 (1935). forests have been felled debating statutory provisions that specify a fixed term in office for agency heads and limit the grounds on which they may be removed.375At the Supreme Court level, major post–Humphrey’s Executor cases include Seila Law LLC v. CFPB, 140 S. Ct. 2183 (2020); Free Enterprise Fund v. Public Co. Accounting Oversight Board, 561 U.S. 477 (2010); Morrison v. Olson, 487 U.S. 654 (1988); and Wiener v. United States, 357 U.S. 349 (1958). Compare Saikrishna Prakash, New Light on the Decision of 1789, 91 Cornell L. Rev. 1021, 1072 (2006) (suggesting that majorities in the First Congress believed the Constitution’s grant of executive power enabled the president to remove executive officers at will), and Christopher S. Yoo, Steven G. Calabresi & Laurence D. Nee, The Unitary Executive During the Third Half-Century, 1889–1945, 80 Notre Dame L. Rev. 1, 7 (2004) (arguing that from 1789 to 1889, presidents asserted power to remove subordinate executive officials for any reason), with Peter M. Shane, Madison’s Nightmare: How Executive Power Threatens American Democracy 41–42 (2009) (reviewing constitutional arguments and historical practice and concluding that the removal power posted by unitary-executive proponents does not exist), and Daniel D. Birk, Interrogating the Historical Basis for a Unitary Executive, 73 Stan. L. Rev. 175 (2021) (contending that the unitary-executive theory of the president’s removal power fails to account for the British Crown’s lack of a similar power).

Debates over statutory removal protections are premised on the assumption—prominently articulated in Humphrey’s Executor—that protecting agency heads from being removed by the president encourages faithful policy execution by giving them independence to act in the public interest.376Humphrey’s Executor, 295 U.S. at 625–26 (reasoning that the Federal Trade Commission Act sought “to create a body of experts . . . independent of executive authority, except in its selection, and free to exercise its judgment without the leave or hindrance of any other official or any department of the government”). But as the same passage in Humphrey’s Executor recognizes, this “independence” does not extend to the president’s selection of agency heads.377Id. When a president installs an agency head opposed to a program she administers, removal protections do not ensure faithful policy implementation but protect agency saboteurs.

These dynamics highlight the shortsightedness of focusing on ex post removal protections to the exclusion of checks that ensure agency heads are capable and well-motivated in the first place. People concerned with the faithful implementation of statutory mandates should focus more on ex ante qualifications for agency heads and less on the protection that appointees enjoy once in office.378For a similar proposal concerning acting officials, see Nina A. Mendelson, The Permissibility of Acting Officials: May the President Work Around Senate Confirmation?, 72 Admin. L. Rev. 533, 605 (2020).

This suggestion invites an objection that limiting the selection of agency heads infringes the president’s authority under the Appointments Clause to “nominate” and “appoint” officers of the United States.379See Common Legis. Encroachments on Exec. Branch Auth., 13 Op. O.L.C. 248, 250 (1989) (claiming that such restrictions intrude into the functions and responsibilities assigned by the Constitution to the executive branch). But there is a rich history supporting the constitutionality of ex ante appointment qualifications, beginning with the Judiciary Act of 1789.380Judiciary Act of 1789, ch. 20, § 35, 1 Stat. 73, 92–93 (current version at 28 U.S.C. §§ 503, 541). There, Congress specified that the attorney general and United States attorneys were to be “learned in the law,” substantially limiting the president’s choice of appointees.381Id.; see Henry B. Hogue, Cong. Rsch. Serv., RL33886, Statutory Qualifications for Executive Branch Positions 2–3 (2015).

Ironically, a leading precedent supporting the constitutionality of statutory appointment qualifications is Myers—the high-point of the Court’s unitary-executive jurisprudence. Dissenting from the Court’s holding that the president had inherent authority to fire first-class postmasters, Justice Brandeis protested that “since the foundation of the Government,” Congress had enacted “a multitude of laws” that “limit the President’s power to make nominations.”382Myers v. United States, 272 U.S. 52, 265 (1926) (Brandeis, J., dissenting). Brandeis then compiled ten pages of statutory citations detailing such restrictions.383Id. at 265–74.

The Myers majority disagreed with Brandeis over the president’s power to remove executive officers,384Id. at 176 (majority opinion). but it did not dispute Congress’s authority to impose appointment qualifications. To the contrary, Chief Justice Taft disavowed the suggestion that “denial of the legislative power to regulate removals in some way” entailed “the denial of power to prescribe qualifications for office, or reasonable classification for promotion.”385Id. at 128 (emphases added). As David Lewis has documented, Congress has since enacted hundreds of ex ante appointment qualifications.386See David E. Lewis, Presidents and the Politics of Agency Design: Political Insulation in the United States Government Bureaucracy, 1946–1997, at 50 (2003). While a Reagan administration lawyer active in the movement to revivify the unitary-executive theory asserted that these qualifications infringe the Appointments Clause,387See Common Legis. Encroachments on Exec. Branch Auth., 13 Op. O.L.C. 248, 248 (1989) (William P. Barr). the history is squarely to the contrary. In light of this history, the case against the constitutionality of appointment qualifications is unimpressive.388See Trump v. Mazars USA, LLP, 140 S. Ct. 2019, 2031 (2020) (“[L]ongstanding practice ‘“is a consideration of great weight’” in cases concerning ‘the allocation of power between [the] two elected branches of Government[]’. . . .” (second alteration in original) (quoting NLRB v. Noel Canning, 573 U.S. 513, 524 (2014))). See generally Curtis A. Bradley & Trevor W. Morrison, Historical Gloss and the Separation of Powers, 126 Harv. L. Rev. 411 (2012).

The more important practical question is how appointment qualifications should be enforced if Congress uses them. Such qualifications were traditionally enforced by the Senate, leaving them vulnerable to nullification by the president’s co-partisans.389See Hogue, supra note 394, at 2–3; see also supra note 198 and accompanying text (noting the 115th Congress’s confirmation of an agency head who openly opposed the mission of the agency he led). But more powerful enforcement mechanisms are possible.

Following the model of the APA and the Federal Register Act, legislation could provide that agency action has no legal effect unless authorized by an official who meets statutory appointment qualifications.390See 5 U.S.C. § 552(a)(1) (“Except to the extent that a person has actual and timely notice of the terms thereof, a person may not in any manner be required to resort to, or be adversely affected by, a matter required to be published in the Federal Register and not so published.”); 44 U.S.C. § 1507 (“A document required . . . to be published in the Federal Register is not valid as against a person who has not had actual knowledge of it until the duplicate originals or certified copies of the document have been filed with the Office of the Federal Register and a copy made available for public inspection . . . .”). In recent years, the Supreme Court has recognized common law causes of action and remedies for parties affected by the actions of an “unconstitutionally structured” agency.391See Zaring, supra note 218, at 728–34. Litigants might invoke those precedents to challenge action by appointees who do not satisfy statutory appointment criteria. In addition, Congress has prohibited expenditures of federal funds to pay salaries of officials who do not satisfy statutory appointment qualifications—a particularly strong enforcement mechanism given that spending federal funds in the absence of an appropriation is a felony.392Cf. Public Works for Water, Pollution Control, and Power Development and Atomic Energy Commission Appropriation Act, 1970, Pub. L. No. 91-144, § 502, 83 Stat. 323, 336–37 (barring the use of appropriated funds to pay compensation to persons lacking specified immigration status); 31 U.S.C. § 1350 (authorizing criminal penalties for knowing and willful violations of federal antideficiency laws).

Each of these enforcement mechanisms raises litigable issues. The key point is that focusing exclusively on removal protections overlooks other tools Congress could use to ensure that agency heads are motivated to administer statutory programs in good faith. In a sense, removal protections are irrelevant unless Congress does more to ensure qualified people are appointed in the first instance.

2.       Limiting Statutory Delegations

Another choice affected by the risk of administrative sabotage involves the scope of statutory delegations to agencies. Since the New Deal, conventional wisdom has held that Congress’s limited capacity makes broad statutory delegations to agencies practically inevitable.393U.S. Dep’t of Just., Final Report of the Attorney General’s Committee on Administrative Procedure 14 (1941) (observing that the “total time” available to Congress means that “[t]ime spent on details must be at the sacrifice of time spent on matters of broad public policy”). Going further, many scholars defend delegation as normatively desirable on a variety of grounds.394See, e.g., Jerry L. Mashaw, Reasoned Administration and Democratic Legitimacy (2018); Gillian E. Metzger, The Supreme Court, 2016 Term—Foreword: 1930s Redux: The Administrative State Under Siege, 131 Harv. L. Rev. 1, 7, 88 (2017) (“[D]elegation is ‘the dynamo of modern government.’” (quoting Louis L. Jaffe, Judicial Control of Administrative Action 33 (1965)).

Perhaps it is. But broad delegations also increase the risk of administrative sabotage. First, statutory delegations create the basic authorities that agencies use for sabotage. Second, broad delegations create ambiguity about the scope of agency authority, allowing agencies to present actions designed to kill or nullify statutory programs as ordinary exercises of agency discretion.395See, e.g., Gresham v. Azar, 950 F.3d 93, 97 (D.C. Cir.) (noting the government’s argument that Medicaid work requirements advanced the purposes of Medicaid by “improving health outcomes; . . . address[ing] behavioral and social factors that influence health outcomes; and . . . incentiviz[ing] beneficiaries to engage in their own health care and achieve better health outcomes”), cert. granted, 141 S. Ct. 890 (2020). Third, that ambiguity makes it less likely that administrative sabotage will be checked by judicial and congressional oversight, because an agency can claim that its actions are within the scope of its statutory authority and entitled to judicial deference.

The risk that broad delegations will be used for sabotage does not mean lawmakers should never delegate broadly. In their leading account, David Epstein and Sharon O’Halloran show that Congress’s choice to delegate is usefully analyzed from the perspective of transaction-cost economics.396See Epstein & O’Halloran, supra note 23, at 34–35. Just as a firm considering whether to produce a component in-house or to outsource production will evaluate the transaction costs of both strategies, lawmakers compare the costs of legislating with those of delegating statutory authority to an agency. Neither strategy is costless. The choice depends on which offers the greatest net benefits.397See id. at 49 (“When legislators feel that delegation will increase their electoral opportunities relative to policy making in Congress, then the executive will be given discretionary authority. When legislators perceive greater advantage in making policy themselves, despite the investment of their own scarce time and resources that this involves, they will write specific legislation that leaves the executive with little latitude.”).

Viewed within this framework, the risk of administrative sabotage shows that the costs of delegation are higher than commonly assumed. Epstein and O’Halloran, for example, focus on the risks of agency slack and drift.398See id. at 47–48. The risk that agencies will affirmatively attack the programs they administer does not figure prominently in their account of Congress’s delegation decisions. All else being equal, the possibility that an agency such as the CFPB will attack statutes and programs that it administers implies that Congress should delegate less. In doing so, Congress reduces agencies’ freedom of action, reduces ambiguity about the scope of lawful agency action, and increases the likelihood that sabotage will be checked through judicial review and congressional oversight.

The risk of administrative sabotage thus points to a progressive case for limiting the scope of statutory delegations—a position generally associated with conservative efforts to undermine the administrative state.399See, e.g., Philip Hamburger, Is Administrative Law Unlawful? (2014). The rationale is not that congressional delegation abdicates legislative power, obscures responsibility for policy decisions, or sets up unaccountable agencies. It is that delegation contains the seeds for undoing the very policies Congress wishes to advance.

3.       Fragmenting Policy Implementation

A third question raised by the risk of sabotage involves the choice to concentrate or divide authority to implement statutory programs. In the classic New Deal model of federal legislation, statutory policies were implemented by a single agency whose actions were subject to judicial review.400See Robert L. Rabin, Federal Regulation in Historical Perspective, 38 Stan. L. Rev. 1189, 1247–54 (1986). In the decades since, however, Congress has increasingly divided implementation authority—both among federal agencies and between federal, state, local, and tribal governments.401See, e.g., Bijal Shah, Congress’s Agency Coordination, 103 Minn. L. Rev. 1961, 1978 (2019); Daniel A. Farber & Anne Joseph O’Connell, Agencies as Adversaries, 105 Calif. L. Rev. 1375 (2017); Jody Freeman & Jim Rossi, Agency Coordination in Shared Regulatory Space, 125 Harv. L. Rev. 1131, 1134, 1155 (2012); Tribal Self-Governance Program, Indian Health Serv., []. As a 2006 article observes, “[S]tatutes that parcel out authority or jurisdiction to multiple agencies may be the norm, rather than an exception.”402Jacob E. Gersen, Overlapping and Underlapping Jurisdiction in Administrative Law, 2006 Sup. Ct. Rev. 201, 208.

As a matter of formal theory, dividing implementation authority can lead either to underimplementation or to a “race to the top,” in which agencies competing for resources, attention, or prestige compete to out-regulate one another.403See id. at 214; Freeman & Rossi, supra note 414, at 1150. Still, commentators tend to view statutes that divide implementation authority as the ugly stepsister of statutes that are implemented by a dedicated federal agency.404See, e.g., David A. Super, Laboratories of Destitution: Democratic Experimentalism and the Failure of Antipoverty Law, 157 U. Pa. L. Rev. 541, 545 (2008); Richard B. Stewart, Pyramids of Sacrifice? Problems of Federalism in Mandating State Implementation of National Environmental Policy, 86 Yale L.J. 1196, 1196 (1977). But cf. Richard L. Revesz, Rehabilitating Interstate Competition: Rethinking the “Race-to-the-Bottom” Rationale for Federal Environmental Regulation, 67 N.Y.U. L. Rev. 1210, 1253 (1992) (arguing that interstate competition in environmental regulation is “presumptively beneficial”).

This skepticism stems in part from the fact that proponents of federal health and welfare programs have historically used state implementation to defuse opposition to the expansion of those programs.405See, e.g., Paul Starr, Remedy and Reaction: The Peculiar American Struggle over Health Care Reform 177–81 (rev. ed. 2013) (describing Democratic lawmakers’ decision to pursue a strategy of “minimally invasive reform” in the lead-up to the ACA); Shanna Rose, Financing Medicaid: Federalism and the Growth of America’s Health Care Safety Net 46–50 (2013) (describing Medicaid’s origins in an uncontroversial federal-state program). In large part, however, the skepticism reflects the fact that compared to a world in which policy is zealously administered by a single federal agency, dividing implementation authority creates new opportunities for ideologically motivated actors to contest federal policy.406See generally Jessica Bulman-Pozen & Heather K. Gerken, Essay, Uncooperative Federalism, 118 Yale L.J. 1256 (2009). As Jessica Bulman-Pozen and Heather Gerken argue, nominally “cooperative” structures institutionalize state opposition, giving state implementers vehicles to “resist, challenge, and even dissent from federal policy.”407Id. at 1307.

The literature on uncooperative federalism focuses on state opposition to programs that federal agencies support, but this configuration can be reversed. When federal agencies attack statutory programs, state implementation can be a counterweight to federal sabotage. Two recent case studies highlight these dynamics. In a sweeping study of the ACA’s first decade, Abbe Gluck and Thomas Scott-Railton show that the ACA’s reliance on state implementation helped the statute withstand the Trump administration’s assaults.408See Gluck & Scott-Railton, supra note 25, at 552. State implementation created instability in the ACA’s early years, but it put states in constant negotiation with HHS and created a need for hundreds of state laws and regulations to implement the ACA’s reforms.409Id. at 573. Although states remain sites of resistance and opposition to the ACA, thirty-eight states and the District of Columbia have elected to participate in the Medicaid expansion at the time of writing.410See Status of State Action on the Medicaid Expansion Decision, Kaiser Fam. Found., []. The investments necessary to expand Medicaid and the political settlements they reflect have made undermining the ACA more difficult, whether through formal legislation, executive action, or litigation.

Similar dynamics have helped the Supplemental Nutrition Assistance Program (SNAP) withstand administrative sabotage. As Andrew Hammond recounts, the Trump administration attempted to undermine SNAP through both legislative and administrative action.411Andrew Hammond, Litigating Welfare Rights: Medicaid, SNAP, and the Legacy of the New Property, 115 Nw. U. L. Rev. 361, 364–65 (2020). In the lead-up to the passage of the 2018 farm bill—an $860 billion piece of legislation that determines SNAP funding and eligibility—the House attempted to limit the time that childless, able-bodied adults who do not meet a work requirement can receive SNAP benefits.412Id. at 403. These restrictions would have excluded some 1.2 million people from SNAP but failed in the face of a threatened Senate filibuster.413See id. at 403–04. The same day that Trump signed the 2018 farm bill, Secretary of Agriculture Sonny Perdue proposed a regulation that imposed the same access restrictions the House wanted.414Id. at 404. States swiftly challenged the rule in court.415Id. at 419. Before it took effect, a district court enjoined the rule on the ground that the USDA violated the APA in promulgating it.416Id. at 421. An appeal threat was mooted by an April 2020 stimulus bill that lifted all SNAP work requirements until a month after the end of the COVID-19 public health emergency.417Id.

Hammond argues that SNAP’s federalist structure, combined with “new property” rights that allow states and nongovernmental organizations to challenge administrative changes to benefits and eligibility, played a central role in the failure of the proposed benefit cuts.418See id. at 426. SNAP benefits are funded entirely by the federal government, but the costs of administering the program are split between the federal government and the states.419Id. at 427. This gives rise to “an unholy, but not unstable alliance of state government, federal courts, and public interest lawyers.”420Id. SNAP’s ideological opponents work across federal and state lines—and across Congress and the administrative state—to impose new eligibility requirements and cut benefits. Yet when states acting in concert with federal agencies erect new procedural requirements, the “unholy alliance” ensures that the lawfulness of the requirements will be vigorously challenged.

Gluck, Scott-Railton, and Hammond focus on the antisabotage effects of dividing implementation authority between the federal government and the states, but similar dynamics can check sabotage when implementation authority is divided across federal agencies. Sharon Jacobs, for example, traces the division of authority between the Department of Energy (DOE) and the Federal Energy Regulatory Commission (FERC) in the Department of Energy Organization Act of 1977.421Department of Energy Organization Act, Pub. L. No. 95-91, 91 Stat. 565 (1977) (codified in scattered sections of 42 U.S.C.); see Sharon B. Jacobs, The Statutory Separation of Powers, 129 Yale L.J. 378 (2019). Enacted in the aftermath of Watergate and the 1973 oil crisis, the Act divides authority over electricity rates between DOE and FERC in an effort to limit presidential influence over energy policy.422See Jacobs, supra note 434, at 411. For example, DOE may “propose rules, regulations, and statements of policy of general applicability with respect to any function” within FERC’s jurisdiction, 42 U.S.C. § 7173(a). But apart from acting on the proposal “in an expeditious manner in accordance with such reasonable time limits as may be set by the Secretary,” the decision to adopt, reject, or delay action on DOE proposals is left to FERC. Id. § 7173(b). As Jacobs explains, this structure checked the Trump administration’s efforts to prop up the ailing coal industry.423Jacobs, supra note 434, at 418 (describing FERC’s rejection of DOE’s fuel-secure energy pricing rule).

As with delegation, the point is not that dividing implementation is a cure-all for administrative sabotage. That the ACA and SNAP withstood administrative sabotage was not preordained, and one can imagine how the programs would have fared worse under slightly different conditions. Gluck, Scott-Railton, and Hammond all caution that institutional design must pay careful attention to the specific factors and historical developments that led to those programs’ ability to withstand attacks. Nonetheless, if advocates of federal entitlements have traditionally resisted programs that disperse implementation authority among agencies and governments, the risk of administrative sabotage is a reason to revisit that choice.424There is a long history of dividing authority to enforce and administer the laws within the executive branch, which dates to the earliest legislation enacted under the Constitution. That history would seem to foreclose most constitutional objections to fragmenting implementation authority as a means of checking administrative sabotage. See Blake Emerson, The Departmental Structure of Executive Power: Subordinate Checks from Madison to Mueller, 38 Yale J. on Regul. 90, 119–24 (2021).


This Article has argued that delegating authority to agencies does not only create risks that they will exceed their statutory authority, shirk, or depart from Congress’s policy preferences. It also creates a risk that agencies will attempt to kill or nullify the very programs they administer.

For students of the administrative state, understanding the causes and mechanisms of administrative sabotage is important because sabotage is an increasingly important part of what agencies do. To understand “what government agencies do and why they do it,”425Wilson, supra note 367 (cleaned up). scholars must account for the phenomenon of sabotage and the unique legal, institutional, and normative questions it raises.

For courts, the payoffs from understanding sabotage are less immediate. Because of courts’ reluctance to examine the subjective reasons for agency action, whether specific agency actions constitute sabotage will often be unclear in the context of specific cases. Still, courts should be sensitive to the dynamics of administrative sabotage and willing to use the tools that are available to them to address it. They should also recognize when invitations to deploy constitutional doctrines are part of an effort to sabotage a statutory regime and think twice about participating in the project.

For policymakers involved in the design of statutory programs, however, the payoffs from understanding sabotage are enormous. Particularly since the public interest era of the late 1960s, the design of federal statutory programs has been sensitive to the agency costs created by Congress’s reliance on agencies to elaborate and enforce statutory policy. Thus, it is common for Congress to legislate reporting requirements, consultation requirements, private rights of action, deadlines, and other checks on faithless execution of the law when it enacts or revises a program. Yet despite all of these checks, statutory programs continue to be premised on the assumption that they will be implemented in good faith by agencies who are committed to advancing the program’s policy objectives.

This Article’s overarching contribution is to highlight the dangers of that assumption. Law can no more eliminate the risk of agency sabotage than it can eliminate the risk that agencies will act beyond their statutory authority. But if Congress is to continue relying on agencies, lawmakers must be sensitive to the risk that agencies will attack the very programs they administer.

* Professor of Law, Rutgers Law School. This Article benefited from presentations at the American Association of Law Schools’ “New Voices in Administrative Law” program, the Rutgers Law School faculty workshop, the Seton Hall Law School faculty workshop, and from discussions at an ad hoc Zoom workshop organized by the indomitable Zachary Clopton. Thank you to participants at those events and to Nick Almendares, Pamela Bookman, Jessica Clarke, Bethany Davis Noll, Daniel Deacon, Blake Emerson, Andrew Hammond, Brian Highsmith, Ron Levin, Jon Michaels, Michael Morley, Justin Murray, William Resh, Ganesh Sitaraman, Jed Steiglitz, Peter Strauss, Dan Walters, Reid Weisbord, and Adam Zimmerman for feedback and discussion. Matt Bodi, John Meyer, Carly Siditsky, and the Michigan Law Review staff provided invaluable research and editorial assistance.