Responding to Alternatives

This Article is the first to comprehensively analyze administrative agencies’ obligation to respond to alternatives to their chosen course of action. The obligation has been around at least since the Supreme Court’s decision in Motor Vehicle Manufacturers Ass’n of the United States, Inc. v. State Farm, and it has mattered in important cases. Most recently, the Supreme Court invoked the obligation as the primary ground on which to invalidate the Trump Administration’s rescission of the Deferred Action for Childhood Arrivals (DACA) program. The obligation to respond to alternatives is also frequently invoked in the lower courts and in the D.C. Circuit in particular. But courts lack a consistent framework for analyzing the obligation, providing agencies with little guidance regarding which alternatives require analysis as part of their decisionmaking process. And to the extent that the obligation allows courts a backdoor opportunity to flyspeck agencies’ policy analysis, it runs the risk of displacing agencies’ expert judgments for the courts’ own, often informed only by the parties’ briefing.

This Article interrogates the obligation to respond to alternatives and proposes a more stable framework for its implementation. After rooting the obligation in agencies’ general obligation to give reasons for their actions and in the values associated with agency reason-giving, the Article turns to two questions. First, to which alternatives must agencies respond? And second, what counts as a response? In answering these questions, the Article draws on broader goals associated with administrative law, which include not only promoting the values associated with reason-giving but also respecting the need for agencies to perform their tasks effectively and within a relatively stable system of judicial review that recognizes their comparative expertise advantage over the courts. In addition, it seeks to develop a framework that fits with and helps to explain the results in most cases invoking the obligation to respond to alternatives. At the same time, the framework allows us to more clearly identify occasions where courts have reached incorrect results or been overbroad in their framing of the obligation to respond to alternatives.


In 2020, the Supreme Court invalidated the Trump Administration’s rescission of the Obama-era Deferred Action for Childhood Arrivals (DACA) program.1See DHS v. Regents of the Univ. of Cal., 140 S. Ct. 1891 (2020). In doing so, the Court drew on the principle, usually attributed to Motor Vehicle Manufacturers Ass’n of the United States, Inc. v. State Farm,2Motor Vehicle Mfrs. Ass’n of the U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29 (1983). that an agency must respond to alternative options when selecting a course of action.3The obligation actually precedes State Farm by quite a bit. E.g., Pillai v. Civ. Aeronautics Bd., 485 F.2d 1018, 1029–30 (D.C. Cir. 1973). For the Court, DACA had two main components: a conferral of certain benefits on individuals participating in the DACA program, and a decision by the Department of Homeland Security to forbear from removing those same individuals during the period of their participation. The Trump Administration had gotten rid of both. Its reasons for doing so, however, went only toward the conferral of benefits. And it had failed to consider the alternative option of retaining DHS’s forbearance decision while rescinding other parts of the program. For the Court, that was fatal.4See Regents, 140 S. Ct. at 1913 (explaining that the “omission alone render[ed] [the agency’s] decision arbitrary and capricious”).

Although agencies’ obligation to respond to alternatives has been a core requirement of administrative law for decades, it has attracted little sustained scholarly attention.5Plenty of articles mention the obligation or discuss aspects of it as part of more general inquiries. See, e.g., Jacob Gersen & Adrian Vermeule, Thin Rationality Review, 114 Mich. L. Rev. 1355, 1388–93 (2016). The case law, while voluminous, has also done little to meaningfully define the contours of the doctrine. There are really two questions involved. First, to which alternatives must agencies respond? And second, what passes as a response? On the first, the courts have explained that “[a]n agency is not required to ‘consider all policy alternatives’ 6MediNatura, Inc. v. FDA, 998 F.3d 931, 943 (D.C. Cir. 2021) (quoting State Farm, 463 U.S. at 29, 51). or “every alternative device and thought conceivable by the mind of man.”7Id. (quoting Vt. Yankee Nuclear Power Corp. v. Nat. Res. Def. Council, Inc., 435 U.S. 519, 551 (1978)). To the contrary, the “obligation extends only to ‘significant and viable’ alternatives,”8Am. Radio Relay League, Inc. v. FCC, 524 F.3d 227, 242 (D.C. Cir. 2008) (quoting Farmers Union Cent. Exch., Inc. v. Fed. Energy Regul. Comm’n, 734 F.2d 1486, 1511 n.54 (D.C. Cir. 1984)). or, perhaps, to “reasonably obvious” ones.9Nat’l Shooting Sports Found. v. Jones, 716 F.3d 200, 215 (D.C. Cir. 2013) (quoting Nat. Res. Def. Council, Inc. v. SEC, 606 F.2d 1031, 1053 (D.C. Cir. 1979)). But beyond such platitudes, the courts’ approach has been largely ad hoc—unguided by crisp rules or even a general framework for determining which alternatives merit consideration.

The question of what passes as a response has received similarly few answers.10See Am. Radio Relay League, 524 F.3d at 242 (explaining that an agency must “give a reasoned explanation for its rejection of . . . alternatives” (quoting City of Brookings Mun. Tel. Co. v. FCC, 822 F.2d 1153, 1169 (D.C. Cir. 1987))). Courts occasionally suggest that agencies may satisfy their obligation by implicitly addressing alternatives,11See, e.g., Am. Iron & Steel Inst. v. EPA, 115 F.3d 979, 1005–06 (D.C. Cir. 1997); FBME Bank Ltd. v. Lew, 125 F. Supp. 3d 109, 124–25 (D.D.C. 2015). but others seem to conflate responding to alternatives with mentioning them.12See, e.g., Walter O. Boswell Mem’l Hosp. v. Heckler, 749 F.2d 788, 797–98 (D.C. Cir. 1984). A host of other questions—such as whether agencies may adopt strategies, such as satisficing, that may operate to limit the number of alternatives considered13Jacob Gersen and Adrian Vermeule say yes. Gersen & Vermeule, supra note 5, at 1403. As explained more fully below, satisficers choose an option that is revealed to be “good enough” and then suspend investigation into other alternatives.—also remain unaddressed.

This Article theorizes the obligation to respond to alternatives and proposes a more concrete way to implement it. The obligation to respond to alternatives is part of an agency’s general duty to give a reasoned explanation for its actions when choosing among legally available options. At the same time, no one would spend all their time dreaming up every possibility for what they could do on a given day, detailing the benefits and drawbacks of each option, and carefully selecting which would be optimal. And in the context of administration, requiring an agency to do the equivalent would grind government to a halt. Thus, sensible limits must be drawn.

Much of the project of this Article is to seek out these limits. To do so, I critically synthesize the case law and propose a structure to guide the courts’ decisionmaking. When it comes to identifying alternatives to which an agency must provide a response, the task breaks down as follows. First, in many cases, it makes good sense to start with alternatives raised by outside parties—for example, as part of notice-and-comment proceedings. But agencies often cannot practically respond to every alternative raised by others, and therefore I propose a set of considerations to guide the courts in determining when an agency needs to respond to a given proposed alternative.

Second, certain alternatives may require consideration regardless of whether any outside party has raised them.14Cf. Ronald M. Levin, Making Sense of Issue Exhaustion in Rulemaking, 70 Admin. L. Rev. 177, 187 (2018) (explaining that “an agency is expected to explore alternatives to its proposed rule if the alternatives were obvious or if they were suggested by commenters” and that this “necessarily means that some alternatives must be considered if and only if someone asks the agency to adopt them”). Identifying such alternatives is particularly important for contexts where the agency decision has been made without any process allowing for public input.15During the COVID-19 pandemic, for example, various agencies promulgated rules on an emergency basis and without soliciting public comment. See, e.g., Nat’l Fed’n of Indep. Bus. v. Dep’t of Lab., Occupational Safety & Health Admin., 142 S. Ct. 661, 663 (2022) (per curiam); Health Freedom Def. Fund, Inc. v. Biden, 599 F. Supp. 3d 1144, 1154 (M.D. Fla. 2022). I propose several categories of alternatives to which an agency must respond even if they were not raised by outside parties. For example, and least controversially, an agency generally must respond to alternatives when expressly required to by the underlying statute. But I also argue that there may be cases where the given statute implicitly requires an agency to respond to certain alternatives—a question on which agencies should receive Chevron deference, if applicable. I also suggest that an agency must also normally explain why it has rejected the option of retaining the status quo. Similarly, where the agency has jettisoned an entire policy without considering retaining individual, standalone components of it—as in the DACA example—it probably has not provided a satisfactory explanation for its chosen course.

Finally, I argue that courts should rarely, if ever, require agencies to respond to alternatives that may seem promising as a policy matter but were neither raised by the parties nor fall within one of the categories of alternatives to which agencies normally must attend. This conclusion is admittedly most difficult to swallow in situations in which the agency has not received outside input at all and therefore the only alternatives to which the agency must respond are those that fall within the categories of alternatives that inherently mandate a response. But where the Administrative Procedure Act does not require an agency to go through public processes, the agency action in question is likely to be either easier to change than officially promulgated regulations are, or it must be justified by emergency conditions. Under such conditions, my preliminary position is that if the agency has explained why its approach represents an incremental improvement on the status quo—which will always be required—then the benefits of remanding to the agency to consider alternatives that were posed for the first time as part of the litigation process are unlikely to outweigh the costs. As a fallback, however, I explore several frameworks and tweaks to existing doctrine for those who believe that there must be at least some instances in which the obviousness of an alternative as a policy option should be relevant.

The second question—what counts as a sufficient response—involves how an agency must respond to an alternative that it has an obligation to consider. Boilerplate language to the effect that the agency has considered a given alternative and rejected it will not do. Rather, in the words of one court, the agency must at least “respond in a way that allows the Court to see why it did not opt for . . . alternative proposals.”16Am. Great Lakes Ports Ass’n v. Zukunft, 296 F. Supp. 3d 27, 44 (D.C. Cir. 2017). At the same time, courts have made clear that “[a]s long as ‘the agency’s path may reasonably be discerned,’ [they] will uphold the decision even if it is ‘of less than ideal clarity.’ 17Casino Airlines, Inc. v. Nat’l Transp. Safety Bd., 439 F.3d 715, 717 (D.C. Cir. 2006) (quoting Bowman Transp., Inc. v. Ark.-Best Freight Sys., Inc., 419 U.S. 281, 285–86 (1974)).

I propose that an agency fulfills its basic obligation to respond when it specifies the criteria by which an alternative was evaluated and explains, using those criteria, why the agency believes its chosen course is superior. This requirement enables courts to review whether the agency has used the correct criteria and whether its choice was reasonable and supported by substantial evidence.

Importantly, an agency’s explanation may reveal why it rejected a given alternative even when the agency does not mention the alternative specifically. Rather, the “agency’s path may be reasonably discerned” when, after having read the agency’s decision in conjunction with the administrative record, the court can understand what criteria the agency employed and why that criteria ruled out a given alternative.18See Bowman Transp., Inc. v. Ark.-Best Freight Sys., Inc., 419 U.S. 281, 286–89 (1974). Relatedly, I argue that an agency should be allowed to give global reasons for rejecting certain categories of alternatives, thus relieving the agency of the need to address variations of alternatives falling within the category. Agencies should also be allowed to announce that they are pursuing strategies of decisionmaking that by their nature restrict the range of alternatives explored. Such situations might occur, for example, where the decision costs associated with a fuller exploration of alternatives are high and the chosen policy is “good enough”—a decisionmaking strategy called “satisficing.” If the statute in question allows the agency to act pursuant to such a strategy—an important qualification—an agency may satisfy its obligation to respond to alternatives by embracing it. Finally, I argue that, in certain cases, even a complete failure to respond to an alternative to which the agency normally must respond may be dismissed as harmless error. In particular, an agency should be able to argue in litigation that a given alternative is illegal, even if the agency had not provided such an explanation in the decision under review.

A word on my own considerations. First, a big problem with the current system, at least how I see it, is the uncertainty surrounding the obligation to respond to alternatives.19Cf. Wendy E. Wagner, Administrative Law, Filter Failure, and Information Capture, 59 Duke L.J. 1321, 1410 (2010) (noting that “[a] number of commentators have suggested that the uncertainty associated with judicial review causes some of the most serious problems in agency behavior”). Thus I have tried, to the extent possible, to give the obligation sharper form in order to allow agencies to better predict which alternatives the courts will find merit a response and what kinds of explanations courts will find satisfactory.20Indeed, in fleshing out the obligation to respond to alternatives, I do not mean to align myself against those who worry about the problem of “ossification” or the potential overproceduralization of agency decisionmaking. See Thomas O. McGarity, Some Thoughts on “Deossifying” the Rulemaking Process, 41 Duke L.J. 1385 (1992); Nicholas Bagley, The Procedure Fetish, 118 Mich. L. Rev. 345 (2019). I hope instead to take a chisel to the doctrine in an attempt to give it some form, recognizing that the core obligation is here to stay. That said, some aspects of the obligation defy easy “rulification,”21See generally Michael Coenen, Rules Against Rulification, 124 Yale L.J. 644 (2014). and some purported rules will have fuzzy edges.

Relatedly, and perhaps counterintuitively, I have attempted to provide a framework that minimizes courts’ need to inquire into the policy merits of the alternative in question. That’s particularly important, I will argue, when the alternative in question has not been raised and, for that reason, has not been developed in the administrative record.

Further, in seeking to shape agencies’ obligation to respond to alternatives, I frequently draw on broader principles of administrative law. For the most part, I take the basic aims associated with the field for granted and seek to justify aspects of the doctrine using shared terms. Although aspects of the field are currently characterized by tumult, certain goals—such as designing the system to ensure that agencies remain faithful to their statutory mandates—remain broadly accepted by courts and scholars alike.22Indeed, Thomas McGarity, perhaps the leading voice raising concerns about ossification, concedes that “judicial review can steer agencies back on the track when they stray from their congressionally assigned roles,” a beneficial outcome. McGarity, supra note 20, at 1452. In attempting to shape the contours of the obligation to respond to alternatives, I have kept these goals in mind.23Cf. Thomas W. Merrill, The Common Law Powers of Federal Courts, 52 U. Chi. L. Rev. 1, 8 (1985) (applying “a norm of legitimacy drawn from positive law: the existing set of legal principles that participants in our legal system consider binding and authoritative”).

Finally, the structure I supply roughly fits with both the extant case law on the obligation to respond to alternatives and with the courts’ broader administrative law jurisprudence.24Cf. Adrian Vermeule, Deference and Due Process, 129 Harv. L. Rev. 1890, 1894 (2016) (applying Dworkin’s theory of fit and justification in administrative law context). Two caveats qualify this statement. First, although the results of the cases generally align with the framework I propose, the courts don’t speak in the terms I do. Indeed, as suggested above, the courts’ approach has been generally ad hoc and not well-theorized. Second, I argue that, judged by my criteria, courts have occasionally made overbroad pronouncements or have reached incorrect results. That is, although consistent with the bulk of the case law, the theory has bite in that it helps identify correct and incorrect applications of the obligation to respond to alternatives.25Cf. Charles Silver, A Restitutionary Theory of Attorneys’ Fees in Class Actions, 76 Cornell L. Rev. 656, 663 (1991) (aiming to “show that the theory has bite” by “enabl[ing] judges to decide practical issues in coherent and defensible ways”).

Now is an important time to refine the contours of the obligation to respond to alternatives. Although the basic obligation serves the values associated with reason-giving requirements more generally, there are signs that courts may increasingly be keen to use the obligation as justification for substituting the courts’ own views about the nature of the problem at issue and the range of responses that might be justified as a policy matter.26See infra notes 91–95 and accompanying text. This approach would usurp agencies’ expertise and lead to wasteful remands for agencies to consider alternatives that are unworthy of consideration. A primary motivation for this Article is thus to devise a framework to implement the obligation to respond to alternatives in a way that minimizes courts’ need to engage in their own policy analysis in order to determine whether the obligation has been satisfied.

The balance of the Article proceeds as follows. Part I explores the basic aspects of agencies’ obligation to respond to alternatives, primarily through a discussion of three Supreme Court cases addressing the subject. Part II begins by questioning why agencies should have to respond to alternatives at all. It proceeds to provide guidance on identifying the alternatives to which an agency must respond. Part III addresses the various forms that an agency’s response might take and considers when an agency’s failure to provide such a response may be considered harmless error.

I. The Basics of the Obligation

This Part sketches the basic contours of agencies’ obligation to respond to alternatives, drawing on three illustrative Supreme Court cases. These cases establish that there are at least certain alternatives that an agency must consider. To fulfill this obligation, the agency must answer why the alternative in question was not pursued. At the same time, the Supreme Court has made clear that an agency need not respond to every conceivable alternative. And when an agency has supplied reasons why a given alternative had been rejected, the most recent indication from the Court is that courts should review those reasons deferentially. The Part ends by discussing concerns some have raised in the wake of the Supreme Court’s latest pronouncement on the subject.

A. State Farm and the Core of the Obligation

The Supreme Court’s decision in Motor Vehicle Manufacturers Ass’n of the United States, Inc. v. State Farm solidified the framework for review of agencies’ discretionary choices and fully introduced, at least at the Supreme Court level, the obligation to respond to alternatives. Some brief background: the Administrative Procedure Act (APA) directs courts to set aside agency action determined to be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.”275 U.S.C. § 706(2)(A). When the APA was passed, it’s quite possible that “arbitrary-and-capricious review,” as the test is typically formulated, was understood to require only that the agency action in question had a rational basis.28Jeffrey A. Pojanowski, Neoclassical Administrative Law, 133 Harv. L. Rev. 852, 893 (2020) (“There is reason to believe that arbitrary and capricious review was understood when the APA was enacted as closer to rational basis review under constitutional law than contemporary hard look review.”); Gillian E. Metzger, Foreword: Embracing Administrative Common Law, 80 Geo. Wash. L. Rev. 1293, 1299 (2012). But in the 1960s and 1970s, courts—and in particular the D.C. Circuit—fashioned a set of more elaborate requirements to govern agency decisionmaking, deploying a “hard look” approach that extended beyond mere rationality review.29Kathryn A. Watts, Proposing A Place for Politics in Arbitrary and Capricious Review, 119 Yale L.J. 2, 15–16 (2009). Soon, the Supreme Court began to move in the same direction. In Citizens to Preserve Overton Park, Inc. v. Volpe, decided prior to State Farm, the Court reviewed a decision by the secretary of transportation to approve the construction of a highway through Overton Park in Memphis.30Citizens to Pres. Overton Park, Inc. v. Volpe, 401 U.S. 402 (1971). Although that decision was the product of an informal adjudication subject to very minimal requirements under the APA, the Court required that the secretary produce a record upon which the courts could decide, among other things, “whether the [agency’s] decision was based on a consideration of the relevant factors.”31Id. at 416.

The full flowering of the Court’s modern approach to arbitrary-and-capricious review—and its fullest endorsement of agencies’ obligation to respond to alternatives—came in State Farm.32Motor Vehicle Mfrs. Ass’n of the U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29 (1983). State Farm involved a decision by the Reagan-era National Highway Traffic Safety Administration (NHTSA) to rescind certain regulations requiring manufacturers to install “passive restraint” systems in new cars.33Id. at 38. Under the prior regulations, manufacturers could comply with the passive restraint requirements in one of two ways: by installing airbags or by installing automatic safety belts.34Id. at 37. In jettisoning the passive restraint requirements altogether, NHTSA explained that fewer manufacturers than expected were choosing to comply by installing airbags and therefore “the lifesaving potential of airbags would not be realized.”35Id. at 38. Moreover, the agency had become disillusioned with automatic seatbelts. It turned out that those belts could be easily detached and, once detached, were no longer automatic but instead required an affirmative act to reengage.36Id. at 38–39. In light of the compliance costs associated with the passive restraint rules and their now-uncertain benefits, the agency concluded the rules were no longer justified.37Id. at 39.

On review at the Supreme Court, Justice White’s opinion, which was unanimous on this point, brushed aside the manufacturers’ argument that NHTSA’s action was not reviewable because it had merely rescinded the prior rules.38See id. at 40–42. In rejecting that argument, the Court held that “normal” arbitrary-and-capricious review applied to rule rescissions.39Id. at 29 (describing the factors upon which courts “[n]ormally” rely for arbitrary-and-capricious review). The Court also clarified that such review entails more than is required from Congress under the rational-basis test.40Id. at 42–43, 43 n.9. The Court stated, in a passage that is now canonical:

The scope of review under the “arbitrary and capricious” standard is narrow and a court is not to substitute its judgment for that of the agency. Nevertheless, the 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.” In reviewing that explanation, we must “consider whether the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment.” Normally, an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.41Id. at 43.

Equally important, however, was the Court’s application of the newly synthesized arbitrary-and-capricious test. In that application, and as Paul Verkuil has explained, “State Farm is really two opinions: the unanimous opinion that condemned the National Highway Transportation Safety Administration’s failure to consider the airbags-only alternative; and the five-to-four opinion that invigorated a hard-look version of the arbitrary-and-capricious clause.”42Paul R. Verkuil, The Wait Is Over: Chevron as the Stealth Vermont Yankee II, 75 Geo. Wash. L. Rev. 921, 923 (2007).

First, when it rescinded the prior passive restraint rules, NHTSA had not said anything at all about a different option: requiring manufacturer to install airbags in all new cars.43State Farm, 463 U.S. at 46. For a unanimous Court, that was fatal. The agency had, in adopting the passive restraint rules, found airbags to be beneficial.44Id. at 47. Even in rescinding those rules, the agency had continued to tout the safety benefits of airbags.45Id. at 47–48. The automatic seatbelt option had been added as an alternative way to comply with the passive restraint rules at a time when automatic seatbelts were also thought to provide important safety benefits.46Id. at 46–47. If that belief was no longer valid, the question remained: why not require airbags? The agency had not attempted to answer that question.

In the Court’s view, NHTSA had thus entirely failed to respond to an important alternative to its chosen course of action. As the Court explained, “[a]t the very least [the airbag-only] alternative way of achieving the objectives of the Act should have been addressed and adequate reasons given for its abandonment. But the agency not only did not require compliance through airbags, it also did not even consider the possibility. . . .”47Id. at 48.

At the same time, the Court marked out some limits on agencies’ obligation to respond. Quoting its previous decision in Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council, Inc. where it had opined on the National Environmental Policy Act’s express requirement to explore alternatives, the Court stated that an agency action “cannot be found wanting simply because the agency failed to include every alternative device and thought conceivable by the mind of man . . . regardless of how uncommon or unknown that alternative may have been.”48Id. at 51 (alteration in original) (quoting Vt. Yankee Nuclear Power Corp. v. Nat. Res. Def. Council, Inc., 435 U.S. 519, 551 (1978)). But, the Court declared, “the airbag is more than a policy alternative to the passive restraint [s]tandard; it is a technological alternative within the ambit of the existing [s]tandard.”49Id. For that reason, the agency was obligated to consider it.50Id.

Second, although the agency had said something about the option of retaining an automatic safety belt option, the Court found the agency’s explanation to be deficient. Even if it was within the agency’s discretion to discount the significance of certain studies demonstrating increased use of safety belts when automatic belts were installed, the agency’s reasoning in the absence of hard evidence demonstrating the belts’ effectiveness or ineffectiveness did not withstand scrutiny.51See id. at 51, 53–54. Essentially, the agency had concluded that because most automatic belts were detachable and lost their efficacy once detached, the safety benefits associated with automatic belts were substantially uncertain.52Id. at 51–53. The Court, however, accused the agency of overlooking the fact that, once reattached, automatic safety belts would work as intended until disengaged in a further affirmative act. For that reason, given that 20% to 50% of drivers wear seatbelts on at least some occasions, “inertia—a factor which the agency’s own studies have found significant in explaining the current low usage rates for seatbelts—works in favor of, not against, use of the protective device.”53Id. at 54. Thus, the Court explained, there are “grounds to believe that seatbelt use by occasional users will be substantially increased by the detachable passive belts,” and the agency “must bring its expertise to bear on the question.”54Id.

B. Trump-Era Applications

State Farm puts on the table the two aspects of agencies’ overall obligation to respond to alternatives. First, agencies must respond to alternatives to their chosen course of action, or at least to some of them. And second, the response provided must not flunk arbitrary-and-capricious review. Two high-profile cases involving the Trump Administration show these two requirements in action.

In Department of Commerce v. New York, the Court suggested that, when an agency has addressed an alternative, its explanation for rejecting it will normally warrant deference. That case involved Secretary of Commerce Wilbur Ross’s decision to reinstate a citizenship question on the 2020 census questionnaire.55Dep’t of Com. v. New York (Census Case), 139 S. Ct. 2551, 2561 (2019). The secretary claimed that his decision stemmed from a request by the Department of Justice, which purportedly wanted more accurate citizenship data for purposes of enforcing the Voting Rights Act.56Id. at 2562. Although the Court ultimately found that justification pretextual,57Id. at 2573–76. it also examined whether the secretary’s action was arbitrary and capricious on its own terms.58See id. at 2569–71. The District Court had found that the secretary had not adequately supported his decision to reject an alternative suggested by the Department’s own Census Bureau involving the use of administrative records, and not a census question, to collect improved citizenship data.59Id. at 2569–70. Those records, the Bureau said, would yield higher quality data while avoiding the downside of a citizenship question—namely, that including such a question would decrease census response rates among noncitizens.60Id. at 2570.

The secretary considered and rejected the Bureau’s preferred approach, and so the question for the Court was not whether the secretary was required to respond to the alternative in question—he had—but whether his response was adequate. In assessing whether “[t]he evidence before the Secretary supported” his decision to reject the Bureau’s preferred alternative, the Court spoke in deferential terms.61Id. at 2569. At the time of the decision, the secretary had explained that administrative records did not exist for about 10% of the population and that a citizenship question, paired with the use of records, would more accurately capture the status of these approximately 35 million people.62Id. As for response rates, the secretary concluded that the lower response rates the Bureau believed were caused by the inclusion of a citizenship question on prior forms could have been caused by other factors.63Id. at 2570. In finding the secretary’s explanation adequate, the Court stressed that “the choice between reasonable policy alternatives in the face of uncertainty was the Secretary’s to make.”64Id. The implementation of the census “called for value-laden decisionmaking and the weighing of incommensurables under conditions of uncertainty.”65Id. at 2571. The secretary was required only “to consider the evidence and give reasons for his chosen course of action,” and it is not for the courts “to ask whether [the secretary’s] decision was ‘the best one possible’ or even whether it was ‘better than the alternatives.’ 66Id. (quoting Fed. Energy Regul. Comm’n v. Elec. Power Supply Ass’n, 136 S. Ct. 760, 782 (2016)). The secretary had given his reasons, the reasons were reasonable, and that was enough.

In the second case, Department of Homeland Security v. Regents of the University of California (Regents), the Supreme Court dealt with a situation in which, as the majority conceptualized it, the agency had failed to grapple with an important alternative altogether. Regents involved a challenge to the Trump Administration’s rescission of the DACA program.67DHS v. Regents of Univ. of Cal., 140 S. Ct. 1891, 1901 (2020). Issued under President Obama, DACA allowed certain undocumented immigrants who had entered the country at a young age to apply to participate in the DACA program.68Id. Participants enjoy a two-year forbearance from removal as well as authorization to work and various forms of benefits, including Social Security and Medicare benefits.69Id. at 1901–02. Fast forward to the Trump Administration. In 2017, Attorney General Sessions sent a letter to Acting Secretary of Homeland Security Elaine Duke advising her that DACA suffered from the same legal defects that had led the United States Court of Appeals for the Fifth Circuit to invalidate the Deferred Action for Parents of Americans and Lawful Permanent Residents (DAPA) program and urging DHS to wind the DACA program down.70Id. at 1903. Duke obliged, issuing a memorandum a day later announcing that the DACA program would be terminated following a transition period.71Id. Challenges ensued.

Chief Justice Roberts’s majority opinion in Regents conceded that DHS had the legal authority to wind down DACA but took issue with the agency’s reasoning, likening its defects to NHTSA’s failure to consider the airbag-only alternative in State Farm.72See id. at 1910–15. The majority’s logic unfolded roughly as follows. Both Sessions and Duke had relied on the Fifth Circuit’s reasoning in invalidating DAPA.73See id. 1911. But as the Court saw it, that reasoning had invalidated DAPA based on DAPA’s extension of federal benefits to program participants.74Id. It had not spoken to the legality of a program that provided only forbearance from removal.75Id. And that was important, the Court explained, because forbearance from removal was an important—and indeed the main—element of the DACA program.76See id. at 1911–12. Thus, even assuming the validity of the Fifth Circuit’s decision with respect to DAPA, the secretary failed to recognize—and thus consider—another option: retaining DACA’s grant of forbearance while excising those aspects of the program that had caused the legal problem.77See id.

For the Court’s majority, the action by DHS was therefore arbitrary and capricious for the same reasons that NHTSA’s rescission of the passive restraint rules was.78Id. at 1912–13. In both cases, the agency had rescinded a program with multiple components. In both cases, the agency had identified problems with one component but had not provided reasons justifying the rescission of others. And in both cases, that amounted to a failure to consider an alternative to its course of action, a failure sufficient to make the agency’s choice arbitrary and capricious. As in State Farm, the Court in Regents grasped towards limits on its rule. In response to Justice Thomas’s accusation that the Court was “ ‘dissect[ing]’ agency action ‘piece by piece,’ ” the majority responded that retaining forbearance was not just any old alternative.79See id. at 1913 (quoting id. at 1930 (Thomas, J., concurring in the judgment in part and dissenting in part)). Rather, in the Court’s view, “forbearance was not simply ‘within the ambit of the existing [policy].’ 80Id. (quoting Motor Vehicle Mfrs. Ass’n of the U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 51 (1983)). Rather, forbearance “was the centerpiece of the policy: DACA, after all, stands for ‘Deferred Action for Childhood Arrivals.’ 81Id. Because “the rescission memorandum contain[ed] no discussion of forbearance or the option of retaining forbearance without benefits,” DHS’s decision was arbitrary and capricious.82Id. at 1913. The majority also rejected DHS’s additional explanation, offered as a memorandum from Secretary Kirstjen M. Nielsen after litigation was underway, as a “post hoc rationalization.” Id. at 1908 (quoting Citizens to Pres. Overton Park, Inc. v. Volpe, 401 U.S. 402, 420 (1971)). The Court went on to separately fault DHS for rescinding the DACA program without taking into account the “potential reliance interests” at stake.83Id. at 1913–14. I’ll discuss the interrelationship between reliance interests and alternatives further below. See infra notes 248–253 and accompanying text.

C. Responding to Alternatives in the Wake of Regents

Regents placed agencies’ obligation to respond to alternatives back in the spotlight. It also reaffirmed, in line with Overton Park and D.C. Circuit case law,84See supra notes 30–31 and accompanying text (discussing Overton Park); Laclede Gas Co. v. Fed. Energy Regul. Comm’n, 873 F.2d 1494, 1498 (D.C. Cir. 1989) (applying obligation in the context of an adjudication). that the obligation extends to all reviewable forms of agency action and not just to rulemaking.

The response to Regents was mixed, including among those who supported DACA on policy grounds. Zachary Price, for example, predicted that Regents’ “meticulous standards” would come back to bite progressives specifically in their efforts to roll back Trump Administration policies and perhaps more broadly.85See Zachary Price, DACA and the Need for Symmetrical Legal Principles, SCOTUSblog (June 19, 2020, 3:51 PM), [].

At least with respect to Regents’ holding regarding DHS’s obligation to consider the forbearance-only alternative, however, Regents was arguably no more stringent than what came before.86The Court had also decided, as an independent ground for setting aside the agency’s rescission, that the rescission was arbitrary and capricious because the agency failed to discuss reliance interests. Regents, 140 S. Ct. at 1913–15. The analogy to State Farm is fairly straightforward, and any distinctions between the two cases don’t require a different end result. Unlike State Farm, Regents involved a decision to rescind a program based on legal considerations and not policy ones. But the agency’s legal reasoning in Regents spoke to the option of retaining forbearance no more so than NHTSA’s policy reasoning did to the option of retaining an airbags-focused standard. And even if DHS had thought, incorrectly, that the legal defects it identified spoke to the legality of the entire program and not only to the benefits, such mistakes in an agency’s legal reasoning provide a proper basis for setting aside agency action.87Sea-Land Serv., Inc. v. Dep’t of Transp., 137 F.3d 640, 646 (D.C. Cir. 1998) (“An agency action, however permissible as an exercise of discretion, cannot be sustained ‘where it is based . . . on an erroneous view of the law.’ ” (quoting Prill v. NLRB, 755 F.2d 941, 947 (D.C. Cir. 1985))); Daniel Deacon, No, Agencies Are Not Allowed to Be “a Little Bit Wrong” About the Constitution, Yale J. Regul. Notice & Comment (June 22, 2020), []; Daniel J. Hemel & Aaron L. Nielson, Chevron Step One-and-a-Half, 84 U. Chi. L. Rev. 757, 783–88 (2017).

In a broader sense, though, a free-floating obligation to respond to alternatives, whether rooted in State Farm or Regents, does raise a set of issues that should give us pause. Agency decisionmaking is oftentimes complex and the range of potential alternatives vast. While the Supreme Court had indicated that agencies do not have to consider all alternatives, it is not self-evident what separates alternatives that the agency must consider from those they needn’t. The lower courts, for their part, have not provided much crisper standards, while at the same time extending State Farm outside its rescissionary context.88See infra Section II.C (describing circumstances where lower courts have found the obligation to apply).

Under such conditions, an obligation to respond to alternatives may have real costs to agency administration and society in general. That’s especially so if whether an alternative requires a response—and whether the response is sufficient—is tightly linked to policy considerations. Generalist judges may, due to lack of relevant knowledge or subconscious results-orientation, require agencies to respond to alternatives that appear attractive to the judges but that are not really within the range of reasonable policy choices.89See infra notes 124–126 and accompanying text. Remanding to agencies for them to consider implausible alternatives or alternatives they may have already implicitly rejected is an inherently wasteful enterprise. And even if courts could do a good job parsing through which alternatives were worthy of a response, the whole endeavor naturally introduces a well-known set of costs by diverting agency resources, creating the potential for informational overload, and delaying important agency policy initiatives.90See infra notes 127–129 and accompanying text.

Several recent opinions involving the Biden Administration may lend credence to such fears. Two occurred in the context of the COVID-19 pandemic. Dissenting from a Sixth Circuit panel decision upholding the Occupational Safety and Health Act’s “vaccine-or-test” emergency temporary standard, Judge Joan Larsen faulted the agency for, among other things, failing to explore “more tailored solutions” that were “easy to envision.”91In re McP NO. 165, 21 F.4th 357, 394, 398 (6th Cir. 2021) (Larsen, J., dissenting). These included measures such as “a standard aimed at the most vulnerable workers; or an exemption for the least.”92Id. at 394. The Supreme Court ultimately reversed the Sixth Circuit and found the OSHA mandate invalid, but it did not reach whether OSHA had violated the reasoned decisionmaking requirements of the APA. See Nat’l Fed’n of Indep. Bus. v. Dep’t of Lab., Occupational Safety & Health Admin., 142 S. Ct. 661 (2022) (per curiam). Similarly, and seemingly in the absence of briefing on the issue, a district court set aside the Center for Disease Control’s transportation masking requirements, in part because the agency had failed to consider alternatives like “testing, temperature checks, or occupancy limits in transit hubs and conveyances.”93Health Freedom Def. Fund, Inc. v. Biden, 599 F. Supp. 3d 1144, 1174 (M.D. Fla. 2022). And outside the COVID-19 context, the Fifth Circuit faulted DHS for failing to consider alternatives when the Biden Administration rescinded the Trump-era “Remain in Mexico” program.94See Texas v. Biden, 10 F.4th 538, 554–57 (5th Cir. 2021). Among other things, the court chastised the agency for its failure to discuss potential “modifications” to the program in lieu of a complete rescission, though it did not explain which modifications, exactly, DHS was under an obligation to consider.95See id. at 555.

Even outside high-profile cases involving salient public policies, the courts’ decisionmaking regarding what alternatives must be considered has an “I-know-it-when-I-see-it” flavor. If that’s troubling, the question remains whether anything can be done. Should the courts jettison agencies’ obligation to respond to alternatives? Or, alternatively, can a sensible doctrine be molded from the clay? The next Part turns to those questions.

II. What Alternatives Must Be Addressed?

Section II.A explains why agencies should have to respond to any alternatives at all. The question then becomes, which ones? Requiring courts to make determinations about which alternatives deserve attention and which do not creates uncertainty about how the agency is to fulfill this obligation. And that uncertainty may cause agencies to act as if they must respond to any and all alternatives if they wish to avoid having their actions set aside.96Wagner, supra note 19, at 1410. The balance of the Part thus seeks to provide some contours to the obligation to respond to alternatives. Section II.B addresses alternatives raised by outside parties—for example, in notice-and-comment proceedings. Section II.C then turns to alternatives that have not been raised—perhaps because the agency solicited no outside views—and builds out categories of alternatives to which agencies presumptively must respond. Section II.D considers whether courts should ever fault an agency for failing to address alternatives that may seem intuitively attractive as a policy matter, but that do not fall within the categories of alternatives to which agencies normally must respond and were not raised by an outside party.

A. Assessing the Core Obligation

One possible answer to the question of which alternatives require a response from the agency would be simple enough: none. We should abandon the enterprise, and courts should demand at most that an agency give a reason in favor of its preferred policy.

Such a rule would not be in keeping with the justifications for requiring agencies to provide a reasoned explanation for their actions. The obligation is the product of two basic starting points. First, as a matter of arbitrary-and-capricious review, agencies must justify their actions by giving reasons.97See Dep’t of Com. v. New York (Census Case), 139 S. Ct. 2551, 2575–76 (2019); SEC v. Chenery Corp. (Chenery I), 318 U.S. 80 (1943). Second, arbitrary-and-capricious review only applies where an agency has chosen between legally available options.98See Nat’l Tire Dealers & Retreaders Ass’n, Inc. v. Brinegar, 491 F.2d 31, 37 (D.C. Cir. 1974); Jerry L. Mashaw, Reasoned Administration and Democratic Legitimacy 3 (2018) (discussing case).

Then, the question is, when an agency has made this choice, what kinds of reasons must it give to support its decision? The Supreme Court has explained that the “reasoned explanation requirement” requires agencies to provide “genuine justifications for important decisions” that can be “scrutinized by courts and the interested public.”99Census Case, 139 S. Ct. at 2575–76. Typically, that means that the agency must display some awareness of the tradeoffs it is making and provide an explanation for why the balance of considerations favored the agency’s preferred course.100See Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 166–67 (1962) (faulting agency for failing to explain why “in the circumstances the public interest” supported the agency’s choice and “outbalances whatever public interest there is in [other paths]”). The philosophical literature on reasons contains differing views on what it means to provide a reason, some of which may be closer to what I take to be the administrative law account. See, e.g., Justin Snedegar, Reason Claims and Contrastivism About Reasons, 166 Phil. Stud. 231 (2012) (discussing different perspectives). I don’t mean to take a stand on this broader debate. That’s a comparative endeavor.101Of course, justifying a given course of action need not logically require balancing or a comparison of options. For one, agencies may say that they are legally required to take one and only one course of action. That adequately explains why the agency has done what it has, but the agency must be right about the law. See generally Hemel & Nielson, supra note 87. It’s also possible that an agency could take the position that one course of action is compelled by nonlegal considerations, perhaps because it is morally required. In my view, that would provide a sufficient explanation for the agency’s action even if one thinks that agencies are under an obligation to provide reasons for rejecting alternatives—the fact that one action is compelled reveals why other actions were not taken. That said, my sense is that agencies rarely take the position that they are strongly obligated to follow one particular course of action outside of situations where they understand themselves to be legally obligated to do so. After all, the judicial review provisions of the APA protect those “aggrieved by” the agency’s action.1025 U.S.C. § 702. Each challenger would have preferred the agency do something else, something that would not have harmed them or would have harmed them less. The “why” question in administrative law is thus properly geared toward explanations that answer why the agency did what it did in light of such alternatives.

What’s more, the more traditional justifications for the agency reason-giving requirement generally support an obligation to respond to alternatives. First, the reason-giving requirement may be justified as a way to improve the quality of agencies’ decisions.103See, e.g., Ashley S. Deeks, Secret Reason-Giving, 129 Yale L.J. 612, 627; Michael A. Livermore & Richard L. Revesz, Rethinking Health-Based Environmental Standards, 89 N.Y.U. L. Rev. 1184, 1233 (2014). The belief that attention to alternatives improves the quality of agency decisionmaking is also reflected in the cost-benefit executive orders that have been adopted by presidential administrations since President Reagan. See Exec. Order No. 12,866 § 1(a), 58 Fed. Reg. 190 (1993) (Clinton); Exec. Order No. 13,422 § 4(c), 72 Fed. Reg. 2763 (2007) (W. Bush); Exec. Order No. 13,563, 76 Fed. Reg. 3821 (2011) (Obama). In the context of alternatives, an agency that knows it must formulate additional ways to satisfy its goals may discover superior means to do so than those it first considered. Given the Trump Administration’s stated intent to allow DACA recipients to stay in the country,104See Remarks by President Trump in Meeting with Bipartisan Members of Congress on Immigration, National Archives: Trump White House (Jan. 9, 2018), []. DHS might have found that the alternative suggested by the Supreme Court—decoupling benefits from forbearance and retaining the parts of DACA regarding the latter105See supra notes 72–82 and accompanying text.—better achieved its express aims.

Another common rationale for the reason-giving requirement is accountability.106See Maria Ponomarenko, Administrative Rationality Review, 104 Va. L. Rev. 1399, 1449 (2018) (“[S]cholars and courts have typically justified reason giving in federal administrative law as a way of facilitating political accountability over agency decisions, and have grounded the requirement in the Constitution’s separation of powers.”). In administrative law, “accountability” typically conjures up notions of political (i.e., electoral) accountability.107See generally Lisa Schultz Bressman, Beyond Accountability: Arbitrariness and Legitimacy in the Administrative State, 78 N.Y.U. L. Rev. 461 (2003) (describing administrative law’s preoccupation with political accountability). But political accountability may be more fictional than real when it comes to a wide range of decisions made by public officials.108See Glen Staszewski, Reason-Giving and Accountability, 93 Minn. L. Rev. 1253, 1266 (2009). When those officials are within agencies, concerns about the government’s lack of political accountability typically only heighten.109See Nicholas Bagley & Richard L. Revesz, Centralized Oversight of the Regulatory State, 106 Colum. L. Rev. 1260, 1285 (2006) (describing how agencies became “increasingly eyed with distrust as politically unaccountable incubators of narrow interest-group politicking”). And of course, some raise particular alarm when it comes to independent agencies. See Seila L. LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183, 2218 (2020) (Thomas, J., concurring in part and dissenting in part). Political accountability, however, is only one way to hold agencies to account for their decisions. Glen Staszewski argues, for example, “that reason-giving requirements are a way to ensure that agencies transparently articulate public-regarding justifications for their actions that can be scrutinized and debated by affected constituencies, thus enabling a form of accountability.”110Staszewski, supra note 108, at 1279–82.

Crucially, reason-giving also fosters legal accountability.111Deeks, supra note 103, at 633 (arguing that “the requirement to give reasons facilitates judicial review, which is the most significant way to hold agencies accountable”). That was the logic of Overton Park.112Citizens to Pres. Overton Park, Inc. v. Volpe, 401 U.S. 402, 416 (1971). Agencies are creatures of Congress, and they must make decisions based on the factors that Congress instructs them to.113Id. Without a record that reveals the reasons the agency had for acting, it is impossible to judge whether it followed its statutory mandate or instead based its decision on factors that are legally out of bounds.114See id. at 420.

To see how accountability concerns play out in the context of the obligation to respond to alternatives, return to the DACA case. Let’s imagine that the Trump Administration, despite its public pronouncements, in fact desired the ability to remove DACA recipients from the country. Regents no longer allowed DHS to hide that preference behind a broad-brush judgment about the legality of DACA as a whole. Rather, by articulating its reasons for rejecting the forbearance-only alternative, the agency would have had to disclose the judgments behind that choice for the public and courts to judge.115See generally Benjamin Eidelson, Reasoned Explanation and Political Accountability in the Roberts Court, 130 Yale L.J. 1748, 1758–73 (2021) (putting forward an accountability-forcing reading of Regents).

Scholars have also argued that the reason-giving requirement is rooted in broader notions of government legitimacy. Sometimes the argument is tied back to accountability concerns. For example, Ashley Deeks has argued that “[t]here is an inverse relationship between popular authority and reason-giving: the greater the popular authority and democratic legitimacy an actor has, the less she needs to give reasons.”116Deeks, supra note 103, at 629. That’s one possible reason why Congress, which may seem politically accountable in a way agencies aren’t,117See Michael Herz, Deference Running Riot: Separating Interpretation and Lawmaking Under Chevron, 6 Admin. L.J. Am. U. 187, 187, 189 n.13 (1992). But see Bagley, supra note 20, at 377 (“Nor is it even obvious that agencies are less democratic than Congress.”). is exempt from a reason-giving requirement, while agencies are not. Reason-giving may promote legitimacy in a number of other ways as well.118For example, Jerry Mashaw has argued that reason-giving legitimates the bureaucracy by treating “persons as rational moral agents who are entitled to evaluate and participate in a dialogue about official policies on the basis of reasoned discussion.” Jerry L. Mashaw, Reasoned Administration: The European Union, the United States, and the Project of Democratic Governance, 76 Geo. Wash. L. Rev. 99, 118 (2007).

In the context of alternatives, different alternatives will affect different constituencies differently, and some people will invariably be treated less favorably than they would have been under arrangements that might have been chosen but were not. Disclosing public-regarding justifications for why a given option has been selected over other options treats those persons as entitled to an explanation for why their interests were subordinated to the broader public’s, and it provides a focal point for contestation and debate.119See Staszewski, supra note 108, at 1278 (“Most fundamentally, reason-giving fosters democratic legitimacy because it both embodies, and provides the preconditions for, a deliberative democracy that seeks to achieve consensus on ways of promoting the public good that take the views of political minorities into account.”).

These are some of the more prominent justifications for the reason-giving requirement,120See Deeks, supra note 103, for a discussion of additional justifications. but there are potentially powerful objections to the requirement as well.121I’ll bracket here strictly legal objections. See Kathryn E. Kovacs, Rules About Rulemaking and the Rise of the Unitary Executive, 70 Admin. L. Rev. 515, 543–45 (2018) (arguing that various obligations, including the obligation to respond to alternatives, are “inconsistent with the text and history of the APA”); Am. Radio Relay League, Inc. v. FCC, 524 F.3d 227, 248 (D.C. Cir. 2008) (Kavanaugh, J., concurring in part and dissenting in part) (similar).

For one, the benefits of robust administrative law requirements are notoriously slippery, and whether particular requirements advance them, or may even undermine them, is often unclear. Take “legitimacy.” As Nicholas Bagley has argued, it could be that some requirements enhance agencies’ legitimacy in some respects.122See Bagley, supra note 20, at 379. But to the extent that placing obligations on agencies undermines their ability to do their jobs effectively, such obligations might also harm agencies’ standing with the public, leading to a loss of “sociological legitimacy.”123See id. at 379–80.

And there is a varied literature arguing that excessive proceduralization—and State Farm in particular—does in fact impair agencies’ effectiveness. One set of concerns involves the courts. As nonexperts, courts may, in assessing the quality of agencies’ reasoning, be prone to error.124See McGarity, supra note 20, at 1452 (pointing out that “there are clear limits to judicial competence in the area of highly scientific and technical rulemaking”); Martin Shapiro, Administrative Discretion: The Next Stage, 92 Yale L.J. 1487, 1507 (1983) (“Courts cannot take a hard look at materials they cannot understand nor be partners to technocrats in a realm in which only technocrats speak the language.”). Richard Pierce has argued that courts have “a remarkable instinct for the capillary,” meaning that they may mistake trivial failures for significant ones and, in faulting agencies for small mistakes, invalidate programs with large benefits.125See Richard J. Pierce, Jr., Unruly Judicial Review of Rulemaking, Nat. Res. & Env’t, Fall 1990, at 23, 24–25. And others have urged that State Farm and other forms of review open the door to second-guessing of agencies based on judges’ own values and beliefs.126See William H. Rodgers, Jr., Judicial Review of Risk Assessments: The Role of Decision Theory in Unscrambling the Benzene Decision, 11 Env’t L. 301, 302 (1981) (“[T]he suspicion has arisen, certainly among practitioners who can say such things, that the grand synthesizing principle that tells us whether the court will dig deeply or bow cursorily depends exclusively on whether the judge agrees with the result of the administrative decision.”); Thomas O. McGarity, The Courts and the Ossification of Rulemaking: A Response to Professor Seidenfeld, 75 Tex. L. Rev. 525, 549 (1997). But whether through willfulness or simple error, courts’ ability to police agency reasoning entails costs—on average, and over time, the quality of government may suffer as courts needlessly invalidate publicly valuable programs.

Even if courts could be counted on to do a reasonably good job enforcing the reason-giving requirement, it could be that the requirement itself imposes excessive burdens when compared to their uncertain benefits. Critiques in this vein often march under the “ossification” banner. The basic idea is that State Farm requirements burden agency decisionmaking, slow down agency action while staff attempt to predict and preempt any possible line of attack, and, in doing so, substantially slow the pace of agency action, particularly with regard to rulemaking.127See generally, e.g., McGarity, supra note 20; Richard J. Pierce, Jr., Judicial Review of Agency Actions in a Period of Diminishing Agency Resources, 49 Admin. L. Rev. 61 (1997). Some scholarship has suggested that the ossification thesis is overstated. See, e.g., Jason Webb Yackee & Susan Webb Yackee, Testing the Ossification Thesis: An Empirical Examination of Federal Regulatory Volume and Speed, 1950–1990, 80 Geo. Wash. L. Rev. 1414 (2012). The result may include less socially beneficial regulation, a decreased ability to respond to changed circumstances, and incentives for agencies to act through channels that escape judicial review entirely.128See, e.g., McGarity, supra note 20, at 1386.

Consider agencies’ obligation to respond to alternatives. If courts set aside every agency action in which the agency did not consider and thoroughly vet every possible alternative—in many cases, an infinite or near-infinite set—agencies would be hamstrung. Fearing the prospect of judicial invalidation, agency staff would be forced to devote significant resources to exploring alternatives even when such alternatives were believed unlikely to bear fruit. Because such resources are limited, agencies would have to carefully choose where to focus, and certain important areas might suffer from neglect. And although, as a formal matter, courts have never purported to require agencies to consider every possible alternative, an amorphous obligation to explore “reasonable” or even “obvious” ones might be just as bad from the standpoint of a risk-averse agency.129Cf. Vt. Yankee Nuclear Power Corp. v. Nat. Res. Def. Council, Inc., 435 U.S. 519, 546–47 (1978) (stating that “agencies, operating under [a] vague injunction to employ the ‘best’ procedures and facing the threat of reversal if they did not, would undoubtedly adopt full adjudicatory procedures in every instance”).

So where does that leave us? On the one hand, there are potentially good justifications for imposing a reason-giving requirement on agencies—justifications that also reach the particular obligation to respond to alternatives. Moreover, the basic obligation to respond to at least some alternatives is unlikely, as a purely descriptive matter, to go anywhere soon. On the other hand, there may be significant costs associated with a reason-giving requirement and the obligation to respond to alternatives. Most concerningly, such obligations may significantly hamstring agencies in the performance of their duties.

The proper course forward is to explore whether there are ways to craft the doctrine such that it retains the core features of the obligation while incorporating countermeasures aimed at its drawbacks. The balance of this Part performs such an exploration with respect to the first question implicated by agencies’ obligation to respond to alternatives: to which alternatives must agencies respond?

B. Alternatives to Which an Agency Must Respond Because They Were Raised

The first category of alternatives to which agencies must respond is made up of significant alternatives raised by outside parties—most prominently, during notice-and-comment proceedings.130Conversely, when a given alternative is not raised by outside parties, that should generally relieve the agency of an obligation to respond to it. See infra notes 249–251 and accompanying text (discussing issue exhaustion). Agencies’ general obligation to respond to comments, sometimes considered an aspect of arbitrary-and-capricious review and sometimes rooted in Section 553, is well established in the case law.131See, e.g., Gresham v. Azar, 950 F.3d 93, 103 (D.C. Cir. 2020); United States v. Nova Scotia Food Prods. Corp., 568 F.2d 240, 252 (2d Cir. 1977); Rodway v. U.S. Dep’t of Agric., 514 F.2d 809, 817 (D.C. Cir. 1975). And requiring agencies to respond to alternatives raised by outside parties finds support in several of the rationales for agency reason-giving requirements generally. Those with an interest in a proceeding have incentives to bring important issues to the agency’s attention, including by raising alternatives that may be attractive as a policy matter and legally relevant.132Am. Bar. Ass’n Section of Admin. L. & Regul. Prac., Comments on H.R. 3010, The Regulatory Accountability Act of 2011, 64 Admin. L. Rev. 619, 633–34 (2012). Requiring agencies to respond to such alternatives thus sheds light on agencies’ policy and legal rationales for acting, serving accountability goals, and provides recognition to the interests of affected entities. Moreover, this requirement may improve agencies’ decisionmaking from the start of the process before any comments have actually been filed. Because agencies know that courts may invalidate their actions if they have not responded to raised alternatives, agencies will work to anticipate alternatives that may be raised.133See id.

Thus, there are good reasons to shape an agency’s obligation to respond to alternatives by reference to the outside input it has received. But it also cannot be that an agency must respond to every alternative proposed by an outside party, no matter how undeveloped or outlandish. As a matter of the law as it is, courts do not purport to require agencies to respond to each and every proposed alternative.134E.g., Nat’l Shooting Sports Found., Inc. v. Jones, 716 F.3d 200, 215 (D.C. Cir. 2013). Limiting agencies’ obligation to respond prevents parties from peppering the comments with alternatives in order to impede the agency in its decisionmaking or with the hope that the agency’s inevitable failure to respond will lead to eventual judicial invalidation. It is also in line with general administrative law principles. In the context of notice-and-comment rulemaking, agencies must only respond to comments that are “relevant and significant.”135Grand Canyon Air Tour Coal. V. FAA, 154 F.3d 455, 467–68 (D.C. Cir. 1998). The D.C. Circuit has stated that the “requirement of agency responsiveness to comments is subject to the common-sense rule that a response be necessary.”136Nat. Res. Def. Council, Inc. v. EPA, 859 F.2d 156, 188 (D.C. Cir. 1988). The closest the D.C. Circuit has come to telling us what that means is suggesting that the inquiry be directed toward the “ ‘materiality’ ” of the comment—in other words, the comment, and the agency’s failure to respond, must be “of possible significance in the results [the agency reaches].”137Northside Sanitary Landfill, Inc. v. Thomas, 849 F.2d 1516, 1519 (D.C. Cir. 1988) (alteration in original) (quoting Portland Cement Ass’n v. Ruckelshaus, 486 F.2d 375, 394 (D.C. Cir. 1973)). One way to think about it is: is it possible that the comment might have changed the agency’s mind or affected its decision in any way?

In the context of alternatives, the question thus becomes whether consideration of the proposed alternative might have altered the agency’s decision. In making that determination, I propose that the courts take the following considerations—supported both by the case law and by common sense—into account.

First, where the agency’s approach suffers from significant flaws apparent from the record—and especially where the agency has acknowledged such flaws—agencies are more likely to be required to respond to proposed alternatives that address those flaws.138See City of Brookings Mun. Tel. Co. v. FCC, 822 F.2d 1153, 1169 (D.C. Cir. 1987) (“This court has been particularly reluctant to blink at an agency’s ignoring ostensibly reasonable alternatives where it admits, as the Commission has here, that the choice embraced suffers from noteworthy flaws.”); Farmers Union Cent. Exch., Inc. v. Fed. Energy Regul. Comm’n, 734 F.2d 1486, 1511 (D.C. Cir. 1984) (stating that the agency’s “responsibility becomes especially important when the agency admits its own choice is substantially flawed”). By contrast, alternatives designed to solve problems that do not exist need not be considered. Consider Sociedad Anonima Viña Santa Rita v. U.S. Department of Treasury, which dealt with the Bureau of Alcohol, Tobacco and Firearms’ decision to allow winemakers to use “Santa Rita Hills” as an appellation of origin for wines.139Sociedad Anonima Viña Santa Rita v. U.S. Dep’t of Treasury, 193 F. Supp. 2d 6, 8 (D.D.C. 2001). When challengers later argued that the agency should have chosen a different appellation, the court faulted them for not having raised alternatives at the agency level.140See id. at 24–25. Independently, however, the court stated that there was “simply no need to consider any alternatives” because the agency had reasonably concluded that the Santa Rita Hills designation was not likely to confuse consumers, the purported flaw that justified a search for different names.141Id.

Second, and relatedly, where the agency’s reasoning—especially in the decision under review but potentially also by reference to past actions—supports a given alternative, that alternative is likely more worthy of consideration and thus deserving of a response.142See Int’l Ladies’ Garment Workers’ Union v. Donovan, 722 F.2d 795, 816 (D.C. Cir. 1983) (“Perhaps the clearest indication that [the alternatives] warranted serious consideration is provided by the Secretary’s explanation . . . .”).

Third, agencies are under a greater obligation to respond to alternatives that have been “sufficiently described,” and where the commenter has provided “sufficient detail and support” to allow the agency to understand and evaluate the alternative.143City of Brookings, 822 F.2d at 1169–70. But where a proffered alternative is not supported by evidence in the record, that alternative is less likely to warrant a response.144See Nat’l Shooting Sports Found., Inc. v. Jones, 716 F.3d 210, 216 (D.C. Cir. 2013); Home Box Off., Inc. v. FCC, 567 F.2d 9, 36 (D.C. Cir. 1977). That makes good sense. Functionally, this limitation on agencies’ obligation to respond means that the burden for providing support for an alternative is placed on the party proposing it. And it prevents parties from flooding an agency with numerous bare-bones alternatives and then requiring the agency itself to devote scarce resources to evaluating them.

Finally, where the problem is complex and the number of potential alternatives vast, the agency should bear a lesser burden to respond to every alternative it receives.145The case closest to holding so is Oceana, Inc. v. Evans, 384 F. Supp. 2d 203, 241 (D.D.C. 2005). Rather, in such situations, the agency discharges its burden if it considers a reasonable range of alternatives and incorporates “the thrust” of the parties’ proposals into its decisionmaking, even if it does not consider the “exact alternatives” proposed by all parties.146See id. at 242. Such a limitation is a nod to the practical limitations that beset agency decisionmaking under complex and uncertain conditions. In such situations, it’s sensible to consider the “materiality” of an individual proposed alternative not in isolation but by reference to the decisionmaker’s overall cognitive load. Where the agency has considered representative alternatives and has shown awareness of the nature of the problem and “the thrust” of potential solutions, it’s unlikely that failure to consider this or that exact proposal would alter the agency’s ultimate decision.

Where the number of possible alternatives facing an agency is large, ensuring that the agency has examined representative alternatives while relieving the agency of the burden of responding to them all also partially offsets one potential downside of linking the obligation to respond to the notice-and-comment process. Wendy Wagner has argued that requiring parties to raise objections in their comments, and requiring agencies to respond to comments that provide sufficient detail to warrant a reply, creates a dynamic in which parties can “wear the agency down” by flooding it with information.147See Wagner, supra note 19, at 1363–65. If that’s a problem, it attaches to the entire legal structure built around judicial review of the products of informal rulemaking. “Wearing the agency down” is not necessarily unique to the problem of the obligation to respond to alternatives. It is thus likely to require more fundamental reforms to that structure.148See id. at 1403–31 (exploring potential solutions). But in the context of alternatives, requiring agencies to respond only to the alternatives that are necessary for the court “to see what major issues of policy were ventilated and why [the] agency reacted to them as it did,”149Pub. Citizen, Inc. v. FAA, 988 F.2d 186, 187 (D.C. Cir. 1993). and thus judge whether the agency acted in conformance with the statute, can partially address the dynamic Wagner identifies.

One note on the above: in the next Sections I will be concerned—perhaps excessively so—with divorcing courts’ judgment about which unraised alternatives are worthy of consideration from the policy attractiveness of those alternatives. Similarly here, I have tried to tie the above considerations to something other than the abstract policy merits of the alternative under review. But they are still a little less rule-like, and more tied to policy analysis, than I might prefer. That is partly because the nature of the inquiry—whether a raised alternative is material—is sufficiently sensitive to the particular policy context that easy rules are difficult and judgment is inevitable. And it’s also because, where an alternative has been developed in the administrative record, courts at least have that to go on, as opposed to situations where the alternative has been raised for the first time in litigation.

C. Alternatives to Which an Agency Presumptively Must Respond

This Section proposes several categories of alternatives that agencies presumptively must address, regardless of whether they were raised by an outside party. Identifying such alternatives is particularly important for cases where the agency has not followed a process allowing for outside input. In crafting the presumptions, I’ve been guided by the following considerations.

First, the presumptions allow courts to identify alternatives in a way that does not require the court to undertake its own policy analysis. Courts often speak of agencies’ obligation to address “obvious” or “reasonable” alternatives, by which they seem to mean alternatives that appear attractive as a policy matter.150See e.g., Off. of Commc’n of United Church of Christ v. FCC, 779 F.2d 702, 712–14 (D.C. Cir. 1985). But the attractiveness of a given option may be difficult to assess on a cold administrative record that, necessarily, does not include information about the alternative in question. Therefore, I have tried to limit such alternatives to those that can be revealed by other means—through statutory interpretation, for example, or by comparing agency positions over time. Because such alternatives are discoverable in a more straightforward way than by performing an assessment of an alternative’s reasonability in the abstract, they can be considered “well-known at the time, including to the agency itself,” thus limiting judges from imposing their own sense of what alternatives are “obvious.”151Oakbrook Land Holdings, LLC v. Comm’r, 28 F.4th 700, 722 (6th Cir. 2022).

Second, the presumptive obligations represent a reasonable accommodation among various administrative law values, some of which conflict with each other if pursued single-mindedly. On one side are the values associated with reason-giving itself, which would push toward a maximalist understanding of the obligation to respond to alternatives. But other values—such as the need for effective governance and for a relatively predictable framework of judicial review—cut in the other direction. My sense, and hope, is that the presumptions will, in most cases, do a good job in revealing a sense of the tradeoffs the agency understands itself to be making and the factors that guided its ultimate decision—thus serving the first set of values—while allowing agencies to effectively administer government programs in a relatively more stable legal environment.

Third, the presumptions roughly explain the results that courts have reached when assessing agencies’ obligations. In other words, they broadly fit with the extant case law. But my analysis also reveals occasions on which courts have gone astray.

1. Alternatives Made Relevant by Statute

The first category of alternatives to which an agency must respond are those that are made relevant by the underlying statute. Discovering such alternatives involves an act of statutory interpretation.

The simplest case in which a statute requires the consideration of a given alternative is when the statute says so—when it explicitly commands the agency to consider certain alternatives prior to making a decision. Take Overton Park. In that case, the secretary of transportation was provided with discretion to authorize the use of federal funds to finance the construction of highways running through public parks.152Citizens to Pres. Overton Park, Inc. v. Volpe, 401 U.S. 402, 405 (1971). But prior to doing so, the secretary was required to consider whether there were alternative routes that did not run through parkland and assess whether such alternatives were “feasible and prudent.”153Id. (citing 49 U.S.C. § 1653(f) (Supp. V 1964) (current version at 49 U.S.C. § 303(a)–(c)); 23 U.S.C. § 138 (Supp. V 1964) (current version at 23 U.S.C. § 138). The “feasible and prudent” standard may or may not have been a stringent one.154Overton Park, 401 U.S. at 411–12. But no one doubted that the secretary was under a basic obligation to identify some range of alternative routes in order to satisfy his statutory obligation, and the Court’s holding ensured the availability of an administrative record upon which to judge whether that obligation was satisfied.155See id. at 411, 420. And for good reason. It is bedrock administrative law that agencies must follow the analytic method required of them by statute because agencies are creatures of Congress and, accordingly, must follow congressional commands.156See, e.g., J.R. DeShazo & Jody Freeman, The Congressional Competition to Control Delegated Power, 81 Tex. L. Rev. 1443, 1456 (2003) (describing how Congress may “establish substantive standards or limits that the agency must implement (such as a health or safety standard), designate decisionmaking procedures for the agency to follow (such as notice and comment rulemaking with all ‘centrally relevant’ materials docketed), and mandate analytic tools the agency must employ (such as cost-benefit analysis)”). An agency that fails to consider an alternative when required to do so by statute has thus committed legal error.

There may also be situations in which a statute implicitly requires an agency to consider a certain alternative and the case law, though sparse, supports that conclusion.157See FBME Bank Ltd. v. Lew, 125 F. Supp. 3d 109 (D.D.C. 2015); see also Allied Loc. & Reg’l Mfrs. Caucus v. EPA, 215 F.3d 61, 80 & n.20 (D.C. Cir. 2000); Spirit Airlines, Inc. v. U.S. Dep’t of Transp., 997 F.3d 1247, 1255 (D.C. Cir. 2021); Ramirez v. U.S. Immigr. & Customs Enf’t, 471 F. Supp. 3d 88, 178 (D.D.C. 2020). The most on-point case, FBME Bank Ltd. v. Lew, dealt with a provision of the USA PATRIOT Act allowing the Treasury Department to require domestic banks to take “[one] or more special measures” against a foreign financial institution that may be operating as a conduit for money laundering.158FBME Bank, 125 F. Supp. at 124 (alteration in original) (quoting 31 U.S.C. § 5318A(a)(1)). In the decision under review, the agency had chosen the most drastic of the identified measures by banning domestic institutions from opening or maintaining so-called “correspondent accounts” on behalf of a foreign bank, effectively precluding that bank from operating within the United States.159See id. As the court explained, the statute authorized the agency to require alternative measures, including “the imposition of conditions on the opening or maintaining of correspondent accounts, rather than the full prohibition of opening or maintaining such accounts.”160Id. at 124–25. And although the agency had “implicitly” responded to some of the statutorily identified alternatives, it had not provided any explanation for why it had chosen to ban rather than condition the maintenance of correspondent accounts.161See id. The court held that it was incumbent on the agency to do so, especially given the draconian nature of the agency’s choice.162Id.

When an agency selects from among a number of statutorily authorized alternatives, it makes sense that its decision must disclose why it chose the alternative it did over the others. First, as highlighted by the particular context of FBME Bank, such a requirement is a way to ensure an agency considers intermediate, less costly ways to achieve the agency’s purpose without the drawbacks of a generalized obligation to consider some amorphous set of “less drastic” alternatives.163See infra notes 197–198 and accompanying text (discussing the pros and cons of requiring agencies to consider less drastic measures). More broadly, when a statute includes a set of options, those options are relevant to the choice being made, and evaluating them is typically required as part of an agency’s broader obligation to rest its decisions on statutorily relevant considerations.164See Fed. Energy Regul. Comm’n v. Elec. Power Supply Ass’n, 577 U.S. 260, 292 (2016) (arbitrary-and-capricious review includes whether the “agency has ‘examine[d] the relevant [considerations]’ ” (alterations in original)). Consider a principal who tells her agent to “fetch some soupmeat or some greens from which a salad can be made.”165Cf. William N. Eskridge, Jr., “Fetch Some Soupmeat, 16 Cardozo L. Rev. 2209 (1995) (discussing Francis Lieber, Legal and Political Hermeneutics, or Principles of Interpretation and Construction in Law and Politics 18 (William G. Hammond ed., St. Louis, F.H. Thomas & Co. 3d ed. 1880) (1837)). If the agent buys greens and the principal inquires why, normally she would be asking for some reason why the greens were chosen over the explicitly identified alternative: soupmeat. And the answer may reveal whether the agent has followed the set of meta-rules that govern their principal-agent relationship—in administrative law speak, whether the agent has based its decision on relevant factors. For example, if the agent replies “because the greens were the first thing I saw,” the principal may rightly be cross if her general instructions are to select soupmeat unless the price is prohibitively high. Requiring an agency to disclose the reasons they selected one alternative over another “obvious” one—a statutorily provided one—can thus help ensure that the agency has made the choice among alternatives that were at the top of Congress’s mind by using considerations that the statutes places in bounds.

One might object at this point that, in certain circumstances, even statutorily identified alternatives might be so implausible that they do not warrant discussion. To test the limits of agencies’ obligation to consider statutorily identified alternatives, let’s return to the CDC’s order requiring masking on airplanes and most other forms of public transportation in response to the COVID-19 pandemic. The statutory basis for that order was a provision allowing the surgeon general to “make and enforce such regulations as in his judgment are necessary to prevent the introduction, transmission, or spread of communicable diseases.”16642 U.S.C. § 264(a). The statute then lists a number of particular measures that the CDC may impose in the exercise of its authority, including “fumigation” and “destruction of animals or articles,” and goes on to authorize “other measures, as in [the agency’s] judgment may be necessary.”167Id. You might plausibly question whether, in the course of promulgating a transportation masking rule as an “other measure[],” the agency should have to provide reasons why it rejected statutorily specified alternatives such as fumigation or the destruction of animals—alternatives that are facially implausible given the nature of the COVID-19 pandemic. I want to suggest that the answer is yes, because the statute makes such alternatives relevant and “obvious,” and then ask that you momentarily suspend your objections to that answer until Part III. There, I’ll argue that what matters is whether a court would understand why the agency chose the course it did over others, even if the other options were not explicitly mentioned. If the court would understand, the agency’s obligation is satisfied. But if at the end of the day a statutorily identified alternative has not been addressed, even implicitly, it is not.

2. Individual Components of Rescinded Programs

The second category of alternatives to which agencies presumptively must respond, exemplified by State Farm and Regents, is the most firmly rooted in the Supreme Court’s case law: when rescinding a program, the agency must give reasons sufficient to explain why it chose not to retain each individual component of it.

The canonical Supreme Court cases illustrate this obligation well. In State Farm, the agency repealed the passive-restraint rule without considering the alternative of keeping an airbags requirement.168Supra notes 44–48 and accompanying text. In Regents, DHS repealed the DACA program without considering the alternative of retaining the parts of it dealing with forbearance from removal.169Supra notes 74–85 and accompanying text. In both cases, this led the Court to vacate the agencies’ actions. Picking up on language from State Farm, the lower courts have generally characterized the agency’s obligation as one to “consider alternatives ‘within the ambit of the existing standard.’ 170E.g., MediNatura, Inc. v. Food & Drug Admin., 998 F.3d 931, 943 (D.C. Cir. 2021) (quoting Motor Vehicle Mfrs. Ass’n of the U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 51 (1983)).

An agency program with multiple, independent parts is of course not unusual. Consider net neutrality regulation. As formulated by the FCC, and at a high level of generality, net neutrality regulation has involved two components: first, a set of restrictions governing the conduct of internet service providers (ISPs) (i.e., things such providers cannot do) and second, a set of transparency obligations requiring ISPs to publicly disclose their network management practices.171See In the Matter of Protecting and Promoting the Open Internet, 30 FCC Rcd. 5601 (2015). Those components sometimes march together.172See id. (imposing both). But they needn’t. In fact, when the FCC rescinded most of the Obama-era net neutrality regulations following the election of Donald Trump, the Republican-majority Commission retained a somewhat narrowed version of the transparency-related components of the rules. The Commission explained that the ills the FCC identified with the conduct restrictions did not extend to all disclosure obligations.173See In the Matter of Restoring Internet Freedom, 33 FCC Rcd. 311 (2018). Imagine, by contrast, that the Commission had explained why it believed that restricting marketplace conduct by ISPs was no longer justified and had rescinded all the previous net neutrality rules, including transparency-related ones. If the agency had not explained separately why it was jettisoning the components of the prior regime involving transparency obligations, State Farm may have provided a basis for invalidating the agency’s action.

This core State Farm obligation makes sense for a few reasons. In promulgating the original program, the agency had presumably given reasons, rooted in its expertise and in the factors made relevant by the statute, for each individual component of that program. If it had not done so for a particular component, that would have provided the basis for a valid legal challenge.174See, e.g., Carlson v. Postal Regul. Comm’n, 938 F.3d 337, 352 (D.C. Cir. 2019) (vacating one component of the agency rule for lack of reasoned decisionmaking while letting others stand). Then, the question is, having given independent reasons for the various parts of a regulatory regime, may the agency then rescind that regime in toto without speaking to its individual components? To answer yes would introduce an asymmetry: in promulgating a regulation, the agency would have had to provide an explanation for each of its individual components. But in rescinding that same regulation, the agency need not.

And that asymmetry would require courts to bless the rescission of rules in which significant gaps exist in the agency’s reasoning. Where an agency has given valid reasons for a component at Time 1 and has said nothing about it at all at Time 2, the agency’s Time 1 judgment remains the agency’s officially expressed view. Airbags are still good. So are transparency obligations. It is therefore valid to ask, and to expect the agency to answer, why it has nevertheless decided not to retain the aspects of the program that are still supported by the agency’s original rationale. Standards of consistency and rational decisionmaking, also reflected in other parts of administrative law, would seem to require no less.175See Child.’s Hosp. Ass’n of Tex. v. Azar, 933 F.3d 764, 773 (D.C. Cir. 2019) (“An ‘unexplained inconsistency’ with an earlier position renders a changed policy arbitrary and capricious.” (quoting Encino Motorcars v. Navarro, 579 U.S. 211, 222 (2016))). That is at least true when there is an unexplained inconsistency between an agency’s new policy and the particular policy it purports to supplant. A different set of considerations may attach to inconsistencies in an agency’s reasoning that become evident from examining wholly different proceedings where the policies in question are not alternatives to each other.

Not requiring agencies to give reasons for why it rescinded an entire regulation would have other negative consequences. For one, it would introduce arbitrariness based on when the prior rules had been promulgated. Let’s say the FCC imposes transparency obligations first and then several years later promulgates marketplace conduct restrictions. No one would think that the agency could subsequently get rid of both without giving reasons that go to the rescission of each. The obligation shouldn’t change just because the agency instead promulgates them at the same time under the umbrella of a single “rule.” Finally, as discussed above,176See supra notes 117–118 and accompanying text. allowing agencies to provide reasons that go to the rescission of some aspects of a program but not others can undermine accountability. By ignoring the availability of airbags, the agency in State Farm, for example, was not truly owning up to the extent it was choosing to depart from its prior judgments regarding the proper tradeoff between safety and cost.177See Ronald M. Levin, Administrative Discretion, Judicial Review, and the Gloomy World of Judge Smith, 1986 Duke L.J. 258, 271–72 (explaining that after State Farm “the agency had to choose between changing its policy and telling the public more candidly why requiring devices that did work was nevertheless not in the public interest”). The obligation also facilitates legal accountability by allowing courts to observe whether the agency’s proffered reasons for jettisoning an individual component comport with the statute.

One potential objection to requiring agencies to consider the alternative of retaining components of a rescinded program is that it requires judges to identify what the “components” of the prior program that agencies must consider retaining are. Because whether an alternative can be well-defined is an important consideration,178See supra note 152 and accompanying text. this objection has some force.

The problems involved with identifying the various components of preexisting regulatory regimes are not so intractable that it makes sense to abandon the obligation in light of its benefits. Recall that the normative roots of the obligation to respond to alternatives lie in agencies’ obligation to give reasons for their actions.179See supra Section II.A. For our purposes, therefore, what matters is whether different aspects of a regulatory regime stand or fall together based on the same reasons. If not, those aspects are distinct (i.e., individual) components, and an agency that wishes to rescind the program must speak to each. Moreover, this determination needn’t be made by seeking to intuit whether two aspects of the same program might be justified by the same reasons. Instead, judges can examine what the agency itself said in promulgating the original program. This makes the inquiry dependent on the agency’s original rationale. For example, let’s say that at Time 1 an agency adopts (1) transparency rules and (2) direct restrictions on marketplace conduct, but it justifies the transparency rules solely as a means to enforce the conduct restrictions. If at Time 2 the agency rescinds the entire program after concluding that the conduct restrictions were no longer justifiable, it has given a valid reason for getting rid of the transparency rules as well: the transparency rules were only ever justified by reference to the conduct restrictions. By contrast, if at Time 1 the agency had given distinct reasons for adopting the transparency rules—say, that they facilitated consumer choice even absent other forms of direct regulation—the agency would be obligated to speak to them independently if it wishes to later rescind the entire program.180Admittedly, this inquiry might lead courts to find that the agency acted arbitrarily and capriciously by failing to speak to relatively minor (but still independent) aspects of the prior regulatory regime. In my view, the downsides associated with that prospect are better dealt with by remedial doctrines such as remand without vacatur and not through tinkering with agencies’ substantive obligations. See, e.g., Mozilla Corp. v. FCC, 940 F.3d 1, 86 (D.C. Cir. 2019) (remanding without vacating the FCC’s rescission of its net neutrality rules despite flaws in the agency’s explanation).

Although the core of the State Farm requirement is defensible, courts have taken it too far in two contexts. First, in the context of rescissions, courts have occasionally required agencies to consider not only the option of retaining components of the prior regime but also the alternative of making “modifications” to that regime.181See State v. Biden, 10 F.4th 538, 554–57 (5th Cir. 2021); Off. of Commc’n of the United Church of Christ v. FCC, 707 F.2d 1413, 1438–41 (D.C. Cir. 1983). The most prominent recent example arose from the Biden Administration Department of Homeland Security’s rescission of the Trump-era Migrant Protection Protocols (MPP). The MPP required the “return” of certain noncitizens to Mexico during the pendency of their immigration proceedings.182See State v. Biden, 10 F.4th at 543. The district court found that the Department, in a June 2021 memorandum rescinding the program, had “failed to meaningfully consider more limited policies than the total termination of MPP.”183Texas v. Biden, 554 F. Supp. 3d 818, 849 (N.D. Tex. 2021). The district court faulted the agency for not having “identif[ied] . . . what a modified MPP would look like or what kind of investment would be needed to modify or scale back MPP.”184Id. And it vacated the Department’s rescission on that basis (among others). The Fifth Circuit largely adopted the district court’s reasoning.185See Texas v. Biden, 20 F.4th at 992. The Supreme Court ultimately reversed the Fifth Circuit. Biden v. Texas, 142 S. Ct. 2528 (2022). But because the Court found that the explanation given by the Department in the June 2021 memorandum was not controlling given subsequent agency action, it did not reach the alternatives issue. See id. at 2544–48.

Applying arbitrary-and-capricious review, courts should not require agencies to consider some unspecified number of modifications when the agency has given reasons why it no longer believed the program as a whole was worth retaining—i.e., when it has explained why it was rejecting the status quo, including its various component parts. Considering modifications to a program prior to rescinding it often makes good sense as a matter of regulatory best practices. 186And of course, discussion of various modifications may properly be part of the back-and-forth that occurs between the agency and the Office of Information and Regulatory Affairs prior to finalizing significant regulatory actions. See Donald R. Arbuckle, Collaborative Governance Meets Presidential Regulatory Review, 2009 J. Disp. Resol. 343, 350 (describing such back-and-forth). See generally Cass R. Sunstein, The Office of Information and Regulatory Affairs: Myths and Realities, 126 Harv. L. Rev. 1838 (2013) (describing OIRA’s various roles). And an agency must attend to modifications offered by significant comments in the context of notice-and-comment proceedings.187See supra Section II.B. But as a matter of court-enforced obligation, agencies should not have to cook up this-or-that modifications in order for their actions to survive review.

That is so for three reasons. First, the number of possible “modifications” to a regulatory program is potentially large, often approaching the infinite, and the agencies’ obligation to respond to them is therefore difficult to keep within reasonable bounds. In rescinding Remain in Mexico, the secretary had explained that “addressing the deficiencies identified in [his] review would require a total redesign that would involve significant additional investments in personnel and resources.”188State v. Biden, 10 F.4th at 555 (quoting secretary’s June 1 memorandum). That kind of global assessment—as opposed to a line-by-line accounting of various “modifications” and their costs and benefits—should generally be enough.189See infra notes 311–316 and accompanying text (defending agencies’ ability to give global justifications for the rejection of categories of alternatives).

Second, modifications not reflected in the preexisting regulatory regime are not necessarily supported by the agency’s prior exercise of reasoned decisionmaking, insofar as the agency might not originally have considered those modifications. Therefore, one of the primary justifications for requiring agencies to consider retaining the individual components of prior programs—that each of those individual components were previously justified—does not apply.190See supra note 177 and accompanying text. Of course, some modifications might be justified by reference to the reasoning the agency gave for the original program. But attempting to delineate which modifications are supported by the agency’s prior reasoning and which ones are not takes us into a territory where courts may be likely to perform poorly.

Third, where the agency proceedings involve outside input, an agency’s obligation to assess the alternatives of modifying the program in question can be funneled through, and cabined by, the agency’s obligation to respond to significant comments.191See supra Section II.B (assessing agencies’ obligation to respond to alternatives suggested by outside parties). If a particular modification is worthy of sustained attention, the public (including regulated entities) can be expected to advocate for that alternative at the agency level. Where interested-party advocacy has not occurred, there is likely little additional benefit from a court imposing an amorphous obligation to respond to some unknown set of unspecified alternatives.192See Ass’n of Pub.-Safety Commc’ns Offs.-Int’l, Inc. v. FCC, 76 F.3d 395, 400 (D.C. Cir. 1996) (relieving agency of obligation to consider every potential modification to rescinded program when the agency “did clearly address the alternatives that had been raised during the comment periods,” and stating that “[t]he fact that the Commission might not have addressed and rejected every conceivable approach to the challenge . . . does not render its decision invalid”).

The second way in which courts have pushed the State Farm obligation too far extends outside the context of rescission. Several courts have intimated that, whether in promulgating rules for the first time or altering existing ones, an agency is under a general obligation to consider “less drastic” or “less far-reaching” alternatives to its chosen path.193See, e.g., Yakima Valley Cablevision, Inc. v. FCC, 794 F.2d 737, 746 (D.C. Cir. 1986); Nat’l Black Media Coal. v. FCC, 775 F.2d 342, 357 (D.C. Cir. 1985); Int’l Ladies’ Garment Workers’ Union v. Donovan, 722 F.2d 795, 815 (D.C. Cir. 1983). Again, as a matter of regulatory policy, considering less drastic alternatives often makes good sense (though so does considering more drastic alternatives). Consider an agency that is deciding whether to set an air quality standard for a certain pollutant at a level lower than the current 10ppm limit. If the agency initially believes that a new standard of 8ppm would be appropriate, it should likely give attention to whether its goals could instead be met with a “less drastic” 9ppm standard, which would invariably involve fewer compliance costs. Doing so would simply be good cost-benefit analysis.194See Off. of Mgmt. & Budget, Exec. Off. of the President, Circular A-4, at 26 & n.10 (2003) (“For each alternative that is more stringent than the least stringent alternative, you should estimate the incremental benefits and costs relative to the closest less-stringent alternative.”).

The problem is that, when taken to an extreme, an obligation to rebut less drastic alternatives is an invitation to regulatory standstill, particularly when agencies are presented with a choice along a continuum. Does the agency in the above example need to assess an 8.1ppm standard? An 8.01ppm standard? Perhaps the most notorious example of a court requiring too much of an agency in this regard is Corrosion Proof Fittings v. EPA, which considered the EPA’s authority to ban asbestos.195Corrosion Proof Fittings v. EPA, 947 F.2d 1201 (5th Cir. 1991). There—in the context of a statute requiring the agency to adopt the “least burdensome” requirements—the court seemed to place the onus on the EPA to consider whether there was “any other regulation” imaginable that would achieve an acceptable level of risk while imposing a lesser burden.196See id. at 1216 (emphasis added). As Rachel Rothschild has explained, “[a] requirement that EPA show that no other rule would achieve a similar level of risk reduction is unreasonable and likely contributed to fears that any future toxics regulation would simply lead to paralysis by analysis.”197Rachel Rothschild, Unreasonable Risk: The Failure to Ban Asbestos and the Future of Toxic Substances Regulation, 47 Harv. Env’t L. Rev. 529, 580 (2023). A similar rule imposed under the auspices of arbitrary-and-capricious review would suffer from the same serious flaw.

For an approach that helps reconcile the commonsense notion that regulators should consider whether more modest approaches might yield similar benefits with the potential for “paralysis by analysis,” it’s fruitful to look to the executive branch’s own practices for performing cost-benefit analyses. The U.S. Office of Management and Budget’s Circular A-4 recognizes that “[t]he number and choice of alternatives selected for detailed analysis is a matter of judgment” and that “[t]here must be some balance between thoroughness and the practical limits on your analytical capacity.”198 Off. of Mgmt. & Budget, supra note 194, at 7. As Rothschild points out, the EPA similarly notes that “there is often a trade-off between ‘considering more alternatives and developing more detailed, quantified, and reliable benefit and cost estimates for fewer alternatives.’ 199Rothschild, supra note 197, at 580 (quoting U.S. Env’t Prot. Agency, Guidelines for Performing Regulatory Impact Analysis M-5 (1983), []). The practice on the ground is therefore to select some reasonable number of alternatives that can be analyzed in more depth. According to Circular A-4, those alternatives may, in appropriate cases, include only three: the preferred option, a more stringent option, and a less stringent option.200 Off. of Mgmt. & Budget, supra note 194, at 16. I would add, as further developed below, a fourth option: retaining the status quo, or the “do nothing” alternative.201See infra Section II.C.3.

Better-reasoned cases thus recognize that “the duty to consider . . . alternatives does not extend to situations where the possibilities are so numerous and the goals of the action so complex that the agency cannot possibly consider every significant alternative in a reasonable time period.”202See, e.g., Oceana, Inc. v. Evans, 384 F. Supp. 2d 203, 241 (D.D.C. 2005). In such situations, rather, “the agency has discretion to choose a manageable number of alternatives to present a reasonable spectrum of policy choices that meet the goals of the action.”203Id.; see also Loyola Univ. v. FCC, 670 F.2d 1222, 1227 (D.C. Cir. 1982) (“The fact that there are other solutions to a problem is irrelevant provided that the option selected is not irrational.”).

Here, Circular A-4’s guidance, with my tweak, would seem to work for many cases: the agency should analyze (1) the status quo, (2) a less stringent option, (3) the agency’s preferred approach, and (4) a more stringent option.

If an agency does so, that should satisfy the agency’s obligation for purposes of judicial review. If the agency describes through the use of “example” alternatives why it has chosen not to go further in its regulatory approach, and why it has chosen to go as far as it has, its reasons for doing so can be analyzed for conformity with the statute and with the other requirements of reasoned decisionmaking. For example, if the agency has rejected less stringent options because it believes that cost considerations are irrelevant under the statute, that belief will come to the fore as long as the agency describes why any single less stringent option was rejected.204Cf. Michigan v. EPA, 576 U.S. 743 (2015) (finding that agency had unreasonably concluded that cost was not a relevant factor). That kind of decisionmaking abandons the pretense that the agency’s choice can be made with scientific precision, particularly if that choice is occurring along a continuum and not every point along that continuum is considered. As the Fourth Circuit has explained, in such circumstances the selection may seem “arbitrary” in some sense, but it is “the kind of arbitrariness that is inherent in the exercise of discretion amid uncertainty and not in the kind of arbitrariness that the [APA] condemns when it exists in tandem with capriciousness.”205Fishermen’s Dock Coop., Inc. v. Brown, 75 F.3d 164, 172 (4th Cir. 1996).

3. The Status Quo

One “less drastic” alternative to which agencies presumptively must respond is the one alluded to above: doing nothing. Retaining the status quo comes in different forms. Where there are rules or policies in place, considering the status quo involves assessing whether they should be kept, repealed, or modified. Where the agency is writing on a blank slate, considering the status quo means assessing whether the problem is best handled through the market and, importantly, background rules such as contract law, the tort system, and antitrust, as opposed to through the agency’s own authorities.206C.f. Exec. Order No. 12,866, 3 C.F.R. 638, 639 (1994).

The reasons for requiring agencies to consider retaining the status quo are similar to those for requiring agencies to consider retaining individual components of existing programs. Where a policy is already in place, that policy represents the agency’s (prior) exercise of reasoned decisionmaking, and principles of consistency require the agency to explain why its old rationales are no longer dispositive if it wishes to repeal or modify it. Where there are no rules in place, that’s not the case, but principles underlying the judicial review of agency action still support requiring agencies to disclose why they have departed from the status quo. Every APA case involves a challenger who has suffered a legal wrong because of, or been “adversely affected or aggrieved by,” an agency action.207See 5 U.S.C. § 702 (limiting the APA’s right of review to such persons). In most instances, that means that they would have preferred the agency do nothing over whatever the agency chose to do. When they demand reasons from the agency for what it did—when they ask why did you hurt me?—what they’re really asking is, why did you not leave things in place, in the way that I preferred? An agency that has no answer to that question has given a substantially incomplete accounting for its action.

That an agency must attend, in at least a general sense, to whether the status quo should be retained may seem unexceptional—even obvious. And indeed, there’s a substantial body of case law in that direction.208See, e.g., Am. Fed’n of Lab. & Cong. of Indus. Orgs. v. Brock, 835 F.2d 912, 918 n.7 (D.C. Cir. 1987) (faulting agency for “fail[ing] to consider the obvious alternative of maintaining [its] enhancement policy”); Greater Bos. Television Corp. v. FCC, 444 F.2d 841, 852 (D.C. Cir. 1970) (“[A]n agency changing its course must supply a reasoned analysis indicating that prior policies and standards are being deliberately changed, not casually ignored, and if an agency glosses over or swerves from prior precedents without discussion it may cross the line from tolerably terse to the intolerably mute.” (footnotes omitted)). The cost-benefit executive order is in accord, requiring agencies to consider the option of “not regulating” and admonishing agencies to regulate only when necessary and in response to identified deficiencies with the status quo.2093 C.F.R. at 638–40.

There is one complication, but I do not believe its consideration requires relieving agencies of their burden to respond to a status-quo option. The complication comes with the Supreme Court’s 2009 decision in FCC v. Fox Television, which interfered with the above-referenced body of case law to some uncertain degree in the context of agency reversals of position.210FCC v. Fox Television Stations, Inc., 556 U.S. 502 (2009). Fox Television dealt with the FCC’s decision to abandon its prior permissiveness when it came to “fleeting expletives” broadcast over the airwaves—henceforth, the FCC declared, such expletives may provide the basis for fines against broadcasters.211Id. at 505–13 (describing evolution of FCC’s approach). In the course of finding that the agency shouldered no heightened burden when explaining reversals of prior policies, the Court wrote that an agency

“need not demonstrate to a court’s satisfaction that the reasons for the new policy are better than the reasons for the old one; it suffices that the new policy is permissible under the statute, that there are good reasons for it, and that the agency believes it to be better, which the conscious change of course adequately indicates.”212Id. at 515.

That language would seem to relieve the agency of the obligation to say anything regarding why the status quo was abandoned. Instead, the agency need only give reasons for the new policy and then assert, rather than explain, that it “believes” that policy to be superior to what came before.

Notwithstanding the opinion’s language, I do not believe that Fox Television can possibly stand for the proposition that the agency need not give any explanation for why it’s jettisoning the status quo. What Fox Television did was reject the proposition that agency reversals are subject to a “heightened standard”—something above and beyond what is normally required of agencies under the APA.213Id. at 514. Even so, the Court indicated that the agency must provide a “more detailed justification” when adopting policies that contradict previously found facts or threaten “serious reliance interests.”214Id. at 515. But even outside such situations, it would be illogical to read Fox Television as permitting agencies to decline to provide any reasons for why it was rejecting the status quo option. Rather, agencies should have to treat the status quo as they would any other alternative they must address: by providing a reason to depart from it. To conclude otherwise would mean that agencies have an obligation to provide reasons for rejecting individual components of the status quo regime—as the Supreme Court reaffirmed in Regents, which postdates Fox Television—but not the status quo in toto. That makes no sense.

Admittedly, my position on agencies’ obligation with respect to the status quo lands somewhere between Justice Scalia’s broad pronouncements in the Fox Television majority opinion and the position taken in the principal dissent, authored by Justice Breyer, which stressed that an agency must explain “why it changed” its position.215Id. at 546–47 (Breyer, J., dissenting). But not only does the maximalist reading of Fox Television sit uneasily with Regents—it is also out of step with other cases postdating Fox Television. First, the Supreme Court in Encino Motorcars, LLC v. Navarro affirmed that “[a]gencies are free to change their existing policies” but must “provide a reasoned explanation for the change.”216Encino Motorcars, LLC v. Navarro, 579 U.S. 211, 221 (2016) (emphasis added). Second, the D.C. Circuit rejected the EPA’s ultrabroad reading of Fox Television in Physicians for Social Responsibility v. Wheeler.217Physicians for Soc. Resp. v. Wheeler, 956 F.3d 634, 646 (D.C. Cir. 2020). There, the EPA announced a new policy prohibiting persons receiving EPA grants from serving on EPA advisory committees.218Id. at 638. Even assuming that the EPA’s announcement was sufficient to indicate that it understood itself to be making a change and that it believed the new policy to be better than the old, the court found the EPA’s prohibition arbitrary and capricious because the EPA had said nothing at all about the merits of the policy it was abandoning.219See id. at 646–47. The court emphasized that “an agency changing its course must supply a reasoned analysis indicating that prior policies and standards are being deliberately changed, not casually ignored,”220Id. at 647 (quoting Lone Mountain Processing, Inc. v. Sec’y of Lab., 709 F.3d 1161, 1164 (D.C. Cir. 2013)). and that this analysis must address glaring inconsistencies between the agencies’ old policy and the new.221See id.

All of this is not to say that agencies must provide a painstaking justification for departing from the status quo, either when reversing the agency’s own policies or when writing on a blank slate. Again, all that is required of the agency is that it treat the status quo as it would any other relevant alternative, a burden that, as argued below,222See infra Part III. may be discharged in any number of ways. And it remains true that the agency “need not demonstrate to a court’s satisfaction” that the new policy is superior to what came before.223FCC v. Fox Television Stations, Inc., 556 U.S. 502, 515 (2009). The agency simply needs to provide a reason—consistent with standards of reasoned decisionmaking—for the rejection of the status-quo option. What’s more, in the vast majority of cases, the agency will have already given sufficient reasons for rejecting the status quo simply by focusing on the reasons for the new policy. Even in Fox Television, the agency had given reasons for rejecting the FCC’s prior, more permissive approach to fleeting expletives because those reasons were implicit in the affirmative justification for the new policy. Namely, the FCC found that such expletives, if allowed, represented “harmful ‘first blows’ to children.”224Id. at 518 (cleaned up). Thus, perhaps outside of some exotic scenarios, an agency that provides satisfactory reasons for its new policy will have satisfied its obligation to address the alternative of retaining the status quo.

Finally, there also may be circumstances where an agency must explain why it is retaining the status quo by giving reasons for why options other than the status quo were rejected, and where an agency’s failure to do so would properly be held arbitrary and capricious. Administrative law has no regularized way of requiring agencies to periodically consider acting in order to capture benefits currently left on the table. The limited opportunities parties have for pushing agencies to revisit the status quo is a big reason why administrative law is sometimes accused of embedding a bias toward keeping things as they are.225See, e.g., William W. Buzbee, Asymmetrical Regulation: Risk, Preemption, and the Floor/Ceiling Distinction, 82 N.Y.U. L. Rev. 1547, 1594–95 (2007). But there are some situations where agencies must consider options other than the status quo and provide reasons for ultimately sticking with it. Some statutes require agencies to periodically review requirements and determine whether they should be kept.226E.g., 47 U.S.C. § 548(c)(5); Cablevision Sys. Corp. v. FCC, 597 F.3d 1306 (D.C. Cir. 2010) (reviewing the FCC’s decision to retain the ban on certain kinds of exclusive contracts); see also Am. Paper Inst., Inc. v. EPA, 996 F.2d 346, 349 (D.C. Cir. 1993) (discussing state regulators’ obligation to revisit their current water quality standards at least once every three years). Agencies sometimes announce in advance that they will revisit certain rules on a set schedule.227E.g., Implementation of the Local Competition Provisions of the Telecommunications Act of 1996, Third Report and Order and Fourth Further Notice of Proposed Rulemaking, 15 FCC Rcd. 3696, 3766 ¶ 151 (1999) (“We expect to reexamine our national list of network elements that are subject to the unbundling obligations of the Act every three years.”). When agencies undertake such revisitation via notice-and-comment proceedings, courts have reviewed their decisions to retain or to change aspects of the regulatory status quo for conformance with the APA, including its reason-giving requirements.228E.g., U.S. Telecom Ass’n v. FCC, 359 F.3d 554, 592 (D.C. Cir. 2004).

An agency should also have to give reasons for why options other than the status quo were not pursued when a party has petitioned it to take a certain action. The APA requires that agencies “give an interested person the right to petition for the issuance, amendment, or repeal of a rule.”2295 U.S.C. § 553(e). In Massachusetts v. EPA, the Supreme Court clarified that, when undertaking notice-and-comment proceedings in response to such a petition, agencies must give valid reasons for rejecting the alternatives to the status quo identified by the petitioner.230Massachusetts v. EPA, 549 U.S. 497, 527, 532–35 (2007). Such an obligation provides a potentially pro-regulatory check on agencies’ tendency toward the status quo, but the petitioning mechanism does have important limitations. First, agencies often take a long time in responding to petitions, and the courts are notoriously unwilling to police agency action as unreasonably delayed.231Michael 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 (2011); see also Am. Anti-Vivisection Soc’y v. U.S. Dep’t of Agric., 946 F.3d 615, 621 (D.C. Cir. 2020) (finding plaintiffs had adequately alleged that agency had failed to promulgate required standards but remanding to district court to determine whether the eighteen-year delay was unreasonable). Moreover, agencies sometimes refuse to initiate notice-and-comment proceedings in response to a petition because of resource constraints, and an agency’s decision to do so is reviewed deferentially.232 1 Kristin E. Hickman & Richard J. Pierce, Jr., Administrative Law Treatise § 4.10 (6th ed. 2019) (discussing the different levels of review that apply when an agency has rejected a requested rule following notice and comment and when it has denied the petition to open a rulemaking altogether). Second, where the petition is construed as asking an agency to alter its enforcement priorities, judicial review may be altogether precluded under Heckler v. Chaney.233Heckler v. Chaney, 470 U.S. 821 (1985).

4. Alternatives Raised by the Agency, Agency Subunits, and Dissenting Commissioners

The final category of alternatives to which an agency presumptively must respond under my framework are those that were raised by the agency itself, or by subunits within the agency. This category is a cousin of the agency’s responsibility, rooted in State Farm, to consider retaining components of existing agency programs. But here, the agency itself has not endorsed the alternative definitively, as it does when it promulgates a regulatory program with multiple, independently justified components. Rather, such agency-articulated (but not endorsed) alternatives include alternatives raised or proposed in a Notice of Proposed Rulemaking (NPRM),234E.g., Dist. Hosp. Partners, L.P. v. Burwell, 786 F.3d 46, 58–59 (D.C. Cir. 2015) (faulting agency for unexplained departure from methodology proposed in NPRM); Int’l Ladies’ Garment Workers’ Union v. Donovan, 722 F.2d 795, 816 (D.C. Cir. 1983) (finding agency failed to respond to relevant alternatives in part because the agency’s explanation “does not provide the slightest indication that [it] gave any consideration to the alternatives raised in [its] original notice”). as well as alternatives identified in the explanation for the agency’s decision.235E.g., TikTok Inc. v. Trump, 507 F. Supp. 3d 92, 111 (D.D.C. 2020) (finding agency failed to consider alternative that was “so obvious that it appears in the first sentence of the ‘Recommendation’ section of the Secretary’s decisional Memorandum”). Also included are recommendations by lower-level agency units that are before the relevant decisionmaker at the time of her decision and therefore are part of the agency record. An example is the Census Bureau’s recommendation that Secretary Ross use administrative records, and not a citizenship question, to gather citizenship information.236See supra notes 61–64 and accompanying text. And in a similar vein, the D.C. Circuit has placed an obligation on multimember agency commissions to consider alternatives (at least nonfrivolous ones) proposed by dissenting commissioners.237E.g., Chamber of Com. v. SEC, 412 F.3d 133, 144–45 (D.C. Cir. 2005); Am. Gas Ass’n v. Fed. Energy Regul. Comm’n, 593 F.3d 14, 19–20 (D.C. Cir. 2010).

The obligation to consider agency-articulated alternatives is justified when examined in the context of agencies’ internal dynamics. Let’s lump together alternatives raised in an NPRM, those identified in the agency’s final decision, and those advocated by agency subunits (leaving aside alternatives proposed by dissenting commissioners). Agencies include, of course, different individuals performing different roles. Alternatives identified in a notice, in a final decision document, or by an agency subunit may be more likely to be those that career staff, representing the bulk of the agency’s body of experts, believe to be at least worthy of consideration.238The Census Case, where the relevant alternative had been developed by agency staff within the Census Bureau, provides one example. See supra notes 57–63 and accompanying text. The ultimate decision may, by contrast, be more likely to be driven by high-level agency staff, including political appointees more directly responsive to the president.239See, e.g., supra notes 57–63 and accompanying text. To the extent that the law requires considerations of agency-articulated alternatives, therefore, it allocates some amount of influence to lower-level agency staff in situations where there may be a divergence between the alternative chosen by the agency and the preferences of the career bureaucracy.240See generally Elizabeth Magill & Adrian Vermeule, Allocating Power Within Agencies, 120 Yale L.J. 1032 (2011) (examining administrative law doctrines that allocate power within agencies).

Even if one doesn’t think (as I don’t) that lower-level agency staff should always, or even usually, “win” in conflicts between them and political appointees, it still makes sense that agencies should bear an explanatory burden when it comes to alternatives articulated by agency staff. To understand the utility of this requirement, all one needs to accept is that it is valuable to have an agency go “on record” with its reasons when it disagrees with a proposal that its staff believes to be worthy of consideration. Nina Mendelson has documented the various benefits that may flow from requiring agencies to disclose oversight by political officials of their decisionmaking.241See Nina A. Mendelson, Disclosing “Political” Oversight of Agency Decision Making, 108 Mich. L. Rev. 1127 (2010). And Jon Michaels has argued that the career civil service provides a kind of “countermajoritarian counterweight” to political officials at the top of the agency, a role that promotes a less formal version of the separation of powers envisioned by the Framers and helps to ease lingering anxieties about the administrative state.242See Jon D. Michaels, Constitutional Coup: Privatization’s Threat to the American Republic 60–65, 70 (2017).

Requiring agencies to provide public reasons for rejecting internal proposals of agency staff may serve similar functions. Consider Secretary Ross’s decision to reject the Census Bureau’s administrative-records alternative.243See supra notes 61–63 and accompanying text. Having the reasons for that rejection on the table allows the public (and the courts) to assess the extent to which the agency was rejecting the staff-level alternative based on a disagreement about the underlying facts or based on differing value judgments.244See Mendelson, supra note 241, at 1165 (noting that disclosure can reveal “the extent to which executive supervision is directed to value-laden issues or, by contrast, the extent to which it amounts to second-guessing expert decisions made within the agencies”). To the extent that such disclosure reveals that the agency based its ultimate determination on considerations the statute places out of bounds, the agency’s decision would be vulnerable to challenge. But even if, as is often the case, the agency has broad discretion, disclosing the reasons for rejecting agency-articulated alternatives reveals how the agency evaluates tradeoffs among various considerations and provides important information about the factors that are driving the agency’s decisionmaking. And it requires political officials to take seriously the views of agency staff in anticipation of the prospect of judicial review.

Agencies with multimember heads should also generally have to respond to alternatives raised by dissenting commissioners. Such alternatives raise a similar set of considerations, but in a slightly different context. There again, requiring the majority to respond to alternatives proposed by dissenters may reveal how the agency is reconciling competing values advanced by different constituencies within the agency and therefore provide useful information to the public. To the extent that dissenting commissioners represent or provide voice to constituencies that have “lost” in the regulatory process, acknowledging those constituencies’ interests and explaining why they have been subordinated provides a kind of accounting that is said to justify reason-giving requirements in general.245See supra notes 116–121 and accompanying text. Sharon Jacobs has also argued that requiring a response to dissenting commissioners may be “deliberation-forcing.”246See Sharon B. Jacobs, Administrative Dissents, 59 Wm. & Mary L. Rev. 541, 593–94 (2017). If commissioners know that they have to respond to dissenters’ views, they may take those views more seriously and consider them more carefully as part of the predecision back-and-forth among colleagues, producing a dynamic that “might produce better-quality decisions.”247See id. at 594. Jacobs identifies other potential drawbacks to requiring a response to dissenting commissioners, some of which can be partly addressed through tweaks to the doctrine. First, regarding alternatives proposed by dissenting commissioners that are “ ‘frivolous,’ ‘out of bounds,’ or otherwise ‘unworthy of consideration,’ see id., I will argue that the “frivolity” of an alternative should be reconceptualized as whether it would be apparent that a certain alternative is a nonstarter given the agency’s own discussion of the features of the problem. Second, regarding requiring agencies to respond to alternatives proposed by dissenting commissioners produced very late in the process and when it is no longer possible to significantly revise the majority’s statement, see id. at 595–96, 595 n.264, I would simply relieve the agency of the obligation to respond to alternatives raised by dissenters where the agency did not have an opportunity to respond. Jacobs acknowledges that this deliberation-forcing rationale relies on a somewhat caricatured view of decisionmaking by multimember agencies: members typically work less closely with each other than judges on multimember courts do, and the actual writing of a response is likely to be delegated.248See id. at 595–96. But even if the action occurs at a level below the members themselves, having to respond to the views of dissenters (who themselves employ staffs of experts) may plausibly improve the quality of agency decisionmaking on average.

D. Truly “Obvious” Alternatives?

This Section tackles whether agencies should ever have to respond to an intuitively plausible alternative that neither falls within one of the presumptions described in Section II.B nor was raised by a party. That question may arise in two contexts. First, it may come up when an agency has taken outside input, but no party has raised the alternative in question. Second, it might arise where the agency has not allowed for outside input and the alternative in question is not among those covered by the presumptions described in Section II.B.

In the first context, where outside parties have been afforded an opportunity to be heard, the courts have generally resisted requiring agencies to respond to alternatives not raised to the agency itself. Thus, where the agency has engaged in notice-and-comment processes, failing to raise an alternative in comments effectively precludes the raising of that alternative in future litigation.249See Nat’l Shooting Sports Found., Inc. v. Jones, 716 F.3d 200, 215–16 (D.C. Cir. 2013); Oakbrook Land Holdings, LLC v. Comm’r, 28 F.4th 700, 721–22 (6th Cir. 2022). There is also precedent for relieving agencies of the obligation to respond to alternatives not raised in an adjudication where the party involved had a chance to do so.250See Clark-Cowlitz Joint Operating Agency v. Fed. Energy Regul. Comm’n, 826 F.2d 1074, 1085 n.11 (D.C. Cir. 1987) (stating that an “alternative was not so ‘obvious’ as to occur to” the party, which had failed to raise it to the agency); see also Ramirez v. U.S. Immigr. & Customs Enf’t, 471 F. Supp. 3d 88, 179 (D.D.C. 2020) (recognizing an “obligation on agencies to consider ‘facially reasonable alternatives’ ” that were “presented by a party to an adjudication” (quoting Laclede Gas Co. v. Fed. Energy Regul. Comm’n, 873 F.2d 1494, 1498 (D.C. Cir. 1989))).

Such results make sense. Courts are in a poor position to judge alternatives that have been raised for the first time in litigation and that have not been supported by evidence at the agency level. Barring litigating parties from raising alternatives not presented to the agency is also in line with the general rules of administrative issue exhaustion and encourages parties to raise alternatives to the agency at a time at which the agency can respond, including potentially by adopting the alternative in question.251On issue exhaustion, see generally Administrative Conference of the United States, 80 Fed. Reg. 60611–13 (Oct. 7, 2015). Issue exhaustion requirements, at least when they have not been derived from statutory text, have their critics. See Ronald M. Levin, Making Sense of Issue Exhaustion in Rulemaking, 70 Admin. L. Rev. 177, 180–85 (2018) (surveying the debate over the legitimacy of judge-made issue exhaustion rules and ultimately concluding that such rules are justifiable). But the legitimacy concerns about issue exhaustion rules in general do not necessarily attach when a court uses whether a party has raised an alternative to assess whether the agency was under an obligation to respond to it. As long as the basic obligation is accepted, courts need some way to delimit the range of obvious/reasonable alternatives to which an agency must respond. Using whether a party has raised a given alternative seems just as legitimate as any other criterion. Allowing consideration of such alternatives, by contrast, would allow parties to sandbag agencies by withholding alternatives from the agency, thus making it more likely that the agency will not respond, and by subsequently using those same alternatives to invalidate agency decisions in court.

A harder case is when the agency has not allowed for public participation and a court is later confronted with an alternative that does not fit within one of the presumptions and that was not addressed by the agency.252See supra note 15 and accompanying text. One plausible view would be that in such circumstances the courts should labor under a special obligation to review whether the agency has carefully considered its options, precisely because the agency’s decision has not otherwise been publicly vetted. That is, in such situations, courts should examine an alternative proposed by the parties in litigation and require agencies to respond to those alternatives that reach some requisite level of plausibility (or “obviousness”) as a policy matter, such that it can be said that the agency was at fault for not responding.

Despite the initial attractiveness of that position, I want to make a tentative case for an opposing view. Where the relevant agency decision has not followed a process allowing for outside input, courts should rarely, if ever, require agencies to consider alternatives not falling within one of the presumptions. After defending this presumptions-or-bust approach, I’ll then discuss potential alternatives to it for those who may not want to follow me over the cliff.

The tentative case begins like this: under the presumptions, an agency will always have to explain why it believes its preferred approach is superior to the status quo.253In addition, the agency will have to consider whether the statute in question requires the agency to evaluate certain other alternatives and address any other alternatives falling into one of the other presumptions. Here, I focus on an agency’s obligation with respect to the status quo because a consideration of the status quo alternative will be required as part of any decision. That will entail identifying the criteria by which alternatives are to be judged and explaining why, according to those criteria, the agency’s chosen option represents at least an incremental improvement.254See infra Section III.A. As part of its explanation, the agency will also have to consider any reliance interests at stake, because the Supreme Court has made clear that such interests are always relevant criteria when an agency seeks to disturb the status quo.255DHS v. Regents of Univ. of Cal., 140 S. Ct. 1891, 1915 (2020). In the words of Regents, agencies are “required to assess whether there [are] reliance interests [at stake], determine whether they [are] significant, and weigh any such interests against competing policy concerns.”256Id.

As I read Regents, one way that an agency may adequately respond to concerns about reliance interests is by assessing alternative approaches that may disturb those interests to a lesser degree.257See id. at 1914–15. If the agency explains why, nevertheless, its chosen option is still the superior one, it has satisfied its burden. This means that, whenever reliance interests are at stake, the agency will have an incentive to attend to certain alternatives, not because consideration of those specific alternatives are required per se, but rather to fulfill its related but distinct obligation to consider reliance interests. Now, an agency might validly respond to the presence of reliance interests in other ways as well, such as by explaining why the interests at stake are not all that strong and the benefits of the agency’s preferred approach make the decision worth it. But either way, the agency will need to have given a reason for its departure from the status quo—whether the status quo be nonregulation or a preexisting regulatory regime—that takes such interests into account.

The question then becomes: when, if ever, should a court disturb an agency’s decision where the agency has (a) provided reasons for why its chosen course is superior to the status quo and (b) has adequately considered the reliance interests at stake, but (c) did not respond to an alternative that, as a matter of policy analysis, appears attractive or even “obvious”?

To my mind, the answer is never. Consider the kinds of agency decisions that are most likely to be made without public input. Those decisions will generally fall into two categories: rules where the agency has invoked one of the exceptions to notice and comment, and informal adjudications where the agency has not provided an opportunity for public participation. Rules excepted from notice and comment as “general statements of policy”2585 U.S.C. § 553 (2023). are rarely subject to pre-enforcement judicial review anyway.259E.g., Nat’l Mining Ass’n v. McCarthy, 758 F.3d 243, 251–52 (D.C. Cir. 2014). And when an agency follows a guidance document in a subsequent proceeding, it may not treat the guidance as the legal basis for acting.260Id. at 252. Importantly, in that subsequent proceeding, the agency must be prepared to defend the policies reflected in the guidance as if the guidance had never been promulgated.261Pac. Gas & Elec. Co. v. Fed. Power Comm’n, 506 F.2d 33, 38 (D.C. Cir. 1974). That justification presumably must include a response to alternatives raised by the party to the action, subject to the limitations described above. The upshot is that, prior to the guidance being applied in a manner that has formal legal significance, the agency will shoulder the burden of responding to alternatives offered by parties being subjected to the guidance and any covered by the presumptions.262The considerations attaching to rules excepted as “interpretative rules” are similar, though such rules are somewhat more likely to be subject to pre-enforcement judicial review. See Nat’l Mining Ass’n, 758 F.3d at 251–52. Interpretive rules typically must be backed by a preexisting statute or regulation that is being interpreted. See Am. Mining Cong. v. Mine Safety & Health Admin., 995 F.2d 1106, 1107 (D.C. Cir. 1993); Chrysler Corp. v. Brown, 441 U.S. 281, 315 (1979). Interpretive rules themselves lack the force of law. See In re Fed. Bureau of Prisons’ Execution Protocol Cases, 955 F.3d 106, 132 (D.C. Cir. 2020). If the interpretive rule is backed by a regulation, that regulation must itself have satisfied requirements of reasoned decisionmaking, including the obligation to respond to alternatives. Motor Vehicle Mfrs. Ass’n of the U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 48–51 (1983). In theory, if the interpretive rule departs somewhat from the regulation being interpreted, it might add something to require agencies to respond to alternatives to its chosen interpretation. But requiring agencies to respond to such alternatives would likely have little practical effect. I have not seen a case in such a posture.

Rules excepted from notice and comment for “good cause” present a slightly different context, but even in these cases, there are still plausible reasons supporting a minimal obligation to consider alternatives. Such rules, unlike guidance documents and interpretive rules, have legal force and effect and are binding.263Osamudia R. James, Breaking Free of Chevron’s Constraints: Zuni Public School District No. 89 v. U.S. Department of Education, 56 Kan. L. Rev. 147, 166 (2007). Even so, however, there are good reasons for courts to hesitate before requiring agencies to respond to alternatives raised for the first time during litigation. The most important category of “good cause” rules are rules justified by emergency conditions, where a protracted notice-and-comment process would be “impracticable” or “contrary to the public interest.”2645 U.S.C. § 553 (2023). And in that context, we may have less reason to worry about unexplored alternatives falling outside the presumptions. First, courts have consistently found that such rules are only justified as a response to true emergency situations,265Sorenson Commc’ns Inc. v. FCC, 755 F.3d 702, 706 (D.C. Cir. 2014) and that the exception should be “narrowly construed and only reluctantly countenanced.”266Am. Fed’n of Gov’t Emps. v. Block, 655 F.2d 1153, 1156 (D.C. Cir. 1981) (quoting N.J. Dep’t of Env’t Prot. v. EPA, 626 F.2d 1038, 1045 (D.C. Cir. 1980)). Thus, prior to considering whether the rule itself is arbitrary and capricious, the court will have already determined that the agency’s claim to be acting under emergency conditions survives review. And the same conditions justifying an agency acting without going through notice and comment may also justify demanding a less comprehensive analysis of various options prior to proceeding. As just explained, under the presumptions, the agency will always have to give reasons for departing from the status quo. And in an emergency, that may be good enough. Second, courts have expressed skepticism when an agency promulgates an “emergency” rule without simultaneously providing a post-promulgation opportunity for notice and comment prior to the rule being treated as permanent.267See, e.g., Mid-Tex Elec. Coop., Inc. v. Fed. Energy Regul. Comm’n, 822 F.2d 1123, 1132 (D.C. Cir. 1987). Thus, prior to promulgating a permanent rule, agencies will typically have to respond to public comments, including those raising alternatives, subject to the limitations above. Of course, in the period between promulgation of the original rule and the rule becoming permanent, persons may have been subject to a less-than-perfect regulation. But under emergency conditions, that’s a price likely worth paying.

Perhaps the most difficult cases are those involving informal adjudications where the agency has not provided an opportunity for outside comment. Informal adjudications represent a residual category that includes every agency action that is neither a rule nor required to be undertaken through formal adjudication.268Izaak Walton League of Am. v. Marsh, 655 F.2d 346, 361 n.37 (D.C. Cir. 1981). It therefore includes much more than what might typically be thought of as adjudication in the judicial setting. Requests for licenses or for the dispersal of government benefits, for example, are often processed through informal adjudication.269See Evan D. Bernick, Is Judicial Deference to Agency Fact-Finding Unlawful?, 16 Geo J.L. & Pub. Pol’y 27, 40 (2018). Such things, of course, are often quite important for the parties involved. But when such adjudications are challenged, there are still several reasons to not subject agencies to an obligation to respond to alternatives raised for the first time during litigation. First, although the APA does not require outside input prior to an informal adjudication,270Emily S. Bremer, The Rediscovered Stages of Agency Adjudication, 99 Wash. U. L. Rev. 377, 388 (2021). agencies often allow for such input under their own rules, and they may be required to allow for such input by particular, non-APA statutes or by the Constitution.271On the Constitution, see Goldberg v. Kelly, 397 U.S. 254, 254–55 (1970). On the extensive procedures that may apply even outside the “formal adjudication” context, see Christopher J. Walker & Melissa F. Wasserman, The New World of Agency Adjudication, 107 Calif. L. Rev. 141, 143–46, 153 (2019). Therefore, the number of agency decisions in which no outside input is received, including from the directly affected parties, is smaller than might first appear. Second, policies rendered in the course of informal adjudication are more easily changed than those embodied in regulations. At most, the agency must treat statements in its prior adjudications as establishing precedent, which can be overruled if done forthrightly.272FCC v. Fox Television Stations, Inc., 556 U.S. 502, 514–16 (2009). Thus, from society’s perspective, the error costs associated with less-than-perfect agency decisionmaking may well be lower when it comes to informal adjudication.

These are all reasons why it may be “good enough” for an agency to consider only the status quo alternative in the contexts in which an agency is most likely to have acted without allowing for public participation. Still, one might ask, why not strive for better? Surely there are some alternatives that, as a policy matter, are sufficiently plausible, obvious, or whatever else, that we would expect a good agency to address them.

The problem stems from the costs associated with allowing courts to conduct a policy analysis of alternatives raised for the first time during litigation and that were not evaluated in the administrative record. Those alternatives are, by their nature, ones identified by the challengers to the agency action in question. Through briefing, those challengers would have the opportunity to argue that their favored alternatives were sufficiently obvious that the agency should have considered them. Agency lawyers, however, may have limited ability to respond to such alternatives under Chenery I, which prevents agencies from supplying post hoc rationales for their decisions.273SEC v. Chenery Corp. (Chenery I), 318 U.S. 80, 92–94 (1943). In State Farm itself, the Supreme Court broadly suggested that Chenery I may preclude an agency’s lawyers from responding to alternatives not previously considered in litigation.274Motor Vehicle Mfrs. Ass’n of the U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 49–50 (1983). Thus, under a strict application of Chenery I, a court may be left with a one-sided view of the plausibility of a given alternative and a lack of material in the administrative record by which to evaluate the alternative. The risks created by that dynamic are at least twofold. First, courts may needlessly invalidate agency actions when an alternative appears facially plausible but would in fact be easily rejected. Second, the prospect that a judge will later find a nonviable alternative facially plausible may lead an agency to “over-respond” to alternatives, compounding some of the ills associated with hard look review, including in emergency situations where the need for swift action may be great.

I realize that readers (and courts) may not be fully willing to accept the above and may want to preserve the ability of courts to intervene and require an agency to respond to obvious alternatives not covered by the presumptions and in cases where the agency has not allowed for outside input. And indeed, there is at least one context in which I am drawn to that view—namely, where the agency has undertaken an informal adjudication without outside input and where, although the principles adopted by the agency in that adjudication may be easy to depart from in future adjudications, the agency decision at hand has permanent societal consequences. In Overton Park, for example, once the highway was approved and built, there was presumably no going back. And in the Census Case, the citizenship question either would or would not appear. In those situations, although the agencies may be free to shift their normative commitments when the next highway is proposed or census is designed, the particular decision at issue cannot easily be undone.

The question, then, is whether a framework might be devised to help courts identify alternatives that are sufficiently plausible as a policy matter that a response is required. I’ll discuss two potential frameworks, and then suggest a reform that should be adopted if the policy attractiveness of a given alternative is ever to factor in to whether a given alternative should have been addressed.

First, courts might borrow from fields of law such as products liability and hold agencies accountable when they fail to address alternatives that were “knowable” to an expert in the field.275See, e.g., Alani Golanski, Paradigm Shifts in Products Liability and Negligence, 71 U. Pitt. L. Rev. 673, 682 (2010). One problem with this inquiry, however, is that it sits uneasily with the broader structure of administrative law. Sellers of products are under a general duty to “stay abreast of scientific and technical knowledge and discoveries relevant to their products, and are thereby presumed to be aware of any risks that are knowable by any expert in that field.”276Id. (citing George v. Celotex, 914 F.2d 26, 28 (2d Cir. 1990); Borel v. Fibreboard Paper Prods. Corp., 93 F.2d 1076, 1089–90 (5th Cir. 1973)). Agencies are allowed to be more passive. Although the Supreme Court has sometimes sent different signals, it’s more consistently been of the view that agencies shoulder no obligation to generate their own studies as a general matter.277See FCC v. Prometheus Radio Project, 141 S. Ct. 1150, 1160 (2021). To the extent that agencies have an obligation to “stay abreast” of developments in their fields, that obligation is largely limited to a duty to consider data, proposals, and evidence brought to them by outside parties.278See id.

Moreover, agencies are judged based on a cold record by generalist judges.279Paul M. Bator, The Constitution as Architecture: Legislative and Administrative Courts Under Article III, 65 Ind. L.J. 233, 261 (1989). The only documents that courts typically consider are those that constitute the administrative record.280See Dep’t of Com. v. New York (Census Case), 139 S. Ct. 2551, 2573–74 (2019). There is therefore no opportunity to develop the kinds of evidence (involving expert testimony and the like) that would help determine whether a given alternative was knowable to someone in the field.

Second, another way to address the issue (which may have more promise) is to adopt something like a “deliberate indifference” framework. In some situations, scrutiny of the administrative record itself might allow courts to infer not simply that a given alternative was “knowable” but that it was, in effect, known to the agency and deliberately ignored. To adapt loosely from one of the disputes in Fox Television, let’s say that there is ample record evidence that if the FCC adopted a certain policy, it would have a devastating effect on small, local, and predominantly rural broadcasters.281See FCC v. Fox Television Stations, Inc., 556 U.S. 502, 557–61 (2009) (Breyer, J., dissenting). Let’s further suppose that the FCC had, in other contexts, adopted carve-outs in order to protect such broadcasters. Even if no party has specifically asked for such a carve-out because no public participation rights had been afforded, a court might conclude that the agency was deliberately indifferent to that alternative. That conclusion would perhaps only differ in degree from finding that a given alternative was “knowable,” but it would more squarely focus the question on what reasonable inferences could be drawn about the agency’s level of actual knowledge based on the administrative record as read against past agency actions.

Finally, under any framework used to assess whether an alternative is sufficiently attractive as a policy matter to require a response, I would suggest one reform to address the possible downside consequences. Namely, courts should clarify that, with regard to alternatives raised for the first time in the litigation process and not covered by a presumption, an agency’s lawyers are not barred by Chenery I from providing reasons why a given alternative is sufficiently implausible as a policy matter that responding to it was not required. Such a result would not actually be inconsistent with Chenery I. Whether an alternative is sufficiently plausible or obvious enough to require a response is a litigation matter that goes to what had to be part of the agency’s affirmative rationale in the first place. It is not a required aspect of the agency’s affirmative rationale that is being supplied ex post, which is the domain with which Chenery I was concerned. However, because the reasons for why an alternative is implausible will largely overlap with the reasons the agency might have given in rejecting it, an overbroad application of Chenery I may lead courts to close their ears to anything that sounds like a new response to an alternative. In order to reduce the possibility of needless remands for agencies to consider nonstarter alternatives, courts should welcome, and not prohibit, arguments from agency lawyers that proposed alternatives are implausible.

III. What Counts as a Response?

This Part turns to the second question raised by agencies’ obligation to respond to alternatives: when an agency must respond to a certain alternative, what counts as a response? Section III.A describes the minimum elements of a response that an agency must provide and introduces the idea that an agency may respond to an alternative without mentioning it. Section III.B introduces variations on the forms that a valid response might take and develops the insight that an agency may rely on strategies of decisionmaking that, if disclosed, provide an adequate response to an alternative without requiring the agency to fully evaluate that alternative on the merits. Section III.C considers situations where, even though the agency has failed to respond to an alternative, its failure may be excused as harmless.

A. The Minimum Requirements

Again, what exactly is required of an agency when it must respond to a given alternative has not been well theorized. The basic obligation is simply to provide a reason why the alternative was rejected in favor of the agency’s preferred option. As one district court put it, when an agency is under an obligation to respond to a given alternative, the agency fulfills its responsibility if it “respond[s] in a way that allows the Court to see why it did not opt for” that alternative.282Am. Great Lakes Ports Ass’n v. Zukunft, 296 F. Supp. 3d 27, 43–44 (D.D.C. 2017). The agency’s response may not be “conclusory.”283Int’l Ladies’ Garment Workers’ Union v. Donovan, 722 F.2d 795, 817–18 (D.C. Cir. 1983). The agency may not, in other words, identify an alternative and then assert that, in the agency’s judgment, its option is the better one. But the response can be pretty short. Agencies are often able to provide adequate responses to alternatives in just a paragraph or two.284See, e.g., Globalstar, Inc. v. FCC, 564 F.3d 476, 488 (D.C. Cir. 2009) (citing Second Order on Reconsideration, 22 FCC Rcd. 19733, 19740–41 (2007)).

That’s all pretty general, and to add some specificity I propose that an agency’s response should include two things in order to survive judicial review. First, the agency should announce the criteria that it used to evaluate the alternative. Second, it should explain why, according to the supplied criteria and based on the administrative record, it concluded that the alternative should be rejected.

Importantly, including those two things in the agency’s response allows a reviewing court to fulfill its responsibilities under the APA. Supplying the criteria used by the agency allows the court to understand what factors were relevant to the agency’s decisionmaking and judge whether those factors are among those that the statute places within bounds.285See Massachusetts v. EPA, 549 U.S. 497, 533 (2007) (setting aside agency’s decision not to regulate certain greenhouse gas emissions because the agency had made that decision based on reasons irrelevant under the statute). To return to an example from above, if an agency states that it has rejected an alternative because it believes it to be too costly, the court can set aside the agency’s decision if it determines that “cost” is not among the criteria available to the agency in selecting between policies.286See Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 486 (2001) (“The EPA may not consider implementation costs in setting primary and secondary NAAQS under § 109(b) of the CAA.”). If an agency need not disclose the criteria it used in evaluating policy options, that review function would be impossible. Under the second required element of a response, the agency must explain how the criteria led it to the decision it made. This will involve disclosing the basic facts or judgments underlying the agency’s determination. Doing so thus allows the court to determine whether the facts the agency relied on are supported by “substantial evidence,”287See 5 U.S.C. § 706(2)(E); Allentown Mack Sales & Serv., Inc. v. NLRB, 522 U.S. 359, 361 (1998). Technically, agency findings of fact made in the course of informal proceedings are subject to arbitrary-and-capricious review, but the substance of the standard is the same as that applied under the substantial evidence test. See Miley v. Lew, 42 F. Supp. 3d 165, 170 n.2 (D.D.C. 2014). as well as whether there is a “rational connection between the facts found and the choice made.”288See Motor Vehicle Mfrs. Ass’n of the U.S., Inc. 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)).

So far, I’ve been assuming that the agency has specifically referenced a given alternative and addressed it explicitly, but I want to now suggest that a specific reference isn’t required and an agency’s response may be implicit.289See FBME Bank Ltd. v. Lew, 125 F. Supp. 3d 109, 124–25 (D.D.C. 2015). The case law often gives the impression that to respond to an alternative, an agency must actually mention it.290Walter O. Boswell Mem’l Hosp. v. Heckler, 749 F.2d 788, 797 (D.C. Cir. 1984). That view is understandable but mistaken. If a court, having read the agency’s decision in conjunction with the record, would understand the criteria used by the agency to evaluate a given alternative and why that criteria led the agency to exclude it, that should be enough to satisfy the agency’s obligation. Such an exercise may involve applying a dose of common sense. But the alternative would be needless remands to agencies for them to more explicitly state grounds for decisionmaking that should be apparent given the agency’s description of the problem, its stated goals, and the affirmative justification offered by the agency for the option it chose.

One situation in which an agency’s response to an alternative should be considered implicitly supplied is when the agency’s stated understanding of its goals and objectives naturally takes the alternative off the table. Then-Judge Clarence Thomas articulated such an understanding of agencies’ obligation to respond to alternatives when he was sitting on the D.C. Circuit in Citizens Against Burlington, Inc. v. Busey.291See Citizens Against Burlington, Inc. v. Busey, 938 F.2d 190 (D.C. Cir. 1991). The case involved the FAA’s obligation to respond to alternatives under NEPA, which explicitly requires agencies to consider alternatives when taking certain actions that affect the environment.292Id.; 42 U.S.C. § 4332. In the decision under review, the FAA had approved the city of Toledo’s plan to expand one of its airports.293Citizens Against Burlington, 938 F.2d at 191. Environmental groups challenged the approval, claiming that the FAA had not provided a “detailed statement” on alternatives, as required under NEPA for certain federal actions.294Id. at 198–200. Thomas began with a general observation on the problem of requiring agencies to respond to alternatives. Deploying a hypothetical, he stated: “Suppose, for example, that a utility applies for permission to build a nuclear reactor in Vernon, Vermont. Free-floating ‘alternatives’ to the proposal for federal action might conceivably include everything from licensing a reactor in Pecos, Texas, to promoting imports of hydropower from Quebec.”295Id. at 195. In his view, therefore, some sense of feasibility or reasonability needed to be brought to bear in order to limit the sheer number of alternatives that required specific discussion. On this point, he was in accord with NEPA’s implementing regulations, which require that agencies evaluate and provide a detailed discussion of “reasonable” alternatives.29640 C.F.R. § 1502.14 (2021).

Thomas’s solution was to focus on the agency’s stated purposes and on whether the proposed alternatives could be viewed as reasonably accomplishing those goals.297See Citizens Against Burlington, 938 F.2d at 198. If not, the agency is not obligated to, in NEPA’s words, provide a “detailed statement” regarding them.29842 U.S.C. § 4332(2)(C) (1988). For example, in Citizens Against Burlington itself, the agency’s stated goals included supporting the city of Toledo’s economy.299Id. Since that was a legitimate goal, alternatives that did not further it—such as building airports outside of the Toledo area—were not required to be specifically addressed.300Id.; see also City of New York v. U.S. Dep’t of Transp., 715 F.2d 732, 743 (2d Cir. 1983) (“The scope of alternatives to be considered is a function of how narrowly or broadly one views the objective of an agency’s proposed action.”). Thomas’s approach has been influential and has been extended outside the NEPA context. For example, in a recent case involving the Trump Administration’s banning of TikTok from the United States, the court broadly stated that “[a]n agency is only required to consider ‘reasonable’ alternatives that ‘will bring about the ends of the federal action.’ 301TikTok Inc. v. Trump, 507 F. Supp. 3d 92, 111 (D.D.C. 2020).

Although the Citizens Against Burlington approach provides a sensible limit on agencies’ obligation to provide a full discussion of alternatives, I would conceive of that limitation slightly differently than the courts have, and in a way that would bring greater clarity to the inquiry. Courts have proceeded on the assumption that agencies need not respond to unreasonable alternatives, in the sense that they would not further the agency’s stated aims. I would put it as follows: where an agency has stated its goals and a court examining the agency decision and the record would understand why the agency excluded the alternative in question, the agency has implicitly responded to that alternative. Remember that the obligation to respond to alternatives is fundamentally about disclosing the reasons for why the agency chose the course it did.302See supra Section II.A. Where it should be clear, applying a modicum of common sense, that a given alternative would not further the agency’s purposes, the public or courts have been supplied such a reason.

To see the concept in practice, return to the CDC’s COVID-19-era decision requiring masking on public transportation—a decision that did not explicitly consider alternatives, such as the destruction of animals or articles, although the statute made those alternatives relevant.303See supra notes 166–167 and accompanying text. A person who reads the agency decision in question and the record materials described therein would not be left scratching their head wondering why the agency excluded destruction of animals or articles from the ambit of reasonable policy choices. Nor would someone wonder why, if the agency wished to stimulate the economy of Toledo, it eliminated projects in California. In each case, a reason has been supplied for why the agency acted as it did and not in another way.

Nor should this way of viewing an agency’s obligation to respond to alternatives run afoul of Chenery I’s prohibition on upholding agency action on a basis other than that supplied by the agency at the time of its decision. The Chenery I principle is not a straightjacket that disallows courts from drawing reasonable conclusions about the reasons for an agency’s decision—as long as those conclusions are supported by the agency’s explanation for the decision itself and not subsequent litigation documents. Rather, courts have consistently stated that “[a]s long as ‘the agency’s path may reasonably be discerned,’ we will uphold the decision even if it is ‘of less than ideal clarity.’ 304Casino Airlines, Inc. v. Nat’l Transp. Safety Bd., 439 F.3d 715, 717 (D.C. Cir. 2006) (quoting Bowman Transp., Inc. v. Ark.-Best Freight Sys., Inc., 419 U.S. 281, 286 (1974)). What’s important is that the court “can discern a reasoned path from the facts and considerations before [the agency] to the decision it reached.”305Neighborhood TV Co. v. FCC, 742 F.2d 629, 639 (D.C. Cir. 1984) (cleaned up). And the basis for the agency’s decision “must be clear enough to permit effective judicial review.”306Int’l Longshoremen’s Ass’n v. Nat’l Mediation Bd., 870 F.2d 733, 735 (D.C. Cir. 1989). Where consideration of the agency’s decision alongside the record (i.e., the “facts and considerations before the agency” at the time it made its decision) would allow a reasonable person to understand why the agency chose the course it did, those requirements are satisfied.

B. Global Reasons and Strategies of Decisionmaking

This Section further unpacks variations on what counts as a valid response to a given alternative. It develops two related points. First, agencies should be able to give “global” reasons for their actions that may operate to eliminate multiple alternatives without having to speak to each individually. Second, such global reasons may include adopting strategies of decisionmaking that, by their nature, reveal why the agency has not fully considered and therefore not adopted a given alternative. However, this only works if the agency gives a reason for pursuing the decisionmaking strategy in question and the chosen decisionmaking strategy comports with the agency’s statutory mandate.

In a number of cases, agencies have satisfied their obligation to respond to alternatives by providing explanations for their actions that reveal why they did not pursue certain categories of alternatives, without being required to speak to each and every alternative falling within that category. In my view, that makes good sense. Consider the example of a person who believes they are under a moral obligation not to lie. Having disclosed that belief, the person needn’t provide further detail on why they chose not to tell this or that falsehood in order for others to understand why they chose to tell what they believed to be the truth. One example in the agency context is found in National Ass’n of Manufacturers v. SEC.307Nat’l Ass’n of Mfrs. v. SEC, 956 F. Supp. 2d 43 (D.D.C. 2013), aff’d in part, rev’d in part and remanded, 748 F.3d 359 (D.C. Cir. 2014), adhered to on reh’g, 800 F.3d 518 (D.C. Cir. 2015). That case involved an SEC rule that would have required companies to disclose their use of certain “conflict minerals.”308Id. at 46. In the notice-and-comment proceedings preceding the rule’s promulgation, commenters suggested that the agency adopt various and varied proposals aimed at exempting companies whose use of such minerals could be considered de minimis.309Id. at 51. In its subsequent decision, the SEC explained that it believed any de minimis exception would have the potential to swallow the rule because conflict minerals are often used in small amounts.310See id. at 64, 65–66. Given the provided explanation, the court found that the agency was under no further obligation to explicitly consider every variation of de minimis exception offered to it: “[G]iven its ‘broader conclusion’ that conflict minerals are often used in minute amounts, the SEC believed that any type of categorical de minimis exception had the potential to swallow the rule and would be inappropriate. This analysis was sufficient to satisfy the Commission’s obligations under the APA.”311Id. at 66. The court in National Ass’n of Manufacturers stated that the agency need not consider or respond to certain alternatives given its categorical statements. Id. at 66. As above, I would reformulate that conclusion and say that when an agency has provided a categorical reason that excludes a number of given alternatives at once, the agency has responded to those alternatives, albeit implicitly, because the agency has adequately disclosed why it has not chosen any one of them. Other cases are in accord.312See Globalstar, Inc. v. FCC, 564 F.3d 476, 488 (D.C. Cir. 2009) (having concluded that spectrum sharing was infeasible, agency was not under a further obligation to analyze a specific alternative involving such sharing); Allied Loc. & Reg’l Mfrs. Caucus v. EPA, 215 F.3d 61, 80 (D.C. Cir. 2000) (agency’s explanation for rejecting one “labeling” alternative disclosed reasons why it rejected others).

Somewhat more broadly, agencies may permissibly adopt strategies of decisionmaking—tied not only to the substantive goals associated with the statute in question but to broader considerations such as administrability and the need to make decisions under conditions of uncertainty—and those strategies may themselves provide a reason why certain alternatives were not pursued. Jacob Gersen and Adrian Vermeule have explored one such strategy: satisficing.313See Gersen & Vermeule, supra note 5, at 1389–90. See generally Herbert A. Simon, A Behavioral Model of Rational Choice, 69 Q.J. Econ. 99 (1955). A decisionmaker acting as a satisficer does not search exhaustively for the best possible option among all options theoretically available.314See Gersen & Vermeule, supra note 5, at 1389. Rather, the satisficer chooses an aspiration level, a “minimum threshold[] that must be reached to be satisfied with a decision.”315Daniel Stevens, Satisficing in Political Decision Making, Oxford Rsch. Encyc. of Pol. (2019), []. They then consider alternatives until they discover one that satisfies the aspiration level.316Id. Having found an option that is “good enough,” the decisionmaker ceases deliberation and abandons further search for the option that is truly best. Satisficing is a particularly attractive decisionmaking strategy when the costs and benefits of given policies are unknown ex ante and an exhaustive process of information acquisition would be too costly to be worth it.317See Gersen & Vermeule, supra note 5, at 1389–90. Policymaking, including at the agency level, often occurs within such conditions.318See id. at 1390–91.

Choices between other decisionmaking strategies may also present themselves. Satisficing is related to a distinction that Colin Diver has made between two policymaking paradigms: comprehensive rationality and incrementalism.319See Colin S. Diver, Policymaking Paradigms in Administrative Law, 95 Harv. L. Rev. 393, 396–400 (1981). Comprehensive rationality involves specifying a goal, identifying all possible methods of achieving that goal, evaluating each identified method, and selecting that which will most fully achieve the decisionmaker’s goal.320Id. at 396. Incrementalism involves evaluating a limited number of choices that tend not to be very different than the status quo.321Id. at 399. It is therefore a kind of satisficing strategy used to delimit the range of alternatives an agency considers. It is also designed to be dynamic and decentralized: “Policymaking becomes a series of small adjustments and avowedly temporary ‘fixes’ ” over time.322Id. Finally, incrementalist decisionmaking strategies tend to be decentralized, with actors at various levels of the hierarchy contributing to the development of policy.323See id. Diver associates incrementalism with agency “policymaking by adjudication” and characterizes agency decisionmaking through the mid-1960s as largely dominated by incrementalism, with comprehensive rationality becoming the more popular paradigm in the 1970s.324See id. at 401–13.

For present purposes, what is important is that if an agency announces it is pursuing a certain decisionmaking strategy, that itself may provide a reason why certain alternatives were not fully considered and thus may properly satisfy the agency’s obligation to respond to such alternatives. An agency engaged in satisficing, for example, has given a reason for rejecting every alternative except the one chosen. Each alternative was either evaluated and did not meet the aspiration level or was not evaluated prior to arriving at a solution that did meet the aspiration level and was therefore excluded under the decisionmaking strategy employed. That an agency’s obligation is satisfied in such circumstances is consistent with the D.C. Circuit’s occasional refrain that an agency must either expressly consider reasonable alternatives “or give some reason . . . for declining to do so.”325Laclede Gas Co. v. Fed. Energy Regul. Comm’n, 873 F.2d 1494, 1498 (D.C. Cir. 1989) (emphasis in original); see also Spirit Airlines, Inc. v. U.S. Dep’t of Transp., 997 F.3d 1247, 1255 (D.C. Cir. 2021); Chamber of Com. v. SEC, 412 F.3d 133, 144–45 (D.C. Cir. 2005).

It’s also in line with broader administrative law principles. Agencies that choose to satisfice or otherwise not to aspire to comprehensive rationality have made a policy choice—namely, that the benefits of a limitless search for alternatives are not justified by its costs.326See Diver, supra note 319, at 433 (arguing that the choice between policymaking paradigms should be for the agency to make because the considerations involved often present “close calls of the sort that agencies are generally best equipped to make”). Often that choice will be informed by the agency’s understanding of the limits of its staff’s current knowledge and the resources that can be devoted to a particular problem given other demands facing the agency. Courts are characteristically deferential when such considerations are at play, and they are right to be.327Cf. Heckler v. Chaney, 470 U.S. 821, 831 (1985) (“[A]n agency decision not to enforce often involves a complicated balancing of a number of factors which are peculiarly within its expertise.”). Nor are agencies generally barred from invoking administrability or resource considerations when announcing policies that might be less than ideal from the standpoint of comprehensive rationality. For example, the D.C. Circuit has frequently stated that agencies may “employ bright-line rules for reasons of administrative convenience, so long as those rules fall within a zone of reasonableness and are reasonably explained.”328Emily’s List v. FEC, 581 F.3d 1, 22 n.20 (D.C. Cir. 2009).

That an agency’s choice of decisionmaking strategy should generally be respected by the courts is also supported by the Supreme Court’s rationale in Chenery II.329See SEC v. Chenery Corp. (Chenery II), 332 U.S. 194, 203 (1947). That case stands for the proposition that agencies are generally free to choose between two different procedures—rulemaking and adjudication—when formulating policy. But although the choice directly at issue in Chenery II was a procedural one, the reasons the Court gave for deferring to the agency’s choice to operate through adjudication went to policy. Policymaking by adjudication and by rulemaking represent different analytical methods for attacking public policy programs. Although, to be sure, an agency might make decisions based on comprehensive rationality while adjudicating, adjudication tends to represent a more incrementalist, dynamic approach than rulemaking. And it was on that premise that the Chenery II Court broadly blessed agencies’ ability to set policy by adjudication, explaining that it is the agency itself that knows the best strategy by which to tackle a particular problem.330See id. at 202–03.

So far, I have suggested that agencies should have wide latitude to adopt decisionmaking strategies that may themselves provide reasons why certain alternatives were not chosen. But an agency’s ability to do so is subject to important limits as well. First, courts should ensure that an agency gives reasons for adopting the strategy it has. In other words, an agency in incrementalist or satisficing mode must say why that strategy is appropriate given the criteria the agency believes to be relevant. Second, courts should review the agency’s explanation to make sure that the strategy adopted, and the reasons supporting it, comport with the relevant statute. This may prove an important limit. Where a statute instructs an agency to do what’s “best” for public health, or to adopt the rule with the “least” burden to industry, that may operate to constrict agencies’ latitude to perform a more rough-cut analysis. Whether given language does set such limits is ultimately a matter of statutory interpretation (and, if relevant, Chevron). But if a court determines that a certain decisionmaking strategy is available to an agency as a general matter, the agency gave a valid reason for adopting it, and the decisionmaking strategy reveals why the agency pretermitted full inquiry into certain alternatives, that alone should be sufficient to satisfy the agency’s obligation to respond to those alternatives.

C. Harmless Error

Finally, might there be situations in which an agency’s failure to respond at all to an alternative it was obligated to address may be dismissed as harmless error? The APA directs courts to take “due account . . . of the rule of prejudicial error” when reviewing agency action. 3315 U.S.C. § 706. In other words, agency action should not be set aside if the agency’s error was harmless. Although the standards courts use to assess whether an error is harmless vary,332See Hickman & Pierce, supra note 232, at § 12.2. the basic idea is that courts should “uphold[] unsound agency decisions when they are confident that the agency would reach the same decision on remand.”333Nicholas Bagley, Remedial Restraint in Administrative Law, 117 Colum. L. Rev. 253, 302 (2017).

As applied to an agency’s obligation to respond to alternatives, there are two situations in which the agency’s failure might be judged harmless. The first, which I believe is unlikely to be fruitful, would involve a more substantive judgment that the alternative in question is so frivolous or unworthy that the agency obviously would not behave differently if forced to confront it. There are two problems with this judgment. First, it puts judges back in a position of deciding the “frivolity” of a given alternative in a relative vacuum, which much of the above analysis is meant to avoid. Second, it is unlikely to do much work that could not be done in an analytically clearer way through the methods described earlier in the Article. An alternative might be frivolous because the party advocating it has not supplied support for it. But in that case, the agency is under no initial obligation to respond at all.334See supra notes 146–147 and accompanying text. Or an alternative might be frivolous because it is clear that it would not further the agency’s stated goals. But in that case, the agency has responded to the alternative, albeit implicitly.335See supra notes 295–310 and accompanying text. In neither case does an extra level of review for harmless error add anything.

There is one category of alternative where harmless error analysis can add value: where an alternative is revealed to be illegal, an agency’s failure to respond to it is always harmless.336Really, the failure to respond to an illegal option is not an error at all, harmless or otherwise. Illegal options are not part of an agency’s set of available choices, and so an agency is never required to do a comparative assessment with respect to them. That said, “harmless error” is something of a misnomer when it comes to administrative law. The text of the APA seems to say that lack of prejudice is a reason for a court to conclude that there was no error at all, not that the error is excused. Cf. Daniel Epps, Harmless Errors and Substantial Rights, 131 Harv. L. Rev. 2117 (2018) (arguing that for purposes of constitutional law, the harmless error doctrine is best understood as an inquiry into the substance of the right at stake). And because, as explained further below, the harmless error frame provides the most intuitive path around Chenery I in situations where the agency had not taken the position an alternative was illegal prior to litigation, I believe it’s the handiest for the courts to use. That is so even if the alternative is one to which the agency was otherwise (absent the illegality) obligated to respond. And it’s so even when the alternative is likely to be substantively superior to the agency’s chosen course. Because the agency is legally barred from adopting the alternative,337See 5 U.S.C. § 706(2)(A) (courts shall set aside agency action “not in accordance with law”). there is no chance that requiring an agency to respond to it would lead to a different result. Importantly, an agency’s failure to respond to an illegal alternative should be deemed harmless even where the agency did not argue that the alternative was illegal in the decision under review. That is, agency lawyers should be able to raise an alternative’s illegality for the first time in the litigation process. Wait, you might ask: does that not violate Chenery I’s command that “an administrative order cannot be upheld unless the grounds upon which the agency acted in exercising its powers were those upon which its action can be sustained”?338SEC v. Chenery Corp. (Chenery I), 318 U.S. 80, 95 (1943). Indeed, many forms of harmless-error analysis are in tension with Chenery I. Bagley, supra note 333, at 302. The answer is no, because even under Chenery I, courts are not required to remand where to do so would be “an idle and useless formality,” which is the case when an error is harmless.339Morgan Stanley Cap. Grp. Inc. v. Pub. Util. Dist. No. 1, 554 U.S. 527, 545 (2008) (quoting NLRB v. Wyman-Gordon Co., 394 U.S. 759, 766 n.6 (1969) (plurality opinion)). Thus, the Supreme Court has held that when a given policy is required by statute, it doesn’t matter that the agency failed to recognize that when it adopted that policy.340See id. Similarly, a remand based on an agency’s failure to consider an illegal alternative would be an idle and useless formality. As long as there was no other error that may have affected the agency’s decision, that decision should stand.

IV. Conclusion

Although agencies have long been obligated to respond to alternatives to their chosen course of action, the courts have struggled to give crisper guidance to agencies regarding the contours of the obligation. Worse, some courts have used the obligation in order to second-guess the outputs of agency policymaking, and in ways that may invite the unnecessary invalidation of agency programs. This Article has attempted to give the doctrine a cleaner shape, informed not only by the values associated with agency reason-giving but also by the need for effective administration and a predictable system of judicial review.

*Daniel T. Deacon, Assistant Professor of Law, University of Michigan Law School. Many thanks to Nicholas Bagley, Nicolas Cornell, Benjamin Eidelson, Sam Erman, Barry Friedman, Jacob Goldin, Margaret Hannon, Don Herzog, Matthew Lawrence, Ronald Levin, Leah Litman, Nina Mendelson, Julian Mortenson, Tejas Narechania, David Noll, Emily Prifogle, Rachel Rothschild, Matthew Stephenson, Rebecca Stone, Christopher Walker, and workshop participants at the University of Michigan Law School and the Administrative Law New Scholarship Roundtable held at the University of Texas School of Law. Thank you to Justin Hill for excellent research assistance.