When it comes to judicial review of administrative action, the scales of justice are heavily weighted on the agency’s side. The process could benefit from a more rigorous and transparent approach to building and testing the decisionmaking record

The fate of environmental regulations often turns on esoteric principles of administrative law. This is the law that governs government and its relationship to the governed. Administrative law covers a lot of ground, including enforcement, due process claims, and appointment and termination of agency personnel. Our focus here, however, is the law of judicial review of agency action. This law applies to challenges to an agency’s promulgation of regulations (“rulemaking”), issuance of orders (“adjudication”), or failures to engage in rulemaking or adjudication. Judicial review cases are different from typical civil lawsuits. This essay highlights some key differences and the edge that they provide to government litigators. Of course, nobody challenging the government expects a completely fair fight. But when pushed too far, the government’s litigation advantages can prevent a reasonable vetting of agency decisions.

The law of judicial review largely comprises judge-made rules which are only loosely related to statutory text. That text is found either in the Administrative Procedure Act of 1946 or in regulatory laws that codify APA-like procedures. The rules of administrative law are at least a little contradictory. For example, “courts may not accept appellate counsel’s post hoc [i.e., after the fact] rationalizations for agency action.” This is because the “Chenery doctrine” (from 1943 and 1947 Supreme Court cases, as repeated in the Court’s much-cited Motor Vehicle Manufacturers Association v. State Farm case from 1983) tells us that administrative decisions “must be upheld, if at all, on the basis articulated by the agency itself.”

Nonetheless, courts may consider those same post hoc arguments when determining whether to uphold decisions “of less than ideal clarity” because “the agency’s path may reasonably be discerned.” Similarly, the “Chevron doctrine” (from a 1984 Clean Air Act case) generally requires courts to defer to an implementing agency’s reasonable interpretation of legislation when the law at issue fails to resolve “the precise question at issue.” But agencies receive no deference when courts can divine Congress’s intent using (sometimes convoluted and controversial) “traditional tools of statutory construction.”

When it comes to judicial review of administrative action, the scales of justice are heavily weighted on the government’s side. Those challenging government decisions win often enough to keep things interesting, but you are unlikely to hear private-practice administrative litigators brag about their win-loss records. A major governmental advantage is rooted in deference doctrines (i.e., Chevron, Skidmore, or Auer deference, and the arbitrary-and-capricious standard of review) which — to various and evolving degrees — implement presumptions that government bureaucrats know what they are doing and that politically responsive agencies are better situated than appointed judges to fill the gaps that Congress leaves in regulatory laws. But deference aside, administrative law procedures provide a decided advantage to the government.

A fair reading of the federal rules of civil procedure suggests that lawsuits, in the normal course, should culminate in either settlement or trial. Lawyers are trained to believe that the best path to revealing truth is through cross-examination of witnesses and the adversary process of presenting evidence to a neutral fact-finder — a judge or jury. Settlement is always encouraged since litigants are likely to have a better understanding of their dispute than a judge or jury ever will. But absent an acceptable compromise, it is comforting to believe that courts will test factual assertions and the credibility of witnesses — including purported experts — in the crucible of trial.

Administrative law cases are different. Even when those cases concern important issues of public policy, they generally proceed with no witnesses, no cross-examination, no jury, and no consideration of evidence that was not — at least in theory — before the agency when it made the decision under review. For better or worse, administrative law is a gentleperson’s undertaking — conducted largely on paper, unrestricted by most evidentiary rules, and culminating in “argument,” i.e., speechifying, rather than a trial with actual testimony subject to objections.

All this is to the government’s advantage because judicial review of agency decisions starts with a presumption of regularity. It is the challenger’s job to rebut that presumption and convince a court that truth and justice require reversal of an agency’s decision. To pull this off without cross-examining the government’s witnesses is difficult. Challengers rely on comments they added to the record — often on short notice — before the government finalized its decision and hope they can find a legal or procedural error that a court will deem significant enough to disrupt the agency’s plans. Reliance on flaky data is not necessarily fatal to the government’s case. For example, when upholding an EPA decision not to increase public health protections from hazardous air pollutants in a 2008 case, the D.C. Circuit accepted EPA’s calculation of the risk to public health based on seven-year-old responses to a trade association’s questionnaire, to which a majority of regulated facilities had not bothered to respond at all. The court noted, “We generally defer to an agency’s decision to proceed on the basis of imperfect scientific information, rather than to invest the resources to conduct the perfect study.”

The deck was not always so stacked. As Justice Thomas noted in his 2015 concurrence in Perez v. Mortgage Bankers Association, the APA seems to contemplate “a much more formal process,” providing for “trial-like hearings in which proponents of particular rules would introduce evidence and bear the burden of proof in support of those proposed rules.” But in a pivotal 1973 opinion, United States v. Florida East Coast Railway Company, the Supreme Court signaled that agencies usually need not follow the APA’s instructions for “formal” procedures, but instead — under most circumstances — may engage in “informal” rulemaking and adjudication. The primary difference is that informal proceedings are not “trial-like”; instead, they are geared more toward collecting feedback than testing the strength of evidence or policy decisions.

For informal rulemaking, agencies normally publish notices of their proposed regulations in the Federal Register. Potential challengers (and supporters) typically get a 20- to 90-day period in which to comment on those proposals. Savvy and well-funded potentially affected parties, however, keep tabs on agency plans and do not wait for formal proposals before swinging into action. Usually, judge-made exhaustion-of-administrative-remedies requirements bar challengers from arguing about issues that nobody raised before the agency during the comment period. (Some regulatory laws have codified this exhaustion rule.) Agencies risk judicial disapproval if they play bait-and-switch by adopting a final rule that is not a “logical outgrowth” of the published proposal. They must explain rulemaking decisions in “statement[s] of basis and purpose.” For informal adjudication, the agencies’ few obligations under the APA include issuance of “prompt notice” of any denial and a “brief statement of the grounds for denial.”

During the 1970s, the D.C. Circuit took a stab at imposing more rigor. (The D.C. Circuit is especially important in administrative law because Congress directs many judicial review cases to that court.) D.C. Circuit opinions suggested that cross-examination before the agency might be required in “particular cases of need, on critical points where the general procedure proved inadequate.” Even where cross-examination was not required, the D.C. Circuit demanded “sensitive use” of “alternative procedural techniques” to ensure the record before the court “discloses a thorough ventilation of the issues.”

The Supreme Court put the kibosh on that approach, however, in its Vermont Yankee opinion of 1978 (about radioactive waste from nuclear power plants). The Court held that, “generally speaking,” the APA’s sketchy provisions provide “the maximum procedural requirements” that courts may “impose upon agencies in conducting rulemaking procedures.” Nobody has taken this “maximum procedural requirements” language completely seriously. For example, courts interpret the APA’s mandate for “a concise general statement of [a final rulemaking’s] basis and purpose” to include a requirement that agencies respond to all significant public comments under most circumstances. Courts also require more specific notice of proposed rulemaking than the APA’s provision for “general notice” implies. Nonetheless, after the Vermont Yankee opinion, informal administrative procedures have remained distinctly informal.

When a rule or order is challenged, courts determine its validity based on review of “the whole record or those parts of it cited by a party,” as per Section 706 of the APA. This reference to the “whole record” is straightforward when applied to the rare agency decisions that follow the APA’s instructions for formal procedures. Those specifically require a “record for decision” comprising “transcript[s] of testimony and exhibits, . . . all papers and requests filed in the proceeding,” and “findings and conclusions . . . on all the material issues of fact, law, or discretion.” But informal procedures do not necessarily generate what many lawyers would recognize as a record. Under these circumstances, how does a court determine from the “whole record” whether an agency’s decision is “based on a consideration of the relevant factors,” and untainted by “a clear error of judgment” — and thus is not “arbitrary and capricious” (as the Supreme Court explained that phrase in its 1983 State Farm opinion)?

The Supreme Court tackled this problem in the Citizens to Preserve Overton Park v. Volpe case, decided in 1971. The Court’s answer was embodied in a remand to the district court to consider “the ‘whole record’ compiled by the agency,” which is “the full administrative record that was before the secretary at the time [of the] decision.” If the agency failed to compile a record sufficient to provide a basis for review, the district court had discretion to require testimony, although “inquiry into the mental processes of administrative decisionmakers is usually to be avoided.” Thus, the Overton Park case provides a powerful incentive for agencies to compile records that justify and explain their decisions, assuming they wish to avoid extra-record inquiry into decisionmakers’ mental processes. In modern cases, a failure to explain or other inadequacies in the record would result in remand — sending the decision back to the agency for further proceedings — rather than agency testimony. Agencies, however, also have an incentive to avoid the disruption of remand, especially a remand accompanied by voiding of the challenged rule (which lawyers call “vacatur”).

Where does the “whole record” come from? We learned from Overton Park that it is “compiled by the agency.” There are few mandatory standards that govern the compilation process, although the record generally must include material submitted during the public comment period in a rulemaking. Public comments aside, the plaintiff must build its case out of evidence compiled by the opposing litigant — the agency whose conduct is under challenge. Overton Park describes the record as that which “was before the secretary at the time [of the] decision.” One might think, therefore, that compiling the record would be easy and uncontroversial; someone could simply sweep the contents of the decisionmaker’s desktop into a redrope folder — or the digital equivalent — as soon as a decision is made. That is not, however, what happens. Even after having 60 days, or more, in which to file an answer to a challenger’s complaint in a district court APA case, the government typically asks for 90, 120, or more additional days in which to compile the record. Why does it take so long? The compilers have decisions to make about what to deem relevant, irrelevant, or privileged.

Consistent with various judicial opinions, Justice Department guidance from 1999 says the record should contain “all documents and materials directly or indirectly considered by the agency.” What is indirect consideration? In 2010 Judge John L. Kane, from the District of Colorado, explained, “If the agency decisionmaker based his decision on the work and recommendations of subordinates, those materials should be included in the record,” if that information “was so heavily relied on in the recommendations that the decisionmaker constructively considered it.” Standards like “so heavily relied on,” however, are difficult to administer consistently.

For good reasons, lawyers tend to distrust standards that rely on adjectives or adverbs. Under the rules of professional conduct that govern lawyers, an agency lawyer would have a basis for withholding a document from the record if that lawyer had a “good faith” argument that the document was not “so heavily relied on” by agency staff as to have been “constructively considered” by the final decisionmaker. The American Bar Association has explained in a comment to its Model Rules of Professional Conduct that there is no requirement that lawyers believe their good-faith arguments would actually prevail in court. Essentially, what is required is that the lawyers “inform themselves” as to the facts and the law and stick to arguments they can make with a genuinely straight face. Nonetheless, those who challenge agency action usually must rely on the good faith of agency employees and lawyers who compile administrative records. The results tend to vary.

Granted, an agency’s challengers have an opportunity to ask the court to complete, supplement, or strike portions of a government-compiled record. The trick, however, is in marshalling proof that the government’s compilation is faulty. In typical (non-administrative law) civil lawsuits, lawyers routinely gather information through the “discovery” process. Discovery rules allow litigants to demand information of their opponents and others by propounding written questions — to be answered under penalty of perjury — or demands to depose witnesses or inspect documents or property. Neither the Federal Rules of Civil Procedure nor the APA purport to preclude discovery in judicial review cases. Nonetheless, those who challenge government decisions typically find that discovery is unavailable.

One barrier to discovery is the Supreme Court’s ruling in Overton Park that “inquiry into the mental processes of administrative decisionmakers is usually to be avoided.” The policy justification for this is unclear, since a biased decisionmaker can affect the fairness of the administrative process. For example, the D.C. Circuit has long disfavored promulgation of regulations by agency heads with “unalterably closed” minds. In a 2019 Supreme Court case affirming vacatur of a governmental effort to ask about citizenship on the census — Department of Commerce v. New York — the Court found that the agency’s justification of its decision was “contrived” and affirmed a district court decision that labeled the agency’s explanation “pretextual.” The opinion, however, may nonetheless amplify the rule announced in Overton Park. The Court held that the district court was justified in invoking an exception to the precedent’s avoidance policy in light of a “strong showing of bad faith or improper behavior.” But while Overton Park discussed that exception in terms of a requirement that agency decisionmakers testify “explaining their action,” the Department of Commerce case speaks more broadly, suggesting that such a showing may be necessary to “justify extra-record discovery.”

Another argument that the government uses to avoid discovery is that the APA’s mandate that courts “review the whole record or those parts of it cited by a party” somehow constitutes an implicit command that courts refrain from considering facts gleaned in discovery. This argument is often effective but can be pushed too far: In one case, the U.S. Army Corp of Engineers argued that facts relating to Hurricane Katrina’s impact on a project were irrelevant because they were not in a record that the Army compiled before the hurricane. In a 2006 opinion, U.S. District Court Judge Eldon Fallon ruled that to ignore such facts “is to ignore reality. For the law to have any credibility or respect, it must be grounded in reality.”

Doubts about the probity of an agency’s record compilation may be fueled by the government’s “deliberative process privilege.” In 2021, the Supreme Court decided U.S. Fish and Wildlife Service v. Sierra Club, ruling that this privilege relieves agencies from the pressure of operating “in a fishbowl” (which some might describe as “transparency”). The Court explained that the privilege protects “advisory opinions, recommendations and deliberations comprising part of a process by which governmental decisions and policies are formulated.” It “blunts the chilling effect that accompanies the prospect of disclosure.” In other words, the privilege encourages sensitive souls to speak frankly when they might otherwise fear the consequences of their views becoming public. “Documents are ‘predecisional’ if they were generated before the agency’s final decision . . . and they are ‘deliberative’ if they were prepared to help the agency formulate its position.” A challenger can defeat the privilege by showing a need for the information that outweighs the agency’s interest in secrecy. Making such a showing, however, is difficult if the plaintiff does not know that a document exists.

The government has a policy that can keep opponents from learning that an agency is withholding documents that are arguably deliberative. EPA asserts in its 2011 Administrative Records Guidance that “the exclusion of deliberative documents from the record is based on relevance, not privilege.” Similarly, DOJ announced in guidance from 2017 that predecisional deliberative documents “generally are not relevant to APA review.” As a result, those documents “generally should not be produced as part of the administrative record . . . nor listed in a privilege log.” So under this scheme, the government not only withholds purportedly deliberative documents, it withholds them quietly. Further — as discussed above — lawyers need only a “good faith” basis (as opposed to a good or substantial basis) when making decisions on behalf of their clients. So if a government lawyer with a good-faith (but shaky) basis designates an EPA document as deliberative, the very existence of that document may be hidden from opposing parties. Granted, EPA’s guidance suggests that “where the [deliberative] document contains factual information or records policy decisions found nowhere else in the record . . . the relevant information should be extracted and placed in a non-deliberative document.” Assuming that the agency will always follow this guidance, however, requires a degree of trust that is foreign to the litigation process.

Possibly the hardest truth for nongovernmental administrative law practitioners to explain to their clients is that victory is rarely complete. Usually, the most a challenger can hope for is a remand with vacatur. This is less than wholly satisfying because the agency often can correct the specific errors identified in the court’s ruling, issue the same decision again with a revised record, and move the previously victorious challenger back to square one. This is exactly what the Securities and Exchange Commission did in the 1943 and 1947 Chenery Supreme Court cases. Having overturned the agency once, the Court upheld the same substantive SEC decision the second time around, based on a new justification in the record. This is not, however, what always happens. A judicial setback for the agency may empower voices within that agency to push for a different outcome. Elections may occur or other circumstances may change during the delay that a remand causes. The issue may fall off the agency’s priority list. So what seems like a mere re-set may end up fully achieving a litigant’s goals. But then again, maybe not — so administrative lawyers are well advised to savor each victory and collect their fees before the agency makes its next move.

Even when government agencies are caught with a blatantly inadequate record (e.g., failure to perform a mandatory step in the decisionmaking process), the government has been known to request (and too often receive) a stay of the litigation while the decision is remanded — without vacatur — for the agency to shore up its record. This practice of providing the agency with a do-over in the middle of a lawsuit is especially problematic in light of the Supreme Court’s 2020 ruling in Department of Homeland Security v. Regents of the University of California, a case about immigration law which affirmed vacatur of the Trump administration’s rescission of the Deferred Action for Childhood Arrivals program. In that case, the Court declined to consider the homeland security secretary’s supplementation of the record with a new memorandum to provide what she hoped would be legally sufficient reasons for the rescission. The Court viewed the secretary’s supplemental memo as “impermissible post hoc rationalizations.”

Courts also sometimes disapprove a regulation but fail to vacate it, for example in the litigation over the Mercury and Air Toxics Standard governing air emissions from coal-fired power plants. In a 2015 Supreme Court case, Michigan v. EPA, the coal industry appeared to win a huge victory: the Court ruled that the agency’s regulation was illegal because it had failed to consider the costs of compliance. But neither the Supreme Court nor the D.C. Circuit vacated the illegal rule. Instead, MATS was remanded without vacatur and, as of 2021, remains the law of the land, despite various supplemental EPA rulemakings.

Perhaps it is churlish to complain that the administrative law playing field is not level. After all, those who challenge U.S. governmental decisions have the privilege of doing so — by and large — without risk to their lives and livelihoods. Throughout world history, powerful rulers have not always been so tolerant of opposition to their decisions. And, as illustrated by some of the cases discussed above, the government does not always win in court. The judicial review process could nonetheless benefit from a more rigorous and transparent approach to building and testing the administrative record. TEF