Census symposium: A place for pretext in administrative law?

Jennifer Nou is Professor of Law and Ronald H. Coase Teaching Scholar at University of Chicago Law School.

With Department of Commerce v. New York, the Trump administration continues its losing streak in court under the Administrative Procedure Act. Many have ascribed this poor record to some combination of incompetence, vacancies, and a greater interest in tweeting policies, rather than implementing them. In this case, the government lost because it invoked a rationale that was simply implausible: It needed to add a census-citizenship question to enforce the Voting Rights Act. But the record was painfully clear that Secretary of Commerce Wilbur Ross had clumsily manufactured this story — cajoling the Department of Justice to make the request. The Supreme Court thus affirmed the district court and remanded the case back to the agency.

Chief Justice John Roberts, for a fractured court, upheld the plaintiffs’ standing, the suit’s reviewability, and Ross’ constitutional and statutory authority to include the citizenship question. But when it came to APA arbitrariness review, Roberts was Solomonic. On the one hand, he agreed with the conservatives that the agency had not been arbitrary under the traditional standards: The secretary “considered the relevant factors, weighed risks and benefits, and articulated a satisfactory explanation for his decision.” On the other hand, Roberts sided with the liberals in concluding that, nevertheless, the agency’s action was ultimately arbitrary because its “sole” voting-rights-related reason was pretextual.

But is there a proper role for pretext in administrative law? It would be easy to interpret the Supreme Court’s answer as a firm no. Agencies, in this view, must reveal their genuine motivations to survive judicial review. This case hardly stands for the proposition, however, that bureaucratic honesty is the best policy. Agencies can still give reasons that obfuscate the political machinations behind them. In Roberts’ words, “a court may not set aside an agency’s policymaking decision” only because it was informed by unstated “political considerations or prompted by an Administration’s priorities.” To the contrary, he continues, “typical” agency decisions can and do rest on both “stated and unstated” reasons.

Indeed, as Katherine Watts notes, agencies under arbitrariness review have long had an incentive “to dress up” their explanations “in technocratic terms and to hide political influences.” In other words, agencies often provide expertise-based rationales, even when their genuine motivations are political. More recently, Thomas McGarity and Wendy Wagner document the ways in which appointees imperceptibly skew scientific work to justify deregulatory ends. Similar accounts abound of cost-benefit analyses written to support preferred outcomes. However pervasive these phenomena, these practices are arguably forms of pretext — rationales masking the genuine motivations for decisions.

The perception of charade only persists when agencies reimpose the same policies after remand, which the Department of Commerce seems likely to attempt here. One study, for example, found that agencies in its sample were able to pursue the same policy in about 80 percent of the rulemakings sent back after arbitrariness determinations. In other words, agencies simply carried on with their decisions after better dotting their i’s and crossing their t’s.

Why has the law allowed, perhaps even encouraged, this state of affairs? The likely answer has something to with the administrative state’s well-known struggle to reconcile politics and expertise. Perhaps a bit of pretext allows agencies to maintain public legitimacy at a time when political accountability is more suspect (think low voter information or partisan gerrymandering). Or perhaps a little required pretext ultimately helps agencies make better decisions, at least some of the time.

Whatever the explanation, administrative law has and will likely continue to tolerate some forms of pretext. A potentially new principle introduced in this case, however, is the idea that such pretext must at least be plausible. How else to give legal content to the idea that courts are “not required to exhibit a naiveté from which ordinary citizens are free?” In other words, Roberts — quoting Judge Henry Friendly — refuses to accept a wholly implausible rationale, one that is not credible. On this view, the kinds of pretext that agencies normally provide to meet the expertise-driven demands of arbitrariness review are fine because they are plausible; there is generally enough in the administrative record to justify them. By contrast, when the available evidence suggests that the only stated rationale is implausible, not credible, then agencies run afoul of the APA.

To be sure, the concept of plausibility itself is ill-defined, but it is not foreign to the law. Civil pleadings, for example, must state “plausible” claims, which the Supreme Court distinguishes from those that are merely “conceivable.” This is not to suggest that another body of law be imported here, but rather to search for some analogous concept in an effort to distinguish between proper and improper administrative pretext. As applied here, the pretext provided by the Department of Commerce was conceivable, but it was not plausible: There was just too much evidence that DOJ did not sincerely seek citizenship data. The secretary was trying to pull a fast one. Roberts makes clear that Ross would have been justified in adding the citizenship question had the trial proceedings suggested otherwise.

Of course, this now raises the related question of whether those trial proceedings were appropriate: What evidence can a reviewing court properly consider? The APA provides that courts must engage in arbitrariness review based on the “whole record,” the administrative record. Different agencies have varying conceptions of what should go into that record. In Citizens to Preserve Overton Park v. Volpe, however, the Supreme Court recognized that agency-furnished records can be insufficient for judicial review. There, the Department of Transportation allowed a highway to be built through a park without generating any contemporaneous record at all. The court thus remanded to the district court to figure out what to do. In passing, the Overton Park court noted that a “strong showing of bad faith or improper behavior” could justify extra-record testimony by agency officials.

Based on that language, in the case at hand, Roberts rightly endorsed the district court’s decision to allow for additional discovery from the Commerce Department. The government’s constant need to supplement and clarify the record demonstrated its bad faith in constructing the record in the first place. But, going forward, just how much of a showing of bad faith would be necessary for courts to look beyond the initial record? A whiff, a “reasonable” amount, more? Both Roberts and Justice Clarence Thomas in partial dissent are unclear on this question.

There is a good argument, however, that Thomas was right to emphasize that whatever the standard is, it should be high. As he writes: “Virtually every significant agency action is vulnerable to the kinds of allegations the Court credits today.” The majority thus “enables partisans to use the courts to harangue executive officers through depositions, discovery, delay, and distraction.” The more discovery is allowed, in turn, the more litigants will be able to sift through depositions in search of any evidence of pretext. Such fears justify the general presumption of regularity, the normal rule against probing the mental states of busy executive-branch actors.

In the final analysis, Department of Commerce v. New York contains several ideas in danger of being read too broadly. One is to demand full administrative candor. Another is to allow discovery at the slightest trace of subterfuge. As Roberts concedes, such approvals should be “rare.” Allowing litigants to pounce at any hint of substantive or procedural bad faith threatens ossification and delay. Lower courts should therefore cabin this precedent’s principles, either to the case’s extraordinary facts or to its narrow context. As proposed elsewhere, perhaps this form of harder-look review should apply only in election-related decisions.

After all, some forms of pretext already have a place in administrative law. The pressing questions now are what kinds of pretext are tolerable, which are not, and why. The answers may determine whether the citizenship question ultimately appears on the 2020 census. This court may very well accept any plausible pretext from the Trump administration if called upon to review this case again. Only time — and the census printing deadlines — will tell.