On March 9, the U.S. Court of Appeals for the D.C. Circuit issued two concurring opinions in an earlier per curiam ruling that had vacated a district court injunction against the military’s restrictions on service of transgender persons. As Judge Robert Wilkins’s concurrence observed, neither the D.C. Circuit’s ruling nor the Supreme Court’s stay of injunctions in two related cases decides the merits. Each nonetheless signals that deference to military judgments may well emerge victorious. But the lack of tailoring—the careful fitting of means to chosen ends—in the military’s transgender policy should prompt the courts to look more closely at the Trump administration’s rescissions of measures followed or implemented by the Obama administration, including DACA and the long-standing omission of a citizenship question on the U.S. census. In a new paper, I discuss the courts’ role in adjudicating these actions by the Trump administration.
The paper describes threshold rescissions, which limit participation based on race, gender, or sexual orientation, restrict the ability to work legally and remain in the United States or adversely affect representation in the political system. To qualify as a threshold rescission, an agency shift must undermine reliance interests of recipients or trigger severe collateral impacts. That is, the shift must undercut the expectations of recipients or third parties who invested time, effort, and reputation in the rescinded reforms and now stand to lose those investments in whole or part.
The military’s new transgender policy supplanted a series of inclusive measures initiated during the Obama administration by Secretary of Defense Ash Carter, including broadened rules on accession (i.e., joining up) and retention of transgender individuals. These measures are inclusive because they shifted the default position from a ban on transgender individuals who revealed their status to acceptance of many transgender individuals. After President Trump’s inauguration and an initial tweet by the president that called for a return to the ban, then-Secretary of Defense James Mattis coordinated a review of Carter’s changes.
As the D.C. Circuit held and Marty Lederman explained in an invaluable post, Mattis’s recommendations—subsequently embodied in a memo by President Trump—did not reinstate a comprehensive ban on transgender service or merely parrot Trump’s first tweet on the subject. Rather, Mattis grandfathered in service members who had commenced medical transitions in reliance on Carter’s policy. In addition, the Mattis policy permitted service by persons who identified with a gender other than that which they were initially assigned but were still willing to serve under assigned-gender standards.
These components of Mattis’s plan did not wholly alleviate the rescission’s harm to reliance interests. For example, under the Mattis plan, a grandfathered-in transitioning service member might well experience both harassment within his or her unit and a passive command response, since the military’s overall position had shifted from inclusion to restriction. Under traditional constitutional doctrine, the absence of a total ban might trigger deferential judicial review, particularly given the tradition of deference to military judgments. However, features of the Trump rescissions call for reconsideration of this deferential posture.
First, as noted in the previous paragraph, threshold rescissions undermine reliance interests. Even when the rescission grandfathers in a cohort that relied on the previous reforms, that modification may leave the grandfathered cohort vulnerable. Second, adjudicating rescissions does not prompt the same concern about institutional competence that underwrites most applications of deference. As Robert Chesney wrote in a much-cited paper, courts often note that they lack the fact-finding resources of the political branches. But in rescissions, an earlier administration has found the facts. The real issue should be whether new facts justify a rescission of the policy. Assessing a rescission’s means-ends fit is a good way to push the rescinding administration to cite new facts.
Third, a rescinded reform often enjoys substantial bipartisan support in Congress. Courts adopting deference often cite the political branches’ electoral accountability, as compared with federal judges’ lifetime tenure. Yet, a rescinded reform has often garnered significant legislative favor on both sides of the aisle. For example, former Speaker of the House Paul Ryan and Sen. Orrin Hatch spoke in favor of DACA (see Kevin Johnson’s article here). Admittedly, that legislative support may not translate into the gold standard: a veto-proof majority for codifying the earlier reform. However, legislative support does signal to courts that a significant number of politically accountable actors endorse the reform, making the courts less isolated than they would be in protecting reforms without such support. In light of these factors, the lack of tailoring in other components of the Mattis policy should give courts pause.
Consider the Mattis policy’s bar on service by transgender individuals who require accommodations in bunking, bathroom use or showering. Judge Stephen Williams, concurring in the D.C. Circuit ruling, cited this component of the policy as a paradigmatic exercise of military judgment. In justifying this bar, the study commissioned by Mattis invoked the privacy rights of other service members. However, as Carter’s handbook on his reforms explained, commanders have many less restrictive fixes available for this situation, including subdividing showering areas and staggering shower times. In light of these common-sense solutions, broader restrictions on accommodations are not tailored to the military’s privacy goal and thus fail the test of means-ends scrutiny. While the 2018 Defense Department study asserted that Secretary Carter had misread reports on the Canadian military’s experience with transgender inclusion, Carter had added his guidance in a direct response to the problems that Canada’s lack of guidance had caused. (See p. 54 of the report issued by the RAND Corporation supporting Carter’s decision.)
The 2018 military study also cited increased costs associated with transgender medical treatment. However, as the RAND report noted (pp. 10-11), the case for transgender restrictions often fails to recognize that transgender persons already serve in the military and have done so for a long time. An impartial analysis needs to balance the cost of transgender treatment (a pittance when compared with the military’s huge overall health-care costs) against the costs that the Carter policy avoids, including the cost of mental health treatment for transgender persons who will have to conceal their identity under the Mattis restrictions. In addition to these costs, Carter’s policy would have lessened the suicide risk linked to lack of acceptance of transgender persons. The 2018 military study failed to adequately address these trade-offs. It also thus failed the test of means-ends fit.
This same lack of tailoring—of means-ends fit—is evident in Secretary of Commerce Wilbur Ross’s decision to insert a citizenship question in the decennial census. The March 6 decision by Judge Richard Seeborg of the Northern District of California in California v. Ross joins an earlier district court ruling from New York invalidating Ross’s insertion of a citizenship question.
Even against the backdrop of judicial deference to the secretary of commerce’s discretionary decisions on conduct of the census, Ross’s precipitous decision raises special concerns. Admittedly, the government asked a citizenship question on the census for well over a century. However, it last did so in 1950. Since then, as Judge Seeborg observed, Census Bureau experts from both Republican and Democratic administrations have repeatedly informed Congress that a citizenship question would “seriously jeopardize” the distributive accuracy of the census, by undercounting undocumented people and others who decline to participate. According to government experts, that refusal would stem from fear that results of the census would be shared with immigration enforcement officials. Since undocumented people tend to congregate in “gateway” states, such as New York, California and Texas, an undercount would disproportionately affect those jurisdictions, leading to malapportionment of congressional representatives and other adverse effects. Hence, the loss in distributive accuracy.
The demographic and statistical literature supporting this view is overwhelming. Consistent with the Constitution’s Enumeration Clause and the 14th Amendment, which require a decennial census, and with the Administrative Procedure Act’s bar on “arbitrary and capricious” agency action, one would expect that the secretary of commerce would order an array of tests before inserting a citizenship question. However, Ross appears to have done no testing at all. Instead, Ross, in a March 2018 memo on his decision, baldly asserted that “no empirical data existed” on the adverse impact a citizenship question would have on census participation rates and hence on distributive accuracy.
Ross’s characterization blatantly misrepresented the state of expert opinion. In an earlier proceeding in the Supreme Court that resulted in a stay of a district court order compelling Ross’s deposition, Justice Neil Gorsuch, joined by Justices Clarence Thomas and Samuel Alito, dismissed this gap between the Ross memo and expert judgment as a garden-variety dispute between a senior official and his staff. This assessment fails to recognize the chasm between Ross’s view and the phalanx of expert analyses, backed up by almost 60 years of agency practice. To paraphrase the late Sen. Daniel Patrick Moynihan, a cabinet secretary is entitled to select her staff, but she isn’t entitled to her own facts. Justice Gorsuch’s opinion fudges this crucial distinction; Judge Seeborg’s decision sees things more clearly.
As Judge Seeborg observed, Ross’s decision was all the more arbitrary because of the manifest lack of fit between his stated goal and his insertion of the citizenship question. According to Ross, his objective was to ensure that the Justice Department had accurate data on the citizen voting-age population (CVAP), which the department needs to enforce the Voting Rights Act. Here, again, Ross’s position conflicted with the facts, including those detailed in Ross’s own March 2018 memo. As the memo noted, self-reports of citizenship status on the American Community Survey, which is regularly conducted by the Census Bureau, are seriously inaccurate—about 30 percent of noncitizen respondents get this question wrong. Importing a 30 percent error into the short-form decennial census would compound the problem.
Fortunately, there’s a good solution to the Justice Department’s legitimate need for CVAP data: Use other administrative records readily available to the government. Ross’s memo acknowledges the benefits of using this reliable resource. However, in an abrupt twist of logic that the memo failed to adequately explain, Ross then contended that combining administrative records and a census citizenship question would further improve the accuracy of the count. Basic statistics and common sense suggest the opposite: A statistical study is only as good as its weakest link. Combining an inaccurate source with an accurate one will downgrade accuracy, not improve it. However, this fundamental truth eluded Ross, as Judge Seeborg explained. Here, too, the palpable lack of means-ends fit in Ross’s decision should count more than the courts’ tradition of deference.
As further evidence of lack of tailoring in the Trump administration’s campaign to rescind Obama administration inclusion efforts, consider the attempt by the Department of Homeland Security (DHS) to rescind the Deferred Action for Childhood Arrivals (DACA) program. As I have written elsewhere, there are legitimate arguments that DACA exceeded Congress’s delegation of power to the executive branch. But those arguments turn on statutory interpretation: ascertaining the quantum of discretion that Congress afforded the executive branch in the Immigration and Nationality Act. The Trump administration has announced that its primary reason for rescinding DACA is concern about its legality. However, in three tries—one by former Attorney General Jeff Sessions, one by former Acting DHS Secretary Elaine Duke and one last June by current DHS Secretary Kirstjen Nielsen—the Trump administration has not adequately explained the basis for its legal doubts.
Both Sessions and Duke stressed constitutional issues, which are not directly relevant, since DACA presents a statutory question. Nielsen focused on the statutory question but failed to address parallels between DACA and past practice regarding discretion to grant deferred action in hardship cases. Perhaps DHS could distinguish those prior cases, which involved far smaller numbers of grantees than the 800,000-strong DACA program. But DHS has not even tried to make this case. If DACA’s putative illegality has driven the Trump administration’s effort to rescind it (which is now subject to a nationwide injunction in the lower federal courts), the Trump administration with the benefit of three tries should be able to articulate a clear rationale for its position. Its failure to do so shows the lack of tailoring in the DACA rescission.
The Supreme Court’s historic deference to executive judgments on such questions may eventually result in upholding the DACA rescission, unless Congress and the president act together to provide lasting relief for DACA recipients. But if neither Congress nor the president acts, the Supreme Court should consider the lack of means-ends fit in the latest iteration of the DACA rescission, as well as similar problems in the transgender and census rescissions. Addressing that gap on either constitutional or statutory grounds would not mean that the validity of any given administrative measure would be up for grabs. Rather, dialing down deference would promote a more deliberate approach to the reliance interests created by previous reforms. That is an objective the courts should embrace.