Yesterday the Supreme Court issued its decision in Department of Homeland Security v. Regents of the University of California, which concerns the Trump Administration’s attempt to rescind the Obama Administration’s Deferred Action for Childhood Arrivals immigration relief executive action (DACA). In a 5-4 decision, Chief Justice Roberts penned the majority opinion and likely cast the deciding vote to find that DHS Acting Secretary Duke’s decision to rescind DACA was arbitrary and capricious under the Administrative Procedure Act (APA). This does not mean, Roberts explains, that DHS cannot rescind DACA. “All parties agree that it may.” (p. 9) Instead, it just means that the Court found DHS’s approach to rescission procedurally defective and thus remanded the entire matter back to DHS for a potential redo (if DHS decides to attempt to rescind DACA again).

There is a lot to unpack in this opinion, as in some seventy pages it covers so many aspects of the four-credit administrative law course. The Court finds the agency action judicially reviewable (Part II) because the DACA rescission action is not a nonenforcement policy and the statutory nonreviewability provisions in the Immigration and Nationality Act are inapplicable. It finds a way to not weigh in on whether DACA was lawful in the first place (or whether Attorney General Sessions’s conclusion that it was not lawful is correct). It finds insufficient the plaintiffs’ allegations of an equal protection violation (Part IV). (Justice Sotomayor dissented from that holding, but the other eight Justices all rejected the equal protection challenge.) The Court also found that the DACA rescission was arbitrary and capricious under the APA because (1) Duke failed to explain why she rescinded the entirety of DACA when Sessions determined that only half of it was unlawful and (2) she failed to adequately consider the reliance interests at stake. And in the last footnote (fn.7), the Court even seems to bless the use of at least some form of a nationwide injunction while at the same time expressly saying it is “unnecessary to examine the propriety of the nationwide scope of the injunctions….”

I will leave to others to explore those issues. Instead, in this post, I focus on the remedial aspect of the opinion: the Court’s application of the ordinary remand rule and its implicit rejection of the unusual-but-now-familiar administrative practice of remand without vacatur.

As I have written about in the immigration adjudication context, in Chenery v. SEC (Chenery I), the Court held that when a court concludes that an agency’s decision is erroneous, the ordinary rule is to remand to the agency to consider the issue anew (as opposed to the court deciding the issue itself). This Chenery I rule encompasses two related principles. First, the agency itself cannot provide additional, post hoc reasons for the agency decision in the litigation itself. Second, the Court also cannot uphold the agency action based on its own reasons. Instead, the Chenery I Court held, “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.”

This is an unusual case when it comes to Chenery I. After all, in one of the decisions under review, Judge Bates of the U.S. District Court for the District of Columbia stayed his order vacating the DACA rescission memo to give DHS 90 days to reissue the DACA rescission memorandum to provide a “fuller explanation.” (p. 7) Within two months of that order, new DHS Secretary Nielsen issued a responsive memorandum. The Nielsen memo dances between just reaffirming the Duke memo and expressly issuing a new agency action agreeing to continue to rescind DACA that includes additional policy rationales for the decision.

Last November on the administrative law professor listserv, I predicted that the Supreme Court would read the last line of the Nielsen memo and, contrary to Judge Bates’ decision, conclude that Nielsen was independently exercising her authority to set DHS enforcement policies and priorities for the reasons she provides in the memo. After all, she concludes: “For these reasons, in setting DHS enforcement policies and priorities, I concur with and decline to disturb Acting Secretary Duke’s decision to rescind the DACA policy.” In that sense, the Nielsen memo reads a lot like a judicial opinion concurring in judgment with three “separate and independently sufficient reasons” for rescinding DACA. And if the Court concluded that Nielsen was independently exercising her enforcement authority and not just reaffirming the Duke memo on the same grounds, the DHS action to rescind DACA would survive judicial review under the APA. This is essentially how Justice Kavanaugh would have ruled, as articulated in his dissent.

I was flat wrong on the first prediction, and maybe also wrong on the second. On the first, the Court found that the Nielsen memo was not a new agency head nonenforcement decision, but instead an ex post, impermissible articulation of new reasons for the agency action under review, in violation of Chenery I. Because the Court excluded the Nielsen memo from the agency record, it did not consider the reasons given in the Nielsen memo.

It is less clear whether my second prediction was wrong. Based on how Roberts approached State Farm hard look review in this case, the Nielsen memo may still have been found procedurally defective. To be sure, the memo addresses reliance interests and says they do not outweigh the policy interests in rescinding DACA. It is possible, though, that the memo does not say as much about the reliance interests as Roberts would require.

Furthermore, the Nielsen memo does not identify and reject the alternative regulatory action of just rescinding the half of DACA that confers benefits and not the half that forbears removal—the second purported procedural deficiency Roberts identifies in the Duke memo. On the other hand, the Nielsen memo addresses forbearance directly, concluding that “DHS should enforce the policies reflected in the laws adopted by Congress and should not adopt public policies of non-enforcement of those laws for broad classes and categories of aliens under the guise of prosecutorial discretion.” That explanation would surely satisfy ordinary arbitrary-and-capricious review. After all, if the agency head has expressly determined as a matter of policy that DACA-like nonenforcement polices should be enacted in legislation, rather than executive action, then the agency has considered and rejected Roberts’ alternative of rescinding benefits but preserving forbearance. Roberts’ requirement that DHS had expressly recognized and rejected the halfway rescission would be redundant and hypertechnical. But again, it is possible Roberts would require more.

In all events, by rejecting the Nielsen memo as an impermissible ex post justification under Chenery I, Kavanaugh seems quite right that the Court has ventured into a new frontier by expanding this Chenery I prohibition beyond “after-the-fact explanations advanced by agency lawyers during litigation (or by judges).” Kavanaugh notes the oddness of this expanded Chenery I prohibition:

Indeed, the ordinary judicial remedy for an agency’s insufficient explanation is to remand for further explanation by the relevant agency personnel. It would make little sense for a court to exclude official explanations by agency personnel such as a Cabinet Secretary simply because the explanations are purportedly post hoc, and then to turn around and remand for further explanation by those same agency personnel. Yet that is the upshot of the Court’s application of the post hoc justification doctrine today. The Court’s refusal to look at the Nielsen Memorandum seems particularly mistaken, moreover, because the Nielsen Memorandum shows that the Department, back in 2018, considered the policy issues that the Court today says the Department did not consider.

Now even a court-requested official statement by the agency head is impermissible to provide additional reasons for a prior agency action. Instead, the agency head must make abundantly clear that the new memo is an independent agency action that supersedes the prior agency action.

Kavanaugh’s observation, however, underscores another fascinating potential remedial implication of the Court’s decision: Did the Court just eliminate the lower courts’ practice of remand without vacatur?

As Ron Levin explores in the seminal article on the subject, remand without vacatur is a remedial innovation that has developed in the circuit courts over the last few decades, largely driven by the D.C. Circuit in the 1990s and 2000s. This remedial doctrine allows courts to declare an agency action arbitrary and capricious yet still keep it in place while the agency cures the procedural infirmities on remand. Once the agency has attempted to remedy those procedural errors, challengers can then bring the modified action back to the court for further judicial review. If the agency action returns to court, the agency’s post-remand reasoning and actions are considered part of the administrative record for Chenery I purposes.

In 2014, the Administrative Conference of the United States documented that remand without vacatur has been used more than 70 times by the D.C. Circuit and recommended that, despite that the APA does not expressly provide the remedy, it “should continue to be recognized as within the court’s equitable remedial authority.” In making this recommendation, the Administrative Conference noted that “remand without vacatur is not without controversy. Some scholars argue that it can deprive litigants of relief from unlawful or inadequately reasoned agency decisions, reduce incentives to challenge improper or poorly reasoned agency behavior, promote judicial activism, and allow deviation from legislative directives. Critics have also suggested that it reduces pressure on agencies to comply with APA obligations and to respond to a judicial remand.”

If this remedial device sounds familiar, that’s because it is essentially the remedy Judge Bates utilized in this case, by staying his order vacating the DACA action for 90 days to allow DHS to remedy the procedural errors by providing additional reasons for the DACA rescission. Or as the National Immigration Law Center put it, the district court “‘stayed’ (or paused) its own order for 90 days to allow the government to come up with a better explanation than the one it presented to the court for why it ended DACA.”

If remand without vacatur were a permissible administrative law remedy, the Supreme Court here should have had no trouble considering the nonenforcement policy rationales included in the Nielsen memo as part and parcel of the agency’s decision to rescind DACA. Perhaps, as Judge Bates concluded when the government returned to his court with the Nielsen memo, Roberts may still have found the agency’s reasoning lacking in the Nielsen memo. As noted above, I’m not so sure he would have, and it would have required an even harder look standard. But we can’t know for sure, because Roberts, writing for the Court, rejected the Nielsen memo under Chenery I.

Does this mean that Chenery I, as applied in the DACA rescission case, prohibits remand without vacatur? Roberts certainly does not say so explicitly. Yet it is hard to escape the conclusion in how Chenery I was applied to bar the agency head’s supplemental memo. It will be interesting to see how lower courts (and litigants) interpret Roberts’ opinion when considering whether they can or should remand without vacatur in future cases.

Maybe courts will limit this rejection of remand without vacatur to the unique aspects of this case. After all, as Levin explains, remand without vacatur is most commonly used in the notice-and-comment rulemaking process, perhaps as a form of judicial modesty to not delay the substance of a regulation for perhaps years as the agency goes through another rulemaking process. Here, by contrast, the Supreme Court’s remand with vacatur does not require the DHS to spend years to go through another rulemaking to achieve its purported substantive regulatory objectives. To the contrary, the DHS Secretary could issue a new DACA rescission memo hours or days after the Court’s decision that addresses the two procedural flaws Roberts notes and perhaps also incorporates and expands on the policy rationales included in the Nielsen memo.

In many ways, Roberts’ opinion yesterday in the DACA rescission case is reminiscent of his opinion last Term in the census case (my analysis here). It has the potential to dramatically alter bedrock administrative law doctrines to rein in administrative governance. Yet because in both cases Roberts does not “anchor decisions that disappoint one faction today in generalizable principles,” as Zach Price put it earlier today over at SCOTUSblog, it’s quite possible that Justice Thomas’s hope in the census case becomes true here–that Roberts’ opinion is “a ticket good for this day and this train only.”

That will likely not be up to the Supreme Court to decide. Instead, litigants and lower courts will have opportunities to further develop these principles in the years to come.