Notice & Comment

Chevron’s Immigration Exception, Revisited, by Michael Kagan

My reaction to the Supreme Court’s recent decision in Luna Torres v. Lynch has drawn informative responses from Patrick Glen, and from Asher Steinberg.

The spark for this debate was the Court’s silence with regard to Chevron deference in Luna Torres, even though the case seemed to be a prototypical Chevron case. An agency, specifically the Board of Immigration Appeals, had published an interpretation of a statute that the court acknowledged was ambiguous. If I had been a student confronted with this scenario on an Administrative Law exam, I would have said that the tests in Chevron’s Step 0 and Step 1 were met, and the decisive question should be whether the BIA’s interpretation is reasonable.

But that’s not what the Supreme Court did. Not only did the Court fail to even mention Chevron in its decision, it barely discussed the BIA’s decision at all. It treated the case as a de novo exercise in statutory interpretation, as if deference doctrines in administrative law do not exist, or at least have nothing to do with this case. Moreover, Luna Torres does not appear to be entirely unique, though Glen, Steinberg and I disagree on the extent to which it is part of a trend.

The role of Chevron in immigration cases is an issue on which I am endeavoring to examine in much greater depth for a future article – so I reserve the right to change my mind (and I have already benefitted from my colleagues’ rejoinders). In this spirit, there are two points that I hope to refine here, reflecting my current thinking on the subject. My first point is an observation of a trend – what is happening. The second is my argument about what should happen.

What is happening to Chevron in immigration cases?

As a starting point, I see an important degree of agreement between Glen, Steinberg and myself. Chevron deference seems to be doing very little work in certain kinds of immigration cases at the High Court.It is a challenge to precisely define this subset of immigration cases (more on that later). And there is room for some disagreement about precisely how little work Chevron is doing in these cases. Is it a more limited role, a peripheral role, a background role, or is Chevron entirely irrelevant in these cases? We each have somewhat different answers to these questions. But to borrow a phrase that Steinberg uses, I think all three of us agree that “it’s clear that something’s happening.”

Glen, who is the most hesitant of the three of us to draw conclusions from the Court’s failure to explicitly invoke Chevron, acknowledges, “[T]here may be situations where the Supreme Court is inclined to apply a lesser measure of deference to Board decisions interpreting the criminal grounds of removability.” In a post from a year ago, after the Court decided Mellouli v. Lynch, Glen expressed some sympathy for the possibility that “deportation is different,” such that the “room for deference” is more circumscribed than in other cases.

Steinberg, after a case-by-case examination of recent crImmigration decisions, aptly summarized Chevron’s limited role and the questions that raises about what we thought we knew about judicial review of agency decision-making:

[There has been] total silence [from the Court] on Chevron, in cases where (1) the government asks for deference, (2) the parties hotly debate whether Chevron applies, and (3) the Court rules for the agency without explicitly or even, I would argue, implicitly finding unambiguity. [This] is neither common nor easily explicable.

The Inherent Ambiguity of Silence

It’s one thing to say that something is happening. It is another to define exactly what is happening. That is especially difficult when the Court is acting through omission, declining to explain in its own words what it thinks it is doing, or why.

Glen makes an interesting point about some of the cases in which the Court failed to mention Chevron. He notes that the government did not ask for deference in many of these cases, such as Moncrieffe v. Holder, Carachuri-Rosendo v. Holder, and Gonzales v. Duenas-Alvarez.

However, DOJ most definitely argued for deference in Luna Torres (starting at p. 14 of the brief): “Any ambiguity concerning the meaning of the aggravated-felony provisions in the INA should be resolved based on deference under Chevron.” The government went on to claim that failure to embrace this application of Chevron “would severely undermine the administration of the immigration laws.” And the Court still ignored it. (See also Steinberg’s discussion of Nijhawan v. Holder, where the government also asked for deference and didn’t seem to get it.

The fact that the government did not ask for deference in some cases is a fact that swings both ways. It may explain the Court’s silence on the issue (in some, but not all, cases). But it also reinforces the larger point that deference may be less important than we might have thought it should be. It also undermines DOJ’s argument that Chevron is essential to the administration of immigration law.

Steinberg offers a different explanation for the absence of Chevron in certain cases. He notes that Chevron deference is only owed to BIA decisions that the BIA publishes and designates as precedents, which did not occur in some of the relevant cases. But this, too, is not an explanation for all of the decisions. Consider Carachuri-Rosendo. In that case the BIA issued an en banc, published, precedent decision. And yet, as Glen notes, the government did not ask for deference, and the Court made no mention of the possibility for deference. There was a published BIA decision in Luna Torres as well.

If anything, Steinberg’s observation about published v. unpublished BIA decisions may suggest a broader doctrinal shift. Even when the BIA issues a non-precedential decision, there should still be cause for Skidmore/Mead deference to the agency’s opinion. But the Court doesn’t seem to even do this, since it barely considers the BIA’s reasoning at all in most of these decisions. So, perhaps the Court really is treating some immigration cases as exceptions from administrative law generally. Maybe this isn’t just about Chevron.

In short, at present the doctrine seems to be in flux. It also seems to be mildly incoherent, in that no single theory seems able to completely explain the Court’s treatment (or lack thereof) of Chevron in recent deportation cases. Steinberg does an excellent job working through several possibilities about why the Court has not explained itself more clearly. The main thread is that this is “a tough question,” and the justices may be too unsure or divided to announce a new rule just yet. Instead, they proceed case-by-case.

Maybe It’s About the Statute

One possibility is that Chevron may not play a major role in deportation cases involving dual-use criminal statutes. In short form, immigration law makes aggravated felonies deportable, but the same statutory category is also used in federal sentencing law and to define certain federal criminal offenses. Cases like Moncrieffe and Carachuri-Rosendo involved such dual-use statutes. The idea here is that the Court cannot defer to the BIA’s interpretation of such provisions because to do so would lead to inconsistent applications of the same statutes, when they are used in different contexts. The BIA has no jurisdiction to interpret criminal offenses directly. The theory here is that the BIA can only interpret non-criminal provisions of immigration law, and that deference would only extend to these questions.

This theory is plausible. It would limit the exception from Chevron to a more narrow category of criminal-removal cases. But it would nevertheless acknowledge that a significant part of immigration is an exception from administrative law norms. But is this exception really limited to dual-use statutes? Or does it extend to all criminal grounds of removal – or even to all grounds of removal?

Last year’s decision in Mellouli was an interesting, and to my mind inconclusive, test of the dual-use theory (see also Steinberg, making similar points about Mellouli). The dual-use statute theory predicts that the Court should revert to Chevron when the BIA interprets a statutory ground of removal that is not part of a dual use statute. Mellouli concerned such a ground of removal, namely controlled substance offenses. As the theory predicts, the Court did at least mention Chevron in Mellouli, but it did so only in passing, at the end of its reasoning, in a manner that seems peripheral at best. Chevron again seemed to do little or no real work. As Chris Walker noted at the time, the Court made no effort to fit its reasoning in Mellouli into Chevron’s analytical steps. Mellouli is thus inconclusive. It is certainly possible to see Mellouli as fitting into standard Chevron analysis. But it is also possible to see Chevron as irrelevant to the holding.

Another possible theory posits that Chevron does not apply (or applies with less force) when the BIA is applying a removal ground that incorporates a criminal law provision. Mellouli could be explained this way, in that the relevant provision of the INA incorporated the Controlled Substances Act, over which the BIA has no authority. In a narrow sense, this also seems plausible, but combined with the dual use statute exception it raises the distinct possibility that we are simply failing to see the forest for the trees. Each individual holding of the Court is perhaps explainable by a narrow rule – this one is a dual use statute, this one references the CSA, this one references a body of law outside the BIA’s jurisdiction. But in Section 237 of the INA (the section listing a grounds for deportation), many of the provisions would fall into one of these exceptions, especially provisions describing criminal grounds of removal. The result would be a confusing patchwork. The simpler explanation for these cases is that deference just has much less relevance in deportation cases based on criminal convictions than in other areas of immigration law.

The trouble with the cases we have is that the Court only sometimes mentions Chevron, and even when it does it is far from clear that Chevron changes the analysis at all. And if it doesn’t change the analysis, why do we need it? This doctrinal ambivalence should not come as a surprise. For awhile now, a prominent theme of administrative law scholarship has been that the Court does not apply Chevron with the analytical consistency suggested by its famous Step 0, Step 1, and Step 2 formula. Chevron may only be a canon that the justices apply flexibly in different contexts. Or perhaps Chevron simply defines a flexible space that the Court allocates to agencies so that they can take the lead in setting public policy in certain contexts. Whatever the right description may be, Chevron deference appears to be at best a very flexible doctrine, and it may be a doctrine in decline, given that the justices themselves no longer seem universally committed to it.

Lenity v. Deference

In addition to the question of judicial deference to the BIA, there is a plausible argument that criminal grounds of removal (or at least dual use statutes) should be interpreted through the rule of lenity. Roughly speaking, the rule of lenity in criminal law requires that ambiguous statutes be interpreted narrowly, in favor of the defendant. This possibility overlaps with the deference question, in that if lenity were to be required Chevron deference would be practically meaningless, since the only acceptable interpretation would be the narrowest one available.

The lenity question was raised energetically by Mr. Luna Torres, but the Court did not address it. The question is not likely to go away, however. [Glen has written an excellent essay on the subject (co-authored with Kate E. Stillman).]

While it overlaps, lenity is really a different question than deference. Deference doctrines focus on who should decide an interpretive question. Lenity is about how such questions should be decided. Glen and Stillman do an excellent job illustrating why this matters. In short, if lenity is required, it could be applied by a court, or by the agency. Or, a court could defer to the agency about whether to apply lenity.

I have not actually argued for lenity, though I am not prepared to argue against it, either. But I do think that keeping deference and lenity distinct is important. One way to understand cases like Luna Torres is that the Court is turning away from deference to the BIA, but is not imposing lenity. This explains a number of cases (Nijhawan also comes to mind) in which the government won – the Court rejected the narrowest possible interpretation – and yet there is no sign that the Court deferred to the BIA in the process.

What should the Court do with Chevron in immigration cases?

My view on what the Court should do – at least my current thinking – is quite close to the deportation-is-different theory (a phrase and insight borrowed from Peter Markowitz). Deference may make sense for a case involving someone outside the country asking for permission to enter, but I would suggest it is far more problematic when someone is already in the United States and facing deportation. This distinction follows naturally from a longstanding principle in immigration law, by which a person asking for entry has less of a claim for due process than someone facing deportation. As Markowitz explained, the Roberts Court has pulled back from the traditional view that immigration enforcement is purely civil, seeing it more ambiguously as “quasi-criminal” or a hybrid of civil and criminal law.

My starting assumption is that courts should normally interpret the law. To depart from this norm, there should be a convincing reason. In the Chevron decision, there were two major reasons why deference was justified in certain contexts. One of these is political accountability. If an interpretation of a statute boils down to a policy choice about which we want the electorate to have a say, then it makes sense to defer to an agency that is accountable to voters through the election of the president. The other rationale is technical expertise. If a statutory question is really a technical question on which the agency is more likely than a generalist federal judge to reach the right decision, then it makes sense for a court to defer to the agency.

In addition to political accountability and technical expertise, I would add a third factor: physical liberty. I argue that judicial deference is less likely to be appropriate when an agency’s decision directly implicates a person’s physical liberty. This explains the deportation-is-different phenomenon. In our democracy, review by an independent judiciary is a critical safeguard whenever the government infringes liberty. To ask the judiciary to defer to the executive on how to interpret a law that allows the executive to arrest, detain and expel a person would simply give the executive branch too much unchecked power over individuals. Or, as Dean Kevin Johnson wrote about Mellouli, “the decision once again demonstrates that the Roberts Court will not rubber-stamp the removal decisions of the executive branch, even those involving immigrants convicted of drug-related crimes that the immigration laws target for harsh treatment.”


These factors do not lead to a very compelling case for deference in the case of the BIA, at least in deportation cases. First, the BIA is structured to be insulated from politics, as Steinberg explains. Second, I am skeptical about claims to the BIA’s technical expertise. Immigration law is complicated, but it is law. It is not a matter of engineering, or medical science, or military preparedness. Immigration law questions should be well within the core competence of the federal courts – especially given that immigration cases occupy a substantial part of the dockets of several circuit courts. Moreover, the persistent resource problems that plague administrative immigration adjudication undermine claims that the BIA has any particular advantage over the federal courts.

To me, the best case for deference is that BIA decisions are subject to review by the Attorney General, which offers some theoretical political accountability (although through a mechanism that is used only in exceptional cases). This political accountability is the only virtue I can see for the agency relative to the courts. But it is a virtue only in some cases. For example, I can see an argument for deference in a case like Scialabba v. Cuellar de Osorio, in which the Obama Administration decided to interpret the statute so as to make certain immigrant families wait much longer for a visa. This decision did not lead directly to anyone being detained or deported. If voters disapprove, they can simply pressure the President to change the policy. There is thus less need for the courts to step in. But when the ambiguous question of law concerns deportation, I see far more reason to be worried about deferring to the political branches. The statutory interpretation in such cases leads immediately to arrest, detention and expulsion. These quasi-criminal consequences call for more rigorous judicial review, and less space for the executive branch to operate unchecked.

For these reasons, I am encouraged that the Court seems not to be meaningfully deferring to the BIA in deportation cases. But I acknowledge that is still unclear why the Court is not doing so. And it is entirely possible that the justices themselves are not yet completely sure.