Notice & Comment

Judge Gorsuch and Chevron Doctrine Part II: The Misuse of Precedent, by Asher Steinberg

This is part two of a three-part series on Judge Gorsuch.

Does Judge Gorsuch care about precedent? The question might seem like asking if Judge Gorsuch cares about stray kittens – of course he cares about precedent. After reading his administrative-law opinions, though, one can wonder.

Padilla-Caldera II

In Padilla-Caldera v. Gonzales (“Padilla-Caldera I”), the Tenth Circuit held that illegal reentrants like Padilla-Caldera were eligible for adjustment of status to lawful permanent residency and ordered the Board of Immigration Appeals to conduct further proceedings on his application for adjustment. Then in Padilla-Caldera v. Holder (“Padilla-Caldera II”), the Tenth Circuit deferred under Chevron to the BIA’s subsequent contrary decision in In re Briones, overruled Padilla-Caldera I under Brand X, and, retroactively applying Briones to Padilla-Caldera notwithstanding the court’s prior mandate to the BIA in Padilla-Caldera’s own case, held Padilla-Caldera was ineligible for adjustment of status. In De Niz Robles v. Lynch, however, Judge Gorsuch held that Briones did not apply retroactively to De Niz Robles and that Padilla-Caldera I controlled De Niz Robles’ pre-Briones application for adjustment—even though Padilla-Caldera II held that Padilla-Caldera I didn’t even control Padilla-Caldera’s own pre-Briones application for adjustment. Finally, in Gutierrez-Brizuela v. Lynch, Judge Gorsuch held that Briones didn’t control postBriones applications for adjustment filed before the Tenth Circuit deferred to Briones in Padilla-Caldera II, even though Padilla-Caldera’s application was obviously filed before Padilla-Caldera II. In sum, by the time Judge Gorsuch was through with Padilla-Caldera II, the only pre-Padilla-Caldera II applicant for adjustment who still lost the benefit of Padilla-Caldera I was Padilla-Caldera himself in Padilla-Caldera II. Judge Gorsuch did not, as circuit rules allow, circulate these opinions to the en banc court before overruling Padilla-Caldera II, so what gives?

One answer might be that Padilla-Caldera II is distinguishable because Padilla-Caldera, unlike De Niz Robles and Gutierrez-Brizuela, didn’t rely on Padilla-Caldera I. But while both of Gorsuch’s opinions and confirmation-hearing commentary thereon talk a lot about reliance, Gutierrez-Brizuela makes it quite clear that neither is limited to reliant applicants; there Gorsuch writes that “De Niz Robles held that Briones was not legally effective in the Tenth Circuit until [Padilla-Caldera II],” does not “apply . . . to conduct in this circuit that predates Padilla-Caldera II,” and that Briones was non-retroactive purely “because . . . [it was] an exercise of delegated legislative policymaking authority.” Besides, Gorsuch’s Charlie Brown-and-Lucy arguments for reliance in Gutierrez-Brizuela are as unserious as they sound. Gutierrez-Brizuela, after all, applied for adjustment after Briones was decided, during a period when four circuits unanimously deferred to Briones’s abundantly reasonable (and probably unambiguously correct) interpretation of §§ 1182 and 1255—making Padilla-Caldera I a virtual lock for a Brand X overhaul, as would indeed happen in Padilla-Caldera II.

Another answer could be that while under Judge Gorsuch’s rule, “Chevron/Brand X adjudications” only apply prospectively, there has to be an exception for the case, like Padilla-Caldera II, in which a court decides whether to defer to those adjudications; otherwise, there would be no controversy before the court in such a case as the adjudication on review wouldn’t potentially apply to the parties. But it turns out that Judge Gorsuch rejected just this argument in Gutierrez-Brizuela, claiming that courts announce purely prospective rules in qualified-immunity cases and could likewise decide whether to defer to a rule announced in a “Brand X adjudication” even if retroactive application of that rule were off the table.

Not only was no distinction available to Judge Gorsuch given the breadth of his rule, Judge Gorsuch actually copped to overruling Padilla-Caldera II’s retroactivity holding in Gutierrez-Brizuela. There, in his concurring opinion, after attacking Padilla-Caldera II’s “unconstitutional” result at some length, he took a step back to survey his handiwork:

[S]ince Padilla-Caldera we have reentered the field and sought to tame some of Brand X‘s more exuberant consequences. So, for example, in De Niz Robles and now again today we have held that an agency’s revision of a judicial decision of what the law is may bear only prospective effect, governing only future cases and controversies. As a result, an executive agency may no longer revise a judicial decision about the law’s meaning with retroactive effect, like the BIA managed to do in the case of Mr. Padilla-Caldera.

I find this passage stunning. How, absent an en banc overruling that never occurred, can the BIA “no longer” be allowed to do what Padilla-Caldera II held it could—retroactively revise Padilla-Caldera I notwithstanding a mandate on the very point at issue? Judge Gorsuch seems to want to suggest, without calling too much attention to the point by forthrightly saying so, that Padilla-Caldera II didn’t quite hold the BIA could do it; the BIA just “managed” to do it. But had he attempted to seriously make that argument, it wouldn’t have withstood a moment’s scrutiny; Padilla-Caldera unsurprisingly argued in Padilla-Caldera II that whatever the consequences of Briones for other immigrants, the BIA at least had to apply Padilla-Caldera I to Padilla-Caldera, and the court spent four pages of its opinion rejecting the argument. The conclusion can’t be escaped: in holding the BIA couldn’t retroactively revise Padilla-Caldera I, Judge Gorsuch unlawfully revised Padilla-Caldera II.

Chenery II and Retail, Wholesale

In addressing Briones’ retroactive effect, Judge Gorsuch didn’t write on a doctrinal blank slate, anymore than he wrote on a precedential blank slate. Since the Supreme Court’s foundational 1947 decision in Chenery II, and certainly since the D.C. Circuit’s 1972 refinement of Chenery II in Retail, Wholesale, federal courts (including the Tenth Circuit) have applied a balancing test for agency-adjudication retroactivity under which agency adjudications are presumptively retroactive absent “manifest injustice” to relying parties. In De Niz Robles and Gutierrez-Brizuela, however, Judge Gorsuch equated agency adjudication for retroactivity purposes with agency regulation, which is presumptively prospective, and concluded that “legislative” agency adjudication is presumptively non-retroactive. Can that equation be squared with Chenery II and Retail, Wholesale? Obviously not.

To be sure, Judge Gorsuch took great pains to show that De Niz Robles and Gutierrez-Brizuela’s specific results were required by Retail, Wholesale, given both petitioners’ reliance on Padilla-Caldera I—and insofar as results go, he’s probably right about De Niz Robles. But as shown above, his rule isn’t limited to reliant parties; it simply forbids agencies from “revis[ing] a judicial decision about the law’s meaning with retroactive effect,” whether as to people who reasonably relied on the decision like De Niz Robles, unreasonably like Gutierrez-Brizuela, or who acted before it was issued or didn’t know it existed. And depending on how one reads the slash in De Niz Robles’ poorly phrased holding that “a new agency rule announced in a Chevron step two/Brand X adjudication” is presumptively non-retroactive, his rule may apply to any precedential agency adjudication interpreting an ambiguous statute as applied to anybody. In either event, neither reading can remotely be squared with Retail, Wholesale’s presumption of retroactivity absent manifest injustice to particular parties.

In disobeying the Court and his Circuit’s framework for agency-adjudication retroactivity while purporting to follow it, Judge Gorsuch engages in more than a bit of sleight of hand. When he first announces his equation between agency adjudication and regulation in De Niz Robles, he claims to be “join[ing] those” who have suggested it, as if it weren’t a novelty. Who are “those?” He first quotes a throwaway line in a Seventh Circuit opinion that appeared to make the same equation, but then proceeded outside of Gorsuch’s quotation to orthodoxly apply the presumption in favor of agency-adjudicative retroactivity—which of course sharply diverges from the presumption against regulatory retroactivity. He then cites a 1991 law review article by Abner Greene, which appears to have been the real source of Gorsuch’s rule and the underlying ingenious argument that all Chevron step two adjudications are necessarily non-retroactive legislation.   What Judge Gorsuch fails to disclose, however, is that Greene accurately and candidly described his proposal as a rejection of “Chenery II’s broad deference to administrative adjudicative retroactivity” that was “at odds with the Chenery II approach,” and credited himself with discovering “the seed of Chenery II’s undoing.”

One would never know, of course, from Judge Gorsuch’s opinions that he made the logical opposite of Chenery II the law of his circuit; rather, Judge Gorsuch claims he’s just found a better way of doing Chenery II. Chenery II’s test, Judge Gorsuch complains, “is like asking us to compare the weight of a stone to the length of a line”; Retail, Wholesale is “an elaborate five-factor ‘balancing’ test” whose factors aren’t “exclusive or even always the most pertinent.” Both, he assures us, get to just the same place as his per se rule, which shouldn’t “come as a surprise,” as both tests merely “provide a doctrinal rubric for assessing the underlying due process and equal protection implications” that Judge Gorsuch’s rule preferably “analyze[s] . . . directly” rather than “indirectly (using these doctrinal rubrics).” Either way, “the answer should of course be the same.” Of course it’s not, though, and the neat trick he pulls to make it look as if it were is constantly shifting feet between the results of his decisions—which happened to be justifiable under Chenery II because of the reliance interests involved—and the rule he’s announcing, which throws Chenery II’s presumption of agency-adjudicative retroactivity in the garbage can.

NLRB v. Wyman-Gordon

However one reads De Niz Robles’ rule on “Chevron step two/Brand X adjudications,” the rule requires agency adjudicators to engage in purely prospective adjudication in a large class of cases. Read most narrowly, any agency adjudicator that disagrees with prior judicial interpretations of ambiguous law is acting as a legislator and cannot apply its decisions retroactively to the parties before it. Read more coherently, any agency adjudicator that makes law at Chevron step two is acting as a legislator on Judge Gorsuch’s logic and cannot apply its decisions retroactively.

In Wyman-Gordon, a plurality of the Court (in agreement with several dissenters) held that the Administrative Procedure Act forbids agencies from “making rules in the course of adjudicatory proceedings”; the precise way in which the NLRB fell afoul of the APA in this respect, the Court said, was that it “did not even apply the rule it made to the parties in the adjudicatory proceeding, the only entities that could properly be subject to the order in that case. Instead, the Board purported to make a rule . . . .” Under Judge Gorsuch’s rule, however, agency adjudicators, were they to continue adjudicating, would purport to make all sorts of rules. How does Judge Gorsuch escape the conclusion that he’s either commanding agency adjudicators to disobey Wyman-Gordon and the APA, or putting agency adjudicators out of business in cases that involve prior judicial precedent or ambiguous statutes?

Judge Gorsuch’s handling of Wyman-Gordon is painful to describe. In a footnote, he opens with the squirrely salvo that “[i]n this case . . . there is no allegation that the BIA is bound by the APA or that it is seeking to evade its dictates,” which carefully stops short of saying the APA doesn’t apply. Whether the BIA was bound by or evading the APA was not a subject for waivable “allegation”; no one alleged anything of the kind because the BIA hadn’t attempted to make prospective rules by adjudication on a regular basis. The only person suggesting that was Judge Gorsuch. Rather, whether the nation’s most active agency adjudicator was bound by the APA or not was a question of law for Judge Gorsuch to look into lest he command it to violate the APA. Had he risked looking, he would have found that in 2011 the Supreme Court set aside a BIA decision under § 706 of the APA without even inquiring into whether the APA was applicable; that then-Judge Breyer once entertained an argument to set aside a BIA decision under Wyman-Gordon and only rejected it on factual grounds; and that the Ninth Circuit has even set aside prospective BIA adjudications under Wyman-Gordon. In any event, Judge Gorsuch’s rule, which applies to all agencies, commands violation of Wyman-Gordon at least as to whatever federal agency adjudicators the APA does bind (which is all of them).

Having wished the APA away, Judge Gorsuch next suggested that Wyman-Gordon, in the view of “many courts,” only forbids prospective agency adjudication in cases of “APA subterfuge.” His citations for that proposition were: (a) Greene’s proposal to overhaul Chenery II, which, like Judge Gorsuch, had to argue away an intractable Wyman-Gordon problem; (b) two cases that involved hybrid adjudications/notice-and-comment rulemakings by FERC, and (c) one case that uncontroversially held Wyman-Gordon allowed an agency to announce a retroactive rule in adjudication and subsequently decline, under Retail, Wholesale, to apply it retroactively to certain parties that relied on a prior rule. No one before Gorsuch, with the exception of Greene, has ever suggested that agencies can engage in whole swaths of prospective adjudications and not apply the rules they announce to the parties before them; as Justice Scalia wrote in Bowen, Wyman-Gordon made it fairly clear “that [agency] adjudication could not be purely prospective, since otherwise it would constitute rulemaking.”


If you want to narrow a precedent while seeding doubts about its legality, it helps to mischaracterize it as being as absurdly overbroad as possible. In denying agencies’ Chevron/Brand X adjudications retroactive effect, and thus drastically limiting agencies’ Chevron powers at least along temporal lines, Judge Gorsuch’s strategy was to grossly exaggerate the discretion Chevron gives agencies. For example, Judge Gorsuch will repeatedly claim that Briones involved “two conflicting statutes,” and that the BIA used its Chevron powers to decide, purely “as a matter of policy discretion,” which one it wanted to “do[] the trumping”; such unfettered discretion could only be described as legislative and could only apply prospectively. And Judge Gorsuch will also claim that this is just the sort of thing Chevron authorizes agencies to do.

In reality, as my last post explained, Briones, as recounted by the Tenth Circuit’s decision in Padilla-Caldera II, said that both statutes clearly mandated the same thing in tandem. But factual niggling aside, Chevron doesn’t even suggest in theory that agencies can or do pick and choose which of two contradictory statutes to follow. As Evan Criddle points out, this is an entirely implausible account of Chevron. It’s also a precedentially foreclosed account. Just one year before Judge Gorsuch wrote De Niz Robles, a three-Justice plurality of the Court suggested the BIA could resolve a statutory tension, if not an out-and-out contradiction, between two divergent clauses of a statute; the Chief Justice and Scalia, concurring in the judgment on narrower grounds, emphatically denied that Chevron was “a license for an agency” to resolve a conflict between statutory provisions. That’s a Marks holding. Chevron may have authorized the BIA to resolve whatever latent ambiguity there was in the two statutes at issue; Chevron didn’t authorize the BIA to choose which of two statutes it wanted to be bound by.

Brand X

Daniel Hemel has devastatingly explained in this space how Gutierrez-Brizuela disobeys Brand X; though Judge Gorsuch has defenders on this score, he hardly helped matters when he boasted that his opinions in De Niz Robles and Gutierrez-Brizuela “sought to tame some of Brand X’s more exuberant consequences,” as if taming decade-old Supreme Court precedent were a job for a lower court. I have only one addendum: the transformation of Brand X into an unconstitutional advisory-opinion engine.

Gutierrez-Brizuela holds that, Padilla-Caldera II and several other circuit precedents notwithstanding, a court overruling its precedent under Brand X can’t retroactively apply its decision to the parties before it. What business, then, does the court have hearing a Brand X case at all? Its opinion will necessarily be advisory as to the parties; if the agency follows Judge Gorsuch’s rule and doesn’t retroactively apply its decision, the loser will have no standing to challenge it, whereas if the agency violates the rule in order to obtain judicial blessing of its decision, the loser will have standing, but no stake whatsoever in arguing the agency’s decision doesn’t deserve deference. Some parties may, as David Feder argues, be repeat players who care about prospective applications, but others, e.g., individual immigrants seeking review of BIA decisions, won’t be.

What does Judge Gorsuch have to say about all this? That the advisory opinion-giving he contemplates isn’t “entirely foreign” to the law, given that in qualified-immunity cases courts will sometimes announce a constitutional rule and not apply it to the parties for lack of prior clear settlement. This analogy is farcically inapposite. Every § 1983 plaintiff argues that the rule he claims the defendant violated is clearly established and should apply in his case; it’s never preordained that the rule he claims was violated won’t apply in his case. A court hearing such cases, then, isn’t engaged in an inevitably advisory enterprise where some or all parties lack standing to argue academic propositions to the court. In Judge Gorsuch’s through-the-looking-glass take on Brand X, though, a court reviewing an agency decision that diverged from prior precedent would know that nothing it could say could ever affect the parties, who would essentially function as unwilling amici, compelled by Judge Gorsuch’s attempt to overrule Brand X from below to participate in an unconstitutional exercise. It is sad that Judge Gorsuch would destroy real Article III values in order to save fake ones.

Asher Steinberg writes The Narrowest Grounds, a law blog. He can be reached at


Print Friendly, PDF & Email