[Note from Daniel Hemel: Guest poster David Feder, currently an associate at Munger, Tolles & Olson in Los Angeles and formerly a law clerk to Judge Neil Gorsuch on the Tenth Circuit, has written a thoughtful response to my post last week criticizing Judge Gorsuch’s opinion in Gutierrez-Brizuela v. Lynch. This post reflects the views only of the author and not any employer — past or present. You can reach David at dfeder@jd14.law.harvard.edu and read more of his work here.]

Funny things happen when you let the executive exercise legislative power in a quasi-judicial proceeding. Funnier things happen still when the executive can exercise this legislative power in a quasi-judicial proceeding to overrule, in effect, a binding Article III decision. In a recent pair of thoughtful opinions, Judge Neil M. Gorsuch of the U.S. Court of Appeals for the Tenth Circuit has confronted this thorny problem and pointed the way forward: applying agency rules only prospectively to citizens who relied on binding circuit court precedent.

*

Alfonzo De Niz Robles is a Mexican citizen who unlawfully entered the United States, married a U.S. citizen, and fathered four children. In 1999 he was caught by immigration officials and agreed to depart the country. After he left, his wife tried for two years to obtain for him a spousal visa but had no success. Mr. De Niz Robles then decided his best shot at lawful residency was to return to the United States and apply for a status adjustment.

This course was made possible by a Tenth Circuit decision, Padilla-Caldera v. Gonzales, 426 F.3d 1294 (10th Cir. 2005) (Padilla-Caldera I). Padilla-Caldera Iresolved a statutory conflict between two provisions in our immigration laws. One provision required people like Mr. De Niz Robles — who had previously unlawfully entered the country — to stay outside the country for ten years before applying for admission. Another provision seemed to give the Attorney General discretion to overlook an unlawful entry and adjust residency status without requiring the applicant to leave the country for ten years. Padilla-Caldera I held that the this latter provision trumped the former and so people like Mr. De Niz Robles could apply for status adjustment without departing the country.

Relying upon this decision, Mr. De Niz Robles chose to take his chances with the Attorney General rather than leave the country — and his wife and kids — for a decade. But after Mr. De Niz Robles applied to adjust his status the Board of Immigration Appeals issued In re Briones, 24 I. & N. Dec. 355, 370–71 (B.I.A. 2007), which confronted the same issue as Padilla-Caldera I but instead reached the opposite conclusion. Briones held that applicants like Mr. De Niz Robles were not eligible for adjustment of status unless they left the United States for ten years.

In Padilla-Caldera v. Holder, 637 F.3d 1140 (10th Cir. 2011) (Padilla-Caldera II), the Tenth Circuit held that Chevron deference was owed to Briones and, accordingly, it overturned Padilla-Caldera I and made Briones’s rule the law of the circuit.

In De Niz Robles v. Lynch, 803 F.3d 1165 (10th Cir. 2015), the Board attempted to apply Briones’s rule to Mr. De Niz Robles. Mr. De Niz Robles, of course, had applied to adjust his status before the Board issued Briones and beforePadilla-Caldera II adopted it as the law of the circuit. In other words, the Board now sought to apply a rule created after Mr. De Niz Robles applied to adjust his status that would disadvantage him based on completed conduct he can no longer undo — his decision to stay in the country rather than depart and start the 10 year wait. By the time the court decided his case he would have completed 8 out of those 10 years.

Judge Gorsuch, writing for the court, rejected the Board’s attempt to applyBriones retroactively and instead held that the agency had statutory authority to apply the rule only prospectively. De Niz Robles relied upon and modestly extended Bowen v. Georgetown University Hospital, 488 U.S. 204 (1988). In Bowen, the Supreme Court held that agencies lacked authority to create retroactive legislative rules unless Congress expressly granted them such authority. De Niz Robles reasoned that at Chevron step 2 an agency is, in substance, creating legislative rules (even if it created the rule in what it called an adjudication) and thus Bowen’s limitation should apply there, too. And because Congress had not granted the Board authority to issue retroactive rules, Briones could apply only prospectively — and not to Mr. De Niz Robles.

Along the way, De Niz Robles also considered the retroactivity problem through the framework of SEC v. Chenery Corp., 332 U.S. 194 (1947) (Chenery II), and its progeny. This rubric tells courts to balance the “ill effect” of applying a new rule retroactively against “the mischief of producing a result which is contrary to a statutory design or equitable principles.” Id. at 203. De Niz Robles concluded that ill effect of retroactive application was significant because Mr. De Niz Robles had strong reliance interests: Briones had no legal effect in the circuit unless and until the court determined that underChevron and Brand X the statutory provisions at issue were ambiguous, that the Board’s resolution of the ambiguity was reasonable, and then formally overruled Padilla-Caldera I and made Briones the law of the circuit. Accordingly, litigants could — and were expected — to rely on Padilla-Caldera I notwithstanding Briones. Meanwhile, the harm to the government was relatively minimal — it had not articulated any harm it would suffer from not applying Padilla-Caldera I to Mr. De Niz Robles.

Shortly after the Tenth Circuit decided De Niz Robles, a similar issue came before the court once again. Hugo Gutierrez-Brizuela, like Mr. De Niz Robles before him, applied for adjustment of status relying upon the Tenth Circuit’s decision in Padilla-Caldera I while it remained the law of the circuit. But unlike Mr. De Niz Robles, Mr. Gutierrez-Brizuela petitioned for relief afterthe Board adopted Briones and before the Tenth Circuit adopted it as the law of the circuit in Padilla-Caldera II. The Board asserted that this distinction made all the difference.

But in Gutierrez-Brizuela v. Lynch, №14–9585, 2016 WL 4436309 (10th Cir. Aug. 23, 2016), Judge Gorsuch rejected the Board’s contention. He reasoned that this distinction made no difference because Padilla-Caldera I was still binding precedent when Mr. Gutierrez-Brizuela applied for relief and, unless and until the Tenth Circuit deferred to it, Briones had no legal effect in the circuit. This led naturally to the conclusion that Bowen also supported the panel’s holding because, unless and until the court adopted Briones the law of the circuit, any application of its rule would constitute retroactive application — and again the Board had no statutory authority to create retroactive rules.

*

These pair of opinions have come under some recent criticism. Professor Daniel Hemel, for instance, offers some thoughtful criticisms of the opinions:

(1) most significantly, the opinions are inconsistent with Brand X’s pronouncement that agencies, and not courts, are the authoritative interpreters of statutes;

(2) they require agencies to generate unlawfully a test case so that the court can uphold an agency’s new interpretation prospectively;

(3) in such a test case, the regulated party challenging the agency’s action has little incentive to argue against the agency’s new interpretation;

(4) the lag between the time the agency adopts the new interpretation and the time the court approves the agency’s new interpretation may be several years;

(5) the opinions will lead to legal disarray across the circuits;

(6) and that it is unclear what should happen when a district court, rather than a court of appeal, has read a statute differently than an agency in a rule entitled to Chevron deference.

I take these concerns as well-founded, if a bit overstated. Nonetheless, they seem to focus on only half the problem — the harm to the government — and not the harm to the individual citizen. And the Constitution is concerned primarily with preserving individuals’ liberty, not ensuring the most expedient government structure. In my view, these criticisms illustrate the necessary flaws of any result when we have strayed so far from the Constitution’s original separation of powers rather than prove that the decisions missed their mark. While the Tenth Circuit’s solution to the problem is not perfect, neither is the alternative solution. And when comparing one imperfect solution to another, I think the Tenth Circuit got it right.

The opinions respect Bowen v. Georgetown University Hospital. It may be out of the frying pan and into the fire when it comes to unlawful agency action — a demonstration of the “strange place in our separation-of-powers jurisprudence” in which we now operate. Dep’t of Transp. v. Ass’n of Am. R.R.s., 135 S. Ct. 1225, 1240 (2015) (Thomas, J., concurring). In Bowen, the Supreme Court held that, unless Congress expressly so provides, an agency lacks statutory authority to promulgate retroactive rules in a rulemaking.Bowen’s reasoning logically encompasses retroactive rules like Briones that are adopted under Chevron and Brand X even if in an adjudication. Accordingly, allowing the Board to apply Briones retroactively in De Niz Robles and Gutierrez-Brizuela would run afoul of Bowen’s logic, notwithstanding the fact that the Supreme Court has not formally extendedBowen to agency adjudication.

The starting point here is Congress’s limitations on legislating retroactively. The Supreme Court has explained that Congress may create retroactive statutes but, to do so, it must explicitly declare the statute retroactive.Landgraf v. USI Film Prods., 511 U.S. 244, 268 (1994). In Bowen, the Court held that this same limitation applied when agencies exercise delegated legislative authority in a rulemaking: an agency cannot create retroactive rules unless Congress explicitly grants the agency such authority. (Bowen left open a narrow exception, which I will discuss more below.) Although Bowenaddressed only rulemakings, its logic applies equally to adjudications. The Court’s reasoning focused on the harsh impact of imposing a disability on completed conduct — indeed, it quoted parenthetically Brimstone Rail Co. v. United States, 276 U.S. 104 (1928), which long ago explained that “[t]he power to require readjustments for the past is drastic” and should not be exercised “without very plain words.” Id. at 122.

The key difficulty, of course, is deciding what constitutes impermissible “readjustment[] for the past” and what is permissible application of preexisting law to new circumstances. But wherever the line between these two poles ultimately rests, an agency creating a new rule pursuant toChevron and Brand X has made a new substantive rule and thus should be subject to Bowen’s limitation. See, e.g., Abner S. Greene, Adjudicative Retroactivity in Administrative Law, 1991 Sup. Ct. Rev. 261, 274; William V. Luneburg, Retroactivity and Administrative Rulemaking, 1991 Duke L.J. 106, 152–53. In that situation, an agency makes a policy determination about what the best rule should be (the rulemaking norm) — not a dispassionate adjudicator providing the best reading of what the law is and applying it to the case before him (the adjudicative norm). Indeed, the entire Chevrondoctrine rests on the premise that statutory ambiguity is a delegation from Congress to the agency to “make … policy choices” because ‘[t]he responsibilities for assessing the wisdom of … policy choices and resolving the struggle between competing views of the public interest are not judicial ones.” 467 U.S. 865–66 (emphasis added).

The opinions do not require an agency to take unlawful action. It might not actually be out of the frying pan and into the fire when it comes to unlawfulness. Professor Hemel contends that an agency acts unlawfully when it generates a test case knowing that, if it wins, the rule will be applied only prospectively.

While I might be missing something here, this strikes me as awkward but not unlawful. Even before De Niz Robles and Gutierrez-Brizuela an agency interpretation under Chevron and Brand X sometimes could be applied only prospectively under Chenery II and its progeny. The primary change is that the odds of a new rule applying only prospectively have increased because in many cases the agency will lack statutory authority to create retroactive rules. Even then, however, it’s not guaranteed that a rule will be applied only prospectively — Bowen leaves open a narrow exception for retroactive rulemaking absent statutory authority when there is “some substantial justification for retroactive rulemaking.” 488 U.S. at 208–09.

If the concern instead is the agency picking and choosing its targets and subjecting them to disfavored treatment, I agree that is troublesome but I do not see how it is any more unlawful than other exercises of prosecutorial discretion. In fact, this strikes me as a point in favor of De Niz Robles andGutierrez-Brizuela. As discussed more below, the due process and equal protection concerns of an agency prosecuting an individual are heightened when it comes to making “readjustments for the past” — exactly what the opinions protect against. As between the two problems, I think that changing the rules after the fact is more worrisome than an agency exercising prosecutorial discretion — though both strike me as troubling.

Lastly, while not directly bearing on the unlawfulness point it’s worth noting that the incentives problem can sometimes be mitigated by the agency choosing to litigate against repeat players. That is, parties that deal often with a regulating agency may have a strong interest to challenge an agency’s interpretation of a statute when they will be dealing with it in the future. The power plants in Chevron itself, for example, would have had incentive to challenge the EPA’s interpretation of “stationary source” even if it were to be applied only prospectively. I concede, though, that in some contexts — such as the immigration context — this will be difficult if not impossible because the regulated parties are rarely repeat players.

The opinions serve important due process and equal protection values.Allowing an agency to impose a new rule retroactively against a private party who expressly relies on a binding Article III decision raises serious due process and equal protection problems. To understand why, it’s useful to begin by examining the traditional legislative and adjudicative paradigms and, in particular, the nature of the acts and actors.

Legislation is designed to create new rules of general applicability and, in doing so, to upset settled expectations. Legislation, moreover, is created by explicitly partisan actors who seek to further policy goals. For this reason, there is a long tradition stretching back perhaps to the beginning of the common law itself that legislation operates only prospectively. This affords citizen fair notice of what the law requires (a due process value) and avoids invidious discrimination in applying new penalties to already completed conduct of disfavored groups (an equal protection value).

By contrast, adjudication is designed to resolve past disputes and so necessarily must apply retroactively. But the same due process and equal protection concerns that animate the presumption of prospectivity in the legislative realm exist in adjudication, too. The Constitution mitigates these problems by circumscribing the nature of both the judicial act and actor. Federal judges may only decide live cases or controversies. And federal judges are insulated from political pressures through life tenure and salary guarantees.

Chevron and Brand X complicate these traditional paradigms. They allow an agency to create new law in an adjudication and attempt to apply it retroactively to someone who has acted in reliance on a prior Article III court decision — and thus can no longer alter his conduct to conform with the new law. This raises an obvious notice problem: individuals cannot conform their conduct with the law when they are not — and could not reasonably be aware — of what the law will be. It also raises equal protection problems — the agency can punish disfavored groups for conduct they can no longer alter. These potential problems for private individuals raised by the potent combination of Chevron step 2 and Brand X, to me at least, outweigh any harm to the government.

The opinions respect Brand X. To Professor Hemel, “[p]erhaps the biggest strike against the Tenth Circuit panel’s view … is that it is plainly inconsistent with Brand X” which held that “a court’s opinion as to the best reading of an ambiguous statute an agency is charged with administering is not authoritative,” because “the agency remains the authoritative interpreter … of such statutes.”

But I think this criticism misunderstands what Brand X meant by “authoritative.” Brand X used that term as synonymous with interpretive supremacy. See 545 U.S. at 983 (explaining that the agency is “the authoritative interpreter (within the limits of reason)”). When the Executive and the Judiciary disagree about the meaning of an ambiguity for which it receives Chevron deference, Brand X explains, the Executive’s interpretive views must prevail.

Authoritativeness, however, does not mean self-executing. Consider the relationship between state legislatures and Congress — the latter which has interpretive supremacy by virtue of Article VI’s Supremacy Clause. If a state legislature creates a statute that conflicts with a federal statute then it will have to give way to the authoritative interpreter. But that requires a court judgment — a court must analyze both statutes and determine that the conditions for preemption are satisfied. Likewise, in the Brand X context, an agency is the authoritative interpreter of the statute within certain conditions. And it is not until a court decides that the conditions for deference under Chevron and Brand X are satisfied does the agency’s interpretation become the law of the circuit. The fact that the authoritative interpreter’s judgments are not self-executing in either context does not detract from the fact that their interpretations are authoritative once certain conditions are satisfied.

Implicit in this criticism, moreover, seems to be an assumption that it is a foregone conclusion that a court will defer to the agency interpretation under Chevron and Brand X. But empirical data suggests this is not the case. A recent study by Professors Kent Barnett and Christopher Walker found that circuit courts defer under Chevron in “only” 74.8% of their administrative law cases. Kent H. Barnett & Christopher J. Walker, Chevron in the Circuit Courts, 115 Mich. Law. Rev. __ (2016) (forthcoming). It is difficult to consider an agency to be “authoritative” in the sense that Professor Hemel suggests when its interpretations only carry the day three-quarters of the time.

The opinions promotes stability and higher-quality rules. One criticism of De Niz Robles and Gutierrez-Brizuela is that the opinions make it difficult, cumbersome, and slow to change the law. In my view, however, making it difficult to change the law is not a particularly persuasive reason for rejecting these opinions. Difficulty in changing the law is a feature and not a bug of our Constitution. See John F. Manning, Lawmaking Made Easy, 10 Green Bag 2d 191 (2007). The Constitution requires both bicameralism and presentment in order to discourage “easygoing, high volume lawmaking.” Id.at 198. The tri-cameral system the Founders created, in effect, imposes “a supermajority requirement upon the enactment of legislation” — which “makes it harder even for a majority to enact self-interested or oppressive legislation because it gives political minorities extraordinary blocking power.” Id. at 199. “[T]his favors the status quo and disfavors legislative output.” Id. at 201. Likewise, “bicameralism and presentment might induce legislators to think and deliberate more about legislation, acting less often in hasty response to a momentary public passion.” Id. As Professor Manning concludes, “[t]o say that certain lawmaking practices have merit becausethey lift those burdens is to rest on perceived virtues that are the converse of those implicit in the legislative decision.” Id. at 204. In other words, difficulty in changing the law is an important aspect of our constitutional structure — and one that was designed to serve the same due process and equal protection values that animate De Niz Robles and Gutierrez-Brizuela.

The opinions reduce uncertainty costs for private individuals. Professor Hemel points out that Chevron and Brand X help to resolve in a timely manner splits in the circuit courts and do so without requiring the Supreme Court to weigh in. Under De Niz Robles and Gutierrez-Brizuela, however, until the circuit court approves the agency interpretation disuniformity will exist. As I understand his argument, the primary issues of this are uncertainty as to the state of law and encouraging forum shopping. While all that is true, I think that the reduced uncertainty for the private individual outweighs these costs. In a world where De Niz Robles and Gutierrez-Brizuelacame out the other way, a private party must:

(1) know of Mead;

(2) know of Chevron;

(3) know of Brand X;

(4) guess whether an agency’s interpretation is entitled to Chevron deference under Mead;

(5) guess whether a prior judicial interpretation of a statute decided that its reading unambiguously was compelled by the statute or was instead its best reading of an ambiguity;

(6) and guess whether an agency interpretation will be deemed a reasonable resolution of the statutory ambiguity.

Even agencies seeking Chevron deference in the circuit courts only win about 75% of the time — and they presumably are much more familiar with Mead,Chevron, and Brand X than most individuals. After an individual manages to correctly navigate the gauntlet, there remains always the possibility that the agency will change their mind tomorrow. De Niz Robles and Gutierrez-Brizuela avoids all this uncertainty: unless and until a circuit court adopts an agency interpretation as the law of the circuit the party can confidently rely on binding circuit precedent.

*

Of course, analyzing all the twists and turns and imperfections of choosing between the above courses of conduct risks losing the forest for the trees. The odd situations that De Niz Robles and Gutierrez-Brizuela were forced to confront were the inevitable consequences of Chevron and Brand X. Perhaps the downstream complications of those cases should give us reason to reconsider them — and reconsider returning to the Constitution’s careful allocations of powers between the three branches of government.