In Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc.,1×1. 467 U.S. 837 (1984). the Supreme Court held that an agency’s interpretation of a statute it administers should prevail if the statute is ambiguous and the agency’s interpretation is reasonable.2×2. See id. at 842–43. In National Cable & Telecommunications Ass’n v. Brand X Internet Services,3×3. 545 U.S. 967 (2005). the Court went a step further and held that, if the Chevron two-step is satisfied, an agency’s interpretation should prevail even when a court has adopted a contrary interpretation in the past.4×4. See id. at 982. In effect, then, Brand X permits agencies to overrule courts when the circumstances are right. This fact invites a question: in a Brand X case, when does an agency’s interpretation displace a court’s and become the applicable law? Recently, in Gutierrez-Brizuela v. Lynch,5×5. 834 F.3d 1142 (10th Cir. 2016). the Tenth Circuit decided that an agency’s interpretation must wait for a court’s say-so — that an agency’s interpretation is not “legally effective” until a court, in deference to the agency, overrules itself.6×6. Id. at 1144; see id. at 1144–45. But the Tenth Circuit’s decision invites its own questions. The most basic, perhaps, is whether the court’s position is tenable, even by its own lights. Put simply, the problem is that the court’s premises appear to be at odds with its conclusion. Because the court’s opinion does not resolve this tension, Gutierrez-Brizuela rests on unsecured foundations.
Gutierrez-Brizuela starts with two hard-to-reconcile provisions of U.S. immigration law. The first, in 8 U.S.C. § 1255, implies that certain people who have illegally reentered the United States can gain lawful residency at any time, provided that the Attorney General grants them adjusted status.7×7. See 8 U.S.C. § 1255(i)(1)–(2) (2012). But the second, in 8 U.S.C. § 1182, states that these same people can’t gain lawful residency unless they first leave the country and wait ten years.8×8. See id. § 1182(a)(9)(C). In 2006, in Padilla-Caldera v. Gonzales9×9. 426 F.3d 1294 (10th Cir. 2005), amended and superseded on reh’g, 453 F.3d 1237 (10th Cir. 2006). (Padilla-Caldera I), the Tenth Circuit attempted to reconcile these provisions. It held that the Attorney General could grant adjusted status to people who would otherwise have to wait.10×10. See id. at 1301. In 2007, however, the Board of Immigration Appeals (BIA) considered the question in In re Briones11×11. 24 I. & N. Dec. 355 (B.I.A. 2007). and reached the opposite conclusion.12×12. See id. at 370–71. In 2011, in Padilla-Caldera v. Holder13×13. 637 F.3d 1140 (10th Cir. 2011). (Padilla-Caldera II), the Tenth Circuit considered the question again. This time, the Tenth Circuit held that the BIA’s decision in Briones was entitled to Chevron deference, and, invoking Brand X, the court adopted the Briones rule and overruled Padilla-Caldera I.14×14. See id. at 1153 (“The BIA’s determination in Briones . . . is a reasonable interpretation . . . to which we owe Chevron deference. As such, the BIA’s interpretation of the statute, not this court’s interpretation in [Padilla-Caldera I], is the authoritative interpretation.”).
As this legal backdrop moved into place, Hugo Gutierrez-Brizuela, a native and citizen of Mexico, found himself forced to choose between these paths to lawful residency.15×15. See In re Gutierrez-Brizuela, No. A200 333 099, 2014 WL 4966418 (B.I.A. Aug. 27, 2014). In 2009 — after the BIA’s ruling in Briones but before the Tenth Circuit’s in Padilla-Caldera II — Gutierrez-Brizuela sought adjusted status.16×16. See id. at *1–2. Then he waited. Finally, in 2013, an immigration judge pretermitted his application for adjustment of status.17×17. See id. at *1. Gutierrez-Brizuela appealed that decision to the BIA, which dismissed his appeal.18×18. See id. at *1–2. The BIA concluded that the Briones rule governed Gutierrez-Brizuela’s application, and that he was therefore ineligible for adjustment of status.19×19. See id. at *2. The BIA rejected Gutierrez-Brizuela’s argument that this was an impermissible retroactive application of Padilla-Caldera II.20×20. See id. The BIA treated Briones as interpretive rather than legislative and found that merely interpretive decisions are “fully retroactive.” Id. (quoting United States v. Rivera-Nevarez, 418 F.3d 1104, 1107 (10th Cir. 2005)). Gutierrez-Brizuela petitioned the Tenth Circuit for review.
The Tenth Circuit granted Gutierrez-Brizuela’s petition for review and remanded the case to the BIA.21×21. See Gutierrez-Brizuela, 834 F.3d at 1149. Writing for the panel, Judge Gorsuch22×22. Judge Gorsuch was joined by Judges McKay and Bacharach. held that Gutierrez-Brizuela was eligible for adjustment of status because Padilla-Caldera I, not Briones or Padilla-Caldera II, governed his 2009 application for adjusted status.23×23. See Gutierrez-Brizuela, 834 F.3d at 1144–45. Judge Gorsuch explained that De Niz Robles v. Lynch,24×24. 803 F.3d 1165 (10th Cir. 2015) (Gorsuch, J.) (unanimous opinion). a year-old Tenth Circuit case, dictated that conclusion.25×25. See Gutierrez-Brizuela, 834 F.3d at 1145. In De Niz Robles, the court had faced a similar question. There, however, the petitioner had applied for adjustment of status before the BIA’s decision in Briones, not after.26×26. See id. at 1143–44 (describing Gutierrez-Brizuela as De Niz Robles with a “twist,” id. at 1144). The question, then, was whether the Briones rule applied to applications filed after Padilla-Caldera I but before Briones. The Tenth Circuit said no: It first determined that the BIA’s decision amounted to an exercise of legislative power and thus, like other exercises of legislative power, should be presumed to have only prospective effect.27×27. See De Niz Robles, 803 F.3d at 1172; see also Gutierrez-Brizuela, 834 F.3d at 1145 (summarizing De Niz Robles). It then found that the BIA’s decision took effect in 2011 when the Tenth Circuit decided Padilla-Caldera II, and thus should be presumed not to apply to the petitioner’s application filed years before.28×28. See De Niz Robles, 803 F.3d at 1174 n.7 (“An agency in the Chevron step two/Brand X scenario may enforce its new policy judgment only with judicial approval. So, for example, the BIA depended on Padilla-Caldera II to render Briones effective.”); see also Gutierrez-Brizuela, 834 F.3d at 1145. The De Niz Robles court concluded, finally, that the presumption was not overcome, because the contrary conclusion would upend the petitioner’s reasonable reliance on Padilla-Caldera I and gain the government little.29×29. See De Niz Robles, 803 F.3d at 1177–80.
In light of “both the rule and the reasoning of De Niz Robles,”30×30. Gutierrez-Brizuela, 834 F.3d at 1145. Judge Gorsuch concluded that Gutierrez-Brizuela had been eligible for adjustment of status when he applied in 2009.31×31. See id. at 1144–49. In the court’s view, the rule’s upshot was clear: because Briones did not “take legal effect” until 2011,32×32. Id. at 1148. its application to Gutierrez-Brizuela would be retroactive.33×33. See id. at 1145. The reasoning’s upshot was clear too: because the BIA’s decision was deemed an exercise of legislative power, and thus presumed not to have retroactive effect, it could govern Gutierrez-Brizuela’s application only if the “costs the petitioner would encounter” were outweighed by “the benefits the agency would enjoy.”34×34. Id. at 1147 (citing and applying the tests developed in SEC v. Chenery Corp., 332 U.S. 194 (1947), and Stewart Capital Corp. v. Andrus, 701 F.2d 846 (10th Cir. 1983)). Judge Gorsuch found that in Gutierrez-Brizuela, as in De Niz Robles, the “balance tips decidedly” toward the petitioner.35×35. Id. Gutierrez-Brizuela had important interests on the line, and, in his attempt to secure them, he reasonably relied on on-point judicial precedent.36×36. See id. It would be unfair to upset his expectations now and to create substantial uncertainty for people in similar positions.37×37. See id. Finally, Judge Gorsuch rejected the government’s argument that the court’s decision would deny Brand X any practical effect in the future.38×38. See id. at 1148 (comparing Brand X litigation to litigation over qualified immunity, where courts can apply one rule to the parties while in effect announcing another for future litigants).
Judge Gorsuch also wrote a concurring opinion.39×39. Id. at 1149 (Gorsuch, J., concurring). In it, he addressed what he saw as the “elephant in the room”40×40. Id. — the possibility that Gutierrez-Brizuela illustrates fundamental constitutional problems with Chevron and Brand X.41×41. See id. Judge Gorsuch first observed that, in Brand X scenarios, executive agencies seem to use legislative power to perform a quasi-judicial function “in a way that seems more than a little difficult to square with the Constitution of the framers’ design.”42×42. Id. In Judge Gorsuch’s view, however, Brand X is just a symptom: after all, Brand X “follow[s] pretty naturally” from Chevron.43×43. Id. at 1151. And this, he argued, “brings the colossus now fully into view.”44×44. Id. If Chevron gives executive agencies the power to interpret law, it “seems no less than a judge-made doctrine for the abdication of the judicial duty,”45×45. Id. at 1152. a doctrine with no foundation in the Administrative Procedure Act or the Constitution.46×46. See id. at 1151. If instead (or in addition) Chevron gives executive agencies the power to make law, it is far from clear that Congress has delegated that power — or even that it could.47×47. See id. at 1152–55. Judge Gorsuch argued that courts recognize these problems in the criminal context and thus do not defer to agencies; he did not see how the same arguments could fail to carry over to the civil context. See id. at 1155–57. The conclusion, in Judge Gorsuch’s view, is that Chevron is incompatible with the constitutional division of powers.48×48. See id. at 1155. What’s more, he noted, Chevron has proven difficult for courts to implement, and few people’s interests depend on it remaining in place.49×49. See id. at 1157–58. Given all this, Judge Gorsuch asked, “what would happen in a world without Chevron?”50×50. Id. at 1158. In practice, he observed, maybe not much — but in principle, he argued, a return to pre-Chevron administrative law would restore the proper constitutional arrangement and the rule-of-law protections it was meant to secure.51×51. See id.
Gutierrez-Brizuela is noteworthy for several reasons: It appears to put the Tenth Circuit at odds with its sibling circuits52×52. In many Brand X cases, courts simply apply the agency’s new rule to the litigants without comment. See, e.g., Fernandez v. Keisler, 502 F.3d 337, 347–48 (4th Cir. 2007) (adopting the Department of Homeland Security’s 2003 rule in place of the court’s 1996 precedent and applying it to the parties); Dominion Energy Brayton Point, LLC v. Johnson, 443 F.3d 12, 16–17 (1st Cir. 2006) (adopting EPA’s 2000 rule in place of the court’s 1978 precedent and applying it to the parties). But see Garfias-Rodriguez v. Holder, 702 F.3d 504, 514–20 (9th Cir. 2012) (en banc) (grappling with whether agencies or courts change the law in Brand X scenarios and how that determination bears on questions about retroactivity). and perhaps with Brand X itself.53×53. See Daniel Hemel, The Tenth Circuit vs. Brand X, Yale J. on Reg.: Notice & Comment (Aug. 24, 2016), http://yalejreg.com/nc/the-tenth-circuit-vs-brand-x [https://perma.cc/R5UQ-Z266] (arguing that the court’s position is incompatible with the view that agencies are authoritative interpreters in Brand X scenarios). But see David Feder, Founders Meet Brand X, Yale J. on Reg.: Notice & Comment (Aug. 31, 2016), http://yalejreg.com/nc/founders-meet-brand-x-by-david-feder [https://perma.cc/SV8U-EBDW] (arguing that the Tenth Circuit’s view is true to Brand X because an agency’s interpretation can be “authoritative” without being “self-executing”). It leaves unanswered important questions about how agencies can administer the law in the future.54×54. See, e.g., Hemel, supra note 53 (explaining several obstacles to implementing the Tenth Circuit’s position, including that agency rules might be delayed many years; that litigants might have odd incentives; and that agencies might have to act unlawfully to prompt litigation). And of course it’s striking for the fact that Judge Gorsuch, author of the court’s opinion, wrote a separate concurrence to argue that Chevron should go.55×55. Commentators noticed. See, e.g., Alison Frankel, Is Court Deference to Federal Agencies Unconstitutional? 10th Circuit Judge Thinks So, Reuters: On the Case (Aug. 24, 2016), http://blogs.reuters.com/alison-frankel/2016/08/24/is-court-deference-to-federal-agencies-unconstitutional-10th-circuit-judge-thinks-so [https://perma.cc/D943-GKXF]. Given all that, however, Gutierrez-Brizuela is also noteworthy because it’s not clear that the court’s position is tenable, even by its own lights. The issue is that the court’s premises — that in a Brand X scenario an agency’s interpretation should be regarded as a legislative act and should be subject to (at least some) legislative presumptions — appear to be at odds with its conclusion that the BIA’s interpretation needs court approval to take effect, because the court’s own analogy to legislation suggests otherwise. It is unclear how the court might resolve this tension, but the court’s concerns with due process and equal protection indicate where it might start: the court might redeploy its concerns about fair notice to show that Briones should have only delayed effect. This can’t be where the court stops, however, because concerns about notice do not adequately distinguish Brand X scenarios from more familiar scenarios where a court’s say-so isn’t needed. As a result, Gutierrez-Brizuela’s foundations won’t be secure unless the court says more.
Start with the court’s premises. First, in both Gutierrez-Brizuela and De Niz Robles, the court took the position that the BIA’s decision, an administrative adjudication, was tantamount to an exercise of delegated legislative power.56×56. See Gutierrez-Brizuela, 834 F.3d at 1146 n.1 (“The fact is, the agency rule a court ratifies in the Chevron step two/Brand X scenario clearly proceeds from the agency exercising delegated legislative authority, not the court.”); De Niz Robles v. Lynch, 803 F.3d 1165, 1171–75 (10th Cir. 2015) (“[C]ourts defer to the agency’s new view [announced in adjudication] because the agency has been authorized to fill gaps in statutory law with its own policy judgments. Form, then, can’t obscure the fact that an agency exercising its Chevron step two/Brand X powers acts in substance . . . like a legislative actor making new policy . . . .” Id. at 1173.). Here the court stood on firm ground. The claim that agencies exercise delegated rulemaking authority akin to (if not quite) delegated legislative power is central to standard accounts of Chevron.57×57. See, e.g., Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843–44 (1984); United States v. Mead Corp., 533 U.S. 218, 226–27 (2001); see also Kathryn A. Watts, Rulemaking as Legislating, 103 Geo. L.J. 1003, 1005–08 (2015) (noting the Court’s official view that Congress empowers agencies to make rules with the force of law without technically delegating legislative power, but concluding that “Chevron . . . implicitly . . . recognizes that agency rulemaking constitutes an exercise of delegated legislative power,” id. at 1008); Adrian Vermeule, No, 93 Tex. L. Rev. 1547, 1556 (2015) (reviewing Philip Hamburger, Is Administrative Law Unlawful? (2014)) (stating that the “official theory” of Chevron turns on delegation). Second, the court took the position that, because the BIA’s decision should be regarded as an exercise of delegated legislative power, “logic suggests it should be bound by the same presumption of prospectivity that attends true legislative enactments.”58×58. Gutierrez-Brizuela, 834 F.3d at 1145; see also De Niz Robles, 803 F.3d at 1172 (drawing on the Supreme Court’s own analogy to legislation in Bowen v. Georgetown University Hospital, 488 U.S. 204, 208 (1988)). So far, so plausible. Courts deploy presumptions for good reasons,59×59. See, e.g., De Niz Robles, 803 F.3d at 1174–75 (explaining the reasons for the presumption against retroactivity). and it makes sense enough to extend them to “functionally equivalent” acts, absent some reason not to.60×60. Abner S. Greene, Adjudicative Retroactivity in Administrative Law, 1991 Sup. Ct. Rev. 261, 274 (arguing that the presumption against retroactive legislation should apply when agencies do the functional equivalent of legislating). Given these premises, the court concluded that the BIA’s 2007 decision must be presumed to have effect only going forward — forward, recall, from 2011.
Notice, however, that the court’s premises might have led it to another date. The court focused on the presumption against giving legislative acts retroactive effect. But courts also tend to presume that laws61×61. See, e.g., Gozlon-Peretz v. United States, 498 U.S. 395, 404 (1991) (unanimous opinion) (“It is well established that, absent a clear direction by Congress to the contrary, a law takes effect on the date of its enactment.”). and agency rules62×62. See, e.g., Administrative Procedure Act, 5 U.S.C. § 553(d) (2012) (implying that agency rules can generally take effect thirty days after they are published in the Federal Register); United States v. Gavrilovic, 551 F.2d 1099, 1106 (8th Cir. 1977) (concluding that, absent good reason for a rule to take effect on the day it was published, it took effect thirty days later). have immediate (or near-immediate) effect, without a court’s preapproval. Wouldn’t “logic [also then] suggest” that the BIA’s decision should be presumed to take effect when (or soon after) it was announced in 2007? The court’s position requires it to say no.63×63. Or to say the presumption is overcome. But given the court’s own analogy to legislative action, this answer calls for explanation. To make the point vivid, imagine that Congress, not the BIA, reconciled the immigration provisions in 2007. In that case, there is little doubt that the Tenth Circuit would have applied Congress’s 2007 rule to Gutierrez-Brizuela’s 2009 application for adjusted status. What’s different here?
The court doesn’t address the question head on, but there is at least one answer implicit in its opinion. The court’s principal concern is that people lack fair notice in Brand X scenarios — that they rely on on-point judicial precedent, only to have the agency change the rules on them.64×64. See Gutierrez-Brizuela, 834 F.3d at 1147. Perhaps this concern can be redeployed to show that an agency’s decision should also wait for the court’s approval. In short, the argument might be that an agency’s decision can’t have immediate effect in a Brand X scenario because that decision’s fate is too uncertain to provide those subject to it with fair notice.65×65. Cf. id. (“Are we really to expect that persons [in Brand X scenarios] should have to bear the cost of ignoring directly controlling judicial precedent in favor of the speculative possibility that an executive revision might ultimately prevail?”). James Dawson articulates a similar worry. See James Dawson, Note, Retroactivity Analysis After Brand X, 31 Yale J. on Reg. 219, 248 (2014) (“The case for applying [the agency’s interpretation] to litigants who acted between [the agency’s decision] and [the court’s approval] is admittedly stronger, since those who acted during this period should have been on notice that . . . [the agency’s interpretation] might soon become the law of the circuit under Brand X. Still, [the agency’s interpretation] should not attach to conduct [until the court weighs in], since litigants could not be sure whether the court . . . would ratify the agency’s decision . . . .”). Dawson takes aim at the rule’s retroactive application rather than its effective date because he concludes that the court changes the law, given that its “ongoing role as gatekeeper” enables it to “reject any agency interpretation that it finds unreasonable.” Id. at 223. But this argument must grapple with the same stubborn fact noted below: that courts play a similar gatekeeping role when Congress acts. If the agency’s decision took effect the day it was announced instead, courts would effectively demand that people read the inscrutable writing on the wall: that they decide whether the on-point precedent foreclosed Chevron step two; and if it didn’t, whether the agency’s interpretation is reasonable.66×66. See Feder, supra note 53 (explaining the steps and the resulting uncertainty in detail). As the Tenth Circuit notes, that’s a lot to ask.67×67. See Gutierrez-Brizuela, 834 F.3d at 1147 (observing that these “[q]uestions . . . frequently provoke deep disagreement even among our most eminent jurists”).
The problem, however, is that this argument seems to work only by making the familiar strange. Congress’s decisions have uncertain fates too: courts can decide years after the fact that Congress exceeded its limited powers.68×68. See, e.g., Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177–80 (1803) (invalidating a provision of the Judiciary Act of 1789). (The same is true, of course, when agencies engage in rulemaking outside Brand X scenarios.) But this long-lingering uncertainty is not thought to undercut fair notice — or to require legislation to wait for a court’s approval. This might be because it’s easier for people like Gutierrez-Brizuela to tell whether a new statute or the existing law — including judicial precedent — will prevail in the end. But that seems unlikely. The writing on that wall is often inscrutable too.69×69. To name just one recent example, NFIB v. Sebelius, 132 S. Ct. 2566 (2012), provoked deep disagreement among lawyers, law professors, judges, and the Justices themselves. Rather, this long-lingering uncertainty is compatible with fair notice because the law doesn’t expect people to see what the future holds; instead it expects them to comply with the announced rule unless courts tell them otherwise. Gutierrez-Brizuela doesn’t explain why the law wouldn’t deal with agency rulemaking in the same way, even in Brand X scenarios.
That leaves an unresolved tension at the center of Gutierrez-Brizuela, one that puts its holding in some doubt. That’s not to say the opinion’s foundations could not be secured: for example, the court might yet locate some crucial difference between agency rulemaking and congressional lawmaking. In that spirit, it’s worth noting that the court stated that due process and equal protection concerns “animated” its holdings,70×70. Gutierrez-Brizuela, 834 F.3d at 1146. so perhaps it should be understood to have broader rule-of-law concerns in mind, despite its focus on fair notice. If so, the court might even share some of the concerns Judge Gorsuch articulated in concurrence. And these concerns highlight a difference worth considering: agency rulemaking and congressional lawmaking occupy different places in the constitutional order, however alike their functions. For example, the constitutional order makes lawmaking difficult by design, in part to secure due process and equal protection.71×71. See id. at 1151 (Gorsuch, J., concurring) (citing John F. Manning, Lawmaking Made Easy, 10 Green Bag 2d 191, 202 (2007)). Perhaps that fact could permit a court to give an agency’s rule delayed effect in a Brand X scenario, even if the court took the legislative analogy quite seriously.72×72. Note, for instance, that courts strive to find duly enacted laws legitimate. See, e.g., Mistretta v. United States, 488 U.S. 361, 384 (1989) (“When this Court is asked to invalidate a [law passed] by both Houses of the Congress and signed by the President, . . . it should only do so for the most compelling constitutional reasons.” (quoting Bowsher v. Synar, 478 U.S. 714, 736 (1986) (Stevens, J., concurring))). Perhaps courts do not extend the same presumption of legitimacy to agency rules. Perhaps. But the argument from that premise to the court’s conclusion is far from obvious. And if Judge Gorsuch’s concurring opinion is any guide, it would be controversial too. To resolve the tension in Gutierrez-Brizuela, then, the court must face two questions. Can its holding be made secure? And given what that might take, should it be?