Harvard Law Review Harvard Law Review Harvard Law Review

Administrative Law

Canada (Canadian Human Rights Commission) v. Canada (Attorney General)

Supreme Court of Canada Clarifies Standard of Review Framework.

How should courts review agency legal interpretations? The U.S. Supreme Court’s answer, known as Chevron1×1. Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984). deference, is that courts should defer to agencies’ reasonable interpretations of ambiguous statutes.2×2. Id. at 843. But hostility to Chevron deference is at its zenith; the Court has implied its willingness to reconsider Chevron,3×3. See SAS Inst., Inc. v. Iancu, 138 S. Ct. 1348, 1358 (2018) (“[W]hether Chevron should remain is a question we may leave for another day.”). and a majority of Justices have expressed skepticism of its reasoning.4×4. See, e.g., Pereira v. Sessions, 138 S. Ct. 2105, 2129 (2018) (Alito, J., dissenting); Michigan v. EPA, 135 S. Ct. 2699, 2712 (2015) (Thomas, J., concurring); Gutierrez-Brizuela v. Lynch, 834 F.3d 1142, 1152–53 (10th Cir. 2016) (Gorsuch, J., concurring); Stephen Breyer, Judicial Review of Questions of Law and Policy, 38 Admin. L. Rev. 363, 373 (1986); Brett M. Kavanaugh, Keynote Address, Two Challenges for the Judge as Umpire: Statutory Ambiguity and Constitutional Exceptions, 92 Notre Dame L. Rev. 1907, 1911–13 (2017). Off the Court, judicialists often favor replacing Chevron with de novo review so that courts, rather than agencies, decide questions of law.5×5. See, e.g., Jack M. Beermann, End the Failed Chevron Experiment Now: How Chevron Has Failed and Why It Can and Should Be Overruled, 42 Conn. L. Rev. 779, 786–87 (2010). And on the other side, deferentialists admit Chevron is not operating optimally.6×6. See, e.g., Cass R. Sunstein, Chevron Step Zero, 92 Va. L. Rev. 187, 248–49 (2006). But as responses to the problem of determining the standard for reviewing agency legal interpretations, both approaches suffer the same absolutism. Recently, in Canada (Canadian Human Rights Commission) v. Canada (Attorney General)7×7. 2018 SCC 31, [2018] 2 S.C.R. 230. (CHRC), the Supreme Court of Canada took a more reserved approach to the problem, clarifying a framework based on a presumption of deference that can be rebutted if the circumstances so warrant. CHRC shows, particularly in one of its concurrences, how courts can escape the deferentialist-judicialist dichotomy by incorporating both interests into a single framework, suggesting an alternate way forward for the American doctrine as the Chevron era ends.

Canada’s Indian Act8×8. R.S.C. 1985, c I-5. provides access to federal programs and benefits to people who qualify for “Indian” status according to criteria in the Act.9×9. Id. Because these criteria do not necessarily align with Indigenous practices, some who consider themselves Indigenous do not qualify for status and are denied the benefits.10×10. CHRC, 2018 SCC 31, para. 4. In CHRC, two sets of applicants who were denied status brought claims before the Canadian Human Rights Tribunal11×11. Id. para. 1. — an administrative body created by the Canadian Human Rights Act12×12. R.S.C. 1985, c H-6. (CHRA) to adjudicate discrimination complaints.13×13. Id. §§ 48.1(1), 53. The applicants argued that denying them status and its benefits constituted a “discriminatory practice in the provision of . . . services,” in violation of section 5 of the CHRA.14×14. CHRC, 2018 SCC 31, para. 10. The tribunal understood the claims to be attacking not a discriminatory provision of services under the Indian Act but rather the passing of the legislation itself.15×15. See Canada (Canadian Human Rights Comm’n) v. Canada (Att’y Gen.), 2015 FC 398, paras. 14, 25. Yet the tribunal could only grant relief under CHRA’s prohibition on providing discriminatory “services” and, on the tribunal’s interpretation, the Indian Act’s status-conferral scheme did not constitute provision of a “service.”16×16. Id. para. 3. The tribunal therefore dismissed the complaints.17×17. Id. paras. 11, 23.

On appeal, the Federal Court consolidated the cases and upheld the tribunal’s decision.18×18. See id. paras. 4, 123. The Federal Court of Appeal also dismissed the challenge. Using the framework for determining the standard of review of agency legal interpretations from the Supreme Court of Canada’s landmark Dunsmuir v. New Brunswick19×19. 2008 SCC 9, [2008] 1 S.C.R. 190. For a description of Dunsmuir’s attempt to simplify the complexity that had plagued the prior jurisprudence of reviewing agency legal conclusions, see Alice Woolley & Shaun Fluker, What Has Dunsmuir Taught?, 47 Alta. L. Rev. 1017, 1021–22 (2010). decision, it held that the appropriate standard of review for the tribunal’s interpretation of whether legislation was a “service” was reasonableness, and that the tribunal’s interpretation was reasonable.20×20. Canada (Canadian Human Rights Comm’n) v. Canada (Att’y Gen.), 2016 FCA 200, [2017] 2 F.C.R. 211, paras. 61, 88–90.

The Supreme Court of Canada affirmed.21×21. CHRC, 2018 SCC 31, para. 3. Writing for the majority, Justice Gascon22×22. Justice Gascon was joined by Chief Justice McLachlin and Justices Abella, Moldaver, Karakatsanis, and Wagner. began by clarifying the two-step framework laid out in Dunsmuir for determining which of the two available standards — “reasonableness”23×23. Comparable to Chevron Step Two, reasonableness review is “concerned with whether the decision falls within a range of . . . acceptable outcomes.” Dunsmuir, 2008 SCC 9, para. 47; cf. Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 843 (1984). For an example of an agency decision that failed reasonableness review, see Delta Air Lines Inc. v. Lukács, 2018 SCC 2, [2018] 1 S.C.R. 6, paras. 12–21. or “correctness”24×24. As with de novo review, a court reviewing for correctness will not defer, but instead will “undertake its own analysis of the question.” Dunsmuir, 2008 SCC 9, para. 50. — will be used to review an agency’s decision. The first step is precedential: Has the standard of review for the question before the agency been settled in the case law?25×25. Id. para 62. If so, the inquiry is over.26×26. Id.; CHRC, 2018 SCC 31, para. 71 (Côté & Rowe, JJ., concurring in the judgment). If not, at the second step, the reasonableness standard presumptively applies to an agency’s interpretation of its home statute.27×27. CHRC, 2018 SCC 31, para. 27 (majority opinion). A “home statute” is the Canadian analog of an agency’s organic statute. See, e.g., id. However, this presumption can be rebutted in favor of correctness review in two ways: if the question falls into any of four “correctness categories” denoting types of questions warranting correctness review; or, alternatively, if the balance of several contextual factors shows legislative intent for correctness review to apply to that particular question.28×28. Id. para. 28; see also Dunsmuir, 2008 SCC 9, para. 64.

As case law had not settled the standard and the reasonableness presumption obtained, Justice Gascon proceeded to step two and held that none of the correctness categories rebutted the presumption. Of the four categories, only those on questions of jurisdiction and on questions that are both “of central importance to the legal system and outside the expertise of the decision maker” were relevant.29×29. CHRC, 2018 SCC 31, para. 28; see id. para. 30. Acknowledging the difficulty in identifying a true question of jurisdiction, Justice Gascon rejected the category on issues of jurisdiction because the question asked whether the tribunal could grant relief to the applicants, not whether it could hear discrimination complaints altogether.30×30. See id. para. 33. Nor was the question outside the tribunal’s expertise under the latter category: the tribunal existed to adjudicate CHRA complaints.31×31. See id. para. 43.

With the categories rejected, Justice Gascon proceeded to the contextual analysis and similarly held that it did not rebut the reasonableness presumption.32×32. See id. para. 49. To avoid unnecessary complication, Justice Gascon cautioned that the contextual analysis should be “applied sparingly” and “limited to determinative factors that [collectively show] a clear legislative intent” for correctness to apply.33×33. Id. para. 46. Accordingly, Justice Gascon’s analysis was brief and dismissive of the four standard contextual factors — the presence of a privative clause, the tribunal’s expertise, the tribunal’s statutory purpose, and the nature of the question at issue.34×34. See id. para. 72 (Côté & Rowe, JJ., concurring in the judgment) (citing Dunsmuir v. New Brunswick, 2008 SCC 9, [2008] 1 S.C.R. 190, para. 64). A privative clause is a statutory instruction that courts should not review an agency’s decision. Cf. 5 U.S.C. § 701(a)(1) (2012). For Justice Gascon, all of these factors were either irrelevant or insufficient to indicate a legislative preference for correctness review.35×35. See CHRC, 2018 SCC 31, paras. 49–53.

Having settled on reasonableness, Justice Gascon applied that standard and upheld the tribunal’s findings that the claims attacked the Indian Act itself and that legislation was not a “service” under CHRA section five.36×36. See id. para. 56.

Justices Côté and Rowe jointly concurred in the judgment, agreeing that the tribunal’s decision should be upheld but disagreeing with the majority’s standard of review analysis.37×37. See id. para. 70 (Côté & Rowe, JJ., concurring in the judgment). Whereas the majority considered the contextual analysis to be reserved for exceptional cases, the concurrence considered that analysis to be fundamental;38×38. See id. para. 78; see also Barreau du Québec v. Quebec (Att’y Gen.), 2017 SCC 56, [2017] 2 S.C.R. 488, para. 23 (collecting cases in which the Court noted the contextual analysis). the correctness categories were simply contextual analyses that had reliably warranted correctness review.39×39. See CHRC, 2018 SCC 31, para. 78 (Côté & Rowe, JJ., concurring in the judgment). Thus, the two tests were equally important and deserving of thorough consideration.40×40. See id. para. 79.

Fully applying the contextual analysis led the concurrence to rebut the presumption of reasonableness.41×41. See id. para. 81. First, the lack of privative clause suggested that the legislature chose not to insulate the tribunal’s decisions from full judicial review.42×42. See id. para. 83. Second, other administrative bodies also had the expertise to interpret the CHRA, so deferring to the tribunal risked creating inconsistent human rights adjudications.43×43. See id. paras. 84–85. Third, the question before the tribunal was a constitutional one regarding the agency’s purpose because it required deciding who could adjudicate challenges to legislation.44×44. See id. para. 89. Collectively, these factors indicated that the legislature intended for courts to answer the question.45×45. See id. para. 90. Justice Brown also wrote separately, preferring not to conduct the standard of review inquiry as the tribunal’s decisions were valid under either standard. See id. paras. 108–09 (Brown, J., concurring in the judgment).

In the United States, the problem of selecting the standard for courts to use when reviewing agency legal interpretations generally sorts scholars into two camps: deferentialists prefer for courts to defer to agencies, whereas judicialists prefer for courts’ interpretations to control. But both are blind to the possibility that the other standard is preferable in some circumstances. By contrast, CHRC undertook a third approach: delaying determination of the standard until the balance of agency deference and judicial supremacy interests implicated by the specific question has become apparent. In particular, the contextual analysis, as emphasized by the CHRC concurrence, tailors the standard of review to the needs of the circumstances. This approach could help the Chevron debate escape the deferentialist-judicialist divide in which it is trapped.

The primary disagreement underlying the Chevron debate concerns whether agency deference or judicial supremacy is a more pressing interest. Chevron favors the former: at Step One, courts determine if Congress has spoken to the question at issue, defaulting to deference at Step Two only if the statute is silent or ambiguous and the agency’s interpretation is reasonable.46×46. See Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 842–45 (1984). The Chevron doctrine presumes that Congress intended for the agency to resolve statutory ambiguities,47×47. See Smiley v. Citibank (S.D.), N.A., 517 U.S. 735, 740–41 (1996). even when courts have supplied alternate interpretations.48×48. See Nat’l Cable & Telecomms. Ass’n v. Brand X Internet Servs., 545 U.S. 967, 982–83 (2005); see also Cass R. Sunstein, Beyond Marbury: The Executive’s Power to Say What the Law Is, 115 Yale L.J. 2580, 2588–89 (2006). But the assumption of congressional intent is a fiction;49×49. See Breyer, supra note 4, at 370. little evidence supports it50×50. Thomas W. Merrill & Kristin E. Hickman, Chevron’s Domain, 89 Geo. L.J. 833, 871–72 (2001). and, judicialists argue, it conflicts with the Administrative Procedure Act.51×51. 5 U.S.C. § 706 (2012) (“[T]he reviewing court shall decide all relevant questions of law, [and] interpret constitutional and statutory provisions . . . .”); see Patrick J. Smith, Chevron’s Conflict with the Administrative Procedure Act, 32 Va. Tax Rev. 813, 814–15 (2013). But see Sunstein, supra note 6, at 196. Moreover, the assumption undermines courts’ constitutional authority “to say what the law is.”52×52. Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803); see Clark Byse, Judicial Review of Administrative Interpretation of Statutes: An Analysis of Chevron’s Step Two, 2 Admin. L.J. 255, 261 (1988); Philip Hamburger, Chevron Bias, 84 Geo. Wash. L. Rev. 1187, 1200 (2016); Raymond M. Kethledge, Ambiguities and Agency Cases: Reflections After (Almost) Ten Years on the Bench, 70 Vand. L. Rev. En Banc 315, 323 (2017); see also Sunstein, supra note 48, at 2589. Respecting the separation of powers therefore seems to require replacing Chevron with de novo review.53×53. See Beermann, supra note 5, at 786; Ronald A. Cass, Vive la Deference?: Rethinking the Balance Between Administrative and Judicial Discretion, 83 Geo. Wash. L. Rev. 1294, 1313 (2015). For the latest congressional attempt to replace Chevron with de novo review, see Separation of Powers Restoration Act, H.R. 5, 115th Cong. § 202 (as received by Senate, Jan. 12, 2017).

However, neither the congressional intent nor the constitutional authority argument inherently favors judicialists. Underlying Chevron was a judgment that Congress considers paramount the agency’s expertise and need for flexibility in policymaking.54×54. See Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 865 (1984); see also Sunstein, supra note 48, at 2596. And Chevron put Congress on notice that courts will defer to the agency. The separation of powers argument can also cut against de novo review: Chevron allows elected members of the executive branch, rather than unelected judges, to pursue policy preferences when interpreting statutes.55×55. See Sunstein, supra note 6, at 206. This argument incorporates both how constitutional authority to make policy decisions rests with the political branches rather than the courts, see John F. Manning, Constitutional Structure and Judicial Deference to Agency Interpretations of Agency Rules, 96 Colum. L. Rev. 612, 617–18 (1996), and how Chevron recognizes that policymakers, but not judges, are accountable to voters dissatisfied with interpretations, see 467 U.S. at 865–66.

Which set of concerns should triumph? The question itself is misplaced because it seeks to resolve the deferentialist-judicialist dichotomy absolutely. Neither approach defeats the other in the abstract, so the approach that triumphs will oscillate according to judges’ policy preferences. To escape that oscillation, an alternative approach is necessary.

Instead of treating agency deference and judicial supremacy as dichotomous, the framework used in CHRC makes either approach available as necessary in the circumstances. The Chevron debate and CHRC ask the same question: Has the legislature delegated authority to interpret the statute to the agency? Under Chevron, the answer is yes, if the statute is ambiguous; under de novo review, the answer is no. By contrast, CHRC reservedly answers, it depends — not on ambiguity, but on interests implicated by the question itself. Whether a discriminatory practice violates section five of the CHRA is within the tribunal’s expertise as an arbitrator of human rights disputes,56×56. CHRC, 2018 SCC 31, para. 43. but whether a nonjudicial court can entertain challenges to legislation’s legality is not.57×57. Id. para. 89 (Côté & Rowe, JJ., concurring in the judgment). In the former circumstance, the agency expertise interest is paramount, whereas combatting agency overreach and maintaining the constitutional structure become more important in the latter. The more pressing interest not only varies, then, but does so at the level of the question before the agency. It therefore cannot be determined in the abstract. Both Chevron and de novo miss this variance: when a statute is ambiguous, Chevron defers even if deference risks agency aggrandizement; de novo review ignores the relevance of agency expertise altogether. Taking either approach as absolute, courts will sometimes apply a standard incongruous with the needs underlying the question. By contrast, CHRC’s presumption of reasonableness minimizes incongruity because it can be rebutted, allowing courts to tailor the standard to the question.

The CHRC concurrence’s emphasis on the contextual analysis is particularly instructive in overcoming the deferentialist-judicialist divide. Rejecting the majority’s subordination of the contextual analysis, the concurrence understood the correctness categories as four circumstances in which the contextual analysis rebutted the reasonableness presumption.58×58. See id. Checking if the categories apply is therefore a proxy for the full contextual analysis, not a replacement for it. Case law reinforces this concern. In Rogers Communication Inc. v. Society of Composers, Authors and Music Publishers of Canada,59×59. 2012 SCC 35, [2012] 2 S.C.R. 283. the Supreme Court of Canada rebutted Dunsmuir’s presumption of reasonableness with the contextual analysis when reviewing a rights complaint underlying the Copyright Board’s interpretation of the Copyright Act.60×60. See id. para. 15. Courts also hear rights challenges implicated by copyright claims, so if the Board received deference on a question of interpretation that implicated applicants’ rights then those claims would be reviewed differently based on whether the parties first brought them to a court or to the Board.61×61. See id. para. 14. Because of this “unusual statutory scheme under which the Board and the court may each have to consider the same legal question at first instance,” the interests implicated by the nature of the question were paramount, requiring correctness review.62×62. Id. para. 15. Not only would Chevron and de novo review have missed this nuance, but so too would the correctness categories. Only the contextual factors caught it, making the CHRC concurrence especially instructive in overcoming the deferentialist-judicialist divide.

One might challenge the concurrence’s theory as merely a totality of the circumstances test. Circumstantial standard of review analyses are not new. Critics of Chevron have long advocated for a return to an analysis like that of Skidmore v. Swift & Co.63×63. 323 U.S. 134 (1944); see also Breyer, supra note 4, at 380–81. Under Skidmore, courts give varying deference based on an indefinite number of factors, including the “thoroughness evident in [the agency’s] consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all those factors which give it power to persuade.”64×64. 323 U.S. at 140. But issues plague Skidmore: it precludes predicting how courts will review agency interpretations;65×65. See United States v. Mead Corp., 533 U.S. 218, 250 (2001) (Scalia, J., dissenting). and it permits unbounded judicial discretion, creating de novo review under the guise of “degrees of deference.”66×66. Thomas W. Merrill, Judicial Deference to Executive Precedent, 101 Yale L.J. 969, 972–74 (1992); see, e.g., Christensen v. Harris County, 529 U.S. 576, 583 (2000) (denying deference because the agency did not make “the proper expressio unius inference” (emphasis added)). As such, even some agency skeptics advocate Chevron’s formalism67×67. See, e.g., Antonin Scalia, Judicial Deference to Administrative Interpretations of Law, 1989 Duke L.J. 511, 517 (calling Chevron a “background rule . . . against which Congress can legislate”). and would likely criticize CHRC for suffering the same faults as Skidmore.

Yet the CHRC concurrence differs from Skidmore in two respects. First, the initial step in the CHRC framework is to follow the standard established in case law. Agencies can thereby rely on courts to treat like questions alike.68×68. Agencies cannot similarly rely under Skidmore, as the factors are so fluid and their application so circumstantial that analogizing to prior analyses is often unhelpful. See Kristin E. Hickman & Matthew D. Krueger, In Search of the Modern Skidmore Standard, 107 Colum. L. Rev. 1235, 1281 (2007). By contrast, the CHRC concurrence emphasized precedent in the standard of review analysis and in developing the correctness categories, see Dunsmuir v. New Brunswick, 2008 SCC 9, [2008] 1 S.C.R. 190, para. 54, limiting judicial discretion in a way unknown to Skidmore. Second, the concurrence’s contextual analysis, while more flexible than the categorical analysis, remained more restricted than Skidmore’s. The conurrence anchored the contextual inquiry on four factors, some entailing a yes-or-no answer, which led to only two possible standards. By contrast, Skidmore considers an indefinite number of factors, including “all those factors which give [the agency] power to persuade,”69×69. 323 U.S. at 140. and results in an unspecified amount of deference.70×70. See Mead, 533 U.S. at 234 (instructing courts applying Skidmore to use “some deference whatever its form”); see also David M. Hasen, The Ambiguous Basis of Judicial Deference to Administrative Rules, 17 Yale J. on Reg. 327, 336 (2000) (noting Skidmore’s “variable deference”). As such, the concurrence’s theory, though contextual, resists Skidmore’s fluidity and thereby avoids the absolutism of the deferentialist-judicialist divide without suffering the ordinary faults of such pragmatism.

The concurrence’s theory is not abstractly better than the American approach. Instead, the challenge is to create a workable test that satisfies the relevant interests as often as possible.71×71. See Beverley McLachlin, Chief Justice, Supreme Court of Can., Administrative Tribunals and the Courts: An Evolutionary Relationship, Address at the 6th Annual Conference of the Council of Canadian Administrative Tribunals (May 27, 2013) (describing “a felt need . . . to settle down and apply Dunsmuir and its progeny and see if what we had . . . could be made to work”). Thus, the CHRC concurrence is less an improvement on the American doctrine and more a demonstration that the deferentialist-judicialist divide is not inevitable. And while translating the framework directly into the American context is neither possible nor desirable, its basic tenets are present in American administrative law. For example, the “major questions” doctrine, used as an exception to Chevron in King v. Burwell,72×72. 135 S. Ct. 2480, 2488–89 (2015); see also FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 159 (2000) (quoting Breyer, supra note 4, at 370). could begin a framework based on categories of questions for which deference is inappropriate regardless of statutory ambiguity.73×73. See Barnhart v. Walton, 535 U.S. 212, 222 (2002) (noting deference depends in part on “the nature of the question at issue”). Several of the CHRC contextual facotrs have appeared in American cases concerned with deference to agency legal interpretations. See, e.g., Gonzales v. Oregon, 546 U.S. 243, 266–67 (2006) (denying deference because the Attorney General was answering a medical question); Rapaport v. U.S. Dep’t of Treasury, Office of Thrift Supervision, 59 F.3d 212, 216–17 (D.C. Cir. 1995) (denying deference because multiple agencies administer the statute). But see City of Arlington v. FCC, 569 U.S. 290, 307 (2013) (declining to treat jurisdictional questions separately for purposes of agency deference). Such a framework can escape the deferentialist-judicialist dichotomy by withholding standard selection until circumstantial interests have been balanced. As Chevron’s days dwindle, CHRC could help U.S. courts construct a new doctrine with which parties move “away from arguing about the tests and [get] back to arguing about the substantive merits of their case.”74×74. CHRC, 2018 SCC 31, para. 27 (citing Alberta (Info. and Privacy Comm’r) v. Alta. Teachers’ Ass’n, 2011 SCC 61, [2011] 3 S.C.R. 654, para. 36).