Dunsmuir – Reflections of a Recovering Judge (Hon. John M. Evans)
John M. Evans is Public Law Counsel at Goldblatt Partners LLP, Toronto, and was formerly a judge of the Federal Court and the Federal Court of Appeal
The tenth anniversary celebration (if that’s the right word) of Dunsmuir has made me again sort through my thoughts on its significance in the development of administrative law and practice in Canada from the perspective of a former judge, a substantial part of whose caseload involved standards of review in administrative law. My remarks here are limited to Dunsmuir’s contribution to the law governing the standard of review.
Dunsmuir as an example of judicial law reform
When embarking on a significant reform of the law judges are not comfortable in making a clean break from the past. They are generally all too conscious of: their limited ability to anticipate problems that may be thrown up by future cases; the constraints inherent in attempting major law reform through adjudication; and their lack of democratic credentials as legislators. The common law model of law reform therefore presents itself as more evolutionary and incremental than revolutionary.
While judges may be forthright in correcting past errors in the law, they are generally more circumspect about providing a detailed set of rules to be followed in future cases. They can, though, provide useful guidance by articulating principles to inform the future development of the law. Landmark judicial decisions should be viewed as a starting point, not the end point in legal evolution or, to change the metaphor, they close one chapter and open another.
Dunsmuir fits this model. It was designed to address complaints from lawyers and judges that the law on selecting the standard of review was not working well. The pragmatic and functional analysis that courts had been refining since the Court first introduced it in U.E.S., Local 298 v. Bibeault,  2 S.C.R. 1048, had come to be regarded as unduly complicated, wasteful of the time and effort of all concerned, uncertain in its outcome, and a distraction from determining the “merits” of the grievance for which the applicant was seeking a remedy.
The Court identified the existence of three standards of review (patent unreasonableness, reasonableness simpliciter, and correctness) as the primary problem to be addressed, which it solved it by replacing them with two standards: reasonableness and correctness. However, it prefaced its consideration of this issue by announcing (at para. 32) that it was time “to re-examine the Canadian approach to judicial review of administrative decisions and develop a principled framework that is more coherent and workable.” It did this by articulating the competing constitutional principles of the rule of law and legislative supremacy that underlie the judicial review of all administrative action and should guide judges in future cases. The Court also purported to provide clearer direction on how to choose between correctness and reasonableness as the applicable standard of review.
How successful has Dunsmuir been in meeting its law reform objectives of making the selection of the standard of review less labour intensive and unpredictable, and of providing a better balance between safeguarding the rule of law and respecting legislative choice? Fair-minded answers to these questions must take into account subsequent, judicial efforts to modify, clarify, or to be confused by Dunsmuir.
A single reasonableness standard
The Court’s elimination of the two reasonableness standards and its substitution of a single standard of reasonableness that admits of no gradations on a sliding scale solved the first problem that Dunsmuir set out to tackle. Neither counsel nor judges have subsequently had to bother themselves with formulating the appropriate deferential standard of review.
However, as Binnie J. perceptively predicted in his concurring opinion in Dunsmuir (at paras. 134-141), reasonableness may connote a single standard of review but is far from being self-applying. It will be easier in some circumstances than in others to establish that a decision is unreasonable depending on the breadth of the discretion that the legislature has conferred. In making this assessment, a reviewing court should take into account the particular statutory, institutional, and factual contexts in which the standard is to be applied, including: the open-endedness of the statutory grant of power; the ambiguity of the statutory text; the political accountability and expertise of the decision maker; the policy content of the decision and the extent to which it is polycentric; and the seriousness of the impact of the decision on the individual.
The discretion exercisable by decision makers in the interpretation of their enabling statute will often be relatively limited. However, judges differ over whether the existence of more than one reasonable meaning attributable to a statutory provision is usual or rare. In Wilson v. Atomic Energy of Canada Ltd., 2016 SCC 29,  1 S.C.R. 770 at paras. 34 and 35, Abella J. stated that a wide range of reasonable outcomes is the norm and a single correct meaning the exception. In contrast, in McLean v. British Columbia (Securities Commission), 2013 SCC 67,  3 S.C.R. 67 at paras. 32-33. Moldaver J. suggested that multiple reasonable interpretations were the exception and existed only “on occasion”.
The Court stated in Dunsmuir (at para. 48) that the elimination of the patent unreasonableness standard was not intended to intensify the level of judicial scrutiny of administrative action that was entitled to curial deference. However, there was a danger that the abandonment of patent unreasonableness might tend to decrease the level of curial deference. Indeed, Abella J. expressed a concern in Wilson (at para. 27) that judicial review for reasonableness can sometimes be hard to distinguish from correctness review.
This is particularly true when the question in dispute is the interpretation of a decision maker’s enabling statute, and the reviewing court starts by examining the statutory text, context, and objectives, an analytical approach designed to enable courts to determine the meaning of a statutory provision when they are the primary decision-maker. But reasonableness review can also merge into correctness when the exercise of an explicit discretion is in issue: see, for example, Kanthasamy v Canada (Immigration and Citizenship, 2015 SCC 61,  3 S.C.R. 909. Recent empirical studies may suggest that judicial intervention has decreased since Dunsmuir. However, when a court reviews a decision on a de facto correctness standard it is narrowing the tribunal’s ability to change its mind in subsequent cases, regardless of whether or not it upholds the decision in that case.
Keeping judicial review to a bare minimum is particularly important when a tribunal’s decisions are protected by a strong preclusive clause, the clearest indicator of legislative intent that courts should generally not interfere with administrative decision-making. I am not altogether convinced by the argument that since something is or is not reasonable, there is no room for a standard of patent unreasonableness. Before Dunsmuir was decided, applicants’ counsel typically devoted much time and effort to trying to persuade reviewing courts that, if correctness was not the applicable standard, reasonableness, not patent unreasonableness, was. It was certainly a widely held view on the Federal Court of Appeal and at the Bar that if the Court concluded that patent unreasonableness was the standard, the applicant had a much steeper hill to climb in order to win than would have been the case if simple reasonableness was the standard.
True, critics could point to uncertainty about what made a decision patently unreasonable: was it the obviousness of the decision maker’s error and/or the perversity of the decision? It might have been wiser for the Court in Dunsmuir to have retained patent unreasonableness as the standard for reviewing a decision protected by a preclusive clause (as s. 58 of British Columbia’s Administrative Tribunals Act, SBC 2004, c. 58, has done), rather than regarding the preclusive clause as simply one reason for applying the reasonableness standard (at para. 52).
Retaining patent unreasonableness as the standard for reviewing the interpretation of its enabling statute by a tribunal protected by a preclusive clause would have given proper weight to legislative supremacy without sacrificing the rule of law or requiring counsel and the court to engage in a lengthy analysis to select the appropriate standard.
It would surely not be beyond judicial wit to come up with factors or descriptive words for identifying when a decision is patently unreasonable. Whether Justice Beetz’s “fraud on the law” or “breach of the duty of fairness” (Syndicat des employés de production du Québec et de l’Acadie v. Canada Labour Relations Board,  2 S.C.R. 412 at 420-421) are apt I don’t know. But they do indicate that a decision may only be set aside if no one who had genuinely heard the argument, and had regard to the law, could have made it.
Reasonableness review: reasons and outcome
Dunsmuir advanced three propositions to guide courts in the conduct of a deferential or respectful reasonableness review: see paras. 47-48. First, a decision maker’s reasons should be the principal focus: do they provide “justification, transparency, and intelligibility” in the decision-making process? Second, a reviewing court may supplement the reasons that the tribunal gave with those that it could have given (but didn’t) by way of justification. Third, a reviewing court must also ask whether the outcome falls within the range of decisions that are reasonably defensible in light of the relevant law and the facts.
I never found it easy to put these pieces together to form a workable template for the conduct of a reasonableness review. While never explicitly renouncing Dunsmuir, the Court subsequently shifted the emphasis of reasonableness review from an examination of the decision maker’s reasons to see if they provide justification, transparency, and intelligibility for the decision, to the Court’s own assessment of the reasonableness of the decision in light of the facts and the law. Again, I focus my remarks on the review of tribunals’ interpretation of their enabling statute.
The first major pull-back from Dunsmuir occurred in Newfoundland and Labrador Nurses’ Association v. Newfoundland and Labrador, 2011 SCC  3 S.C.R. 708, where the Court stated that the inadequacy of a decision maker’s reasons is not a stand-alone basis for setting a decision aside; the reasons given have to be read together with the outcome in an “organic exercise”: para. 14.
While this is not altogether easy to understand, I take from Newfoundland Nurses’ Association that if a reviewing court is satisfied that the outcome is reasonable on the basis of the reasons given, when considered in light of the record, and the reasons that the decision maker could have given, the decision should stand, even though the decision maker’s reasons do not themselves demonstrate the necessary justification, transparency, and intelligibility in the decision-making process. No doubt the Court was intent on putting a stop lawyers’ attempts to frustrate the policy of curial deference by persuading courts to subject tribunals’ reasons to close scrutiny for any error or omission. In addition, setting the standard demanded of reasons too high would undermine the “speed, economy and informality” that were intended to characterize administrative decision-making: Newfoundland Nurses’ Association at paras. 23-25.
The second major departure from Dunsmuir came with Alberta Teachers’ Association (paras, 23-29) and Edmonton East (paras. 36-40), where the decision makers had provided no reasons at all to support their interpretation of the provisions in dispute, because the interpretative question only became an issue in the judicial review proceedings. In both cases the majority came up with reasons that persuaded them that there was a reasonable interpretation of the statute that supported the decision. The Court’s job was made easier in Alberta Teachers’ Association by the existence of previous administrative decisions interpreting the provision in the same way.
Thus, far from being the primary focus of reasonableness review that Dunsmuir had prescribed, reasons are not necessarily needed at all and the reviewing court, assisted by counsel (including counsel representing the decision maker), can produce reasons to justify the decision. Indeed, even when reasons are given, they may attract little attention from courts, which all too often embark on a reasonableness review on the same basis as they determine de novo the “correct” interpretation of a statutory provision.
However, in two decisions released this year, the Court has pulled back from potentially resting reasonableness review solely on the outcome of the administrative process and has imposed limits on the extent to which a reviewing court may support a decision by reasons that it has supplied. In Delta Airlines Inc. v. Lucáks, 2018 SCC 2 at para. 23, the Court held that in reviewing for reasonableness a court is not entitled to ignore the reasons of the decision maker and substitute its own when the decision maker has “provided detailed reasons that enumerated and then strictly applied a test unsupported by the statutory scheme.” Such reasons do not provide the necessary intelligibility, justification or transparency to the process: para. 37. In contrast, a court’s power to “supplement” reasons may be exercised when reasons are “non-existent or insufficient” (para. 23). The all-Ontario dissent of Abella, Moldaver, and Karakatsanis JJ. focussed exclusively on the reasonableness of the outcome and ignored the tribunal’s reasoning.
But how far may judges supplement “thin”, but not otherwise legally flawed reasons for a decision maker’s interpretation of its statute? A 5-4 majority in Williams Lake Indian Band v. Canada (Aboriginal Affairs and Northern Development), 2018 SCC 4, was willing to do this, even though the tribunal’s reasons were “sparse” (para. 108) or apparently “conclusory”. Tribunals, Wagner J. (as he then was) said (at para. 37) must be able to rely on reviewing courts to try to make sense of their reasons by looking to the administrative record and the parties submissions. To require more would unduly burden the decision making process. In contrast, the dissenters found no reasoning by the tribunal on the interpretative issue that could be supplemented.
While the Court continues to espouse the importance of reasons in the conduct of reasonableness review (see , for example, Williams Lake at para 36), the rhetoric of Dunsmuir has come up against reality: the reluctance of reviewing courts to further delay decision-making and to impose additional costs on the parties and the public by remitting a matter for redetermination that they believe they can resolve for themselves, provided only that the decision maker’s reasons are not wrong”. The take-away for decision makers is that their reasons need merely hint at the basis of the interpretation; if necessary, reasons can be elaborated and patched up on judicial review. This is surely not what Dunsmuir intended.
Selecting the standard of review: correctness or reasonableness?
In addition to replacing two reasonableness standards with one, Dunsmuir’s broader mission was to simplify the law governing the choice between reasonableness and correctness as the applicable standard of review. Judicial review of the interpretation of tribunals’ enabling legislation has always been the most contentious issue in Canadian administrative law. I shall focus on what Dunsmuir said on this issue.
The Court’s most significant statement was that reasonableness will usually be the standard of review applicable to a tribunal’s interpretation of its enabling statute or another law closely related to its functions (at para. 54). However, correctness is the standard when the question: (i) is one of constitutional law (para. 58); (ii) involves “a true question of jurisdiction or vires” (para. 59); (iii) is one of central importance to the legal system as a whole and is outside the tribunal’s area of expertise (para. 60); or (iv) demarcates the decision-making authority of competing administrative regimes (para. 61).
Dunsmuir thus seemed to signal a clear change in the law from the previous pragmatic and functional test, under which the fact that the question in dispute was one of statutory interpretation was an indicator that correctness was the standard of review. It also may have been thought that by making reasonableness the usual standard, the burden of further contextual analysis by counsel and reviewing courts would be much reduced. However, having taken a couple of steps forward, the Court shuffled backwards by saying that the four factors of the pragmatic and functional analysis (repackaged at para. 63 as the standard of review analysis) were still to be used to determine the standard of review when previous authority had not satisfactorily resolved the matter: paras. 55-57 and 62. The Court caused confusion by never really explaining the relationship between the new, more categorical approach and the contextual standard of review analysis, and risked forcing courts to undertake exactly the kind of lengthy and indeterminate analysis that it said it wanted to avoid.
Nonetheless, Dunsmuir unlatched the door to further reform by emphasizing the importance of simplifying the selection of the standard of review and on seemingly placing a burden on counsel to identify what it was in the statutory scheme that warranted departing from the usual conclusion that reasonableness is the applicable standard. And in Alberta (Information and Privacy Commissioner) v. Alberta Teachers’ Association, 2011 SCC 61,  3 S.C.R. 654. The Court barged right in by installing deference as the default standard of review for issues of statutory interpretation. It elevated reasonableness from the “usual” standard to the presumptive standard, and made no mention of the need to consider the other contextual factors when previous authority had not satisfactorily settled the matter: paras. 39, 44. Writing for the majority (which, significantly, included LeBel J., co-author of the majority opinion in Dunsmuir), Rothstein J. said that a full standard of review analysis should not be employed when the standard of review “can be determined summarily”: para. 44. Subsequent decisions of the Supreme Court, particularly that of the majority in Edmonton (City) v. Edmonton East (Capilano) Shopping Centres, 2016 SCC 47,  3 S.C.R. 293, indicate that the presumption of reasonableness is a strong one, and that the Court has little inclination to go through each statutory scheme with a fine-tooth comb in order to look for indications that the presumption is rebutted.
The heyday of the controversy about the selection of the standard of review is over. On only two occasions since Dunsmuir has the Court found that the presumption that reasonableness is the presumptive standard for reviewing a tribunal’s interpretation has been rebutted. In neither case did this require a lengthy analysis: Rogers Communications Inc. v. Society of Composers, Authors and Music Publishers of Canada, 2012 SCC 35,  2 S.C.R. 283 at paras. 11-15 (litigant had choice of pursuing remedy in court or before a tribunal); Tervita Corp. v. Canada (Commissioner of Competition), 2015 SCC 3,  1 S.C.R. 161 at paras. 35-39 (uniquely worded right of appeal).
Similarly, exceptions to the presumption that the reasonableness standard applies have been narrowly confined. In only one case has the Court found that the interpretative issue in dispute concerned a question of central importance to the legal system as a whole and outside the expertise of the decision maker: Alberta (Information and Privacy Commissioner) v. University of Calgary, 2016 SCC 53,  2 S.C.R. 555 at paras. 20-27 (solicitor client privilege). Provisions in human rights statutes defining prohibited discrimination and any applicable defence may also fall into this exception as being quasi-constitutional in nature: Mouvement laïque québécois v. Saguenay (City), 20-15 SCC 16,  2 S.C.R. 3 at paras. 48-49.
Dunsmuir’s retention of the concept of “true questions of jurisdiction”, formulated in terms reminiscent of the discredited “preliminary question” doctrine, as a correctness category is an example of the Court’s reluctance to make too clean a break with the past, despite the potential for confusion: para. 59. However, this has done no lasting damage to Dunsmuir’s missions of simplifying the law and aligning it with constitutional principles. The majority of the Court is yet to characterize a provision in an enabling statute as “a true question of jurisdiction” and has doubted whether a creature so rarely seen even exists: Alberta Teachers’ Association at para. 42; Quebec (Attorney General) v. Guérin, 2017 SCC 42 at paras. 32-36.
In short, Dunsmuir may not have established a clear, stable, and manageable set of principles for selecting standard of review of tribunals’ interpretation of their enabling statute and other laws closely related to their function. But it did point the law in the right direction and enabled the Court in subsequent decisions to move closer to the principled simplicity to which Dunsmuir aspired. Is it unduly cynical to wonder whether the growing judicial acceptance of review for reasonableness reflects a realization that a decision maker’s margin of appreciation on interpretative issues will often range from narrow to non-existent, leaving the courts very much in control of interpreting legislation?
Non-adjudicative decision makers
Dunsmuir initiated a welcome move towards simplifying the selection of the standard of review applicable to adjudicative decision makers’ interpretation of their enabling statute. However, the Court has subsequently extended reasonableness as a deferential standard of review to the exercise of all statutory powers, and not just to the decisions of those performing adjudicative functions. Whether this was Dunsmuir’s intention is unclear.
The Court stated that while the appeal “deals with the particular problems of judicial review of an adjudicative tribunal, these reasons will address first and foremost the structure of judicial review as a whole”: para. 33. But it also quoted Le Bel J.’s earlier doubt about whether the former pragmatic and functional approach applied to “the decisions and actions of all kinds of administrative actors”. Further, as Binnie J. observed, the fact that the majority’s discussion of the standard of review focused almost exclusively on adjudicative tribunals belied their more grandiose statements about the scope of their reformulation of administrative law : para. 121.
Curial deference to an administrative interpretation of an enabling statute is only relevant to decision makers whose delegated powers include deciding questions of law. “Traditional” administrative tribunals (labour boards, human rights tribunals, professional discipline tribunals, and the like) generally have an express or implied power to decide any question of law or fact necessary to decide the matter before them. However, this is not true of all those exercising statutory powers who are subject to judicial review. Of course, all administrative actors have to take a view of whether their enabling statute authorizes a proposed action, but this is not the same as having the legal power to make binding determinations of questions of law that are presumptively only subject to judicial review on the reasonableness standard.
The Supreme Court has addressed this issue when considering whether a decision maker is authorized to determine Charter challenges to the validity of its enabling legislation. The first step in this inquiry is to determine whether the decision maker has implied jurisdiction to decide questions of law, if, of course, it does not have an express power. The adjudicative nature of the body in question is the most important, but not necessarily determinative, factor indicating an implied power to decide questions of law: Nova Scotia (Workers’ Compensation Board) v. Martin, 2003 SCC 54,  2 S.C.R. 50 at paras 41, 52-61. In examining the legislation as a whole a reviewing court should also consider the interaction of the decision maker with other components of the administrative scheme, the importance to its mandate of deciding questions of law, and practical considerations, such as its capacity to decide these issues, and administrative efficiency. In a concurring judgment in Quebec (Commission des normes, de l’equité, de la santé et de la sécurité du travail) v. Caron, 2018 SCC 3 at paras. 102-109, Rowe J. recently used this framework, as elaborated in R. v. Conway, 2010 SCC 22,  1 S.C.R. 675, in concluding that the tribunal under review had jurisdiction to grant relief under the Quebec Charter of Human Rights and Freedoms, a quasi-constitutional statute.
It was assumed in Catalyst Paper Corp. v. North Cowichan (District Council), 2012 SCC 2,  1 S.C.R. 5, and Green v. Law Society of Manitoba, 2017 SCC 20,  1 S.C.R. 360, that those exercising delegated legislative powers have the authority to interpret their grant of statutory power. However, I doubt whether, if the Martin factors had been applied, the municipalities in question would have been found to have the power to make binding decisions on the scope of their delegated legislative powers.
In contrast, in Canadian National Railway Co. v. Canada (Attorney General), 2014 SCC 40,  2 S.C.R. 135 at paras. 51-54, the Court took a more limited view of Dunsmuir, holding that its innovation was to extend the presumption of reasonableness to all those exercising adjudicative powers, including the Cabinet, and not just to administrative tribunals in the traditional sense. Similarly, the Court has applied the Dunsmuir presumption to an individualized decision made by a Minister without a formal hearing process, but did not expressly address whether the Minister had the necessary delegated interpretative authority: Agraira v. Canada (Public Safety and Emergency Preparedness), 2003 SCC 36,  2 S.C.R. 559 at paras. 48-50.
Only decision makers who are required by the duty of fairness to afford affected individuals an effective opportunity to make submissions on questions of statutory interpretation should be held to have an implied power to decide questions of law and the benefit of Dunsmuir’s deferential standard of review. While this is a necessary requirement, it may well not be sufficient when other Martin factors are considered. Indeed, Rowe J. suggested in Caron (at para. 102) that the power should be limited to adjudicative bodies that “function in an adversarial setting and [are] presided over by an administrative judge”, and not extended to all whose exercise of administrative powers may be subject to the duty of fairness.
Administrative action of a non-adjudicative character is reviewable for ultra vires: Chamberlain v. Surrey District School Board, No. 36,  SCC 86 ,  4 S.C.R. 710 at paras. 190-194 (per Le Bel J.), and vires is subject to correctness review: United Taxi Drivers’ Fellowship of Southern Alberta v. Calgary (City), 2004 SCC 19,  1 S.C.R. 485, cited with approval in Dunsmuir at para 59. In these contexts, reasonableness may be relevant in two respects.
First, as a ground of review it may somewhat expand the scope of vires review (to include proportionality, for example). This is not the Dunsmuir deferential standard of review for determining whether an applicant has established a ground of review (error of fact or law, for example). Second, it may remind reviewing courts not to take an unduly narrow view of the factors that bodies, especially if elected or otherwise politically accountable, may consider when exercising broad discretionary powers. This comes closer to the concept of judicial deference.
Whether or not Dunsmuir intended its standard of review analysis to apply beyond bodies or officials exercising adjudicative powers is not clear. However, it has sown unnecessary conceptual confusion by encouraging an unthinking application of an analytical framework appropriate for adjudicative decision makers with the power to decide questions of law.
This content has been updated on February 28, 2018 at 22:41.