Corps de l’article

Introduction

Last year, Dunsmuir v. New Brunswick[1] was cited for the ten thousandth time by a Canadian court or an administrative tribunal. To put the number in perspective, Housen v. Nikolaisen,[2] the leading case on standards of appellate review, has been cited over five thousand times, although it was decided six years before Dunsmuir was handed down. Even allowing for the fact that my source is CanLII, which may not adequately cover the early 2000s,[3] the difference is remarkable. Dunsmuir is cited roughly one hundred times a month, twenty-five times a week, five times a (working) day.

Yet, many of Dunsmuir’s features remain somewhat obscure.[4] As Justice Layh put it in Skyline Agriculture Financial Corp. v. Saskatchewan (Farm Land Security Board), “[L]ocating the goalposts of correctness and reasonableness has remained an elusive target for those obliged to follow [the Supreme Court of Canada’s] leadership.”[5] Similarly, Justice Slatter chimed in: “The day may come when it is possible to write [an administrative law] judgment ... without a lengthy discussion of the standard of review. Today is not that day.”[6] And Justice Abella described standard of review as the “prodigal child” of Canadian administrative law.[7] More recently, an appellate judge has taken the unprecedented step of posting on an open-access website a twenty-seven-page “Plea for Doctrinal Coherence and Consistency” in the Canadian law of judicial review, which is referred to as a “never-ending construction site.”[8]

Most of the academic commentary on Dunsmuir has been cautiously supportive, praising the changes that the Supreme Court of Canada sought to effect but suggesting that more may need to be done to develop a workable approach to reasonableness review. Audrey Macklin’s comment that, on the whole, Dunsmuir has made “[t]he job of discerning the appropriate standard of review ... simpler”[9] is probably representative, though the recent complaints from the bench and from a judge writing extra-judicially suggest that any such early optimism is wearing thin.

Indeed, a majority of the Court seems to appreciate the desirability of modifying the current standard of review framework. In Wilson v. Atomic Energy of Canada Ltd., Justice Abella aired in obiter a “proposal” on how to “simplify the standard of review labyrinth we currently find ourselves in,” with a view to “starting the conversation about the way forward.”[10] Four of her colleagues welcomed her “efforts to stimulate a discussion on how to clarify or simplify our standard of review jurisprudence to better promote certainty and predictability.”[11] The dissenting judges commended the “constructive spirit” in which Justice Abella’s suggestions were offered, although they “harbour[ed] concerns about their merits.”[12] Only Justice Cromwell, concurring, firmly took the view that Dunsmuir should not be revisited, commenting that the standard of review framework “does not need yet another overhaul.”[13]

At the root of these difficulties in understanding, applying, and changing Dunsmuir is the attempt to set out a categorical approach to judicial review of administrative action.[14] In doing so, Justices LeBel and Bastarache found themselves swimming against a strong tide. The current of modern administrative law has long been pulling toward context. Gone are the old categories of “quasi-judicial” and “administrative” decisions,[15] replaced by more nebulous notions, such as fairness and reasonableness, which require courts to focus on various contextual factors.[16] The attempt to impose a categorical framework to restore order was as doomed as it was noble.[17] Sure enough, as my review of recent cases on standard of review will reveal, context has returned to the forefront of Canadian administrative law.

A possible manifestation of these difficulties is that the Supreme Court of Canada sometimes avoids standard of review analysis in whole or in part. The most egregious example is surely Febles v. Canada (Citizenship and Immigration),[18] an immigration judicial review in which the standard of review was not mentioned even in passing. This omission is all the more bizarre when viewed in light of a spirited disagreement between Justices Evans and Stratas in the court below on the appropriate approach to questions of international law,[19] a disagreement which the Supreme Court acknowledged but did not deign to resolve in B010 v. Canada (Citizenship and Immigration).[20]

This article is not intended to give a comprehensive overview of the post-Dunsmuir jurisprudence[21] or literature.[22] My modest goal is to analyze recent cases decided under the Dunsmuir framework with a view to determining where Canadian courts might usefully go next.

My focus in Part I will be on the first step in the standard of review analysis: selecting the standard of review. In and subsequent to Dunsmuir, the Supreme Court of Canada explained the required categorical analysis. Correctness applies to: constitutional questions; questions of general law “that [are] both of central importance to the legal system as a whole and outside the adjudicator’s specialized area of expertise;”[23] jurisdictional conflicts between two or more specialized tribunals; and “true questions of jurisdiction or vires.”[24] Meanwhile, the deferential standard of reasonableness “is normally the governing standard” for: the interpretation of an administrative decision-maker’s “home” statute or statutes closely related to its function; matters of fact, discretion, or policy; and “inextricably intertwined legal and factual issues.”[25] 

Lurking on the edges of this new “analytical framework” were the contextual factors that formed part of the discarded pragmatic and functional analysis: statutory language relating to appeals or privative clauses; relative expertise; statutory purpose; and the nature of the question.[26] These contextual factors were retained in Dunsmuir and, in Alliance Pipeline, served to resolve “[a]ny doubt” as to whether the categorical analysis identified the appropriate standard.[27] Yet, as we will see, the precise relationship between categories and context remains uncertain and continues to cause confusion.

In Part II, I will address the second step—applying the appropriate standard of review—with an emphasis on reasonableness review because its application is much more complex than the substitution of judgment permitted by correctness review. In Dunsmuir, Justices Bastarache and LeBel offered an elegant definition of reasonableness as “a deferential standard” that gives administrative decision makers “a margin of appreciation within the range of acceptable and rational solutions” but which nonetheless requires courts to inquire into “the existence of justification, transparency and intelligibility within the decision-making process” and into “whether the decision falls within a range of possible, acceptable outcomes which are defensible in respect of the facts and law.”[28] Turning this definition into concrete guidance has proved difficult, however, and much ink has been spilled on “justification, transparency and intelligibility” and the “range of possible, acceptable outcomes.”[29]

For all the confusion, perhaps we are edging toward coherence in standard of review analysis. I will suggest that both in selecting the standard of review and in applying the reasonableness standard, context is reasserting itself. Moreover, context is reasserting itself in a way that has the potential to be consistent with the two principles said to hold the Dunsmuir project together: “the rule of law and the foundational democratic principle.”[30]

The thesis of this article is that the rule of law and democracy can and should now be used to guide the contextual inquiry required of reviewing courts, a contextual inquiry that would take the form of a flexible but robust standard of reasonableness review. Obviously, much has been said about these two principles as a matter of legal and political theory.[31] Without wishing to sidestep important theoretical issues altogether, for the purposes of this article, I think it is sufficient to say that the understanding of these principles set out in Dunsmuir is relatively straightforward. On the one hand, the rule of law requires courts to ensure that administrative decision makers “do not overstep their legal authority.”[32] On the other hand, the democratic principle “finds an expression in the initiatives of Parliament and legislatures to create various administrative bodies and endow them with broad powers” and, in order to respect legislative intent, the courts must avoid “undue interference with the discharge of administrative functions” duly delegated to administrative decision makers.[33] The rule of law is firmly associated with legality, or the idea that reviewing courts have an important oversight role in ensuring that administrative decision makers stay within acceptable boundaries, and the protection of important individual interests.[34] Meanwhile, democracy means primarily that reviewing courts ought to respect the legislative choice to vest decision-making authority in bodies other than courts; that is, administrative law doctrine should aim to protect the administrative autonomy accorded by legislatures.[35]

I will leave it to other work to consider whether these understandings of the principles of the rule of law and democracy are defensible in theoretical terms and whether they give courts or administrative decision makers roles that are not normatively defensible. The apparent loss of faith in—or at least frustration with—the Dunsmuir framework suggests that there is great wisdom in Matthew Lewans’ comment that Dunsmuir’s “enduring value ... lies in its illustration of two persistent problems with judicial review,” namely, the perennial attraction of jurisdictional error and correctness review, and the inability to articulate a reasonableness standard that is capable of consistent application in different contexts.[36] In this article, I take up the challenge of responding to those problems with the principles articulated in Dunsmuir in hand.

This is not my first contribution to debates about the standard of review in Canadian administrative law. In a pair of essays published in 2012, I strongly attacked the decision in Dunsmuir and its subsequent application by the Supreme Court of Canada.[37] My two lines of attack related to the replacement of context by categories and were neatly summarized in “Dunsmuir’s Flaws Exposed: Recent Decisions on Standard of Review.”[38] First, “the categorical approach is unworkable and ... a reviewing court cannot in fact apply the categorical approach without reference to the much-maligned four [contextual] factors (or some variant thereon).”[39] Second, “the single standard of reasonableness is similarly impractica[ble]” without reference to some version of the four contextual factors.[40] While I continue to believe that context is an inescapable feature of the modern Canadian law of judicial review, I hope to build on my 2012 essays by proposing a means by which the contextual analysis can be cabined. Rather than casting courts adrift on a sea of context, I describe the instruments that they can use to navigate the vast seas of administrative law in a more effective and predictable manner.[41]

Finally, I should acknowledge that, elsewhere, I have developed my own preferred standard of review framework;[42] one which differs from the unified reasonableness standard I advocate below. There are, as Justice Abella observed in Wilson, “many models” for standard of review analysis.[43] Accordingly, this article should be understood as an attempt to articulate a rational next step for Canadian standard of review jurisprudence, not an elaborate scheme developed from first principles.

I. Step One: Selecting the Standard of Review

Several recent Supreme Court of Canada cases merit attention for their treatment of the question of how to select the standard of review. They signal two things: an obvious openness from the country’s apex court to the application of correctness review and a return to context. These cases are the subject of Part I-A. Meanwhile, in Part I-B, I identify lower-court cases that employ contextual analysis and explore the implications of the reasoning there employed.

A. Correctness and Context

Tervita Corp v. Canada (Commissioner of Competition)[44] is a long, complex, and important decision on competition law. It also contains a spirited disagreement between Justices Rothstein and Abella on the appropriate standard of review for determinations of law made by the Competition Tribunal.

The case concerned sections 92 and 96 of the Competition Act,[45] in particular their application to a merger of companies owning secure landfills in British Columbia. Section 92 prohibits mergers that would lessen or prevent competition.[46] This case involved prevention, because the acquired company had not yet begun to operate its landfill but would have at some point in the future. An acquisition by Tervita, an incumbent, would have had the effect of preventing competition in secure landfill services in British Columbia. Section 96 provides a defence to the prohibition, where the efficiency gains from the merger would outweigh its anti-competitive effects.[47]

The resulting issues were considered by the Competition Tribunal. The Competition Tribunal is a slightly unusual creature: its membership is drawn in part from the judiciary and only judicial members are entitled to address questions of law.[48] Moreover, under subsection 13(1) of the Competition Tribunal Act, a decision it makes is appealable to the Federal Court of Appeal “as if it were a judgment of the Federal Court.”[49] Justice Rothstein acknowledged the ordinary presumption that interpretations by an administrative decision maker of its home statute are entitled to deference. Nonetheless, the presumption was rebutted in this case because of this unique statutory provision, which was evidence of “a clear Parliamentary intention that decisions of the Tribunal be reviewed on a less than deferential standard, supporting the view that questions of law should be reviewed for correctness and questions of fact and mixed law and fact for reasonableness.”[50]

Justice Abella disagreed. She would have applied a reasonableness standard. To do otherwise, in her view, would undermine settled expectations:

[J]udges and lawyers engaging in judicial review proceedings came to believe, rightly and reasonably, that the jurisprudence of this Court had developed into a presumption that regardless of the presence or absence of either a right of appeal or a privative clause—that is notwithstanding legislative wording—when a tribunal is interpreting its home statute, reasonableness applies.[51]

Rebutting the presumption in this case would “chip away” at the “precedential certainty” that the Court had developed because it would represent “an inexplicable variation from our jurisprudence that is certain to engender the very ‘standard of review’ confusion that inspired this Court to try to weave the strands together in the first place.”[52] She saw “nothing ... that warrants departing from what the legal profession has come to see as our governing template for reviewing the decisions of specialized expert tribunals on a reasonableness standard.”[53]

Justice Abella is correct in asserting that Tervita will inject uncertainty into the law. Although the Competition Tribunal Act is the only piece of Canadian legislation that contains a provision requiring the appellate court to treat an administrative decision as if it emanated from an inferior court,[54] it relies implicitly on a general principle that statutory language may rebut the presumption of reasonableness. Indeed, in Stewart v. Elk Valley Coal Corporation, the Alberta Court of Appeal wasted no time in employing arguments based on statutory language giving a reviewing court wide remedial powers in appeals from human rights tribunals and creating an enforcement mechanism for tribunal decisions to support a conclusion that the legislature had “indicated that the Court and the Tribunal are dealing with ‘rule of law’ questions” for which correctness was the appropriate standard.[55] It should be said, however, that the clause in the Competition Tribunal Act makes, at best, an oblique reference to standard of review issues and could just as easily be understood as, say, guiding the Federal Court of Appeal in how to deal with the distinction between interlocutory and final decisions or remedial matters.

The post-Dunsmuir framework recognizes that there are certain categories to which correctness and reasonableness apply; the categories, however, are not self-applying, such that in hard cases—or maybe even all cases—courts must rely on contextual factors to determine which category applies.[56] Then, the presumption of reasonableness review has been tacked onto the categorical approach without any explanation of how it might be rebutted, or of its relationship to the categorical approach. As Justice Cromwell observed, “Creating a presumption without providing guidance on how one could tell whether it has been rebutted does not, in my respectful view, provide any assistance to reviewing courts.”[57] Does correctness review apply whenever a case falls into a correctness category, or only when the presumption has been rebutted, so that “rebutting the presumption” is simply shorthand for the conclusion that a case falls into a correctness category based on a consideration of contextual factors? Or does an applicant have two bites of the cherry: one to get into a correctness category and another to rebut the presumption of reasonableness, presumably relying on contextual factors on both occasions?

Justice Gleason considered these issues in her comprehensive judgment in Pfizer Canada Inc. v. Canada (Minister of Health).[58] Justice Gleason was surely right to say:

[T]he inquiry into standard of review does not necessarily end with the determination that the issue being reviewed involves the interpretation of the decision-maker’s home statute or a statute or regulation closely connected with its function and does not fall into one of the four foregoing categories to which correctness applies.[59]

Both categories and context are relevant.

The subsequent Supreme Court of Canada decision in Mouvement laïque québécois v. Saguenay (City of)[60] underlines the importance of contextual analysis, though in some respects it confuses the state of the law further. To begin with, Canada’s human rights tribunals have extensive powers to investigate and redress alleged breaches of fundamental rights by public and private parties. These statutory rights overlap but do not mirror the rights protected constitutionally; the human rights codes they are found in are usually described as “quasi-constitutional”.[61] But Canadian courts have typically been cautious in allowing the tribunals to define the scope of their own mandates—especially the scope of protected rights—a caution that can be traced at least as far back as Canada (AG) v. Mossop.[62] Post-Dunsmuir, the situation has remained the same, either because the scope of fundamental rights is considered a question of general law of central importance to the legal system or because the presumption of deferential review can be rebutted.

The first strategy is found in an early post-Dunsmuir decision from Alberta, Walsh v. Mobil Oil Canada, which uses “existing case law” to justify selecting correctness as the standard of review.[63] The second strategy can be seen in Canada (AG) v. Johnstone.[64] Justice Mainville found in Johnstone that the presumption of reasonableness review had been rebutted, for several reasons. For instance, “labour arbitration boards, labour relations boards and superior courts” often address human rights questions, creating a “concurrent jurisdiction of a multiplicity of decision makers” that calls for correctness review.[65] Moreover, the scope of discrimination on family status is a matter of concern across provincial boundaries. Therefore, “for the sake of consistency between the various human rights statutes in force across the country, the meaning and scope of family status and the legal test to find prima facie discrimination on that prohibited ground are issues of central importance to the legal system.”[66] One might be puzzled about why these factors rebut the presumption of reasonableness rather than indicate that the questions at issue fall into the category of questions of general law of central importance to the legal system; however, the confusions further evidence the uncertain relationship between categories and context.

If anything, the confusion was exacerbated by Justice Gascon’s reasons in the Saguenay case. The substantive aspects of the decision concerned the state’s duty of religious neutrality—violated here by a prayer read by the mayor of a Quebec city before municipal meetings. Ultimately, the Court upheld the conclusion of the Quebec Human Rights Tribunal (Tribunal) that the prayer was an impermissible discriminatory interference with the freedom of religion and conscience of an atheist participant, a breach of the provincial human rights code interpreted by the Tribunal. Reciting the prayer turned meetings “into a preferential space for people with theistic beliefs,” whereas non-believers who participated faced “isolation, exclusion and stigmatization”, a breach of the “right to exercise ... freedom of conscience and religion.”[67]

Justice Gascon identified an “important question” implicated by the Tribunal’s decision, specifically “the scope of the state’s duty of religious neutrality that flows from the freedom of conscience and religion protected by the Quebec Charter.”[68] Correctness was the appropriate standard of review of this question: “[T]he importance of this question to the legal system, its broad and general scope and the need to decide it in a uniform and consistent manner are undeniable.”[69] In addition, the presumption of deference applicable because the Tribunal was interpreting its home statute was rebutted because “the jurisdiction the legislature conferred on the Tribunal in this regard in the Quebec Charter was intended to be non-exclusive; the Tribunal’s jurisdiction is exercised concurrently with that of the ordinary courts.”[70]

The foregoing analysis is confusing because, post-Dunsmuir, a question falls either into a “correctness category” or a “reasonableness category”. General questions of law of central importance to the legal system, which are outside the expertise of the administrative decision maker under review, come under a correctness category. If a question is adjudged to fall into this category, that should be the end of the matter: correctness is the standard and it is up to the reviewing court to resolve the issue. Here, Justice Gascon concluded—without explaining why: his conclusion was said simply to be “undeniable”—that the state neutrality question was of central importance to the legal system. If this is the case, there is no need to rebut the presumption of deference. Nonetheless, although there was no need to do so, Justice Gascon went on to hold that the presumption of deference had been rebutted because of the existence of concurrent jurisdiction: individuals can ask the Tribunal or a court to apply the Quebec Charter.

Justice Gascon’s analysis further muddies the already murky waters of the relationship between Dunsmuir’s categorical approach and context. In order to convince a court to apply a correctness standard, does an applicant now have to demonstrate both that a decision falls into a correctness category and that the presumption of reasonableness can be rebutted? Or is there just one step in the analysis, with the contextual factors used to determine whether a correctness category applies? Conceptually, it makes more sense to think of a two-step process, relying first on the categories before turning to the contextual factors to confirm that the choice of category is appropriate—and, indeed, Dunsmuir envisages this sort of exercise.[71] But this is not the route that Justice Gascon followed.[72]

Justice Gascon’s reference to the concurrent jurisdiction of the Tribunal and the Quebec Superior Court will doubtless give a new lease of life to the gloss applied to Dunsmuir by Rogers Communications Inc. v. Society of Composers, Authors and Music Publishers of Canada.[73] There, Justice Rothstein wrote that concurrent jurisdiction can rebut the presumption of deferential review when an administrative decision maker is interpreting its home statute. Otherwise, “inconsistent” results could arise depending on whether a question of interpretation was raised at first instance (subject to de novo appellate review) or in an administrative setting (subject to deferential review).[74] It had seemed as if the Rogers exception had been limited to its special facts: in Justice Evans’ last set of reasons for the Federal Court of Appeal, he certainly took that view.[75] Indeed, in Rogers itself, Justice Rothstein said, “Concurrent jurisdiction at first instance seems to appear only under intellectual property statutes where Parliament has preserved dual jurisdiction between the tribunals and the courts.”[76] Evidently not! Context underpins the renewed significance of concurrent jurisdiction, though it bears noting that concurrent jurisdiction was not among the contextual factors mentioned in Dunsmuir. Plainly, in rebutting a presumption of reasonableness review, it is permissible to look outside the factors that made up the old pragmatic and functional analysis. It should be noted, however, that Justice Rothstein insisted in Rogers that the basis for rebutting the presumption of reasonableness review was legislative intent—in other words, an invocation of the democratic principle.

The confusion created by the Saguenay decision was entirely unnecessary, as there was unusual statutory language that would have provided a better route to Justice Gascon’s conclusion. Decisions of the Human Rights Tribunal are appealable, with leave, directly to the Quebec Court of Appeal. The relevant statute also provides that the general rules governing appeals are to apply in this context.[77] The Quebec Court of Appeal has split previously on the proper interpretation of its role on appeal from the Tribunal: some judges have applied judicial review criteria (following the well-established rule that appeal clauses do not eliminate deference to specialized tribunals), whereas others have applied appellate criteria based on the apparently plain language of the statute.[78] In Saguenay, Justice Gascon held that judicial review criteria apply: the Dunsmuir framework is the appropriate one.[79]

Nonetheless, one might reasonably think that the leave requirement is designed to ensure that matters of general importance should be addressed by the courts. One might further deduce from the legislature’s reference to rules governing appeals that a differentiation between questions of law (de novo review) and fact (deferential review) is in order. Of course, it has long been the case that the existence of a statutory appeal does not eliminate deference.[80] But a general understanding that deference will often be appropriate even on appeals should not be transformed into an inflexible rule that appeal clauses can never rebut the presumption of deference. All appeal clauses are not created equal. Surely, the better route to Justice Gascon’s conclusion would have been to rely on the unusual statutory language to rebut the presumption of reasonableness, as Justice Rothstein did in Tervita.[81] Certainly, an “appeal with leave” clause seems like a clearer manifestation of legislative intent with respect to judicial control of an administrative decision maker than the unusual clause in Tervita. That said, however, the Supreme Court’s later decision in Kanthasamy v. Canada (Citizenship and Immigration) seems to rule out the possibility that appeal clauses can ever be “determinative” of the standard of review: even a procedure whereby a first-instance reviewing judge can certify a question of general law for authoritative resolution by an appellate court is not capable of rebutting the presumption of reasonableness.[82]

This is not all. Justice Gascon also segmented the question before the Tribunal into two separate parts. On the general question of state neutrality, the standard was correctness. However, “the question whether the prayer was religious in nature, the extent to which the prayer interfered with the complainant’s freedom and the determination of whether it was discriminatory [fell] squarely within the Tribunal’s area of expertise,” as did “the qualification of the experts and the assessment of the probative value of their testimony, which concerned the assessment of the evidence that had been submitted.”[83] Clever lawyers will be licking their lips at the possibility of slicing decisions apart, extracting “general” or—an old favourite—”jurisdictional” issues for intensive judicial review. Justice Abella was quite right to warn in her concurring reasons of problems to come. As she asked, rhetorically, “How many components found to be reasonable or correct will it take to trump those found to be unreasonable or incorrect?”[84]

Treating the discrimination based on religious belief as bound up with the facts as found by the Tribunal is more consistent with current trends in reasonableness review and with the role of administrative decision makers: as Justice Abella put it in Moore v. British Columbia (Education), they are not Royal Commissions, but rather respond to particular factual situations.[85] A deferential approach to judicial review is more respectful of the incremental, bottom-up development of policy to which administrative decision makers are well suited. In Saguenay, we are simply told that the state neutrality question is “undeniabl[y]” of central importance to the legal system, without any explanation of why it is important outside the setting of the Tribunal and without any guidance as to how this category might be applied in future cases.[86]

No further guidance was offered in Canadian Broadcasting Corp. v. SODRAC 2003 Inc., where Justice Rothstein performed a standard of review analysis for five separate issues, applying a standard of correctness to one of them and a standard of reasonableness to the rest.[87] The question of whether broadcast-incidental copies form part of the “reproduction right” protected by paragraph 3(1)(d) of the Copyright Act was one of law that could arise before the Copyright Board or at first instance in enforcement proceedings and so, following Rogers, was to be resolved on a standard of correctness.[88] Most of the other issues related to exercises of discretion or mixed questions of fact and law. In dissent, Justice Abella said, “[T]his takes judicial review Through the Looking Glass.”[89] In her separate dissent, Justice Karakatsanis expressed the concern that Justice Rothstein’s analysis “unnecessarily complicates an already overwrought area of the law.”[90] For Justice Abella, the possibility of segmentation represents a “new and regressive” step that effects a “significant and inexplicable change” in the law of judicial review.[91] Justice Rothstein’s reply—that this was all settled by the Saguenay decision[92]—is unconvincing, because neither there nor in the instant case is there any explanation of why a particular decision should be segmented (beyond the banal observation that one of its elements is general or legal in nature) or how a reviewing court should perform a segmentation operation.

In summary, categories and context continue to exist side-by-side with little or no authoritative guidance on how they relate to one another. The possibility of “segmenting” administrative decisions adds a further layer of complexity. In many cases, there will be no dispute about the standard of review, but as soon as there is, the problems with the Court’s approach to standard of review are all too apparent.

B. The Return of Context

Both Tervita and Saguenay demonstrate an openness on the part of the Supreme Court to correctness and contextual analysis. An unthinking application of the reasonableness standard of review is not to be taken for granted. Several appellate decisions indicate that lower courts have understood this message.

A particularly clear example of the attraction of contextual analysis is Edmonton East (Capilano) Shopping Centres Ltd. v. Edmonton (City of).[93] At issue was “whether an Assessment Review Board can increase a property assessment when a complaint is brought by a taxpayer seeking a reduction of the assessment.”[94] For Justice Slatter, this question had to be answered by the courts: a standard of correctness applied.

Justice Slatter’s comment that “a mechanical and formalistic test for the standard of review is not reflective of the subtlety of the underlying issues”[95] gives a flavour of his preferred approach. He provided six reasons justifying a correctness standard in this case, in particular, to justify his conclusion that this case “presents either an addition to or a variation of the four ‘presumptive’ categories”[96] of correctness review set out in Dunsmuir—a conclusion that, in its equivocation, is indicative of the confused state of the relationship between categories and context. Justice Slatter’s six reasons were the following.

First, the legislation provided for an appeal.[97] Second, the appeal to the courts had “specific mandatory parameters”[98] that commanded correctness review: in particular, where a case is remitted to the board, the legislation binds it to follow the directions given by the court. Third, the appeal is by way of leave, “a signal that the Legislature wishes to have questions of this sort reviewed by the superior courts, and the legislative intent is not fully realized without a correctness standard of review.”[99] Fourth, statutory interpretation “is not the core of [the board’s] expertise.”[100] Fifth, taxation is special: “[T]he existence of a right of appeal is in keeping with the general democratic principle that taxpayers are entitled to have their liability to the government determined by the ordinary courts.”[101] Sixth, “multiple tribunals” are involved in the assessment process, which creates a need for judicial intervention to ensure coherence.[102]

Having “weighed and considered” all of these factors, Justice Slatter applied a standard of correctness.[103] While one might not necessarily agree with each of Justice Slatter’s justifications for correctness review, he surely mounted a formidable argument in this particular case, to which one can respond without having to read between the lines of his judgment. On the other hand, Justice Slatter’s invocation of six distinct contextual reasons for favouring correctness review underlines the open-ended nature of the current contextual inquiry. Even if some of the factors, such as the presence of a statutory appeal, might seem to carry less weight in view of recent Supreme Court jurisprudence, the absence of any guidance on how to weigh competing factors or how to rebut the presumption of reasonableness review makes it difficult to describe Justice Slatter’s approach or conclusions as wrong.

With context so dominant, categories can never be as categorical as their supporters would like.[104] Consider Commission scolaire de Laval v. Syndicat de l’enseignement de la région de Laval,[105] where a school board had to decide whether to end its contractual relationship with a teacher, on the basis that the teacher had serious criminal antecedents. Having heard from the teacher, the executive committee of the board entered an in-camera session during which it deliberated. Once its deliberations were concluded, the board issued a resolution removing the teacher from his position and providing some supporting reasons.

Subsequently, the teacher’s union filed a grievance on his behalf contesting the dismissal, alleging, for instance, that the termination procedure in the collective agreement had not been followed. Notably, the collective agreement provided that a teacher could only be dismissed after “thorough deliberations” by the board. After the board had made its case to the arbitrator, the union called as witnesses the three members of the executive committee who had deliberated in camera. Ruling on the board’s objection, the arbitrator concluded that the testimony would be relevant in assessing whether the deliberations were “thorough”.[106]

The Supreme Court of Canada was unanimous in concluding that it was legitimate for the arbitrator to have the decision makers testify. It was wrong to suggest that the decision makers’s motives were “unknowable”, a principle that “applies only to decisions of a legislative, regulatory, policy or purely discretionary nature made by public bodies.”[107] Here, the board “was acting as an employer,” a situation in which “the principles of employment law that are applicable to any dismissal” apply.[108] Moreover, the principle of deliberative secrecy did not shield the members of the board.[109]

The Court, however, divided six to three on the applicable standard of review. Writing for the majority, Justice Gascon applied a standard of reasonableness, on the basis that labour arbitrators are to be afforded deference on matters of procedure and substance falling within their area of expertise: it was “up to the arbitrator to apply the rule of relevance to the facts of the case in such a way as he or she deems helpful for the purpose of ruling on the grievance.”[110] That the arbitrator was applying general principles, which can be applied in areas other than labour relations did not change this analysis, for their application “to a fact situation characteristic of a dismissal” did not “amount to a question that is detrimental to consistency in the country’s fundamental legal order.”[111]

Justice Côté disagreed. Deliberative secrecy, like professional secrecy, must be interpreted in a uniform and consistent manner across regulatory domains. The interpretation of deliberative secrecy is a question of general law of central importance to the legal system and outside the expertise of an arbitrator:

Where the question relates not simply to the rules of evidence in general, but to the scope of such basic rules as those relating to the immunities from disclosure and deliberative secrecy, a court reviewing an arbitrator’s decision in this regard must be able to go further than merely inquiring into the reasonableness of the decision.[112]

The division is further evidence that the categories alone rarely resolve the question of what standard of review to apply in difficult cases.[113] The dominant considerations were, for Justice Gascon, the relative expertise and scope of authority of the arbitrator, and, for Justice Côté, the essentially legal nature of the question. These, it goes without saying, are contextual factors external to the categories. Context matters even when it is not supposed to.

Andrew Green has been probably the strongest and most sophisticated defender of the Court’s resort in Dunsmuir to categorical analysis. I think it is fair to say, however, that Green’s defence is lukewarm. He sees the “objective of greater simplicity” as a “move in the right direction,”[114] but his analysis of Dunsmuir leads him only to say that “[t]he categorical approach may ... reduce some errors and have some beneficial systemic effects.”[115] Moreover, although his institutionalist analysis sheds valuable light on the costs and benefits of categories versus context, it could be more alive to the doctrinal characteristics of modern administrative law. Green persuasively argues that courts will save time in identifying the appropriate standard of review by using categories rather than context. His focus, however, is on the “institutional impact” of the categorical framework and “less about the application of the [reasonableness] standard.”[116] The key problem is that context simply cannot be eliminated from judicial review: the attempt to remove it from the framework for determining the standard of review has failed (and, as I will demonstrate in Part II, context also seeps into the framework for applying the reasonableness standard).[117] Context is not quite everything in judicial review of administrative action, but it is everywhere.

A way forward is suggested by the Federal Court of Appeal’s approach—implicitly followed by a dissenting group of three judges at the Supreme Court of Canada—in Wilson.[118] The Canada Labour Code (Code) applies to a variety of enterprises falling under the authority of Parliament, (i.e., federally-regulated entities). Section 240 of the Code protects some categories of employees from unjust dismissal.[119] Mr. Joseph Wilson was dismissed without cause. He claimed that he was unjustly dismissed for whistleblowing on his employer’s activities. It fell to an independent adjudicator appointed under the Code to decide.

Unfortunately, there are two distinct streams of arbitral jurisprudence on a critical preliminary question. Some take the view that a dismissal without cause is per se unjust. Others prefer to say that the absence of cause is a factor to be taken into account in a global assessment of whether the dismissal was unjust.[120]

Justice Stratas, writing for the Federal Court of Appeal, applied a standard of correctness on the basis that the question of “whether Part III of the Canada Labour Code permits dismissals on a without cause basis” was a general question of law of central importance to the legal system and outside the decision maker’s expertise.[121] Although the majority judgment at the Supreme Court of Canada, delivered by Justice Abella, applied a reasonableness standard on the basis that the decision maker was interpreting materials “within his expertise”, the dissenting judges applied a correctness standard.[122] Justices Côté and Brown, however, made no attempt to justify their choice of standard in terms of the Dunsmuir framework.[123] Bearing in mind the goal of articulating a rational next step for Canadian administrative law (rather than making a clean break with Dunsmuir), it is therefore useful to consider Justice Stratas’ approach to the standard of review in some detail.

To justify this categorization, Justice Stratas invoked rule of law concerns created by the fact that adjudicators “do not consider themselves bound by the holdings on the other side”,[124] with pernicious results: “Draw one adjudicator and one interpretation will be applied; draw another and the opposite interpretation will be applied. Under the rule of law, the meaning of a law should not differ according to the identity of the decision maker.”[125] In my view, explained in greater detail elsewhere,[126] Justice Stratas’ concern for the rule of law should have been offset by an appreciation of the importance of decisional autonomy, which follows from the democratic principle relied on in Dunsmuir. Accordingly, I agree with the majority of the Supreme Court of Canada that a reasonableness standard was appropriate in this case.[127]

For present purposes, what is interesting about Justice Stratas’ approach is that it relies less on categorizing the question at issue as a “general question of law” than it does on contextual analysis. Justice Stratas appealed not to categories but rather to the dispute resolution function performed by reviewing courts: sometimes, a question needs a uniform answer and, sometimes, a court will be the only one able to provide it.[128] There is much to commend this approach. It replaces metaphysical musings about reviewing courts’ role in keeping administrative decision makers within their “jurisdiction” with an approach that builds on the judicial role in establishing uniform national standards on important matters of principle.

Justice Stratas’ approach also points toward a means of cabining the contextual factors: the two principles at the heart of Dunsmuir—the rule of law and democracy—can provide a framework for the identification of contextual factors. Insisting that contextual factors be drawn only from these principles would provide some structure to the contextual inquiry.

Moreover, although the confused relationship between categories and context complicates matters, reference to the rule of law and democracy as guiding principles may assist judges who must navigate the morass of the standard of review analysis. It is possible to envisage an approach in which the Dunsmuir categories function as “signposts”,[129] with contextual factors providing further guidance where necessary, without sight ever being lost of the need to justify the choice of standard of review by reference to the rule of law and democracy.

As I will explore below, these principles can also assist in structuring reasonableness review, in which case the clear distinction between selecting the standard of review and identifying the boundaries of reasonableness would break down, perhaps resulting in a single contextual inquiry that sets the “range of reasonableness”.

The recognition of this possibility was evidently at the root of Justice Abella’s intriguing suggestion, in Tervita, that the lines between reasonableness and correctness may soon be “completely erased”, a suggestion which became an “option” for future reform in Wilson.[130] Contextual factors would still be important, but would operate only to determine the “range” of reasonable outcomes, thereby eliminating the confusion caused by the uncertain relationship between categories and context.[131]

II. Step Two: Reasonableness Review

Reasonableness is fast becoming the dominant organizing principle of Canadian administrative law. In Wilson, Justice Abella suggested that reasonableness should now become the only standard of review;[132] this standard, however, needs to be properly understood. As one commentator has noted, “[T]he Court continues to veer between two approaches to reasonableness review that are at opposite extremes.”[133] I argue that contextual factors cabined by the rule of law and democratic principles—an idea introduced in Part I—can shape the content of reasonableness review.

Courts of appeal across the country have been putting flesh on the bones of the skeletal definition given in Dunsmuir. There, reasonableness was said to have two components: first, a decision-making process bearing the hallmarks of “justification, transparency and intelligibility”; second, a decision falling within the “range of possible, acceptable outcomes” in respect of the facts and the law.[134] The florid language of the first component has been replaced by a more functional test that asks whether the reviewing court can, from the record and reasons provided, clearly understand how the administrative decision maker reached its conclusion.[135] Satisfying this functional test is necessary to permit a reviewing court to assess the second component for, without a clear understanding of why the decision was reached and on what it was based, it is impossible for the courts to perform their constitutionally mandated function of judicial review.[136]

As to the second component, there is always a range of reasonable outcomes that “must be assessed in the context of the particular type of decision making involved and all relevant factors.”[137] As the Federal Court of Appeal has explained in Canada (Minister of Transport, Infrastructure and Communities) v. Farwaha:

In some cases, Parliament has given a decision-maker a broad discretion or a policy mandate—all things being equal, this broadens the range of options the decision-maker legitimately has. In other cases, Parliament may have constrained the decision-maker’s discretion by specifying a recipe of factors to be considered—all things being equal, this narrows the range of options the decision-maker legitimately has. In still other cases, the nature of the matter and the importance of the matter for affected individuals may more centrally implicate the courts’ duty to vindicate the rule of law, narrowing the range of options available to the decision-maker.[138]

As Gerald Heckman comments, it may be that “Dunsmuir has not really simplified the task of ascertaining the appropriate degree of deference, but has simply left it for a later stage in the analysis.”[139] At times, the range of options might be very narrow indeed, especially where the Charter is involved. Consider Loyola High School v. Quebec (AG).[140] Quebec has a secular religious education course that is mandatory across the province, in public and private schools alike. An exemption is available, “provided the institution dispenses programs of studies which the Minister of Education, Recreation and Sports judges equivalent.”[141] In this case, Loyola, a Jesuit high school in Montreal, applied for an exemption, as it sought to teach a religious education from (primarily) a Catholic perspective. The minister refused the application, essentially because Loyola wanted to teach comparative religion and ethics from a Catholic point of view. The minister wrote, for example, that “[a]ccording to the summary of the program proposed by Loyola High School and transmitted to the department for evaluation, the program does not meet the requirements for the Ethics and Religious Culture program in terms of religious culture, as religions are studied in connection with the Catholic religion.”[142]

The Supreme Court of Canada split on how to review this decision. One might have thought that the analytical framework was settled by Doré v. Barreau du Québec, a unanimous decision written by Justice Abella, in which a deferential approach was preferred for the review of discretionary decisions affecting Charter rights.[143] For a majority of four judges in Loyola, Justice Abella applied the Doré framework to the minister’s refusal to grant the exemption. Yet, without mentioning Doré, the three judges in the minority—including two who signed onto Doré—applied a proportionality test.[144]

As for Justice Abella’s application of the reasonableness standard, it is difficult to discern how it is more deferential than, or analytically distinct from, proportionality. She began by stating that “the task of the reviewing court applying the Doré framework is to assess whether the decision is reasonable because it reflects a proportionate balance between the Charter protections at stake and the relevant statutory mandate.”[145] But she quickly added, “In the context of decisions that implicate the Charter, to be defensible, a decision must accord with the fundamental values protected by the Charter.”[146] Indeed, “in contexts where Charter rights are engaged, reasonableness requires proportionality.”[147]

Analytically speaking, the conclusions of the majority and the minority were almost identical (though they disagreed on an important point about the teaching of ethical issues).[148] For all the judges, the minister’s inflexible position that religion had to be taught from a neutral perspective violated religious freedom and ran counter to the purpose of the exemption provision. A unanimous win for Loyola, this case reflects the lack of consensus on the relationship between constitutional and administrative law, but offers clarity at least on the narrowness of the range of reasonable options open to an administrative decision maker who infringes on Charter rights.[149]

In other recent cases involving garden-variety statutory interpretation rather than Charter issues, the Court has confirmed that, on some occasions there will only be one possible, acceptable outcome.[150]

In general, the range of reasonable outcomes is determined by contextual factors drawn (it seems) from the rule of law and democracy principles invoked in Dunsmuir.[151] On the one hand, statutory language (or the “rationale of the statutory regime”[152]) may restrict the range of reasonable outcomes, a nod to democracy, as might the importance of a decision to an individual, a nod to the rule of law. On the other hand, where a decision-making power has been granted to an expert body or to a politically accountable minister, the range of reasonable outcomes will generally be larger, because of the body’s institutional knowledge or the minister’s democratic credentials. In some cases, the legislature might have drawn the relevant criteria in broad terms, thus further expanding the range. Regardless, the boundaries of reasonableness are drawn in large part by reference to the rule of law and democracy.

Identifying the range of reasonable outcomes is only part of the reasonableness analysis, however. It is necessary to develop the analytical structure of reasonableness to assist courts in determining why a decision falls outside the range of reasonable outcomes. Here again, the principles of the rule of law and democracy are important. The rule of law, with its concern for the maintenance of the precepts of the legal order, requires reviewing courts to police the boundaries of reasonableness. Equally, however, democracy, with its recognition of the decisional autonomy of the administrative decision maker chosen by the legislature to regulate a particular area, imposes important restraints on the type of analysis a reviewing court may legitimately conduct.

A. How to Do Reasonableness Review

There is a fascinating review of Canadian administrative law on reasonableness in Workplace Health, Safety and Compensation Commission v. Allen.[153] Mr. Douglas Allen received benefits for a workplace injury. These benefits were capped at eighty per cent of actual earnings. Allen then retired and was to receive benefits “equal” to the pension he would have received. The Workplace Health, Safety and Compensation Commission (Commission) sought to cap the pension at eighty per cent as well. Allen took a different view but lost in the administrative process. He won, however, on judicial review; the first-instance judge and court of appeal both concluded that there was only one reasonable outcome, that benefits could not be capped.

Building on the Supreme Court of Canada decision in Newfoundland and Labrador Nurses’ Union v Newfoundland and Labrador (Treasury Board), the Commission argued that administrative decisions should be presumed correct.[154] Justice Barry rejected the suggestion that there was any presumption of correctness of administrative decisions. If anything, there is a presumption of validity that places an “onus” on an applicant “to point to some reason, whether stemming from the facts or the words of the statute to question the reasonableness of the tribunal’s interpretation”, for which the administrative decision maker must provide “sufficient justification” in the form of “a convincing explanation why its choice of meanings was reasonable.”[155]

Along similar lines, as least as far as the structure of reasonableness review is concerned, is Delios v. Canada (AG),[156] a straightforward review of a labour adjudicator’s interpretation of a collective agreement. First, Justice Stratas noted that, although the reviewing court had nominally applied a standard of reasonableness, it had “actually performed correctness review.”[157] Reasonableness review does not permit a court to arrive at its own preferred interpretation of a provision and then check to see whether the administrative decision maker’s interpretation matches the court’s preferred interpretation: “[A]s reviewing judges, we do not make our own yardstick and then use that yardstick to measure what the administrator did, finding any inconsistency to be unreasonable.”[158] That would be disguised correctness review, “the court developing, asserting and enforcing its own view of the matter.”[159] Second, Justice Stratas recalled the idea that the range of acceptable and defensible outcomes “can be narrow, moderate or wide according to the circumstances,”[160] before adding, in a passage that needs to be quoted at length:

The evidentiary record, legislation and case law bearing on the problem, judicial understandings of the rule of law and constitutional standards help to inform acceptability and defensibility. Here, certain indicators, sometimes called “badges of unreasonableness,” may assist. For example, a decision whose effects appear to conflict with the purpose of the provision under which the administrator is operating may well raise an apprehension of unreasonableness. In that sort of case, the quality of the explanations given by the administrator in its reasons on that point may matter a great deal. Another badge of unreasonableness is the making of key factual findings with no rational basis or entirely at odds with the evidence. But care must be taken not to allow acceptability and defensibility in the administrative law sense to reduce itself to the application of rules founded upon badges. Acceptability and defensibility is a nuanced concept informed by the real-life problems and solutions recounted in the administrative law cases, not a jumble of rough-and-ready, hard-and-fast rules.[161]

Badges of unreasonableness must be identified in order to justify striking down a decision, as the analyses in Allen and Delios indicate. Notably, the indicia of unreasonableness can be drawn from the same sources as the contextual factors that make up the range of reasonable outcomes: inconsistent decisions, for instance, sound in the rule of law;[162] whereas decisions that fail to take into account important statutory language do violence to the democratic principle.[163] And, in general, ensuring that decisions respect the fundamental precepts of the legal system is a means of upholding the rule of law.

But this is not a laundry list of potential reasons for judicial intervention. Sometimes—perhaps even often—what look like badges of unreasonableness on first glance will turn out, on a patient review of the record, to be perfectly acceptable and defensible ways of expressing a particular thought or justifying a particular conclusion.[164]

Where a decision is indelibly tainted by a badge or badges of unreasonableness, judicial intervention will be more or less appropriate, depending on the range of reasonable outcomes. For instance, the narrower the range, the more that will be required by way of explanation by the administrative decision maker of the badge(s) of unreasonableness tainting the decision. Conversely, the wider the range, the less a reviewing court should require by way of explanation. In searching for these explanations, an administrative decision should be read fairly, not picked apart in a “line-by-line treasure hunt for error.”[165]

To summarize: if the same contextual factors that are relevant to choosing the standard of review are also relevant to determining the range of reasonable outcomes, Canadian law has, despite everything, come a long way toward achieving coherence in the standard of review and, perhaps, toward an all-encompassing flexible reasonableness standard that restricts extremely narrowly or expands very broadly depending on the interplay of the rule of law and democracy in a given case.

Adopting an all-encompassing flexible reasonableness standard would eliminate the problematic relationship between categories and context described above. And it would not necessarily raise constitutional difficulties by eliminating correctness review. In Alberta Teachers’ Association, Justice Cromwell argued that the “constitutional guarantee” of judicial review “does not merely assure judicial review for reasonableness; it guarantees jurisdictional review on the correctness standard.”[166] Justice Abella had a strong response to this argument in Wilson:

Nothing Dunsmuir says about the rule of law suggests that constitutional compliance dictates how many standards of review are required. The only requirement, in fact, is that there be judicial review in order to ensure, in particular, that decision-makers do not exercise authority they do not have. I see nothing in its elaboration of rule of law principles that precludes the adoption of a single standard of review, so long as it accommodates the ability to continue to protect both deference and the possibility of a single answer where the rule of law demands it, as in the four categories singled out for correctness review in Dunsmuir.[167]

More generally, reasonableness review, guided by the rule of law and democracy, is not “unduly deferential”.[168] And, if the range of reasonable answers will sometimes be so narrow as to admit only one possible, acceptable outcome, this narrowness ensures that the constitutional guarantee of judicial review is more than an “empty shell”;[169] in some cases, it will shade into correctness review, but without engendering endless confusion between categories and context. Doing away with this confusion would save litigants and courts precious resources and thus have its own rule of law and, indeed, access to justice advantages.

B. How Not to Do Reasonableness Review

These intelligent efforts to explain and structure the reasonableness inquiry can be contrasted with the less impressive guidance from the Supreme Court of Canada. As Professor Mullan has recently suggested, “If the whole standard of review enterprise is not to fall further into disrepute, the Supreme Court of Canada needs to articulate more fully a template for the conduct of proper or appropriate deferential reasonableness review and to condemn disguised correctness review in all of its various forms.”[170]

One case will be used in this section for the purposes of comparison, but many others could be singled out. In general, these are decisions in which the language of reasonableness performs a primarily rhetorical function while the Court applies a form of review that treats the administrative decision as a decorative ornament rather than the considered position of the legislature’s designated administrative decision maker.[171] The Court has understood this in the past. Justice Iacobucci concisely explained—in part by reference to the democratic principle—the analytical structure of reasonableness review in Law Society of New Brunswick v. Ryan.[172] First, “a court should not at any point ask itself what the correct decision would have been,” because the administrative decision maker, not the court, has been assigned by the legislature “the primary responsibility of deciding the issue according to its own process and for its own reasons.”[173] Building on this invocation of the democratic principle, Justice Iacobucci warned, “Even if there could be, notionally, a single best answer, it is not the court’s role to seek this out when deciding if the decision was unreasonable.”[174]

Although the following discussion is primarily analytical, it is underscored by the democratic principle. A legislative choice to designate a decision maker other than a court as regulator of a specified area should be respected. At the same time, upholding the rule of law requires courts to keep a check on the rationality of administrative decision making. Applying the reasonableness standard appropriately is the primary way for Canadian courts to respect legislative choices to grant decision-making autonomy to administrative bodies.[175]

Considering an example in detail may be helpful. The province of Quebec allows pregnant workers to exercise a right of withdrawal from dangerous work environments. At issue in Dionne v. Commission scolaire des Patriotes,[176] was a supply teacher’s thwarted effort to exercise her right of withdrawal. A unanimous Supreme Court of Canada quashed the decision of the Commission des lésions professionnelles (CLP) and held that the teacher was entitled to withdraw.

Although it may seem unusual to treat schools as dangerous workplaces, it is common and accepted practice in Quebec for pregnant teachers to withdraw from the workplace because of the risk of contracting harmful diseases from their students. Reading between the lines of the present case, the school board and the CLP apparently took umbrage at the teacher’s temerity in claiming her statutory rights, evidence perhaps of a disconnect between law in the books and law in practice and lingering discomfort among employers about assertive employees.

Be that as it may, the most interesting aspect of the case, from an administrative-law point of view, lies in the differing approaches to the task of judicial review taken by the appellate judges involved. In my view, the Quebec Court of Appeal’s stance is more appropriate than that of the Supreme Court of Canada. And, of the Quebec Court of Appeal judges, the dissenting reasons of Justice Dalphond are preferable.

First, the facts. Ms. Marilyne Dionne was a qualified teacher. But she did not have a permanent contract of employment. As a supply teacher, she was contacted on a regular basis by the school board and filled in as requested. Once she learned that she was pregnant, she responded to offers from the school board by saying that she would be happy to teach but that she would have to exercise her right to withdraw due to her pregnancy.

All agree that, when a supply teacher agrees to teach for a particular period of time, a contract is formed between the teacher and the school board. The question was whether, in light of her desire to exercise her right of withdrawal, Dionne was a “worker” for the purposes of the Act respecting occupational safety, sections 40 to 48 of which provide for the right of withdrawal and associated rights.[177] Properly speaking, the right is to be reassigned to other activities that are not dangerous, with a right to withdraw, with benefits, if no reassignment is offered by the employer.[178]

In the Supreme Court, Justice Abella made only fleeting reference to the decision under review. She engaged in an analysis of the text and purpose of the statutory provisions at issue, concluding that the legislation “protects pregnant women in two significant ways: it protects their health by substituting safe tasks for dangerous ones, and it protects their employment by providing financial and job security.”[179] She mentioned and criticized the CLP’s conclusion that Dionne could not be treated as a “worker” because her inability to go to the workplace frustrated the creation of a contract of employment. The whole point of the scheme, in Justice Abella’s view, was “to protect pregnant workers who have a contract to work,” in which case ”[i]t would be anomalous, to say the least, to use the legislated right of a pregnant worker to withdraw from an unsafe workplace to conclude that her withdrawal negates the formation of the contract of employment.”[180] Thus, Justice Abella continued, as soon as Dionne had accepted the offer, she became a “worker” within the meaning of the statute: “Her pregnancy was not an incapacity that prevented her from performing the work, it was the dangerous workplace, and that in turn triggered her statutory right to substitute that work with a safe task or withdraw.”[181]

This was the core of Justice Abella’s reasoning and justified her conclusion that the CLP’s decision was unreasonable. The thrust of Justice Abella’s analysis was that the CLP should have answered the question before it in a particular way. Passing references to unreasonableness[182] cannot obscure the fact that Justice Abella essentially stepped into the shoes of the CLP and rendered what she thought was the most appropriate decision in the circumstances. Reading the judgment from start to finish, one could be forgiven for thinking the Supreme Court was sitting in an appellate capacity, rather than conducting a judicial review.

Contrast this approach with that of Justice Dalphond, dissenting in the Quebec Court of Appeal.[183] Both Justice Dalphond and Justice Abella reached the same result, but by very different means. Justice Dalphond began with the decision of the CLP, underlining its central elements:

Partant, pour qu’il y ait formation d’un nouveau contrat, il faut que la personne soit en mesure de s’obliger à effectuer un travail sous la subordination d’un employeur et qu’elle soit rémunérée en conséquence, selon les termes de l’article 2085 [du Code civil du Québec].

Le tribunal ne peut donc partager l’opinion de la procureure de madame Dionne quand elle allègue qu’une seule offre de suppléance acceptée par madame Dionne entraîne la formation d’un contrat. En effet, il manque une cause essentielle à ce contrat, soit une prestation de travail. Ainsi, les dix fois, en novembre 2006, où madame Dionne accepte une offre de suppléance, il n’y a pas formation de contrat puisque aucune prestation de travail n’est offerte ou ne peut être offerte par elle.[184]

Plainly, the CLP’s conclusion was based on the absence of a contract between the parties.

Justice Dalphond gave three reasons why the CLP’s decision was unreasonable. First, the CLP’s own logic supported the conclusion that offer and acceptance of occasional work triggered the right to withdrawal and related provisions; yet, the school board had made no effort to give Dionne other tasks, such as correcting work or dealing with small groups of students.[185] Second, for the CLP to deny that a contract had been created was contrary to its own factual conclusions and the evidence in the record.[186] Third, the CLP’s position was irrational, because it placed teachers like Dionne in an invidious position of choosing between potential harm to their child and the loss of statutory benefits.[187]

Justice Dalphond’s approach is notable because he carefully examined the reasons given for the administrative decision and, by demonstrating their internal inconsistencies and irrational effects, justified his decision to intervene. Rather than establishing an external benchmark based on his examination of the law and the facts against which to judge the decision, as Justice Abella did, he worked from within the decision to demonstrate why it was untenable. Whereas Justice Abella asserted that, based on the evidence, there was a contract, Justice Dalphond preferred to say that the CLP’s own conclusion was that there was a contract and that its refusal to recognize this led to perverse results.[188]

Conclusion

In the wake of the recent Supreme Court of Canada decisions on standard of review discussed in Part I, it seems fair to say that both correctness and context are firmly in vogue. Lower courts have already fastened onto these decisions to justify more contextual approaches to selecting the standard of review and applying a correctness standard.

Given that the Supreme Court has not yet been clear on the precise relationship between categories and context, or the contextual factors to which reviewing courts might legitimately have regard, there is reason to fear that Tervita and Saguenay might create confusion in Canadian administrative law. The Federal Court of Appeal decision in Wilson and Justice Abella’s obiter comments on appeal, however, indicate that the contextual factors may be confined by the two principles said to underlie the Dunsmuir framework: the rule of law and democracy. In future cases, courts will explain how the contextual factors relied on relate to these core principles. Otherwise, unfortunately, the law will become more confused.

One might quibble that the proposals set out in this article simply represent “the bold innovations of a traffic engineer that in the end do no more than shift rush hour congestion from one road intersection to another without any overall saving to motorists in time or expense.”[189] Nevertheless, it is better “to adapt the framework of judicial review to varying circumstances and different kinds of administrative actors than it is to go through the same checklist of factors in every case, whether or not they are pertinent.”[190]

Another reason to hope that these principles come to structure the standard of review inquiry is that they are already influential in applying the reasonableness standard discussed in Part II. In particular, the range of reasonable outcomes can be determined by reference to contextual factors drawn from the rule of law and democratic principles, and the indicia of unreasonableness that must be identified to justify judicial intervention also sound in these underlying principles. Recognizing the conceptual unity between the selection of the standard of review and the application of reasonableness points the way to a further simplification of the Dunsmuir framework: collapsing the correctness and reasonableness standards into one range of permissible outcomes that expands or contracts depending on contextual factors.

But attention must also be paid to the analytical structure of reasonableness review. If reasonableness is not applied in a way that is respectful of the democratic principle embodied in the legislative choice to grant decision-making authority to a body other than a court, confusion lies ahead. Hopefully, ten thousand more citations to Dunsmuir from now, Canadian courts will have achieved the greater degree of conceptual clarity that now seems possible.