Canada Post: Vavilov’s First Day in the Sun

Vavilov didn’t have to wait long to have its first day in the sun. In Canada Post Corp v Canadian Union of Postal Workers, 2019 SCC 67 (a 7-2 opinion, Abella and Martin JJ dissenting), the Court had its first crack at applying the revised standard of review framework set out in Vavilov. In my view, the Court did an admirable job in Canada Post, showing that at least when it comes to garden-variety administrative law cases, Vavilov is a durable precedent. The devil will be in the details: especially where reasons are non-existent.

In this post, I review the facts of Canada Post and the analysis of the majority and the dissent. Then I demonstrate why the majority had the better of the argument here, particularly on applying the reasonableness standard in statutory interpretation cases.

Facts and Opinions

Canada Post concerned a decision of the Occupational Health and Safety Tribunal of Canada (OHSTC). A union representative at the Burlington mail depot filed a complaint with Human Resources and Skills Development Canada, alleging that Canada Post was not abiding by its obligations under the Canada Labour Code [CLC]. The complainant claimed that Canada Post was failing these obligations because it was only inspecting—for health and safety—the Burlington depot itself, and not “letter carrier routes and locations where mail is delivered” [4]. This had obvious practical impacts, because “letter carriers travel 72 linear kilometres delivering mail to 8.7 million points of call” [4].

Following the complaint, an officer conducted an investigation and found that Canada Post failed to comply with s.125(1)(z.12) of the CLC, which provides that “the employer must ensure that every part of the workplace is inspected once a year” [5]. Canada Post appealed to the OHSTC, which rendered the decision under review. The decision-maker, an Appeals Officer, held that the CLC section only applied to the “parts of the work place over which the employer had control…” which did not include the mail routes and places where mail was delivered [5] The Appeal Officer looked to the text of the relevant section, the definition of “workplace” in the relevant part of the CLC, and the “remedial purpose of health and safety legislation” [17]. He ultimately concluded that since health and safety inspections were designed to “permit the identification of hazards and the opportunity to fix them or have them fixed,” control over the workplace was a necessary condition [19].

The Federal Court affirmed. The Federal Court of Appeal would have affirmed the original officer’s decision.

At the Supreme Court, Rowe J wrote for the majority. He first outlined that the decision would be assessed under the Vavilov framework, even though the result under the Dunsmuir framework would have been the same [24]. After quickly determining the standard of review of reasonableness under the Vavilov framework [27], Rowe J turned to applying the standard. He noted that Vavilov instantiated a “reasons first” [27] approach to review, and so first set out some comments about reasons under the Vavilov framework. Here, he noted that the administrative decision-maker’s reasons were “exemplary,” but that what is required of decision-makers in a given case will depend on the context [30]. Both outcome and reasons matter: a decision must be justifiable in outcome and justified by reasons [28]. He noted that, while Vavilov defined reasonableness as generally mandating two things, (1) an internally coherent set of reasons (2) justified by legal and factual constraints, this analysis is not necessarily germane to every case [34] (see Vavilov, at para 101, 106).

Rowe J first concluded that the decision was internally consistent [35], before turning to the Appeal Officer’s interpretation of s.125(1)(z.12). Rowe J first outlined that, in reviewing a delegate’s legal decision, “a reviewing court should not conduct a de novo interpretation…” by adopting a “yardstick” and measuring a decision up against that yardstick [40]. A “reasons first” approach “requires the reviewing court to start with how the decision maker arrived at their interpretation, and determine whether it was defensible in light of the interpretative constraints imposed by law” [41]. This still requires a decision-maker to be alive to the essential elements of statutory interpretation: text, context, and purpose (Vavilov, at para 120).

Here, the Court first looked to the text of the section. The Appeals Officer concluded that the definition of “workplace” in the CLC, must be interpreted “broadly to account for all areas in which an employee may be engaged in work” [46]. Section 125(1) applies in respect of a workplace [45]. It provides that:

125 (1) Without restricting the generality of section 124, every employer shall, in respect of every work place controlled by the employer and, in respect of every work activity carried out by an employee in a work place that is not controlled by the employer, to the extent that the employer controls the activity,

The core interpretive difficulty was whether this paragraph should be read disjunctively or conjunctively; meaning, do the obligations apply to situations where employers control the workplace, do they apply where they don’t, or do they apply to both situations? The Union made the latter argument, but the Appeals Officer concluded that certain of the listed obligations “apply only where the employer has control over the work place…” [49], whereas others applied when the employer controlled the activity but not the workplace [50]. The statutory context ultimately supported this view [50-53].

Turning to purpose, Rowe J focused on the specific purpose animating Part II of the CLC, under which the core interpretive difficulty arose: that purpose was designed to “prevent accidents and injury to health arising out of, linked with or occurring in the course of employment to which this Part applies…” [54]. The Appeals Officer concluded that a condition precedent to the fulfillment of this purpose, and the specific obligation in the relevant section, was control over the workplace, so that hazards could actually be identify [55]. Practically, it would be difficult for Canada Post to ameliorate hazards in areas it did not control. This practical reasoning was endorsed in Vavilov as an indicium of reasonableness (see Vavilov, at para 93).

Finally, the Court concluded that the evidence and arguments were taken seriously by the Appeal Officer [63].

In dissent, Abella and Martin JJ would have held that the relevant section applies to both activities where control is not present and areas controlled by Canada Post. This reading was supported by the text of the provision and a more general legislative purpose contained in s.124: “Every employer shall ensure the health and safety at work of every person employed by the employer is protected.” The dissent claimed that the Appeal Officer’s reasoning was “deeply flawed” [100, citing Dunsmuir at para 72), because it “read out the words and purposes of the safety inspection duty…” [101].


In my view, the majority had the better of the argument here, mostly because the majority’s analysis is consistent with what the Supreme Court has recently said about statutory interpretation and because it is consistent with the principled deferential posture set out in Vavilov. The dissent makes a fundamental statutory interpretation mistake while arguably unreasonably parsing the Appeal Officer’s decision.

First, I would be remiss if I did not mention the determination of the standard of review. That analysis was undertaken by the majority in one paragraph. This is astounding for Canadian administrative law. Finally, it appears that we have accomplished Binnie J’s admonition in Dunsmuir that we move to merits quickly rather than focusing on determining the standard of review. In Canada Post, the emphasis is rightly where it should be: dealing with the statutory interpretation and the merits, particularly the reasons offered for the decision.

And on that score, there is much to analyze. Most importantly, and first, is the Court’s approach to the reasons of the Appeals Officer. The Court does what it says in this decision: it focuses on the reasons for decision, without parsing them, without creating its own yardstick, and without engaging in disguised correctness review. By focusing first on the coherence of the Appeals Officer’s reasoning, and then moving to legal constraints, the Vavilov framework guides and structures the application of the standard of review—and it did so here, in Canada Post.

Canada Post also demonstrates how Vavilov strikes the correct balance in ensuring that courts do not engage in disguised correctness review. Prior to Vavilov, the Court often engaged in disguised correctness review, sometimes not even mentioning the standard of review at all. But the reasons first methodology focuses the inquiry on what the decision-maker actually said, protecting against disguised correctness review. Instead of the court crafting its own interpretation, and disregarding the administrative reasons, it is for the decision-maker to demonstrate how she engaged with the “interpretive constraints” imposed by law; text, context, and purpose. Those constraints are real and binding as a matter of structuring administrative discretion, but the onus is properly on the delegated decision-maker to abide by them. Reasons are properly the gateway to deference, and its scope.

Here, the Appeal Officer’s reasons engaged with the text, context, and purpose. As the Court noted, the text could be conjunctively or disjunctively read: but the Appeal Officer’s reasoning, when taken in light of the purpose, was consistent with this reading. This meant that the Appeals Officer had some room to maneuver under the standard of review. Given the possible interpretive options, the Appeals Officer was owed deference.

But the dissent seemed to take a more activist approach that is profoundly inconsistent with Vavilov. The dissent instead came to its own interpretation of the relevant statute, and imposed that yardstick on the decision-maker. It did not take a reasons first approach; rather, it first determined the statutory meaning and then analyzed the Appeals Officer’s reasoning in light of that meaning. Instead, to be consistent with Vavilov, the dissent should have structured its analysis by first analyzing the reasons of the decision-maker, and then determining whether they engaged with the text, context, and purpose, which sets the range of interpretive options open to the decision-maker. This is not an approach that should find majority support.

On the merits, another issue cropped up in terms of statutory interpretation: how courts deal with purpose. This has been a common theme at the Supreme Court this year: see Telus v Wellman and Rafilovich. In particular, what divided the majority and dissent in Canada Post was the selection of the purpose that drove the analysis. For the majority, the obligations contained in s.125(1) fulfilled the broader purpose in Part II of the CLC, but the focus was on s.125 and what was required to fulfill those obligations: namely, control. For the dissent, it would have focused on s.124 as a general organizing purpose, and set s.125 as a particular instantiation of that purpose (see para 77).

The majority’s approach is more consistent with the Supreme Court’s recent statutory interpretation precedents, particularly Rafilovich. Rafilovich deals with what I call the horizontal frequency problem in statutory interpretation: where multiple purposes are stated across the statutory context at the same level of abstraction. Rafilovich basically held that courts should choose the purpose most local to the dispute at hand, not other purposes that may have more bearing on other sections of the statute. In essence, this may come to an assessment of weight; which purpose is more relevant?

This approach—choosing the most local purpose—is consistent with the task on judicial review to discover legislative intent and meaning in language. As I noted in reference to Rafilovich:

Applying this sort of thinking to Rafilovich, Justice Martin is clearly in the right. In this case, the most local purposes to the dispute at hand were the purposes speaking of access to justice and the presumption of innocence, assuming these purposes were identified correctly. Why must these purposes be prioritized over the general purpose? Because of the principle of democracy. The use of different language to express Parliament’s law in the legal fees provisions should lead to different interpretive outcomes. By this, I mean that ensuring crimes does not pay may be an overall purpose of the proceeds of crime provision, but Parliament clearly used different language and a different approach in the legal fees provisions. This different approach must, consequently, reflect different legislative purposes, as the legislative history in the case outlines (see para 39 et seq—though I cringe at the reliance on legislative history writ large). The court must give “purpose and meaning to each provision” (at para 20).

In my view, Vavilov does not change any of this, and is arguably actually consistent with Rafilovich by focusing on the ordinary principles of interpretation. In reference to Canada Post, then, the most local purpose to the interpretive difficulty is s.125(1), and its particular obligations. Section 124, taken too far by the dissent, could be read to impose an obligation on employers no matter whether they control the workplace or not. But section 125(1), properly interpreted in its context, would only apply some of the obligations to situations of control. This best effectuates the general purpose of Part II of the CLC and s.125 because of the practical reasoning employed by the Appeals Officer. If s.124 was taken to its literal extent, it would mean Canada Post has an obligation to conduct inspections on all sorts of routes and mail delivery locations. This would undermine the general safety purpose—as the Court noted “[a]n interpretation which imposed on the employer a duty it could not fulfil would do nothing to further the aim of protecting accidents and injury” [59]. This is simply an application of absurdity avoidance; but taking s,124 at its word could conceivably lead to this absurdity.


All in all, Vavilov’s first day in the sun was a good one. While there will be more battles to come, and potentially more ambiguities, for now we must celebrate. Those in the administrative law community do not often get to do so.

Vavilov: A Step Forward

**This post originally appeared on Advocates for the Rule of Law**

Today, the Supreme Court of Canada released its decisions in Vavilov and Bell/NFL. I have previously summarized the facts of these cases and analyzed them here (Vavilov) and here (Bell/NFL).

Overall, today’s decisions (a 7-2 decision, Abella and Karakatsanis JJ concurring in result) are a net positive for the law of judicial review in Canada. The Court has done a service to the many individuals who face administrative power on a daily basis—the real winners of a clear, predictable, and conceptually sound approach. To be honest, I count myself surprised and pleased, which are odd emotions to carry when it comes to the Supreme Court’s administrative law jurisprudence. But on at least three issues, today’s decision is conceptually sound when it comes to selecting the standard of review: statutory rights of appeal, jurisdictional questions and legislative interpretation generally, and expertise. I first describe the new approach before moving to analyze these areas.

The basic takeaway from Vavilov and Bell/NFL is that the standard of review will presumptively be reasonableness across the board [16-17], but the situations in which the presumption will be rebutted take on greater importance than the pre-Vavilov Bell/NFL cases. Start with the presumption of reasonableness. It remains; this is not much of a change from existing jurisprudence, given the entrenchment of that particular rule in pre-Vavilov and Bell/NFL cases: see Edmonton East and CHRC. What is somewhat new are the circumstances in which the presumption will be rebutted. First, because the standard of review analysis is based in the common law, the legislature is free to explicitly legislate the standard of review [17]. But also—and significantly, as I will point out—now statutory rights of appeal will rebut the presumption of reasonableness and attract correctness review on questions of law (under the typical appellate standards of review) [17, 36 et seq]. Finally, where the rule of law requires correctness review—as in constitutional questions, general questions of law of central importance, and boundaries between two or more administrative bodies—the standard will be correctness [17]. Gone is the search for “context” [17], and importantly, gone is that much maligned category of jurisdictional questions [67].

Now, turning to the significant parts of this decision. The first area is statutory rights of appeal. Statutory rights of appeal have previously been considered a big nothingburger from the Supreme Court: see Dr. Q, Saguenay. As I wrote here, this state of affairs was completely unjustified. This is because rights of appeal are implied signals that a legislature intended the judicial review court to interfere with the administrative decision as it would in a normal appeal. This is contrasted with an application for judicial review, where common law standards apply. All of this is a function of the relationship between statutory and common law—statutes override the common law, and so statutory rights of appeal should override the common law standard of presumptive deference. The Court has now recognized this fact, offering the additional reason that the word “appeal” when used in legislation must be interpreted across different statutory contexts consistently [44]. This is a step in the right direction: it corrects the relationship between statute and common law that went awry in the Dr. Q line of cases.

Second, take the issue of jurisdictional questions. In the Dunsmuir era, jurisdictional questions were questions “where the tribunal must explicitly determine whether its statutory grant of power gives it authority to decide a particular matter” (Dunsmuir, at para 59). These questions were to attract correctness review (Dunsmuir, at para 50). No more. The Court has wisely done away this category of correctness review [67]. The Court noted the following:

The arguments that support maintaining this category — in particular the concern that a delegated decision maker should not be free to determine the scope of its own authority — can be addressed adequately by applying the framework for conducting reasonableness review that we describe below. Reasonableness review is both robust and responsive to context. A proper application of the reasonableness standard will enable courts to fulfill their constitutional duty to ensure that administrative bodies have acted within the scope of their lawful authority without having to conduct a preliminary assessment regarding whether a particular interpretation raises a “truly” or “narrowly” jurisdictional issue and without having to apply the correctness standard [67].

This is the approach adopted in the US: City of Arlington. And it is the approach I articulated in my paper, Two Myths of Administrative Law. It is a sound approach. All administrative actors are empowered and created by statute. In turn, the relationship between courts and administrative actors is also regulated by the degree of deference that the legislature prescribes. This is true on all questions that might come before a court—including cases that raise so-called “jurisdictional issues.” In other words, even assuming one can identify a jurisdictional question—a feat the Supreme Court has rarely been able to accomplish—there is no warrant to impose a less deferential standard if the legislature has signalled that it prefers a more deferential one on these questions. As Stratas JA pointed out in Access Copyright, a jurisdictional question “is really a question of legislative interpretation” (Access Copyright, at para 57). Here, again, the Court takes seriously the statutory basis of all administrative law, choosing to do away with a category of question that might undermine that basis.

In relation to questions of legislative interpretation, the Court does an admirable job. In applying the standard of reasonableness, the Court recognizes, first, that all administrative actors are creatures of statute [108]. This means that there is no constitutional aspect to the administrative state–decision-makers can only exercise powers delegated to them. The Court goes on to recognize that the degree of deference owed is primarily–but not totally–a function of statutory interpretation [110]. The Court notes that:

Whether an interpretation is justified will depend on the context, including the language chosen by the legislature in describing the limits and contours of the decision maker’s authority. If a legislature wishes to precisely circumscribe an administrative decision maker’s power in some respect, it can do so by using precise and narrow language and delineating the power in detail, thereby tightly constraining the decision maker’s ability to interpret the provision. Conversely, where the legislature chooses to use broad, open-ended or highly qualitative language — for example, “in the public interest” — it clearly contemplates that the decision maker is to have greater flexibility in interpreting the meaning of such language. Other language will fall in the middle of this spectrum. All of this is to say that certain questions relating to the scope of a decision maker’s authority may support more than one interpretation, while other questions may support only one, depending upon the text by which the statutory grant of authority is made.

This is a justified approach, given that all administrative actors are creatures of statute. Determining the scope of deference owed need not depend on metaphysical categories like “jurisdictional questions.” It is simply a function of determining the scope of power afforded to a decision-maker using the ordinary principles of interpretation [117]. This is a wise move that is conceptually sound and will bring greater clarity to the law.

Finally, I want to say something about the idea of expertise, which divided the majority and the concurrence in Vavilov. As I have written before, expertise is an inherently difficult idea in administrative law. The Court has taken an extreme position on expertise, holding in Edmonton East that expertise inheres in a tribunal as an “institution” (Edmonton East, para 33). But expertise is not necessarily a legal reason for deference—for example, a legislature could delegate to an inexpert tribunal just as easily as it can delegate to an expert one. The presumption of deference—based as it is on expertise—is therefore overbroad and unjustified.

The Court tacitly recognizes this in Vavilov. It notes that expertise is no longer a free-standing reason to justify deference in determining the standard of review [31]. This is partially because “if administrative decision makers are understood to possess specialized expertise on all questions that come before them, the concept of expertise ceases to assist a reviewing court in attempting to distinguish questions for which applying the reasonableness standard is appropriate from those for which it is not” [28]. And this is basically right. We cannot simply assume expertise.

The concurrence takes significant issue with this, and more generally the majority’s reasons. To the concurrence, the majority “ignores the specialized expertise of administrative decision-makers” [230]; “affords no weight to the institutional advantage of specialization and expertise that administrative decision-makers possess in resolving questions of law” [236]; and empowers reviewing courts to “freely revisit legal questions on matters squarely within the expertise of administrative decision-makers…” [251].

The death cries for expertise are unconvincing. As the majority noted, there may be situations in which decision-makers are inexpert. But courts cannot conduct a case-by-case analysis of expertise in each case. So, the best way to deal with expertise is to simply do away with it as a legal reason for deference, as far as selecting the standard of review. As the majority notes, expertise could play a role in the application of the reasonableness standard, especially in the process of reasons giving [93].

I want to briefly deal with one issue raised by the concurrence: the issue of the Rule of Law. The dissent says that “[t]he majority’s approach to the rule of law, however, flows from a court-centric conception of the rule of law rooted in Dicey’s 19th century philosophy” [240]. But this attack misses the mark. Judicial review, the cornerstone of these appeals, is fundamentally about courts reviewing administrative decision-making under the Rule of Law; it is about policing the boundaries of delegated power to ensure its legality (Wall, at para 13). This necessarily implies a hierarchical relationship between courts and decision-makers. It is the preserve of courts to ensure that administrative decision-makers follow statutory rules. The Court recognizes this in its analysis of jurisdictional questions, endorsing Scalia J in Arlington [68]. This is especially so when dealing with the Constitution, as the judiciary is the guardian of the Constitution. Correctness review in those circumstances protects the court’s role in ensuring that individuals are protected against administrative power. The Court has now recognized this, at least implicitly, as an organizing principle of administrative law.

Overall, as I mentioned above, this is a good decision all around. And more could be said about it, especially on the application of the reasonableness standard and the role of reasons. But I think that, for now, administrative law watchers in Canada can breathe a sigh of relief.