Context, Reasonableness Review and Statutory Interpretation: Mason v. Canada (Citizenship and Immigration), 2023 SCC 21

After a hiatus of nearly four years, the Supreme Court of Canada yesterday applied the reasonableness standard for the first time since Vavilov and the companion case of Canada Post. The decision in Mason v. Canada (Citizenship and Immigration), 2023 SCC 21 (Jamal J for the majority; Côté J concurring) is significant as far as the methodology of reasonableness review is concerned, in respect of (1) the need for an administrative tribunal to grapple with submissions decided from the parties; (2) the extent to which implied reasons can support the reasonableness of a decision; and (3) when a reviewing court can consult a statute. I will start, however, by discussing the majority’s choice of standard of review, which may sound the death knell for contextual analysis.

The Death of Context?

Mason is an immigration law case. In the Immigration and Refugee Protection Act, s. 74(d) there is a provision permitting the Federal Court, having heard a judicial review application about an immigration or refugee matter, to certify a question of general importance for resolution by the Federal Court of Appeal. For years, this provision was interpreted as requiring correctness review on questions of law. As the Supreme Court explained in Pushpanathan v. Canada (Minister of Citizenship and Immigration), [1998] 1 SCR 982, at para. 43:

The key to the legislative intention as to the standard of review is the use of the words “a serious question of general importance” (emphasis added).  The general importance of the question, that is, its applicability to numerous future cases, warrants the review by a court of justice.  Would that review serve any purpose if the Court of Appeal were obliged to defer to incorrect decisions of the Board?  Is it possible that the legislator would have provided for an exceptional appeal  to the Court of Appeal  on questions of  “general importance”, but then required that despite the “general importance” of the question, the court accept decisions of the Board which are wrong in law, even clearly wrong in law, but not patently unreasonable?  The only way in which s. 83(1) can be given its explicitly articulated scope is if the Court of Appeal — and inferentially, the Federal Court, Trial Division — is permitted to substitute its own opinion for that of the Board in respect of questions of general importance. 

In 2015, the Supreme Court abruptly changed course, applying the reasonableness standard to a question of interpretation in Kanthasamy v. Canada (Citizenship and Immigration), 2015 SCC 61, [2015] 3 SCR 909. I do not propose to rake over the coals of this dispute (which is well described by Côté J in her compelling concurring reasons at paras. 128-137) but simply to highlight that the approach to the certified question regime has long been a significant source of contention (see further here and here). 

Now, of course, the point of departure is the Vavilov framework. With colleagues I appeared for the intervener Canadian Association of Refugee Lawyers before the Court in Mason. We argued that under the Vavilov framework — as applied notably in Society of Composers, Authors and Music Publishers of Canada v. Entertainment Software Association, 2022 SCC 30 — the correctness standard should apply. We leaned heavily on the apparent return to contextual analysis in ESA (described here), making legislative intent and rule of law arguments consistent with the approach in ESA. I will not repeat the arguments here as they are thoroughly and clearly developed by Côté J in her concurring reasons at paras. 146-176. Suffice it to say that having put our fingers through a door that was opened by ESA, in Mason, the door was slammed shut by Jamal J:

[R]ecognizing a new correctness category here would conflict with Vavilov’s goal of simplifying and making more predictable the standard of review framework by providing only limited exceptions to reasonableness review (para. 47). Treating s. 74(d) as justifying correctness review would effectively reintroduce a “contextual” approach to the standard of review — with the certification of a serious question of general importance being a “contextual” factor suggesting correctness — and thus would revive the approach that Vavilov eliminated because it created “uncertainty” and was “unwieldy” (para. 7; see also P. Daly, “Unresolved Issues after Vavilov” (2022), 85 Sask. L. Rev. 89, at pp. 91-92 (Vavilov is “an exercise in simplification and clarification” that “excised” the “‘vexing contextual factors’ . . . from the standard of review selection exercise”.)) (at para. 53).

Although I let out a yelp of pain for my injured fingers and the definitive demise of correctness review on certified questions under the IRPA, I am not displeased at the overall outcome (as the citation to my work correctly suggests). Vavilov did simplify and clarify the law; rejecting context on the selection of the standard of review was an important part of the simplification and clarification exercise. Our argument for correctness review in Mason was narrowly tailored (see Côté J’s conurring reasons at para. 163) but a firm rejection of contextual analysis will strongly dissuade lower courts from opening up the correctness categories. Indeed, Jamal J did not give an inch on the scope of the existing categories. Certified questions are not, he held, general questions of law of central importance to the legal system:

The proper interpretation of s. 34(1)(e) of the IRPA is not a “general question of law of central importance to the legal system as a whole” under Vavilov (paras. 58‑62). Such “general questions of law” require uniform and consistent answers because of their impact on the administration of justice as a whole or for other institutions of government (para. 59). Examples of general questions of law include questions with legal implications for many other statutes or for the proper functioning of the justice system as a whole (paras. 59‑61). It is not enough for the question to “touc[h] on an important issue” or to raise an issue of “wider public concern” (para. 61). Although the proper interpretation of s. 34(1)(e) is important for the affected persons and the proper administration of the IRPA, it does not affect the legal system or the administration of justice as a whole, have legal implications for many other statutes, or affect other institutions of government. Rather, the issues raised are particular to the interpretation of the conditions for inadmissibility under s. 34(1)(e) (at para. 47).

This is commendably clear. I am happy to lose the battle over the IRPA if it means simplicity and clarity prevail in the wars that have raged for decades in Canadian administrative law.

Nevertheless, we are now in a situation where in its first two discussions of standard of review subsequent to Vavilov, the Supreme Court has taken divergent approaches, leaning heavily on context in ESA, but lurching the other way in Mason. It is, after all, remarkable that a statute that said nothing at all about judicial review or appeals was held in ESA to require correctness review but a statute with elaborate mechanisms relating to judicial review and appeal was held in Mason not to evidence any sort of relevant legislative intent.

I have the distinct feeling that had Mason been heard and decided before ESA, the standard of review outcomes would have been reversed. Yet unexplained divergence from one year to another — of just this sort — was one of the ills that led to Vavilov. If the Supreme Court wishes to achieve the simplicity and clarity promised by Vavilov, it needs to pick an approach and stick with it. I hope (for the Supreme Court’s sake) that yesterday’s rejection of context is definitive.

The Methodology of Reasonableness Review

The standard being reasonableness, Jamal J then turned to the question of whether the decision at issue was reasonable. The particular issue here related to the inadmissibility provisions of the IRPA, which determine who may or may not gain status in Canada. Subparagraph 34(1)(e) makes inadmissible anyone “engaging in acts of violence that would or might endanger the lives or safety of persons in Canada”. In concrete terms, a finding of inadmissibility will typically lead to deportation from Canada. M was charged with attempted murder after discharging a firearm and injuring two people at a concert at the Canadian Legion in British Columbia. The charges were stayed, however, and the applicant was thus not convicted. The Minister argued — and the Immigration Appeal Division agreed — that the applicant’s conduct came within s.34(1)(e). M argued on judicial review that it was unreasonable to interpret s.34(1)(e) as encompassing “acts of violence” which do not occur in the context of terrorism, war crimes or organized criminality, these being the concerns underpinning the inadmissibility provisions of the IRPA. There must be, M argued, a national security nexus to s. 34(1)(e). Pre-Vavilov, the Federal Court (Grammond J) agreed; post-Vavilov the Federal Court of Appeal (Stratas JA; Rennie and Mactavish JJA concurring) found that the tribunal’s decision was reasonable.

Jamal J held that the tribunal’s decision was unreasonable. The starting point for his analysis was the reasons of the tribunal (described at para. 83), understood against the need to for those reasons to “reflect the stakes” (at para. 81): potential deportation from Canada. From here, he worked out to identify three flaws in the decision which, cumulatively, led to the conclusion of unreasonableness. Although the tribunal had “applied several recognized techniques of statutory interpretation” (at para. 84), its decision could nonetheless not withstand reasonableness review.

First, the tribunal had failed to grapple with M’s argument that s. 34(1)(e) requires a security nexus because the availability of discretionary relief (from the responsible minister) was narrower for s. 34(1)(e) than the relief available for serious criminality and criminality offences (which may lead to inadmissibility under s. 36). The logic of M’s argument is that the context of the statute indicates that national security crimes are to be treated more seriously than other crimes, but on the opposing interpretation, less serious crimes (indeed, here, one for which there was not even a conviction) would carry the most serious possible consequences. This was a key argument advanced by M but the tribunal did not address it:

The IAD’s reasons, read with sensitivity and in light of the record, did not address this important contextual argument. While perhaps not in itself determinative, this argument supported Mr. Mason’s position and imposed a significant legal constraint on the interpretation of s. 34(1)(e) (at para. 91).

The tribunal had also failed to grapple with a related argument about the impact of a finding under s. 34(1)(e) on the pre-removal risk assessment to be conducted by the responsible minister before deporting M or a similarly situated person. Again, M’s argument was that the statutory scheme would only make sense if the most serious consequences attached to the most serious inadmissibility ground, namely national security:

Mr. Mason argued before the IAD that these diverging considerations for pre-removal risk assessments suggest that Parliament contemplated that conduct captured by s. 36 poses a danger to the public in Canada, while conduct captured by s. 34 poses a danger to the security of Canada (A.R., vol. II, at pp. 30‑31). He claimed that this supported his position that the security grounds under s. 34 require a nexus to national security or the security of Canada. Again, the IAD’s reasons did not address this important contextual argument, which, while not in itself determinative, supported Mr. Mason’s position and imposed a significant legal constraint on the interpretation of s. 34(1)(e) (at paras. 94-95).

Second, the tribunal failed to grapple with M’s argument that not requiring a national security nexus would lead to absurd consequences that Parliament could not possibly have intended:

The acts of violence caught by s. 34(1)(e) would, he submitted, extend from “domestic altercations” to “bar fights and schoolyard fights” (p. 31). I note that unlike a criminal conviction triggering inadmissibility under s. 36, which must be proven beyond a reasonable doubt, the facts triggering inadmissibility under s. 34 are subject to the much lower standard of “facts for which there are reasonable grounds to believe that they have occurred, are occurring or may occur” (IRPA, s. 33). Thus, s. 34(1)(e) would extend to any “acts of violence” — past, present, or future — that “would or might” “endanger the lives or safety” of “persons in Canada”. Mr. Mason submitted that it would be absurd for such acts to be captured by s. 34(1)(e).

As for the second broad consequence, Mr. Mason submitted to the IAD that if s. 34(1)(e) is interpreted without a nexus to national security or the security of Canada, it would do an “end-run around the limitations under s. 36(3)(e) with respect to youth offences” (A.R., vol. II, at pp. 31‑32). Section 36(3)(e) of the IRPA provides that young persons who are permanent residents or foreign nationals found guilty under the Young Offenders Act, R.S.C. 1985, c. Y-1, or who received a youth sentence under the Youth Criminal Justice Act, S.C. 2002, c. 1, cannot be found inadmissible under s. 36(1) (serious criminality) or s. 36(2) (criminality). In effect, under s. 36(3)(e), Parliament exempted young persons from inadmissibility for most criminal offences. But on the IAD’s interpretation of s. 34(1)(e), young persons can be found inadmissible for any acts of violence that would or might engage the lives or safety of persons in Canada, even without a nexus to national security or the security of Canada, and even without a criminal conviction (at paras. 99, 102).

Again, the tribunal should have considered this point, which was not a “minor aspect” of the interpretive context (at para. 103).

Third, the tribunal had failed to consider Canada’s international law obligations. Jamal J’s analysis is lengthy, but the key point is as follows:

The IAD’s interpretation allows a foreign national found inadmissible under s. 34(1)(e) to be subject to refoulement contrary to Article 33(1) of the Refugee Convention. On the IAD’s interpretation, a foreign national can be deported to persecution once they are found inadmissible under s. 34(1)(e), without a finding that the person poses a danger to the security of Canada or even if they have not been convicted of a serious offence. Such a person would be entitled to the benefit of Article 33(1) of the Refugee Convention, as the exceptions under Article 33(2) would not apply: on the IAD’s approach to inadmissibility under s. 34(1)(e), there need not be “reasonable grounds” to regard the foreign national as a “danger to the security” of Canada, or for them to have been “convicted by a final judgment of a particularly serious crime” (at para. 109).

As the IRPA must always be interpreted with Canada’s international obligations in mind, this was a fatal flaw in the tribunal’s reasons. It “involved the omission of the principle of non-refoulement — “the cornerstone of the international refugee protection regime”, and a critical legal constraint on interpretation of the IRPA, one that Parliament has decreed must be considered in construing and applying the IRPA” (at para. 113). This was a “crucial omission” and the decision was therefore unreasonable (at para. 113).

Indeed, there was in reality only one reasonable interpretation of s. 34(1)(e), namely that a national security nexus is required:

The relevant legal constraints cumulatively point overwhelmingly to only one reasonable interpretation of s. 34(1)(e) — the provision requires a nexus to national security or the security of Canada. This interpretation is supported by the placement of s. 34(1)(e) amid other “security grounds” in s. 34, all of which have a link to national security or the security of Canada. And although the IAD noted that some interpretive constraints point in the other direction, the two critical points of statutory context that the IAD failed to consider, and especially, the legal constraints imposed by international law, overwhelmingly support the opposite conclusion: s. 34(1)(e) can be invoked to render a person inadmissible only when their “acts of violence that would or might endanger the lives or safety of persons in Canada” have a nexus with national security or the security of Canada (at para. 121).

But this conclusion resulted from what one might term an internal rather than external approach, working out from the reasons rather than beginning with the statutory provisions themselves (2021 FCA 156, at para. 19).

There are three broad implications for the methodology of reasonableness review.

First, although it is now almost trite to say this, an administrative decision-maker must grapple in its reasons with the submissions of the parties. Notice that the tribunal decision here was handed down long before Vavilov was decided. I think tribunals from coast to coast to coast now appreciate that it is necessary to address the arguments made to them in order to render a reasonable decision. This is the principle of responsive justification that was central to Vavilov. The only relevant limits are that the requirement to grapple with a submission only relates to a “key” argument. In this instance, the arguments were “core planks” supporting M’s position (at para. 97).

This is a remarkable feature of Vavilov. Not only are the tribunal’s reasons treated as the starting point — elevating the tribunal to (close to) equal partner status as far as the interpretation of law is concerned — but the tribunal’s reasons are shaped by the interaction between the tribunal and the citizen. Vavilov requires, one might say, citizen-led law-making.

Second, in working out whether a tribunal has grappled with key arguments, a reviewing court must exercise significant caution before inferring that an argument has been addressed. The Federal Court of Appeal had held that the tribunal implicitly turned its mind to the arguments. Jamal J firmly disagreed (at paras. 96, 101). As these were key arguments — “core planks” — then they had to be met head-on, by explicit reasons.

Working out clear parameters here is difficult. The issue is how much “sensitivity to the institutional setting” and reference to the “history and context of the proceedings” a reviewing court can engage in (Vavilov, at paras. 94, 96) to deduce that arguments were dealt with. Reference to background context is permissible to plug some holes in a tribunal’s reasons, but evidently not all of them. Here is what I wrote (commenting on the Federal Court of Appeal decision) in A Culture of Justification:

For my part, the term “implied” or “implicit” reasons is too strongly associated with the darkest days of the Dunsmuir decade, so I prefer to simply say that administrative decisions should be read fairly, in their whole context. The point is well explained at paras 93–94 of Vavilov … without mention of the words “implied” or “implicit”…I would say that the best way to summarize this passage is that a decision is not unreasonable because of a failure to expressly mention a particular point, where it is obvious why the decision maker did not consider it. If a test is needed, it should be a test of obviousness (pp. 213-214, note 49).

Put another way, the court “must be sure” that an argument has been addressed (Zeifmans LLP v. Canada, 2022 FCA 160, at para. 10). This chimes with the Supreme Court’s analysis in Mason. Plainly, Jamal J was not “sure” that the tribunal had addressed the key arguments. The lesson for tribunals is to err on the side of dealing with an argument; the lesson for reviewing courts is to err on the side of not inferring that an argument was addressed in the absence of explicit grappling.

Third, the treatment of international law creates a tension in Jamal J’s reasons and, indeed, recalls a foundational tension in Vavilov. The tribunal did not hear arguments about international law. Yet Jamal J held that international law was a significant legal constraint on the tribunal.

On the one hand, Jamal J criticized the Federal Court of Appeal for introducing an additional step into the Vavilov analysis. As the court below described it, the judge should undertake “a preliminary analysis of the text, context and purpose of the legislation just to understand the lay of the land before they examine the administrators’ reasons” (2021 FCA 156, at para. 19). For Jamal J, “Vavilov is clear that a reviewing court must start its analysis with the reasons of the administrative decision maker; starting with its own perception of the merits may lead a court to slip into correctness review” (at para. 79).

On the other hand, Jamal J sought out relevant provisions in the IRPA that made clear that Parliament intended the IRPA to be interpreted in conformity with Canada’s international obligations:

The presumption of conformity with international law assumes added force when interpreting the IRPA, because Parliament has made its “presumed intent to conform to Canada’s international obligations explicit” through two provisions of the IRPA (B010 v. Canada (Citizenship and Immigration), 2015 SCC 58, [2015] 3 S.C.R. 704, at para. 49). First, s. 3(2)(b) of the IRPA expressly identifies one of the IRPA’s objectives as being “to fulfil Canada’s international legal obligations with respect to refugees and affirm Canada’s commitment to international efforts to provide assistance to those in need of resettlement”. Indeed, this Court has described the IRPA as the “main legislative vehicle for implementing Canada’s international refugee obligations” (Németh, at para. 21). Second, s. 3(3)(f) of the IRPA instructs courts and administrative decision makers to construe and apply the IRPA in a manner that “complies with international human rights instruments to which Canada is signatory” (B010, at para. 49). This Court has stated that “[t]here can be no doubt that the Refugee Convention is such an instrument, building as it does on the right of persons to seek and to enjoy asylum from persecution in other countries as set out in art. 14 of the Universal Declaration of Human Rights, G.A. Res. 217 A (III), U.N. Doc. A/810, at 71 (1948)” (para. 49). As a result, the Refugee Convention is “determinative of how the IRPA must be interpreted and applied, in the absence of a contrary legislative intention” (de Guzman v. Canada (Minister of Citizenship and Immigration), 2005 FCA 436, [2006] 3 F.C.R. 655, at para. 87; B010, at para. 49).

This was surely based on a consideration — far from preliminary — of the text of the IRPA, which the tribunal did not consider and which it was not asked to consider. Citizen-led law-making has its limits!

More seriously, I think this issue requires careful unpacking. To begin with, the Federal Court of Appeal was riffing on a theme I developed before Vavilov in a post engaging with Grammond J’s thoughtful, sophisticated reasons at first instance in Mason: the ‘internal’ approach I advocated was designed to produce deferential judicial review and is consistent with Vavilov; the ‘external’ approach I described was explicitly rejected in Vavilov. The Federal Court of Appeal’s preliminary view methodology was therefore not designed to lead to less deferential judicial review. Quite the opposite.

This does not mean (at least, I did not intend it to mean) that a reviewing court should literally develop a preliminary view before it even reads the tribunal’s reasons. My ‘internal’ approach sought to ensure that a reviewing court has some basic understanding of the statutory structure. Without such an understanding, a reviewing court may well be at a loss to know which arguments are important and which statutory provisions might plausibly have figured in the tribunal’s reasons. Indeed, as a practical matter, a reviewing court will be introduced to the relevant provisions by the parties’ written submissions. It has, therefore, to have some idea of how the decision is situated relative to the statutory scheme. Jamal J’s reliance on the international law provisions of the IRPA is proof positive that a reviewing court cannot conduct reasonableness review in abstraction, sealed away hermetically from statutory text, context and purpose.

This recalls a tension within Vavilov itself. As I noted shortly after Vavilov was handed down:

Whilst it is pellucidly clear (and most welcome) that a reviewing court is not to conduct its own statutory interpretation exercise to establish a benchmark or yardstick against which to measure an administrative decision-maker’s interpretation of law (at para. 124), the finer details are murky. On the one hand, the reader is told: “Administrative decision makers are not required to engage in a formalistic statutory interpretation exercise in every case” (at para. 119). On the other hand, a few paragraphs later, the administrative decision-maker’s task is said to be to “interpret the contested provision in a manner consistent with the text, context and purpose, applying its particular insight into the statutory scheme at issue” (at para. 121). That looks awfully like a “formalistic statutory interpretation exercise”, one which judges suspicious of an administrative decision-maker’s ability to issue interpretations of law might well require.

Here, Jamal J is plainly sympathetic to the deferential aspects of what was said in Vavilov about statutory interpretation. I think he would agree that an administrative decision-maker’s approach might “actually enrich and elevate the interpretive exercise” (Vavilov, at para. 119). Nonetheless, Vavilov also says that an administrative decision-maker “must” consider statutory text, context and purpose, and in Mason one of the reasons the tribunal’s interpretation could not stand was that it had failed to consider elements of the statutory scheme that Jamal J adjudged important. This tension is present in Vavilov. Mason does not resolve it.

One thing is clear, however. Under the IRPA, consideration of international obligations is going to be mandatory, or close to mandatory, in statutory interpretation going forward. Vavilov made clear that international law is sometimes a relevant constraint. ESA made clear that international obligations are relevant to the context prong of the statutory interpretation analysis. Mason makes clear that international law is a relevant constraint on decision-makers under the IRPA. Failure to consider relevant international law obligations in the IRPA context will generally lead to unreasonable decisions. Maybe the best way to think about the third aspect of the reasonableness analysis in Mason is that Parliament made a uniquely powerful, textually explicit commitment to implementing Canada’s international law obligations in the IRPA, which is unlikely to have similar force in any other context (save, perhaps, citizenship).

To close, let me draw a link between the discussion of standard of review and the end of the discussion of reasonableness review. There are tensions in the Supreme Court’s post-Vavilov case law already. There are tensions within Vavilov itself, some of which are evident in Mason. I can only applaud Jamal J’s clear and cogent analysis and his commitment to deferential judicial review. But ensuring that the tensions within the Vavilov framework and jurisprudence do not cause serious difficulties will require the Supreme Court to chart a consistent course and stick to it.

UPDATE: I accidentally described Justice Côté as dissenting when, of course, she actually concurred in the judgment. Sorry!

This content has been updated on September 28, 2023 at 21:08.