Yesterday, the Supreme Court decided Mason v Canada (Citizenship and Immigration), 2023 SCC 21, which could turn out to be an important administrative law decision ― though my past assessment of how important administrative law cases are likely to be have often been badly wrong. Mason is noteworthy for being, in some ways, a resounding confirmation of Canada (Minister of Citizenship and Immigration) v Vavilov, 2019 SCC 65, [2019] 4 SCR 653, and in some a clarification of that decision, potentially resolving some of the tensions latent in it in favour of robust judicial review of administrative decisions. At the same time, Mason provides yet another illustration of Vavilov‘s unsoundness in principle, even as it makes a commendable effort at making it work in practice.
At issue in Mason is s 34(1)(e) of the Immigration and Refugee Protection Act (IRPA), which makes non-citizens “inadmissible” to Canada ― or, as here, liable to be expelled if already present in Canada ― “on security grounds for … engaging in acts of violence that would or might endanger the lives or safety of persons in Canada”. Specifically, the question is whether the acts of violence in question had to be connected to “the security of Canada” as a nation, or whether ordinary, albeit perhaps relatively serious, violence that endangers individual Canadians but not the country would suffice. While at first glance the statutory text seems broad, this provision is only one of several that deal with inadmissibility, including inadmissibility on the grounds of “serious criminality” and “criminality” tout court. The latter two require criminal conviction; s 34(1)(e) does not. There are also differences in the recourses available to people declared inadmissible under different provisions, including in that people found to be inadmissible under s 34(1)(e), unlike those inadmissible on grounds even of “serious criminality”, are unable to seek relief on “humanitarian and compassionate” grounds.
The Immigration Appeal Division of the Immigration and Refugee Board considered that s 34(1)(e) can apply to any acts of violence, regardless of any connection to national security. The Federal Court disagreed and set its decision aside, but gave the government permission to appeal by certifying a “serious question of general importance”, as permitted by the IRPA. The Federal Court of Appeal restored the Appeal Division’s decision, holding that, while not explicit on certain points, it was not unreasonable.
Before getting to the substance of the case, the Supreme Court must decide on the standard of review. Generally, under Vavilov, when administrative decision-makers like the Appeal Division interpret legislation, they are entitled to deference, and the reviewing court will only ask if they have sufficiently justified their interpretation to make it reasonable. But this is only a presumption, which can be rebutted if the legislation signals that the reviewing court should impose its own view of the legislation or, even in the absence of such a signal, if, among other things, the case presents a “general question of law of central importance to the legal system as a whole”. Is, then, the certified question mechanism either a legislative signal for correctness review or an indication that a question that has been certified is one of central importance?
The Supreme Court splits on this issue. The majority, in reasons by Justice Jamal, says “no”. It says that precedent supports reasonableness review even when the Federal Court has certified a question and that, important though it may be for the individuals to whom the provision might be applied, the interpretation of s 34(1)(e) is not of central importance to the legal system as a whole, notably because it only matters within the context of the administration of the IRPA. The majority also points out that no certified question exists until the Federal Court gives judgment on a judicial review application, which means that it would, logically, have no reason or even opportunity to apply correctness review in a case that, as it ultimately turns out, will eventually present such a question.
For her part, Justice Côté, who concurs in the result and, on most other points, in Justice Jamal’s reasoning, thinks that certified questions should be reviewed for correctness. The precedents are not nearly as unequivocal as the majority suggests. More importantly, the certification mechanism is a signal that Parliament wanted the courts to be clarifying the meaning of the Immigration and Refugee Protection Act, instead of leaving it to administrative decision-makers; indeed, the certification of questions would be pointless if they did not have to be definitively answered by the Federal Court of Appeal. This makes sense, moreover, in light of the fact that many provisions are subject to concurrent jurisdiction of different decision-makers in the immigration ecosystem. But the Rule of Law principle also requires correctness review. Justice Côté argues that because
[a] question whose answer turns on the unique facts of the case will not be certified … [b]y definition … certified questions concern issues of broad significance or general importance within Canada’s immigration and refugee protection regime. [160]
Correctness review is then in order. Moreover, when expulsion from Canada, possibly to persecution or even torture is at stake, nothing less than a correct legal interpretation can suffice to justify such action by the government and avoid arbitrariness that is antithetical to the Rule of Law.
Turning to the substantive issue, Justice Jamal begins by recapitulating ― at remarkable length ― Vavilov‘s instructions on conducting reasonableness review. He also reproves the Federal Court of Appeal for having suggested that a reviewing court should start with “a preliminary analysis of the text, context and purpose of the legislation just to understand the lay of the land before . . . examin[ing] the administrators’ reasons”. Justice Jamal worries that this is tantamount to the court “starting with its own perception of the merits”, which “may lead [it] to slip into correctness review”. [79] Justice Jamal summarizes the Appeal Division’s reasons, noting that they “applied several recognized techniques of statutory interpretation”. [84] However, he faults them for failure to address some key arguments.
First, it ignored the differences between the recourses available to people found inadmissible under s 34(1)(e) and other inadmissibility provisions, leading to a paradoxical situation where people who fall within the scope of the former on the basis of conduct that might have amounted to a crime for which they have, nonetheless, not been convicted are worse off than those actually convicted of that same crime and found inadmissible on the basis of that conviction. Justice Jamal writes that these considerations might not be “determinative” [91, 95] but failure to address them is a reviewable error, which could not be disregarded by inferring, as the Federal Court of Appeal did, that the Appeal Division considered them implicitly. Second, the Appeal Division did not sufficiently consider the potentially sweeping nature of its interpretation of s 34(1)(e). And third, it did not attend to “the legal constraints imposed by international law on its interpretation”. [104] The interpretation adopted by the Appeal Division could result in Canada expelling people who are not a danger to its security, in violation of its commitments under the Convention Relating to the Status of Refugees, to which IRPA explicitly points as a mandatory interpretive consideration. Justice Côté agrees with these points, emphasizing the significance of expulsion to the person affected.
Having concluded that the Appeal Division’s interpretation of s 34(1)(e) is unreasonable, Justice Jamal addresses the question of the remedy. Normally, deference means that the administrative decision-maker will be asked to interpret the statute anew, albeit drawing on the reviewing court’s guidance. But here, “[t]he 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”. [121]
Let me start with what I see as the positive aspects of this decision. As I have argued here and elsewhere, and as other said too, after Vavilov, there was a great deal of uncertainty about what its actual implications could be. While co-blogger Mark Mancini and I expressed some cautious optimism that it would lead to meaningful judicial review of administrative interpretations of law, it could also, quite plausibly, have been read with an emphasis on deference and judicial restraint, leading to continuing judicial abdication to the administrative state. As I noted at the time, this ambiguity, which was probably the product of compromise in the quest for a broad majority as much as of theoretical confusion, left Vavilov “open to future manipulation by courts that do not share its spirit or find it inconvenient in a given case”. In particular, of course, the danger was that, as the Supreme Court’s composition changed, the deferential reading of Vavilov would come to the fore. Mason suggests, to my mind at least, that this is not happening.
The Supreme Court is essentially unanimous in adopting an intrusive approach to reasonableness review, in which even a comparatively thought-through administrative decision is vulnerable to correction for failing to engage with relatively subtle issues raised by the applicable statutory scheme. The majority also doubles down on Vavilov‘s rejection of the earlier cases inclination to make up administrative reasons and defer to these judicial concoctions ― what I described as a court playing chess with itself and contriving to lose. I think it’s very ironic for the majority to reprove the Federal Court of Appeal for failing to follow this rule. The Court of Appeal’s decision was given by Justice Stratas, who penned the pre-Vavilov locus classicus on this exact point, writing in dissent in Bonnybrook Park Industrial Development Co. Ltd. v. Canada (National Revenue), 2018 FCA 136: “My job is judicial review of the Minister, not judicial impersonation of the Minister. I do not work for the Minister. I am not the Minister’s adviser, thinker, or ghostwriter.” I do not suppose that Justice Stratas, of all people, has decided to take the job of ghostwriter for the Immigration Appeals Division. But, however uncalled for, the Mason majority’s criticism of him at least shows that the Supreme Court will not tolerate others doing so. The majority’s refusal to remit the case to the Appeal Division and its insistence that there is only one sensible reading of the statute also signal that courts should not pretend that they lack the wisdom to read legislation administered in the first instance by some bureaucrat.
All this is happening four years after Vavilov, with three members of the Vavilov majority gone from the Supreme Court. It is happening the agreement of Justice Karakatsanis who, along with the now-retired Justice Abella, attacked that majority for being insufficiently deferential. At the risk of being unduly optimistic, I will take this as a positive sign that the robust interpretation of Vavilov has legs and is likely to stand even as the turnover on the Supreme Court continues. It is probably not a coincidence that, like Vavilov, Mason is an immigration case. In decades past, too many leading administrative law cases featured friendly neighbourhood labour arbitrators solving annoying disputes between big business or big government and big labour. The more these cases are drawn, instead, from the immigration world, where the state red in tooth and claw is baring its fangs at the individual, the more we can hope for robust judicial review doctrine.
That said, Mason shows that, at the level of principle, this doctrine is as unsound as ever. It remains beset with the contradiction between the rhetoric of deference and the practice of robust review. This is illustrated, for instance, by Justice Jamal’s criticism of the Federal Court of Appeal’s “preliminary analysis” for the risk of slipping into correctness review ― combined with also criticizing the Court of Appeal for not conducting reasonableness review with enough vigour. Logically, the Court of Appeal wasn’t insufficiently and excessively deferential at once. The “preliminary analysis” it suggested is just the natural way a flesh-and-blood judge will approach judicial review, whatever the Supreme Court may say. Indeed, Justice Jamal himself begins by reciting the applicable statutory provisions. Are we to suppose he just copied them with no shade of reflection on what they might mean having entered his mind? Sorry, but I don’t believe this. But the posture of largely fake deference to the administrative state forces the Supreme Court into such implausible pronouncements.
More seriously, I’m inclined to think the majority is wrong to insist on deference in the context of certified questions under the IRPA. Justice Côté is right that the legislative scheme indicates that Parliament, for good reason, wanted the courts to sort out the legal issues arising out of IRPA’s interpretation, and that their failure to do so risks creating arbitrariness in high-stakes cases where it is utterly unacceptable from a Rule of Law standpoint. But Justice Côté’s reasons hint at a deeper problem with Canadian administrative law doctrine, which she does not address, and indeed might not see as a problem at all. Justice Côté writes:
[T]he risk of arbitrariness may be acceptable in the context of decisions regarding the extent of an income replacement indemnity during a temporary plant closure … or alleged violations of a provincial collective agreement … to use the two examples referred to in Vavilov … . It is not acceptable when the identity of the individual decision maker is what determines who is permitted to remain in Canada, as in these companion appeals, or in the context of other serious questions of general importance under the IRPA. [164]
But, even allowing, as one must, that the practical stakes of an immigration decision are much higher than many others ― indeed, they are as high as it gets in a legal system that does not impose the death penalty ― why exactly is arbitrariness tolerable in other contexts?
Questions of law ― at least pure questions of law whose answers, to quote Justice Côté, do not “turn[] on the unique facts of the case” ― are by definition of general importance insofar as laws, unlike commands directed at individuals, exist to apply to unknown numbers of unknown future cases. By definition, they have “broad significance or general importance” in whatever legal regime they arise in. On one point, Justice Jamal is actually right: the certified questions regime is, first and foremost, a leave to appeal mechanism rather than a way of distinguishing some special category of legal questions. But in a sound administrative law regime this would not have the consequences he says it does, which are wrongheaded for the reasons Justice Côté explains. (For that matter, even without moving to an entirely sound administrative law regime, progress could be made simply by reversing the fit of judicial madness that was Canada (Citizenship and Immigration) v Khosa, 2009 SCC 12, [2009] 1 SCR 339, and holding that, in accordance with the Federal Courts Act, immigration decisions are reviewable for error of law on a correctness standard. This would eliminate the paradox of differing standards of review at the Federal Court and the Federal Court of Appeal when a question is certified.) Vavilov, with its misbegotten claim that questions of law need not be given unequivocal answers to comply with the Rule of Law, does not a sound administrative law regime ground.
Its unsoundness is, finally, further illustrated by the majority’s decision not to remit the case back to the Appeal Division because there is actually only one reasonable reading of the provision at issue. As I have written here,
I find it odd to say that reviewing courts must start from the position that “respect for [the] institutional design choices made by the legislature” in setting up administrative tribunals “requires a reviewing court to adopt a posture of restraint on review”, [Vavilov, 24] but then insist that respect for legislative choices also requires the courts to be vigilant in case these choices leave only one permissible interpretation. The view, endorsed in Dunsmuir, that deferential judicial review reflects the inherent vagueness of legal language, was empirically wrong (and indeed implausible, as I argued here), but coherent. The recognition in Vavilov that statutory language is sometimes precise and can have a definitive meaning is welcome, but it is logically incompatible with an insistence on deference and judicial restraint.
The conclusion that there is only one justifiable (and hence not only reasonable but also correct) reading of a statutory provision doesn’t need to come at the end of a big, but fake, show of deference. It should be the first thing a reviewing court decides, as part of its duty to say what the law is, and so of giving decisions that are definitionally of general importance.
I remain hopeful that one day, Canada’s administrative law will recognize these truths, even if it does not yet hold them to be self evident. But yesterday was not that day. The absurdity at the core of our law remains. But, at least, the law in action might just make more sense than the law on the books. The Supreme Court seems committed to playing a meaningful role in upholding the Rule of Law in the face of administrative overreach and to instructing other courts to do likewise. As the recent history of Canadian administrative law reminds us, this is no small blessing, and I am grateful for it.
Double Aspect 
One thought on “It’s Nonsense But It Works”