Double Aspect

Last week, the Supreme Court released its judgment in Democracy Watch v Canada (Attorney General), 2026 SCC 28, holding that privative clauses that purport to oust judicial review on questions of fact and law are an unconstitutional violation of the judicature provisions of the Constitution Act, 1867. The practical effects of this decision will be limited (which is not to say nil), because Canadian courts had a longstanding “tendency to read privative clauses into constitutional compliance”, [83] as Chief Justice Wagner, writing for the unanimous court, puts it. But it continues a trend of desperate theoretical confusion in Canadian administrative law.

In effect, Democracy Watch recreates and constitutionalizes the House of Lords’ decision in Anisminic Ltd v Foreign Compensation Commission [1969] 2 AC 147, though with important caveats that highlight the superiority of the British original over the Canadian knock-off. But for reasons that quite escape me, Canadian judges would much rather talk evident nonsense than be caught admitting that Anisminic was right.

Democracy Watch concerns a decision of the Conflict of Interest and Ethics Commissioner pursuant to the federal Conflict of Interest Act. Under the Act, the Commissioner’s decisions are protected by a partial privative clause that preserves review for, notably, procedural and jurisdictional issues, but excludes it for errors of law and fact. The Act also provides that the Commissioner must issue periodical reports to the House of Commons, which the government argued (and the Federal Court of Appeal agreed) provided an adequate alternative to judicial review. Sensibly enough, the Chief Justice rejects the suggestion. He doesn’t put in quite these terms, but political accountability, even if it happens, is not the same as legal accountability, and cannot substitute for it.

But can Parliament limit the scope of legal accountability to which a decision-maker is subject? The Chief Justice says no. He points out that “[t]he supervision of the administrative state through judicial review of the exercise of public powers is one of these essential aspects of the courts’ constitutional function”. [41] In order to protect the constitutional status of the courts referred to in ss 96 et seq of the Constitution Act, 1867, this function cannot be taken away by statute. Here, of course, the court at issue is the Federal Court of Appeal, not a “s 96 court”, but the Chief Justice argues that, when Parliament transferred the judicial review jurisdiction to the federal courts, it had to transfer it whole, and not just in part.

More precisely, though, the question is whether the guarantee of judicial review extends beyond jurisdictional review, which the statute at issue preserves, and applies to non-jurisdictional factual and legal issues. In Crevier v Attorney General of Quebec, [1981] 2 SCR 220, which is the lynchpin of the whole jurisprudence on the constitutionally protected role of the s 96 courts, the Supreme Court seemed to say that it did not. But the appellants (represented by Sujit Choudhry and Paul Daly) argued

that Crevier must be read in its historical context, and that the jurisdictional review which Crevier identified as constitutionally guaranteed encompasses legality review on all aspects of an administrative decision, including questions of fact and law.

The Chief Justice accepts this originalist-ish framing. As he tells the story, Canadian courts used to follow Anisminic, which “was the ‘canonical example’ of the orthodox approach to judicial review, which germinated in the long shadow of Professor A. V. Dicey’s court-centric view of the rule of law”. [45, quoting the concurring opinion in Canada (Minister of Citizenship and Immigration) v. Vavilov, 2019 SCC 65, but failing to acknowledge that it is a minority opinion] On this approach, it was jurisdictional errors that opened the door to judicial review, even in the presence of a privative clause, and the courts took a broad view of what constituted a jurisdictional error. These included “an administrative body … erring in law, making an unreasonable finding of fact, or breaching the laws of natural justice”. [46] But in Canadian Union of Public Employees, Local 963 v New Brunswick Liquor Corp, [1979] 2 SCR 227, the Supreme Court took a different tack. Starting in CUPE and culminating with Vavilov, the touchstone of judicial review would no longer be jurisdictional error, which came to be interpreted increasingly narrowly and then, ostensibly, eliminated as a category distinct from an ordinary error of law, but unreasonableness in one form or another.

Crevier, however, “was decided shortly after CUPE, when the jurisdictional error approach was still firmly entrenched”. [63] Its seeming acceptance of legislative ousters of “judicial review” for non-jurisdictional errors must therefore be understood by keeping in mind that “‘judicial review’ in the Anisminic era meant correctness review”. [64] So read, Crevier does not apply to ousters of reasonableness review:

Crevier cannot have intended that unreasonable findings of fact or law could be shielded from judicial correction by a privative clause. Such a holding would have been incompatible with the whole weight of centuries of English and Canadian jurisprudence. It would also be incompatible with subsequent statements in our jurisprudence to the effect that unreasonable errors of fact and law are jurisdictional errors. [64]

The superior courts’ (or the federal courts’, which are exercsing a jurisdiction derived from them) ability to engage in (some form of) reasonableness review is thus constitutionally protected with respect to what the Chief Justice describes as “legality review — that is, review to ensure that an exercise of public power has not exceeded the boundaries of an administrative decision-maker’s delegated authority”. [67] That said, the Chief Justice insists that this protection does not necessarily require the availability of reasonableness review as articulated in Vavilov: “As products of our Court’s jurisprudence, the common law standards of review can be and have been modified over time”. [67] At a minimum, “rationality review has a constitutional aspect. Rationality is a boundary of legality.” [69] But there may exist a space between what is irrational and what is Vavilov-unreasonable which legislatures might be able to occupy by legislating specific standards of review.

What they may not do is precisely what Parliament attempted in this case: excluding all judicial review on non-jurisdictional questions of law and fact, even when the decision-maker acts unreasonably or indeed irrationally. And while in the past courts tended to read down or simply ignore privative clauses that purported to require them to refrain from reviewing unreasonable decisions, the Chief Justice says that this is the wrong approach: “where legislation is not amenable to a constitutionally compliant interpretation, the legislation must be declared invalid”. [82] So it is here.

There are things to like here, not least the forthright acknowledgment of the unconstitutionality of privative clauses, which until now existed in a peculiar sort of limbo — neither effective nor invalid. More broadly, Democracy Watch‘s idée maîtresse, that in “our constitutional system, every exercise of public authority is subject to the courts’ supervisory jurisdiction to ensure that it does not exceed its legal limits” [8] is sound, and the decision will, to some limited extend, contribute to its practical implementation. At the same time, it is tainted by the theoretical muddle in which the Supreme Court has long found itself in administrative law, and even on a practical level it will introduce further complications into an already messy area. 

As I have said at the outset, I think that — much like Vavilov was a clumsy version of Chevron USA Inc v Natural Resources Defense Council, Inc, 467 US 837 (1984), making similar moves in an unnatural order — Democracy Watch is a bizarre version of that which it tries hard to repudiate: Anisminic. The Chief Justice’s concept of “legality”, which he defines as the requirement that “public power has not exceeded the boundaries of an administrative decision-maker’s delegated authority” is not much different from the kind of review that Anisminic held not to be precluded by a simple privative (or, in the usual English terminology, ouster) clause. In that case, Lord Reid said that

It has sometimes been said that it is only where a tribunal acts without
jurisdiction that its decision is a nullity. But in such cases the word “jurisdiction” has been used in a very wide sense, and I have come to the conclusion that it is better not to use the term except in the narrow and original sense of the tribunal being entitled to enter on the inquiry in question. But there are many cases where, although the tribunal had jurisdiction to enter on the inquiry, it has done or failed to do something in the course of the inquiry which is of such a nature that its decision is a nullity. It may have given its decision in bad faith. It may have made a decision which it had no power to make. It may have failed in the course of the inquiry to comply with the requirements of natural justice. It may in perfect good faith have misconstrued the provisions giving it power to act so that it failed to deal with the question remitted to it and decided some question which was not remitted to it. It may have refused to take into account something which it was required to take into account. Or it may have based its decision on some matter which, under the provisions setting it up, it had no right to take into account. I do not intend this list to be exhaustive. (171)

Like the Chief Justice’s “legality”, this encompasses all manner of legal and procedural errors, and in that sense the Chief Justice is simply reinventing the wheel even is as he is talking a very different type of round objects. To be sure, Anisminic does not explicitly mention unreasonableness, in the sense of an irrational finding of fact, which was not relevant to that case. But I think it is encompassed by its logic because of the inextricable connection between rationality and legality, to which the Chief Justice adverts in Democracy Watch, and which was recognized by common law courts at least as far back as Coke. While Mark Mancini has argued that review of jurisdictional facts should (mostly) not be regarded as part of the constitutionally protected core power of the Canadian judiciary, this has to do with his scepticism of the category of jurisdictional issues. I am much more comfortable with the Anisminic-style everything-is-jurisdictional approach that the Chief Justice recreates.

There is, however, a difference between the Chief Justice’s position and that of the English courts. It concerns not the scope of constitutionally (in Canada) or presumptively (in the United Kingdom) protected judicial review, but its intensity. In the United Kingdom, questions of law are reviewed on a standard of correctness, and it is, I assume, mostly from this that the Chief Justice wants to distance himself when he insists that he is absolutely not following Anisminic at all, oh no, perish the thought. As he puts it, with reference to Vavilov,

Because the legislature intends to give decision-making power to its administrative delegate, not the courts, reasonableness review balances the constitutional duty of the courts to police the boundaries of legality with the constitutional imperative to respect legislative intent. [58]

I will not rehearse all my criticisms of Vavilov here, but suffice it to say that they are all the stronger in the wake of Democracy Watch. With not even the fig leaf of the now-unconstitutional privative clauses to support it, the conclusion that the legislature intends to give its administrative delegates the power to interpet the law has no leg to stand on, and it is belied by the Chief Justice’s entirely correct view that the legality of administrative action must always be judicially reviewable and that political accountability is not a substitute for this. There is nothing to balance here. It is the courts’ role to uphold legality, and they are failing to do this job insofar as they allow administrative say-so to have anything more than a persuasive role in their assessment. It is fair to say that questions of fact are not wrongly answered, from an administrative standpoint, unless they answered unreasonably. But that does not apply to questions of law.

A few words on what comes next. Professor Mancini, in his newsletter, suggests that Democracy Watch‘s expansion of the constitutional guarantee of judicial review “won’t mean much in day-to-day practice … Courts will continue to review decisions as they have.” I think that’s right (and Professor Mancini’s sense of the practical realities of Canadian administrative law is, in any case, far better than mine). Democracy Watch will mostly matter to the application of statutes like the Conflict of Interest Act, which specifically allowed only limited judicial review or appeal rights. I’m not sure about this, but my impression is that these aren’t all that common.

Perhaps the bigger question, as Professor Mancini notes, is what happens to statutory standards of review that are more deferential than Vavilov. The issue does not arise in Democracy Watch but, as we have seen, the Chief Justice is insistent that standards of review are creatures of the common law and thus subject to legislative override. Professor Mancini argues that “there is good reason to fully decline the invitation to recognize reasonableness review as a constitutional floor”, because “I[i]ntensity [of review] should be a constitutional question at one point only: where a legislated standard is so deferential that it is tantamount to no review at all”.

That’s fine in the abstract, but it’s not obvious to me how that actually works. On questions of law, even reasonableness review is already a form of judicial abdication, tantamount to no review in Professor Mancini’s words. Or, in the words of Lord Reid and others in Anisminic, it involves judges asking themselves the wrong question, and thus not doing what it is their job in our legal framework to do. And as for questions of fact, there is a reason why Canadian courts have abandoned the patent unreasonableness, regarding it as unworkable. There is a reason, too, why “Wednesbury unreasonableness”, which is patent unreasonableness’s close equivalent in English law, has been criticized for about as long as it’s been around. Patent unreasonableness should never be a constitutionally acceptable standard of review for legal issues. But I question its defensibility for factual ones too.

One last thought. There are apparently people who are worked up about the idea that the exercise of executive power cannot be shielded from judicial supervision. To the extent I can steel-man this concern and understand it as something other than a desire for arbitrary power, I think it may reflect a preoccupation with the expenses and delays that are inevitably involved in litigation about governmental decisions — about the costs of judicial review, which may well be higher than its benefits in a given case. I have written about this last year when discussing a judicial review of a violation of a police authority’s duty of neutrality in England. There are situations where one can understandably be impatient of the courts seemingly getting in the way of public authorities just doing things. But the appropriate response to this is not to cut the courts out and allow the executive unbridled power subject to, at most, highly uncertain legislative supervision. It is to simply have fewer rules compliance with which can then be judicially reviewed. And one way to have fewer rules is for the government to do and to regulate fewer things.

Of relevance to Democracy Watch, government doing fewer things would also allow legislatures — which are not subject to judicial review — to take a more active role when they focus on the things that are necessary to do. This case wouldn’t have arisen if Parliament had it in itself to hold the executive to account on its own. But it didn’t, and chose to offload this process on a bureaucrat. Oh well. But that choice had to carry its constitutional consequences. Insofar as it imposed these consequences, Democracy Watch is, for all its flaws, a good decision.