Romancing the Law

An ode to formalism and reflections on Runnymede’s Law and Freedom Conference

I had the pleasure of attending last weekend’s Runnymede Society conference in Toronto. As always, the conference was a welcome opportunity to meet with old friends and new, and to reflect on a number of pertinent issues in Canadian law.

Perhaps because of my own research interests in the last year, I was particularly interested in a theme that seemed to run throughout the conference: the degree of confidence that each of us has in law, particularly the statutory law. Justice Stratas’ talk with Asher Honickman highlighted that there are many in the legal community that, if not giving up on law, are questioning its relevance in a society that is now defined by greater calls for context, nuance, individualized application, and discretion.  The virtues of rules—the creation of economies of scale, the structuring of norms and expectations according to positive orders, the costs saved at the ex post application stage—are apparently counterweighed by the potential for overbroad application, rank injustice, and otherwise discriminatory treatment.

The degree to which we are worried about these vices, or encouraged by these virtues, is probably a function of our belief in legislatures and their work product. Even if legislatures do not get things “right,” there are good reasons to believe that what the legislature does is owed a wide degree of respect–because of the value of legislative compromises, the “finely-wrought” legislative procedure, and the representative nature of the legislature . Nowadays, though, a commitment to the law passed by the legislature is labelled pejoratively as “formalist.” In administrative law, offshoots of this belief are characterized, dismissively and without analysis, as “Diceyan” or an unwelcome throwback to the days of “ultra vires” (take a look at the oral argument in the Bell/NFL & Vavilov cases for many examples of this).  In statutory interpretation, a belief that text in its context will generally contain answers is dismissed as a belief in “the plain meaning rule,” mere “textualism”–notwithstanding the important distinction between these two methods. In constitutional law, a focus on constitutional text is “originalism.” None of these are arguments, but they have since infiltrated the orthodoxy of the academy.

The consequences of this argument-by-label should not be understated. Take the case of statutory interpretation. The Supreme Court of Canada tells us that we should interpret statutes purposively, but at the same time, that the text will play a dominant role in the process when it is clear (Canada Trustco, at para 10). This implies that purposes, while helpful to the interpretive process, should not dominate where the text is clearly pointing in another direction.

But a focus on statutory text—especially the contention that text can ever be clear—is often derided as inconsistent with the contingent and “ambiguous” nature of language. So the argument goes, text can never truly be “clear,” and so textualism falls away. But whether the text of a statute will contain answers to an interpretive difficulty is, in part, a function of the judge’s belief in the coherence and determinacy of law—in other words, her appreciation of the point at which “law runs out”. A judge inclined to believe that the tools of statutory interpretation can be used to come to a defensible answer on a matter will commit herself to that task, and will probably not consider legislative language “ambiguous” in its purposive context. For her, law will maybe never run out, or if it does, it will only do so in the extreme case of true ambiguity, where no discernible meaning cognizable to human understanding could be appreciated. A judge less committed to the determinacy of law will be more willing to introduce extraneous materials—legislative history, Charter values—in order to come to a meaning that makes sense to her. For her, the law may “run out” quite early. The risk here, of course, is the enlargement of the scope for judicial discretion. For those who believe in the general soundness of statutory law, this creates the potential for conflict with the generally-elected representative body.

This is not a hypothetical problem. In the United States, Chevron administrative law deference rests on the judge’s appreciation of statutory language. At step 1, courts are asked to apply the ordinary tools of statutory interpretation to determine if Congress spoke clearly on a particular matter. If so, that meaning binds the agency. If not, at step 2, if there is ambiguity in a statute, courts defer to a reasonable agency interpretation. As Justice Scalia said, a judge committed to the text at step 1 will rarely need to move to step 2. In this way, there would be less scope for agencies to exercise virtually unreviewable discretion. Those who believe that law runs out earlier will, ceterus paribus, be more willing to allow multiple decision-makers to come to very different decisions on a matter so long as those decisions are roughly justified by a statute.

The various points on the spectrum of “giving up on law” will be the product of many factors, including factors particular to cases before courts. But at some level, a belief that text can, or should, contain answers seems to undergird the entire process of determining the meaning of a statute. I think there are good reasons to hold the belief that what the legislature produces is generally sound for reasons that are particular to the legislative process and the law in question. To my mind, judges should be wary of letting text “run out,” in part because of what replaces it; more abstract, generally less clear “second-tier” sources of legislative meaning (Note: sometimes text will be truly unclear, and a statutory purpose can be clearly gleaned from the text. Our law sees no problem with this, and neither do I).

This is not to presuppose that legislatures always make sense in their enactments. The process of making law is not designed to be a perfect application of human rationality or even of expertise. Legislatures sometimes don’t make sense. But there are good reasons to respect the legislative process. Importantly, seemingly non-sensical legislative compromises, run through readings in Parliament and the committee process, are sometimes the product of concessions to minority groups, represented through their Members of Parliament. These legislative compromises are sometimes essential, and should be respected even if they do not make sense. Judge Easterbrook puts it well: “If this [an outcome of statutory interpretation] is unprincipled, it is the way of compromise. Law is a vector rather than an arrow. Especially when you see the hand of interest groups.”

If the legislative process is imperfect, so is the process of statutory interpretation. Statutory interpretation will not always yield easy answers, or even the ex ante “correct” answer. The tools of statutory interpretation are often contradictory, some say outmoded, and sometimes unwieldly. But as Judge Posner said in his book Reflections on Judging, the tools of statutory interpretation are designed to impose meaning. Used authentically and faithfully, with a concomitant belief in the legitimacy of the law passed by the legislature, they help courts come to a defensible conclusion on the meaning of a provision; one that is consonant with the universe of laws in the statute book, with the particular statute’s larger purposes, and the immediate context of a statute.

It worries me that some no longer belief in this process—in the formal quality of law as law, in the idea that when the legislature speaks, it does so for a reason. Similarly, I worry that the invitation for judges to rely on values and principles extraneous to a statute—for example, Charter values, legislative history, etc—to impose a meaning on a statute is based on wrongheaded idea that judicial discretion is somehow absolutely better than legislative power. I, for one, think that we should expect judges in a constitutional democracy to believe in the law passed by the legislature. This is not judicial acquiescence, but there is perhaps a value to formalism. Parliament, to be sure, does not always get everything right. But there is a benefit to formalism: the way in which Parliament passes laws is subject to a formal process, interposed with legislative study. The way we elect our leaders and the way Parliament operates is, in a way, formal. The law it creates should be owed respect by those sworn to uphold it.

The debate over rules versus standards or discretion is one that is rife throughout history. But presupposing the debate, I always thought, was a belief in law itself. For those of us at Runnymede this weekend, we were invited to question whether that belief exists any longer.

Statutory Interpretation in Admin Law and the Supreme Court’s Trilogy

Over on Professor Daly’s blog Administrative Law Matters, Professor Audrey Macklin wrote what I would characterize as a confessional: an admission that the law of judicial review in Canada may be beyond repair. What Prof. Macklin proposes, in light of this realization, is a renewed focus on the principles of statutory interpretation, rather than a myopic focus on standard of review categories—a parochical Canadian invention. Prof. Macklin specifically argues that our disagreements should focus on “what ought to count in statutory interpretation, why it ought to count, and how it ought to count.” For Professor Macklin, external considerations such as access to justice and international law should inform statutory interpretation.

I agree with Professor Macklin—to an extent. I see Macklin as making two claims: (1) the rules of statutory interpretation should be used to replace the rigid categories of standard of review and (2) that the rules of statutory interpretation can be massaged to take account of “normative, policy, operational, and other considerations.” The first contention, I think, is desirable. The second, I worry, will lead to more uncertainty, asking judicial review courts to do more than they are designed to do.

Let’s take (1) first. On this front, Professor Macklin finds agreement with many existing allies. Advocates for the Rule of Law, in its submissions before the Court in Vavilov and Bell/NFL, basically made this very point. With submissions expertly argued by Adam Goldenberg, ARL asserted that the Court, in its attempts to simplify the standard of review, had actually made things worse. Instead, ARL argued that the “established tools of statutory interpretation” should be used in a three-step framework to determine the level of deference. First, courts should look to judicial constating legislation to determine the level of deference; explicit direction that exists, for example, in the Federal Courts Act (which, it should be said, was ignored by the Court in Khosa). Second, the decision-maker’s home statute may contain explicit or implicit legislative direction on the standard of review. A statutory provision allowing a decision-maker to grant licences “in the public interest,” if supported by relevant context and purpose, should allow more deference. Finally, courts should look to the specific textual provision under which the decision-maker made her decision. If not contradicted by context, (say, a statutory right of appeal), broad language in this respect should also mean deference.

This strikes me as a good way to proceed. I make similar arguments in a paper to be released in the Western Journal of Legal Studies in January, “Two Myths of Administrative Law.” There, I assert that toying around with presumptions, categories, and even standard of review labels (reasonableness and correctness), should not be the sum of Canadian administrative law. These legal devices should be scrapped, the focus should be on the principles of statutory interpretation, and signals of deference—for example, expertise—should only come into consideration where they exist in statutory text, context, and purpose. I don’t see Professor Macklin as disagreeing with any of this.

At the same time, Professor Macklin takes a first crack at defining the work the principles of statutory interpretation could do in judicial review. She claims that so-called “operational considerations” that influenced the Supreme Court in Alberta Teachers—such as efficiency—also mean that courts should take into account principles of access to justice and international law in statutory interpretation. To Professor Macklin, courts should take “all the arguments” seriously in a given case.

This, I think, is where Professor Macklin’s proposal introduces more uncertainty than it actually solves, and runs afoul of the very basis of the law of judicial review. Her extension seems to extend the task of statutory interpretation beyond its traditional realm of discovering the meaning of the legislature. Instead, she would make statutory interpretation a normative exercise, one that seeks to render an interpretation that best accords with any number of “external” considerations.

But if we are starting from scratch in developing doctrine for the law of judicial review, we should do so in light of that very fact: it is the law of judicial review with which we are concerned. Judicial review is not a free-standing inquiry into the legality of particular government decisions. When governments delegate away power to make decisions, it does so on a narrow basis—to decide which of Party A and B receives Licence X; to decide whether A’s zoning application is legal. This is the run-of-the-mill judicial review case. When courts review these decisions, they are usually bound to the record the decision-maker considered on review. The courts do not move beyond that record, which maps onto the scope of delegative power afforded to the decision-maker. As a result, and by design, the ancient writ of certiorari and the modern act of judicial review are first and foremost designed to “ensure the legality of state decision-making” (Wall, at para 13). Courts are equipped to make those decisions by rules of the record that mandate the production of all information considered by the decision-maker under its statutory mandate. In this respect, judicial review proceedings are designed to be quick and cost-effective.

There are two upshots of this ordering. First are the considerations that should inform the exercise of judicial review. In the Two Myths paper, I argue that such external considerations lack the imprimatur of legislative approval. In other words, they do not exist in the world with which judicial review is concerned. If judicial review is considered primarily with the supervision of state actors—what I call its “policing function”—then it must hew closely to the four corners of a statute. Anything beyond that is a concern for legislative actors, who are legitimately empowered to address broad, systemic concerns such as access to justice because of their particular institutional competencies. Any other positive policy entitlements—a right to access to justice, considerations of normative justice—are second to the immediate statutory context of a decision-maker’s decision in judicial review. That is not to say that these considerations are irrelevant. But they are distinctly second to the goal of a quick determination of legality “on the record.”

But there is a bigger practical concern here, beyond the issue of separation of powers. To my mind, Macklin asks judicial review courts to do more than they are designed to do, from a perspective of information. By using the principles of statutory interpretation to gaze beyond the four corners of a statute and into other complex, external considerations, judicial review becomes more of a legislative act rather than a pure adjudicative one. In this sense, the benefit of the principles of statutory interpretation—that they, as Macklin notes, put us on the same terrain—is lost. For example, what does a presumption in favour of access to justice mean in the context of particular cases? In absence of real information on what would facilitate access to justice in the context of a defined legislative scheme, how do courts make informed decisions on the matter? Even if they could, doesn’t this complicate the narrow and efficient basis of judicial review?

It is no answer to these problems to say, as Macklin does, that courts are informed by policy considerations when they rely, for example, on presumptions in favour of contract or liberty. These presumptions are of a different sort than the type advocated for by Macklin, at least in the context of access to justice. Most notably, they are negative in nature. But there are particular issues that distinguish these traditional presumptions from the sort of rule Macklin seeks to formulate. For example, in terms of contract, the presumption is justified as a policy matter from the perspective of information. Courts, for good reason, assume that parties enter bargains on the basis of mutual gains, which in turn enable social welfare gains. Courts (and legislatures) lack the information necessary to overturn the parties’ agreement and instead impose their own understanding of what would be an efficient outcome; transaction costs of doing so are high because the court lacks information. Macklin’s presumption (in favour of access to justice) is of a different sort; it asks judicial review courts to consider or act in areas where there is a distinct informational deficiency, at least in the context of particular judicial review cases.

The problem of information is not a practical concern that exists in the ether. It is directly related to the roles of various players in our democratic structure. Courts on judicial review are empowered to redress public law grievances caused by organs of the state; no more or less. As a result, they are given the information required to do so on the basis of the record. The realm of legislative action, on the other hand, is broader—and so legislatures are empowered with more information and more resources. None of this is a coincidence.

People might use a laundry list of adjectives to describe my preferred approach: “cramped,” “formalistic,” “Diceyan.” None of this, of course, is an answer. The bottom line: the law of judicial review, and the principles of statutory interpretation, cannot encompass all, or even most, of what we consider good and virtuous in the world. They can only accomplish their narrow but infinitely important goal. Any renovation to the law of judicial review that uses the principles of statutory interpretation cannot forget this.

Statutory Interpretation in Canada from the “Stratasphere”

For those interested in statutory interpretation and its effect on administrative law, I have a new piece coming out in the Advocates’ Quarterly in October. A preliminary version of the piece was posted on the Advocates for the Rule of Law website over the summer. The paper is basically a review of two opinions written by Justice Stratas of the Federal Court of Appeal. I argue that the opinions give us an opportunity to consider an underexplored area in Canada: how statutory purposes should interact with text, and the implications for the level of deference granted on questions of law to administrative decision-makers. I write the following in the introduction of the piece:

Statutory interpretation presents problems of judicial subjectivity. Though it is well-established that courts and advocates must look to the “text, context, and purpose” of a particular statutory provision to determine its meaning, little work has focused on what courts should do when purposes are stated at different levels of abstraction, or where the statute has multiple purposes which are seemingly contradictory. In fact, there are no rules governing how courts should act in these situations. The potential result of this void is the rule of “homunculi sitting in the minds of judges”; judicial subjectivity beyond statutory text.

While these problems remain, Justice Stratas of the Federal Court of Appeal provides guidance on these questions to courts and litigants in two recent cases: Williams and Cheema. After reviewing the cases, I argue that Justice Stratas’ opinions properly warn courts against characterizing highly abstract statutory purposes, outside of what the statutory text prescribes. In the context of judicial review of administrative determinations of law, doing so could facilitate an overly deferential or interventionist posture to administrative interpretations of law, beyond what text actually prescribes. This is a court created distortion. As an antidote, Justice Stratas’ opinions rightly remind us that legislation binds, and that as a matter of the rule of law, courts must enforce statutory language rather than purposes untethered to text.

 

It’s a Dog!

The majority’s pro-regulatory beliefs help make West Fraser a dog of a decision

In previous posts, I have summarized the Supreme Court’s decision in West Fraser Mills Ltd v British Columbia (Workers’ Compensation Appeal Tribunal), 2018 SCC 22, upholding the validity of a regulation of the British Columbia Workers’ Compensation Board imposing safety-related obligations on owners of forestry workplaces, and the legality of a penalty imposed on such an owner under a statutory provision authorizing penalties against employers who do not comply with regulations, and discussed some of the administrative law issues to which this decision gives rise. As previously noted, however, West Fraser is interesting not just for what it can tell us about the finer points of judicial review, but also for what it implicitly says about the Supreme Court’s relationship to the administrative state.

In its overall orientation as well as in some details, the majority opinion, written by Chief Justice McLachlin with the agreement of five colleagues, is reminiscent of R v Comeau, 2018 SCC 15 (further confirmation, perhaps, of the Chief Justice’s likely authorship of that ostensibly per curiam decision). It’s not just that the deferential approach to judicial review is, in practice, in Canada, almost necessarily a pro-regulatory position, though that’s part of the story. It’s also that, on the Chief Justice’s view of statutory interpretation, a statute’s pro-regulatory purpose is to be amplified, while whatever constraints on its pro-regulatory orientation the statute might contain are to be played down. And, most fundamentally, the Chief Justice tells us that regulation is good, and the more of it there is, the better.

As discussed in more detail in my previous posts, the Chief Justice’s approach to both issues in West Fraser is deferential ― or so the Chief Justice says. In reality, I have argued, she engages in disguised correctness review and agrees with the administrative decision-maker. But, in principle at least, it’s the deferential approach that’s binding on future courts. Conceptually, deference might be neutral as between pro- and anti-regulatory outlooks. In the United States, famously, Chevron USA v Natural Resources Defense Council, 467 US 837 (1984), which required a deferential approach to administrative interpretations of legislation, arose out of efforts at deregulation by the Reagan administration. Even so, it seems likely that administrative decisions that reduce the scope or onerousness of regulation are less likely to be challenged, so that in practice a deferential court will be a pro-regulatory court even if it has no particular desire to be one. And, of course, the prospects of serious regulatory roll-back in Canada seem rather remote.

But there is more. Whatever abstract theory might suggest, Canadian deference theorists are unabashedly in the pro-regulatory camp. David Dyzenhaus’s famous chapter on “The Politics of Deference: Judicial Review and Democracy”, from which the Supreme Court in Dunsmuir v New Brunswick, 2008 SCC 9, [2008] 1 SCR 190, plucked the phrase “deference as respect”, [48] urged the courts to defer to administrative decision ― because its author thought that administrative regulation was normatively desirable, notably in that it advances the cause of equality. (Professor Dyzenhaus went so far as to argue that non-deferential review of the decisions of human rights tribunals by independent courts was a “setback[] to the constitutional commitment to equality between Canadians”, (297) as if the courts were not ultimately responsible for upholding this constitutional commitment.) But for Professor Dyzenhaus, deference was a one-way ratchet: if, peradventure, “judges find themselves confronted with administrative determinations of law that flow from” deregulatory impulses, “they should not be embarrassed to ask how those determinations advance the cause of equality” (306) ― and  so to intervene if they do not. Much more recently, in her contribution to the “Dunsmuir Decade” symposium, Kate Glover proposed a novel theory according to which judicial deference to administrative decision-makers is now a constitutional requirement. Professor Glover did so in an attempt to prepare the administrative state’s defences against a (purely hypothetical, as she rightly notes) siege by deregulating anti-administrativists. This should, I think, be a warning to anyone who hopes that deferential courts would in fact be neutral as between more regulation or less.

Statutory interpretation, no less than (rhetorical) deference, is marshalled in support of regulatory expansion by the Chief Justice. She stresses that the Workers Compensation Act, the statute at issue, “is meant to promote workplace safety in the broadest sense”, [18] and discounts the more specific purpose statements that seem to suggest that this purpose is not to be pursued by whatever means necessary. Focusing on them is “formalistic” and “inconsistent with a purposive interpretation of the scheme”. [18] (To be honest, I don’t know what “formalistic” is supposed to mean here. But it’s bad, bad, bad.) When it comes to the issue of whether the statute authorized the imposition of penalties seemingly reserved for “employers” on firms that were, in the context of the events in relation to which the penalty was being imposed, “owners” but not “employers”, the Chief Justice once again favours an interpretation “more supportive of the goal of promoting safety and the overall operation of the scheme”. [38] This interpretation, as I argued in my last post, is strained to the point of rendering the statutory language meaningless. However, what mattered to the Chief Justice is that reading the statute to mean what it said “would undermine [its] goals”, while the strained interpretation “would further the goals of the statute and the scheme built upon it”. [40] In short the statutory purpose, understood in the most pro-regulatory way possible, must be given effect ― other purposes and text itself be damned.

Now, in fairness, to the Chief Justice, she arguably is dealing with a real interpretive difficulty. Probably all, certainly most statutes involve compromises between a number of values or purposes. The Workers Compensation Act promotes workplace safety, of course, but it also accommodates a measure of free enterprise. It could, after all, have imposed  even more invasive regulation that might have done even more for workplace safety ― but the legislature chose to only go so far towards that purpose, because going further might have undermined other purposes that it also valued. Or, to take another example, human rights legislation doubtless aims at achieving equality in society ― but the limits on its scope, for example the fact that it is typically not applicable to personal, non-economic relationships, suggests that it respects a measure of personal liberty ― implicitly anyway. The problem, though, is that if the legislature enacts a provision that specifies the purpose of a statute, it is likely to present  some, perhaps just one, of the values that the statute actually accommodates, as the purpose it seeks to realize, and omit the others. This might be done for political reasons ― it might not look good to tell workers, or voters, “we’re protecting you, but only some, since protecting you more would actually put a bunch of you out of work”. Perhaps more forgivably, this might also be because, relatively to the previous state of the law, the statute does move things in the direction of more protection, so characterizing that as its main purpose is not unfair. But, either way, the legislature is misleading those who read and try to understand the statute ― above all the courts ― by giving them a distorted view of its objectives.

What are the courts to do when the legislature does this? I think they should do what Justice Côté did in West Fraser ― read the whole statute and give effect to its terms, not letting the (one-sided) purpose section override the substantive provisions. By choosing to focus on the purpose indication (and to read it selectively to emphasize its pro-regulatory aspects), the Chief Justice once again implicitly privileges regulation. For the same political reasons I refer to above, it seems likely that the legislatures will systematically overstate the significance of their regulatory purposes, and understate whatever countervailing values might also be animating them. So, a judge who overvalues statutory statements of purpose at the expense  of the text will tend to produce pro-regulatory outcomes even  without setting out to do so. But I doubt that the Chief Justice is such a judge.

In fact, her reasons in West Fraser suggest that the Chief Justice’s basic disposition is in favour of regulation ― the more of it the better. She is comfortable with a legislative mandate to an administrative agency “to enact whatever regulations it deemed necessary to accomplish its goals of workplace health and safety”, [10] going so far as to characterize this as an “unrestricted delegation of power”. [11] Though admittedly it is unlikely that the Chief Justice means this adjective literally, it is remarkable that she appears untroubled by the idea of an unrestricted regulatory mission. Later, when discussing the issue of the penalty, the Chief Justice writes that “[t]he general scheme of the [Workers Compensation Act] is to hold both owners and employers responsible in an overlapping and cooperative way for ensuring worksite safety” [43] by way of justifying holding the ones responsible for violations of obligations the statute only seems to impose on the others. As in the area of what used to be known as division of powers, “cooperation” comes to mean the accumulation of regulatory mandates ― and is seen as a good thing. The Chief Justice’s shows her attitude towards such mandates most clearly when she makes a point of observing that the regulation challenged in West Fraser was adopted

in response to a concern in the province about the growing rate of workplace fatalities in the forestry sector … provid[ing] a clear illustration of why a legislature chooses to delegate regulation-making authority to expert bodies — so that gaps can be addressed efficiently. [20]

It is important that something be done about social problems, and whatever is done about them by regulators ― presumed,  conclusively, to be experts ― must therefore be good. There is only a step, if that, from here to what Sir Humphrey Appleby described as “the politician’s logic”, and what later became known as “the politician’s syllogism“. Something must be done; this is something; therefore this must be done. The Chief Justice has, on the occasion of her already-happened-but-still-impending retirement, and indeed before, been much praised for her statecraft. In West Fraser, she reminds one of The Right Honourable Jim Hacker, MP.

Of course, by criticizing the Chief Justice’s pro-regulatory views ― and those of the other judges in the West Fraser majority ― I do not mean that judges ought to become the flag-carriers of deregulation. They should be neutral and, within constitutional bounds, give effect to the legislation that Mr. Hacker’s colleagues, in their wisdom, enact. Much of this legislation will delegate considerable regulatory powers to administrative agencies. That’s too bad, so far as I am concerned, but this a policy view, not a constitutional argument. However, judges should not, in the name of doing something, be trying to give the regulators freer rein than legislators intended. In Yes, Prime Minister, just before Sir Humphrey formulates the politician’s syllogism, his mentor, the wily Sir Arnold Robinson, exposes its logic by proposing a different one with the same logical structure: “all cats have four legs; my dog has four legs…” “Therefore,” concludes Sir Humphrey, “my dog is a cat.” Well no. And so West Fraser is a dog of a decision.

 

Looking past Dunsmuir: Beginning Afresh

Imagining stable and generally acceptable administrative law doctrine

The Hon. Justice David W. Stratas,
Federal Court of Appeal

The views expressed in this post are the author’s own and are offered for education and discussion purposes only.

Introduction

In my first article for the Dunsmuir Decade series, “A Decade of Dunsmuir: Please No More,” I suggested that Dunsmuir should no longer be followed. But what should take its place?

To some extent, I have already explored this: D. Stratas, “The Canadian Law of Judicial Review: A Plea for Doctrinal Coherence and Consistency,” (2016), 42 Queen’s L.J. 27 (available online here).

To create a reliable, stable law of substantive review, we need to: (i) identify settled doctrine and well-accepted principles; (ii) deduce the operational rules from them, and (iii) take into account relevant, widely accepted judicial policies. 

Identification of settled doctrine and well-accepted principles

This step builds the solid and stable foundation upon which one can deduce operational rules to regulate this area of law.

As explained in my first article, operational rules developed in this way have every prospect of widespread acceptance, permanence and consistency of application. On the other hand, operational rules constructed out of judges’ personal views and say-so are not likely to survive the next set of judges on the court.

For too long in this area of law, judges have set out operational rules based on their own personal views of the proper relationship between the judiciary and administrative decision-makers and their own freestanding opinions—not well-settled doctrine and well-accepted principles of a longstanding and durable nature.

As we shall see, the settled doctrine and well-accepted principles that underlay this area of law aren’t just longstanding and durable. They are fundamental constitutional principles as well.  This adds force to the operational rules logically deduced from them. 

Some doctrine and well-accepted principles

Fortunately, settled doctrine and well-accepted principles are not hard to find. Dunsmuir did some of the work for us (at paras. 27-31).

Dunsmuir told us that two principles lie at the heart of judicial review and animate it: legislative supremacy and the rule of law.

The U.K. Supreme Court recently affirmed this same idea: Michalak v General Medical Council, [2017] UKSC 71; and see commentary here.

Legislative supremacy suggests that laws passed by legislators must be obeyed. The rule of law suggests that administrators must comply with certain time-honoured standards concerning decision-making processes and substantive outcomes.

The principles of legislative supremacy and the rule of law sometimes can pull reviewing courts in different directions: in some cases, the former tells us that legislatively created bodies with legislatively bestowed jurisdiction must be left alone to the extent the legislation provides (e.g., where the legislation contains a privative clause), but the latter tells us that courts can interfere if certain decision-making standards are not met. Sometimes these principles work in harmony (e.g., where the legislation allows for a full right of appeal).

There is a corollary of these two principles. The corollary also has constitutional force. Our law has a fundamental ordering, one so well-accepted that if any judges do not accept it, they ought to quit and run for public office instead. The principle? Unless valid constitutional concerns are present, laws passed by the legislators bind all, including judges: Imperial Tobacco. Judge-made law must work within the parameters supplied by legislation and cannot oust legislation: see the general discussion in D. Stratas, “The Canadian Law of Judicial Review: Some Doctrine and Cases” (October 21, 2017 version) at pp. 9-14 and 77.

Another principle also springs from the principle of legislative supremacy. But for clarity it is best identified as a separate principle. Where the legislator has vested decision-making power in an administrator and judicial review lies to a reviewing court, administrators and reviewing courts have two different, exclusive roles. Administrators are the merits-deciders and reviewing courts are the reviewers: see discussion in cases such as Access Copyright  at paras. 17-20; Bernard at paras. 17-19; Gusto TV at para. 11; Robbins at para. 17; Tsleil-Waututh at paras. 85 and 87. Each, bound by law, must stick to its role.

A fourth principle is the nature of the separation of powers. Like legislative supremacy and the rule of law, this too is a constitutional principle: Judges’ Reference; Babcock at para. 54.

How does separation of powers play out? In our system of government, government decisions can be placed on a spectrum.  At one end are legislators and their decisions to make laws. Then there are decisions by others, sometimes legislative in nature, on truly political matters or tough political questions about who gets what privilege or grant; in consideration here are political determinations and the government’s view, sometimes ideological, of what is best for all. A little further over are decisions, sometimes legislative in nature, on broad policy matters. Even a little further over are decisions on broad policy matters where some of the policies are informed by laws on the books. Further over are decisions based on a mix of policy matters, legal considerations such as the interpretation of legislation and the adjudication of rights and legal entitlements. Even further over are matters even more in the realm of fact-finding and legal adjudication, devoid of policy content. And then there are decisions pretty much indistinguishable from the matters courts typically handle, matters firmly in the realm of fact-finding, ascertainment of the law and application of the law to the facts, with a view to settling rights and legal entitlements without regard to policy considerations. And then, finally, at the other end of the spectrum there are courts adjudicating rights and entitlements.

Logically, at the legislative end of the spectrum, one would expect that courts would be reluctant to interfere. And absent constitutional concerns, they can’t on the grounds of justiciability. But go a little further in and we find administrators whose decisions are justiciable but who decide matters in a way that is quite alien to what courts do. Way at the other end, just before we reach the courts, one would expect that administrators such as these could be policed quite closely by courts; they are deciding matters and deciding in a way that is pretty much indistinguishable from what courts do.

Put bluntly, at one end, the matters being decided are not in the ken of the court or in its wheelhouse. And at the other, they most certainly are. This has obvious relevance to the intensity to which courts should review various administrative decisions. 

Deducing operational rules from the doctrine and principles: an introduction

Before deducing operational rules from the doctrine and principles, we must consider what sort of operational rules we ought to have in this area of law.

The consideration of this, below, draws upon a more complicated and vigorous debate in the United States concerning the desirability of rules, standards and principles to regulate subject-matters: those interested in pursuing this further can begin, e.g., here and here, and this is just the tip of a large iceberg.

For our purposes here and to keep things simple, I will posit that operational rules can come in two types: rigid and tangible rules and looser, more conceptual, flexible rules.

An example of the former is the test for an interlocutory stay or injunction: RJR-Macdonald. There, one must satisfy three rigid, tangible requirements.

An example of the latter is the test in negligence for determining whether a defendant has fallen below the standard of care expected of persons in the defendant’s position. To the extent that definition of the concept is required, the definition develops as cases are decided over time and it is discovered through good common law method. This is not an area where judges are shackled by a multi-branch test with rigid, tangible requirements that must be met. In the end, judges know the concept and they are trusted to apply it to the cases before them.

In the area of standard of review—determining the extent to which reviewing courts should interfere with an administrators’ decision—what’s best?

A scheme where there are a set number of categories—say, just two—with tangible rules about when each applies? And no flexibility within the categories to adjust the intensity of review? Just two intensities of review, and no others?

Or a more flexible, conceptual, non-categorical approach, one where we recognize that there is a spectrum of intensities of review that vary according to the context?

In the area of appellate review under Housen, the binary categories of review (correctness or palpable and overriding error) depending on whether the appellate court is dealing with questions of law, mixed fact and law, or pure fact makes sense because the context is relatively static. But the substantive review of administrators’ decisions must be different. A more flexible, conceptual, non-categorical approach makes sense because, as explained in the discussion of the separation of powers, above, the administrative context varies so sharply.

The contributions to the Dunsmuir Decade series suggest that most of our leaders in administrative law support a greater reliance on context for determining the extent to which reviewing courts should interfere with administrators’ decisions. I have found no wholesale endorsements of the presumptions set out in Dunsmuir. In academic papers, three of our most well-regarded and knowledgeable administrative law scholars agree that a more contextual approach is apt: P. Daly, “Struggling Towards Coherence in Canadian Administrative Law? Recent Cases on Standard of Review and Reasonableness”; L. Sossin, “The Complexity of Coherence: Justice LeBel’s Administrative Law” and G. Heckman, “Substantive Review in Appellate Courts Since Dunsmuir” (2010), 47 Osgoode Hall L.J. 751 at 778-79.

Judicial policies

In designing operational rules to govern this area of law, universally-accepted judicial policies about how legal proceedings should unfold—not the just the say-so’s and on-the-spot opinions of a particular constellation of judges at a particular moment about social and political values—should play a role. Where the doctrine and principles suggest a number of possible operational rules, the one that best advances these policies should be chosen.

The sorts of judicial policies I am thinking about are access to justice, judicial economy and minimization of litigation expense, as explained in leading decisions such as Hryniak, Danyluk and Housen.

And related to this are insights gleaned from a greater appreciation of the nature of administrative law and the values that animate it, values that judges in this area of law need to draw upon in exercising their discretions: see P. Daly, “Administrative Law: A Values-Based Approach” in J. Bell et al (eds.), Public Law Adjudication in Common Law Systems: Process and Substance (Oxford, Hart Publishing, 2015); and see how administrative law values were deployed to shape operational rules in Tsleil-Waututh, Wilson (F.C.A.) at para. 30, and Bernard.

In considering judicial policies, care must be taken not to elevate any one consideration too far and create operational rules that no longer reflect settled doctrine and well-accepted principles.

For example, minimization of litigation expense might cause some impulsively to advocate for operational rules that are hard-and-fast and simple to apply; some contributors to the Dunsmuir Decade series seem all too ready to go there, for the sake of simplicity.

But simple, one-size-fits-all approaches are at odds with the broad and varied nature of administrators, their decisions and their mandates. Nor are they rooted in settled doctrine and well-accepted principles and, thus, they lack persuasive force and permanency.

The result? Injustice is seen, judges start to chafe, they construct ad hoc exceptions upon exceptions to address the injustice, the longed-for simplicity is lost and, ultimately the operational rules, lacking persuasive force and permanency, join the graveyard of failed approaches over the last few decades.

We’ve seen this all-too-many times. The approach Dunsmuir replaced was called “pragmatic and functional.” It was replaced because it turned out to be neither pragmatic nor functional. The desire in Dunsmuir was about greater simplicity. So how did that work out?

Rigid, inflexible rules aimed at simplicity can be a siren song leading to disaster. And do not underestimate for a moment the simplicity that can emerge from clearly stated operational rules of a conceptual nature founded on definite, well-understood, longstanding doctrine and principles. 

Some possible deductions

What can be deduced from all of the foregoing?

1. Let’s begin with the idea that legislatively created bodies with legislatively bestowed jurisdiction must be left alone to the extent the legislation provides. Let’s combine this with the idea that laws passed by legislators bind all, even courts.

This suggests that reviewing courts must take note of what legislators have said in their legislation. In short, legislative interpretation must be a key part of the exercise of judicial review. In terms of when a reviewing court may intervene, what did the legislature intend?  (By the way, Dunsmuir said this at paras. 30-31, but Dunsmuir’s presumptions and contextual factors and cases post-Dunsmuir have largely ignored this.)

We interpret legislation—discover its authentic meaning—by examining its text, context and purpose: Rizzo and Rizzo; Bell ExpressVu; Canada Trustco. The focus is on what the legislators—the people with the exclusive right to make laws under ss. 91 and 92 of the Constitution Act, 1867—enacted, not what we would like to see enacted. We must do this neutrally and dispassionately without injecting into the analysis our own preconceptions about the extent to which courts should interfere with administrators’ decisions, our opinions about what is best for Canadians or our own musings about what we personally think is sensible and practical: Williams at paras. 41-52; Cheema at paras. 76-80.

This is something courts are very experienced in doing.

Implications can follow. While legislative text may permit a full appeal, this does not mean that a reviewing court may interfere whenever it wants. The legislation, properly interpreted, may suggest that restraint by the reviewing court is in fact warranted.

Where a securities commission makes a decision relying upon a complicated, relatively non-legal, policy-based understanding of the securities industry, did the legislature really intend that courts can interfere with the decision whenever they want to? When the legislation says that an administrator operating in a field that is outside of the ken of the courts can grant a licence when “it is of the opinion that it is in the public interest,” isn’t the legislator really saying that it is for the administrator to decide what is in the public interest with a minimum of interference?

2. Next, the rule of law.

Legislative text can bar an appeal or a review, for example by way of a clause called a “privative clause”. But that does not mean that a reviewing court can never interfere. As explained above, as a matter of legislative interpretation, the legislation, properly interpreted, may suggest that interference by the court in limited circumstances is warranted.

And even if the legislation truly bars appeals in all circumstances, the rule of law allows courts to interfere nonetheless: Crevier; Dunsmuir at para. 31.

But rule of law concerns vary in size and shape. Cases show that the courts’ sensitivity to rule of law concerns is accentuated by decisions that have significant impact on the individual. The more drastic the measure upon an individual, the more likely the court will be vigilant and ensure that the administrator has complied with basic and fundamental substantive and procedural standards of decision-making.

All are subject to law. No one can be a law unto themselves. All must be accountable to the public they serve. Accountability is a particularly important concern for administrators who are emanations of executive government, i.e., part of government as a whole.

Thus, administrators cannot be immunized from meaningful review by reviewing courts nor can they conduct themselves in a way that shields their decisions from review: Crevier, above; and see discussion in Tsleil-Waututh. This has implications for the giving of reasons: administrators must explain themselves sufficiently in order to be accountable to the public and they must say enough to make review possible.

3. In developing and identifying operational rules, courts should respect the third principle—that of the separate roles of the administrator as merits-decider and the reviewing court as reviewer. Operational rules cannot place the reviewing court in the position of merits-decider. This must shape some of our operational rules—e.g., rules concerning the admissibility of evidence in reviewing courts and the role that reviewing courts should play in coming up with reasons that the administrator should have given.

4. Let’s recap some of the above. Legislative interpretation is case specific. As mentioned above, the rule of law assumes greater or lesser prominence depending on the nature of the case. Different types of administrators have different types of decisions. And administrators occupy different positions in the spectrum of administrative decision-makers, at one end completely different from courts and what they do and at the other end virtually the same.

This suggests that the extent to which a court can interfere with an administrative decision must be a qualitative, conceptual assessment that results in a sliding scale.

It follows from the foregoing that the determination of the extent to which reviewing courts should interfere with an administrators’ decision—the intensity of review—is like a dimmer switch, not an on-off switch. Slotting cases into rigid single-standard categories like correctness and reasonableness is inapt.

A few words about a longstanding, uniquely Canadian predilection

For decades now, Canada has dealt with the intensity of review by constructing artificial categories of review and forcing judges to ram their cases into them. We also seem to venerate a superstructure of arcane rules decreeing what goes into what category. Chaos and constant revision is the result.

So we’ve suffered through categorizations into jurisdictional error and non-jurisdictional error, legislative, administrative, quasi-judicial and judicial decisions, categories of correctness, reasonableness and patent unreasonableness and, now, categories of correctness and reasonableness.

Consider for a moment what’s more likely to be true: as a result of Dunsmuir, Canada has found the magic elixir, the idea of two inflexible, single-standard categories of review, and everyone else in the world is silly not to do the same, or Canada is an outlier that stubbornly clings to an idea that everyone else has rejected for good reason?

And what’s the point of all the gymnastics and the spilling of so much ink on whether the standard of review is correctness or reasonableness—and then just doing correctness anyway?

And, with apologies to the majority in Edmonton East, what is wrong with contextual approaches anyway? What I advocate (see below) is not unlike the contextual approach of Baker for determining the level of procedural fairness owed. And Baker has brought relative calm to the law of procedural fairness. A contextual approach can do so here as well. The test for negligence is a contextual approach and no one is calling that law a never-ending construction site; it is reasonably settled in 99.9% of the cases it regulates. Over time, case-by-case, the common law gives a good measure of certainty and predictability.

Some suggested operational rules

From the foregoing, I suggest the following operational rules for this area of law. They are supported by settled doctrine, well-accepted principle and judicial policy. Because of that, if they are adopted, have every chance of widespread acceptance and permanence.

1. The intensity of review must be on a sliding scale that varies according to the terms of the legislation, the breadth of the discretion granted, the prominence of rule of law considerations (including the drastic nature of the decision upon the individual), the nature of the decision-maker and the nature of the decision. Administrators have varying margins of appreciation or ranges of what is acceptable depending on these circumstances.

Avoid rigid tests on this. Let reviewing courts stir all these factors into a pot and decide on the intensity, allowing them to express it in general but still helpful terms. Examples: Farwaha; Boogaard; Mills; Re:Sound; Delios; Abraham; Almon Equipment; Erasmo; Walchuk; Emerson Milling; and see P. Daly, “Struggling Towards Coherence in Canadian Administrative Law? Recent Cases on Standard of Review and Reasonableness” and D. Stratas, “The Canadian Law of Judicial Review: Some Doctrine and Cases” (October 21, 2017 version) at pp. 57-73. Of these, perhaps Boogaard contains the most comprehensive analysis and weighing of factors that can affect the intensity of review.

Indeed, some of these bear a remarkable resemblance to cases in other Westminster jurisdictions whose law is not as chaotic as ours: see, e.g., Wolf v Minister of Immigration, [2004] NZAR 414 (N.Z.H.C.); AI (Somalia) (“the lawfulness of the exercise of powers by a body that is unusually constituted must be judged against its nature and functions, powers and duties and environment in relation to those of other bodies”); Rotherham Metropolitan Borough Council v Secretary of State for Business Innovation and Skills, [2015] UKSC 6; R. v. Secretary of State for the Home Department, [2014] UKSC 60. This suggests that the Canadian cases, above, are on to something useful.

And let’s be honest: cases decided under the reasonableness standard do exhibit variable intensities of review. We see fussy review, not-so-fussy review, and correctness review.  This operational rule reflects what judges, cognizant of the relevant factors, are inclined to do anyway.

Under the approach I suggest, the strong, judge-made presumption of reasonableness adopted by the majority in Edmonton East would disappear, with legislative interpretation playing a greater role in the analysis amongst other factors.

Overall, this would result in more correctness review. Legislative regimes where there is no privative clause (often with a full right of appeal) and other legislative indicators of correctness review will often lead to correctness review; on the other hand, the assignment of power to administrators in specialized areas to grant to withhold licences or approvals based on public interest criteria will often lead to a more deferential review. A good discussion of this is here. The dissenters in Edmonton East are far closer to the approach I suggest.

A return to the old, discarded approach of correctness review for “jurisdictional questions,” recently advocated by a minority of the Supreme Court in Guérin, should be rejected for all the good reasons offered both long ago (see N.B. Liquor at p. 233) and recently (Halifax); and see the compelling majority reasons of the U.S. Supreme Court in Arlington.

“Correctness review for jurisdictional questions” is an example of the siren song of simplicity I spoke of above, one that leads to unprincipled word games and arbitrariness. Under my suggested approach, matters that the Guérin minority considers to be “jurisdictional” and that the administrator “must get right” are going to be matters of legislative interpretation that admit of few interpretive options, i.e., matters on which there the administrator will have little or no margin of appreciation. The Guérin minority could easily live with the operational rule I propose here; and it would be better and offer more permanence because it is based on settled doctrine and well-accepted principles.

2. Make more tangible and rigorous the assessment whether a decision passes muster under deferential standards of review. Identify specific circumstances that can take a decision outside the range of acceptability and defensibility. The cases cited under point 1, above, all do this. Examples include: disobedience with constraining legislation such as legislative recipes that must be followed (Almon Equipment and Emerson Milling); disobedience with mandatory principles such as procedural fairness, constitutional standards, other fundamental standards such as the need for reconciliation with Indigenous peoples and First Nations (Canadian Human Rights Commission); disobedience with court cases that are directly on point and cannot be distinguished on the facts or policy (Abraham; Emerson Milling). For a good approach to deferential review of administrators’ legislative interpretations, see Allen. For more, see D. Stratas, “The Canadian Law of Judicial Review: Some Doctrine and Cases” (October 21, 2017 version) at pp. 57-63.

Certain indicia or badges of reasonableness and unreasonableness can be identified from the case law. These can provide reviewing courts with useful illustrations: Delios at para. 27; Farwaha at para. 100; Re:Sound at paras. 59-61; Boogaard at para. 81; Forest Ethics at para. 69; and for a more complete list of badges, see D. Stratas, “The Canadian Law of Judicial Review: Some Doctrine and Cases” (October 21, 2017 version) at pp. 63-66.  This promotes certainty which facilitates access to justice by fostering predictability of result.

3. End de novo appeals to appellate courts. Judicial economy, access to justice and minimization of expense require this. Absent legal error, first instance reviewing courts deserve deference to their heavily fact-based assessments of the intensity of review. In other words, repeal Agraira and import Housen to govern appellate review.

4. Reviewing courts must stick to their role and not delve in the merits. Their job is to ensure that administrators required by legislators to grapple with a problem have in fact grappled with it. They are not to do the administrator’s job. Therefore, it is no part of the reviewing court’s function to redraft, correct or supplement administrators’ reasons for decision. Neither is it their role to detect error on the administrator’s part and then cooper up the outcome reached by the administrator—an outcome that an administrator might not have reached had it known of its error.

5. A corollary of the relative roles of reviewing courts and administrators is that administrators’ reasons must be sufficient to permit review and to fulfil their obligations of public accountability. A standard similar to that proposed in earlier, lower-court cases on adequacy of administrators’ reasons is designed to meet those objectives and, thus, is more doctrinally sound. Note that these cases sit comfortably with this recent U.K. Supreme Court decision on point.

6. The evidentiary and procedural law of judicial review must reflect the contrasting roles of reviewing courts as reviewers and administrators as merits-deciders. To a large extent, this is happening in the Federal Court of Appeal: see Access Copyright at paras. 17-20, Bernard at paras. 17-19, Gusto TV at para. 11, Robbins at para. 17, Tsleil-Waututh at paras. 85 and 87; and see cases from other jurisdictions referred to in these cases.

7. Judicial economy and efficiency require reviewing courts to take more seriously their remedial discretion. Although a decision is liable to be set aside and sent back for redetermination, that needn’t happen: MiningWatch Canada; Mobil Oil. Circumstances may suggest otherwise. Would any purpose be served in sending the matter back? Realistically speaking, is there only one correct or acceptable result and so there is no point in sending it back? Does chronic or severe maladministration by the administrator mean that the matter should not be sent back but rather should be determined, exceptionally, by the reviewing court? Are there important practical reasons why it should not be sent back? If it is sent back, can the reviewing court, while respectful of the administrator as the merits-decider, still offer guidance by giving fulsome reasons or imposing terms. See generally D. Stratas, “The Canadian Law of Judicial Review: Some Doctrine and Cases” (October 21, 2017 version) at pp. 102-109.

Deduced logically from settled doctrine and well-accepted principle and leavened by contemporary judicial policy, these operational rules will have real staying-power. They give us the best chance of finally ending the tumult and chaos that has afflicted this area of law for decades.

Personal comments

I am one of the last contributors to the Dunsmuir Decade series. On behalf of all of the contributors and on behalf of the administrative law community I would like to offer my personal congratulations and thanks to Professor Daly and Professor Sirota.

I congratulate them both for organizing and carrying out this Dunsmuir Decade series. Collectively, the articles are most illuminating. They have assisted the Bar, academia and the judiciary in understanding the latest issues in the law of substantive review of administrators’ decisions. We are all better for this.

Their blogs, Administrative Law Matters and Double Aspect, are such a service to the legal community. They keep us acquainted with important and interesting developments in administrative law and more generally in public law. To get such prompt insights on the latest legal developments from two cutting-edge experts is a real privilege. Even when we disagree with them, they cause us to reflect, question our preconceived notions, and think about what is correct. Because of the generous work of Professor Daly and Professor Sirota, often many of us end up in a better place in our own work.

Congratulations and thanks!

A View from South of the Border

Dunsmuir, Chevron, and what Canadians and Americans can learn from each other about judicial deference and interventionism

Jeffrey Pojanowski, University of Notre Dame

First, I would like to thank Leonid and Paul for inviting me to contribute to this symposium. Reading up on Dunsmuir and its legacy has expanded my horizons on administrative law and introduced me to great Canadian legal scholarship. My sense is that Canadian administrative law scholars are engaged in important conversations with their counterparts in Australia, New Zealand, and the U.K., whereas U.S. scholars, per usual, are doing their own thing. For reasons I discuss below, that separation may make some sense. But I am also convinced that further conversation between these wings of Anglo-American public law is important, for we are all struggling with the tension between the supremacy of law and the need for sound, politically responsive policy in a complex world. To keep within the space allotted, I will focus on only one of the many comparative angles, namely the extent of correctness review in our two systems. (On the U.S. end, I will only be discussing federal administrative law, not the law governing review of agency action in state governments.)

Dunsmuir, especially as interpreted in Edmonton East, indicates a broad presumption against review for correctness. The exception for general legal questions of substantial importance is narrow, deference on Charter interpretations has taken a bite out of the exception for constitutional questions, and jurisdictional review is withering away. As indicated by the 5-4 vote in Edmonton East, however, this broad presumption of reasonableness is controversial, and there is some indication that a return to contextual factors will defeat a strong, rule-like presumption of reasonableness review.

In the United States, standards of review are (sometimes nominally) governed by a statute, the Administrative Procedure Act (“APA”), which separates questions of law, fact, and policy. As a result, unlike Dunsmuir’s transubstantive reach, we have three separate doctrinal hooks for review, though there is some overlap. For findings of fact, the “substantial evidence” standard is similar to the jury review standard, though with a mood that is a little more searching. On questions of policy, the “arbitrary and capricious” standard of reasonableness governs and, while it has its complexities, there is little doctrinal support for anything like correctness review. Thus, on questions of fact and policy, the U.S. tracks Canada in eschewing correctness review.

Judges and scholars in the U.S., however, are obsessed with judicial review of legal questions. Here, the landmark case is Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984). Chevron offers a deceptively simple test. First, courts ask, using the ordinary tools of statutory interpretation, whether the legislation speaks clearly to the question at hand. If so, that interpretation governs. If the question is unclear, the court then asks whether the agency’s interpretation is reasonable. If it is, that interpretation stands, even if it were not the one the court would have adopted under de novo review. Looming large above this two-step doctrine is the “step zero” question: when does Chevron apply, as opposed to a less-deferential standard of review? Even the most zealous judicial advocates of Chevron deference agree that an eligible interpretation must represent the agency’s authoritative judgment over a statute it administers. Without further qualifications, this strong Chevron approach would look much like the presumption of reasonableness review in Edmonton East.

Yet it is not that simple. In United States v. Mead Corp., 533 U.S. 218 (2001), the Supreme Court rejected a broad Chevron rule in favor of a standard. Even when an agency administers a statute, the Court will look for contextual factors to suggest an exercised delegation of interpretive authority from Congress to the agency. Most prominently, Mead links implied delegation to an agency’s power to make policy through reasonably formal measures, such as legislative rulemaking and procedure-heavy adjudication. Agencies that have those powers and use them have a much stronger chance of receiving Chevron deference on an interpretation than those that lack them or do not use them. In subsequent cases, most notably the healthcare case King v. Burwell, the Court has also indicated that on some legal questions of major importance, it would be implausible to infer that Congress intended deference, even if the agency administers the statute and uses formal procedures. Thus, unlike Dunsmuir, Mead carves out for non-deferential review some legal questions that reside under the aegis of agency’s statute.

Therefore, in the U.S. a rough contextualism reigns supreme, with defeasible rules of thumb about when one can imply a delegation of interpretive authority from Congress to the agency. As in Canada, there is substantial (though not unanimous) dissatisfaction with the doctrine from opposite ends of the spectrum. Those who complain about the unpredictability of the doctrine post-Mead would warn a Canadian pushing for contextualism to be careful about what you wish for. On the other side, a more legalist strain has attacked the legitimacy of any legal deference, claiming that it flouts the APA, abdicates judicial duty, or unfairly biases adjudication in favor of the government. Like Alberta is to Canada, this latter chorus is not the dominant voice in American jurisprudence, but it represents the most sustained attack on deference in a long while.

Arguments about deference touch on deep questions of jurisprudence that transcend national boundaries. But it is also possible to ask mid-level questions about whether, given a set of assumptions or features of a legal system, deference on questions of law makes sense. If a legal community has a uniform approach to statutory interpretation, correctness review might be easier to manage; similarly, deciding when an interpretation is beyond the realm of reason is more tractable if judges carry roughly the same measure. In the United States, there can be sharp disagreements among textualists and purposivists about what counts as a good argument, and thus what makes an interpretation “clear” or “unreasonable.” If the Interpretation Act and Elmer Driedger-style-purposivism lead to interpretive practice as uniform in action as it appears on the books, this suggests that, ceteris parabis, Canadian judges could feel more comfortable than their U.S. counterparts in patrolling agency interpretations of law.

But not all else is equal. If the ordinary science of statutory interpretation in Canada is broadly purposive, that could strengthen the case against correctness review on legal questions. As a legal realist would be quick to point out, picking a statute’s purpose, selecting the level of generality at which to describe the purpose, and making the consequentialist judgment about which interpretation promotes that purpose can be a deeply political and policy-laden endeavor, one that looks a lot more like making law than finding it. On those premises, the standard justifications for Chevron ring true; compared to courts, agencies have superior technical expertise and are more accountable to the political branches. Judicial review of law and policy blur in a way less amenable to the distinctly judicial craft.

In systems like the U.S. where interpretive formalism has much greater purchase, a root-and-branch defense of correctness review could have more stable ground. Where inputs like text, structure, and linguistic canons offer substantial guidance, a formalist judge could contend that resolving a disputed question of interpretation can be separated from the consequent policy implications. (She would be wrong if interpretive formalism is illusory, but she would be right on her own premises.) As I have argued, it is therefore telling that Chevron’s most prominent critics today are neoclassical formalists who resist strongly purposive and dynamic approaches to interpretation. This is not to say such formalists maintain that the law never “runs out” on judicial review. There will be questions, like whether an agency’s regulation is “in the public interest,” that are in fact not questions of interpretation amenable to the formalist toolkit, but rather placeholders for delegated policymaking and its accompanying reasonableness review. But for the formalist, the line between law and policy is sharper, or at least legal disagreement crosses into policy choice much further down the line than the standard interpretive legal realist story suggests. If so (a big if!), that would muddy the policy-based case for broad deference on questions of law.

This critique of reasonableness review on law is not the only one available, but it is the one underwriting deference skepticism in the U.S. today. A Canadian deference skeptic who also rejects interpretive formalism would have to pursue other avenues and explain why judicial policy balancing is superior to its agency counterpart. And, as American scholars like Cass Sunstein and Adrian Vermeule have both argued, that is a challenging task. On the other hand, American jurists and scholars who defend Mead’s contextualism and reject interpretive formalism might look northward to bolster their position by reading the burgeoning Canadian literature criticizing Edmonton East. And, thanks to the internet, such exchange does not require a passport, let alone a drive to the Peace Bridge crossing.

The Merits of Dunsmuir

Rightly or Wrongly Decided (Then and Today)?

The Honourable Joseph T Robertson QC, formerly of the Federal Court of Appeal and the New Brunswick Court of Appeal

This digital symposium marks the 10th Anniversary of Dunsmuir’s release. Undoubtedly, attention will focus on whether this “transformative” decision has achieved its stated objective of simplifying the law of administrative deference. Regrettably, the Supreme Court’s post-Dunsmuir jurisprudence has generated allegations of doctrinal “incoherence” and “inconsistency”. It is also alleged that the deferential standard of review has had little impact on case outcomes that hinge on the interpretation of enabling legislation. Too often the Court is caught applying the correctness standard under the banner of reasonableness (“disguised correctness review”). While that contention falls within my terms of engagement, the others do not.

The title of this post is obviously disingenuous. The “rightness” or “wrongness” of any Supreme Court decision is largely a matter of personal judgment. Importantly, I fail to qualify as a disinterested observer, having authored Dunsmuir on behalf of the New Brunswick Court of Appeal. That decision was upheld in the result, but for reasons that bear little upon what was argued and decided in the lower court. In the circumstances, I hope to sustain reader interest by advancing the following thesis: there is an arguable case that Dunsmuir would be decided differently today when it comes to the underlying merits. I also address the procedural fairness issue in a way that others have not.

While Dunsmuir has generated more than its fair share of commentaries, few have looked to the underlying issues that motivated Mr. Dunsmuir to seek leave of the Supreme Court. Most have forgotten the central issue. The adjudicator was asked to decide whether non-unionized employees of the government were to be treated like unionized employees when it came to termination of employment. Mr. Dunsmuir was a non-unionized employee who had been terminated with severance in lieu of notice. The government consciously chose not to allege cause, as is its prerogative under common law principles.

I pause here to draw attention to what every employment lawyer knows. The underlying issue and fact pattern in Dunsmuir parallel those found in a relatively recent decision: Wilson v Atomic Energy Canada Inc., 2016 SCC 29 (“Wilson”). Better still, the same holds true in regard to Knight v Indian Head School Division No. 19, [1990] 1 SCR 653 (“Knight”) which was central to Dunsmuir’s analysis of whether Mr. Dunsmuir was owed a duty of procedural fairness.

After exhausting the government’s internal grievance procedure, Mr. Dunsmuir filed a third-party grievance that was heard by an adjudicator appointed by the labour board. A preliminary issued was raised with respect to whether, on a discharge with notice/severance, the adjudicator was “authorized” under the enabling legislation to look into the reasons underlying the government’s decision to terminate Mr. Dunsmuir’s employment. The government argued that, under the enabling legislation, it retained the right to terminate with proper notice/severance and, therefore, it was irrelevant whether the government had sufficient grounds for dismissal. In effect, the government argued that only employees who had been discharged for cause could invoke the third-party grievance procedure (save in regard to the amount of notice/severance). In response, Mr. Dunsmuir insisted that the same legislation authorized the adjudicator to look into the true reasons for his dismissal.

The adjudicator ruled that “[a] grieving employee is entitled to an adjudication as to whether a discharge purportedly with notice or with pay in lieu of notice was in fact for cause, either disciplinary or non-disciplinary.” Inexplicably, however, the adjudicator made no finding as to whether the underlying facts supported a dismissal for cause. He simply left us with the understanding that the decision to terminate Mr. Dunsmuir’s employment was not “disciplinary” but related to “work performance” and his “suitability for the positions”.

The adjudicator’s preliminary ruling provided the confidence necessary to pursue the procedural fairness issue. It was framed in terms of whether the duty owed to Mr. Dunsmuir, under the principle established in Knight, had been breached. This argument flowed easily from the facts. Not only was Mr. Dunsmuir a contract employee, he was also an “at-pleasure” appointee of the Lieutenant Governor in Council. As Clerk of the Court of Queen’s Bench, Mr. Dunsmuir easily qualified as a public office holder.

The adjudicator summarily accepted Mr. Dunsmuir’s argument that the government had failed to inform him of the reasons for its dissatisfaction with his work and to provide him with an opportunity to respond and, therefore, the duty had been breached. The adjudicator then declared Mr. Dunsmuir’s discharge void and ordered retroactive reinstatement. Finally, the adjudicator acknowledged the possibility of judicial review before increasing the notice period from four to eight months, but without offering a legal justification for the increase.

In the Supreme Court, the newly minted deferential standard of reasonableness was applied to the adjudicator’s interpretative ruling. Correlatively, the Court was unanimous in declaring his interpretation unreasonable: “The decision of the adjudicator treated the appellant, a non-unionized employee as a unionized employee.” But, as the Court quickly noted, the interpretative ruling was “inconsequential to the overall determination of the grievance.” The adjudicator simply failed to decide whether or not the government had cause for terminating Mr. Dunsmuir’s employment. This meant that the success of Mr. Dunsmuir’s appeal hinged on the application of the procedural fairness duty.

The Supreme Court made no mention of the proper review standard for procedural fairness issues and understandably so. The Court had decided to overturn its earlier ruling in Knight. No longer was Mr. Dunsmuir owed the administrative duty. To borrow directly from Dunsmuir: “where a public employee is employed under a contract of employment, regardless of his or her status as a public office holder, the applicable law governing dismissal is the law of contract, not general principles arising out of the public law.”

Surely no one quibbles with the proposition that Dunsmuir got it “right” on the standard of review issue. And for the record, the notion that the interpretative (preliminary) issue qualified as a “true jurisdictional question” never crossed the mind of anyone and properly so! That said, I acknowledge a movement from within the Court to adopt a single unifying standard of reasonableness for all issues: see Wilson, per Abella J. I also acknowledge a movement from outside the Court to have the deferential standard extended to procedural fairness issues. Paul Daly, C. Bredt and A. Melkov argue in favour while (Justice) John Evans argues for the status quo: with respect to the American debate, see Adrian Vermule v (Judge) Richard Posner.

In the space of five paragraphs, the majority of the Supreme Court concluded the adjudicator’s interpretation of the enabling legislation was “deeply flawed” (paras. 72-76). That opinion was shared by those who wrote concurring opinions. Dunsmuir’s analysis of the interpretative issue has attracted criticism for applying the correctness standard under the banner of reasonableness. Both Matthew Lewans and David Mullan insist that the Dunsmuir Court breached its own directives when applying the reasonableness standard to the adjudicator’s interpretative ruling. And indeed, Professor Mullan made the same observation in regard to the decision of the Court of Appeal. On reflection, he writes with a gentle pen and the criticism of both scholars has merit. However, Dunsmuir is not the only Supreme Court precedent to engage in disguised correctness review. There are close to twenty post-Dunsmuir decisions listed on my scratch pad that warrant the same indictment.

Disguised correctness review means that the reviewing court conducts a de novo analysis of the interpretative issue. Little or no meaningful reference is made to the reasoning of the administrative decision-maker; just the interpretative result. Instead, the reviewing court begins its analysis by turning to the template set out in Elmer Dreidger’s modern principle of statutory interpretation; the one the Supreme Court consistently applies. This leads to an unanswered question: Why does the Court pursue correctness review under the banner of reasonableness? While several plausible explanations are available, Paul Daly offers valuable insights (“The Signal and the Noise in Administrative Law”).

Professor Daly observes, for example, that in cases where the issue is perceived as one of public importance, or of significance to the law, the greater the temptation to seek out the correct or preferred answer. There is no precedential significance in cases that say the interpretative result falls within a range of reasonable outcomes. In theory, such cases should never make it through the Court’s screening process. But once the interpretative issue is recast as a question of whether non-unionized employees should receive the same treatment as unionized ones, what judge would not grant leave to appeal? The opportunity to look into power imbalances in the workplace is the kind of stuff that fits nicely within the job description of any court of last resort. [No one who made an appearance in Dunsmuir anticipated the Supreme Court’s move to a single deferential standard of review and with a revised framework for isolating the proper standard; nor did anyone anticipate that Knight would be targeted.]

Additionally, I maintain the Court’s affinity for disguised correctness review most often stems from the failure of administrative decision-makers to offer an interpretative analysis that comports with Dreidger’s template. One has to assume as much, as the Court’s references to the decision-maker’s reasons invariably focus on the interpretative result to the exclusion of anything that might have been penned in support of one interpretation over another. This leads to the inference that the Court is too often presented with an interpretive result but not much in terms of justification. Accordingly, the Court has no practical alternative than to engage in correctness review.

The fact of the matter is that Mr. Dunsmuir never pursued the interpretative issue in terms of whether the legislation only allowed for terminations with cause. It would be unbecoming to engage in “bootstrapping” with respect to the reasons that propelled the Supreme Court to distance itself from the adjudicator’s approach to the interpretative issue (termination for “disguised” cause). In its place, I offer a compromise of sorts. I maintain there is an arguable case that Dunsmuir would be decided differently today.

My thesis rests on one’s acceptance of the premise that the Dunsmuir legislation was ambiguous as to whether all terminations of employment within the civil service must be for cause. As Professor Willis so aptly stated so long ago: “the words are ambiguous enough to induce two people to spend good money in backing two opposing views as to their meaning.” Assuming ambiguity is present and in the absence of compelling evidence that might shine light on the ever-elusive intent of the legislature, then the task of the adjudicator is to settle on an interpretation that it believes is consistent with the legislation’s objectives: see Mclean v British Columbia (Securities Commission), 2013 SCC 67.

Admittedly, any interpretative argument would have to overcome the presumption against legislation that abrogates common law rights such as the right to dismiss with proper notice/severance. That hurdle, however, is not insurmountable. Indeed, Paul Daly argues that administrative decision-makers should not be bound by the interpretative principles that bind the courts, a proposition with which I respectfully disagree. Additionally, he argues that statutory ambiguity should not be the gateway to administrative deference. On that issue we are also in respectful disagreement. Nevertheless, the fact that my real-life hypothetical assumes ambiguity may be sufficient for purposes of deflecting immediate criticism.

In the face of an ambiguity, one would expect the adjudicator’s analysis to embrace a candid consideration of the competing policy arguments in support of competing interpretations. At this point, the adjudicator now has the opportunity to demonstrate the expertise which is otherwise presumed. Surely, it is the expert who is aware of the workplace realities that permeate public sector employment (in New Brunswick) and is able to communicate those realities when explaining why one interpretation should prevail over another. And surely, everyone is entitled to know why the distinction between unionized and non-unionized employees should be dissolved when it comes to termination of employment.

Some will query why any government would want to relinquish its common law right to dismiss with notice in the face of what every employment lawyer knows is an “elevated” threshold for establishing just cause. Arguably, governments must be as fiscally responsible as employers in the private sector. Perhaps the answer lies in the understanding that, as an employer, the Crown owes more to its employees than do private sector employers. Or perhaps it is because of documented abuses surrounding dismissals with notice in the civil service. Surely, these are the kind of questions the labour expert is expected to address (with the assistance of counsel). Unfortunately, that type of analysis is not found in the any of the decisions underlying Dunsmuir (nor for that matter in Wilson).

At this juncture, the legal realist will take over where Paul Daly left off. Cases such as Dunsmuir have nothing to do with deference. The underlying issue is whether the historical power imbalance in the employment relationship is one requiring reform. The rule of law is preserved so long as administrative reform hinges on statutory ambiguity. The non-expert fails to recognize the influence of those who regard the common law as an impediment to achieving justice in the work place and, correlatively, the non-expert fails to recognize that dismissals with notice are unjust dismissals. Of course, this line of reasoning is consistent with that expected of an employee advocate and antagonistic to the majority view expressed in Dunsmuir. Here is what the majority had to say: “In the context of this appeal, it must be emphasized that dismissal with reasonable notice is not unfair per se. An employee’s right to terminate the employment relationship with due notice is simply the counterpart to the employee’s right to quit with notice [para. 105].”

The notion that a dismissal with notice is an unjust dismissal brings into consideration the Court’s post-Dunsmuir decision mentioned earlier. As in Dunsmuir, the issue in Wilson was whether a non-unionized employee could be terminated without cause. Mr. Wilson, an employee of a federally regulated employer, subject to the Canada Labour Code, was dismissed with severance in lieu of notice. Like Mr. Dunsmuir, he sought to grieve the dismissal but his employer objected for the same reason advanced in Dunsmuir. However, the federal legislation differs in wording. Under s. 240(1) of the Code, a person may grieve if she or he believes the dismissal is “unjust”. Relying on earlier arbitral precedents, the adjudicator held that a dismissal with notice/severance is itself an unjust dismissal. A divided Court (5/4) upheld that interpretative result. In short, terminations can only be for cause under the federal legislation. Curiously, the majority opinion did not address Dunsmuir’s observations with respect to dismissals with notice or severance. The minority opinion did!

Parenthetically, the majority in Wilson is also guilty of disguised correctness review and understandably so. The adjudicator based its decision on the earlier arbitral jurisprudence that favoured Mr. Wilson’s interpretation. As a result, there were no reasons upon which to measure the interpretative result. Further, no attempt was made to look for reasons that might lie elsewhere in the arbitral jurisprudence as happened in Alberta (Information and Privacy Commissioner) v Alberta Teachers’ Association, 2011 SCC 61.

If Dunsmuir were to be re-litigated, Wilson is the one case that might reasonably impact on the outcome. Admittedly, the statutory scheme in Wilson differs in material respects from the Dunsmuir scheme, and in particular with respect to the history of the legislation and, correlatively, the evidence of Parliamentary intent. In short, Wilson serves as both a shield and a sword. As a shield, it counters a view expressed in Dunsmuir that a dismissal with notice is not “unfair” or “unjust”. In summary, there is an arguable case that Dunsmuir would be decided differently on the interpretative issue.

I turn now to the procedural fairness issue which had been abandoned in the Court of Appeal but resurrected for the benefit of the Supreme Court. Having regard to the Court’s subsequent decision in Canada (Attorney General) v Mavi, 2011 SCC 30, it appears Dunsmuir’s ruling is still good law. Hopefully, supporters of Knight will acknowledge an arguable case with respect to the incompatibility of administrative and contract law principles, having due regard to the significance of the good faith doctrine, as discussed in Dunsmuir, and more recently in the Court’s now lead decision of Bhasin v Hrynew, 2014 SCC 71. I also assume commentators will draw a distinction between express and implied employment contracts.

Personally, I believe that Dunsmuir got it right when it came to overruling the impact of Knight on employees who are subject to a written contract and who are also public office holders. But I am also convinced that even if Knight were still good law, Mr. Dunsmuir could not have succeeded. The adjudicator applied the “wrong” standard. In Knight the Court held: “The content of the duty of fairness would be minimal where dismissal is at pleasure.” In fact, the adjudicator in Dunsmuir imposed an “elevated” duty. As will be explained, the elevated one too closely resembles the one applied at common law for determining sufficient or just cause in those instances where the employment is terminated because of a failure to meet work performance expectations (“incompetence). But first, I offer a rudimentary understanding of the law surrounding Knight.

At common law, at-pleasure appointees could be terminated without notice, without cause and without ever knowing why the employer terminated the appointment. The duty of procedural fairness was intended to ensure that the reasons for termination were communicated to the appointee and, in turn, the appointee had the opportunity to address those reasons before the termination took effect. In theory, the appointee would have a chance to address misinformation that might reasonably influence the decision to terminate. This is administrative law’s equivalent of the last chance doctrine.

In Knight, the director of education for the school board had entered into a three-year contract of employment that provided for the possibility of a renewal for an additional three years. However, the contract also provided for termination by ether party upon the giving of three months’ notice. The parties began negotiations for a renewal but the school board was only prepared to renew for a year due to employment issues that had arisen over the preceding years. With no agreement in sight, the school board decided to terminate the contract with timely notice. Mr. Knight’s first argument was that under the Education Act he could only be dismissed for cause! As that interpretative argument failed, the Court focused on whether Mr. Knight’s employment attracted a public law duty of fairness.

The majority (4/3) of the Knight Court so found, but went on to hold there had been no breach. As an at-pleasure appointee, Mr. Knight was well aware of the reasons why his contract was not renewed. Further, he was given the opportunity to speak at the board meeting where the motion to terminate the contract on notice was passed. In brief, the minimal standard had been met. I now turn to an abbreviated recitation of Dunsmuir’s facts, as told by the Supreme Court.

Mr. Dunsmuir’s employment relationship was “not perfect”. The probationary period was extended twice to the maximum of 12 months. He was reprimanded on three occasions during his two years with the government. Two of the reprimands are relevant to the procedural fairness issue. The second reprimand led to a one-day suspension and to notice of work performance issues, including complaints from unnamed staff, lawyers and members of the public. The third reprimand embraced three alleged incidents relating to his job performance and contained a warning that a failure to improve would result in further disciplinary action up to and including dismissal. This was followed by a few meetings the Regional Director to discuss court backlogs and organizational problems in response to undocumented complaints lodged by staff. A further meeting was scheduled to deal with work-related issues but cancelled because of the decision to terminate Mr. Dunsmuir’s tenure with the government.

Now recall that Mr. Dunsmuir had argued that the government had breached its duty of procedural fairness by failing to inform him of the reasons for its dissatisfaction with his work and to provide him with an opportunity to respond. Recall also that the adjudicator summarily accepted that submission. With great respect, that is not a “minimal” threshold or standard. In fact, it is too closely aligned to the common law standard applied in cases where the termination of employment is for cause based on the employee’s inability to perform (“incompetence”). In such cases the employer must establish progressive discipline: sufficient calls for improvement and a warning that failure may result in dismissal.

Succinctly stated, the adjudicator in Dunsmuir effectively adopted the common law standard for establishing just cause as the administrative standard for establishing a breach of the procedural fairness duty. Importantly, the common law standard is incompatible with the minimal standard adopted in Knight! Finally, even if one were to apply the Knight standard, can one reasonably conclude that Mr. Dunsmuir had no knowledge of his employer’s concerns over his performance and no notice of possible termination of employment? Personally, I think Binnie J’s concurring reasons in Dunsmuir capture the essence of the case. The adjudicator had “stretched the law too far in coming to [Mr. Dunsmuir’s] rescue.” But of course I am not a disinterested observer!