The Language of Administrative Law IV: Terminological Exactitude

(This is the fourth in a series of posts. Read the first, second and third. The draft can be downloaded in its entirety here.)

Language can be useful in reminding judges of the limits of the judicial role. As any student of administrative law comes quickly to understand, where a power has been assigned by legislation to an administrative body, it is not for a reviewing court to substitute its judgement on the merits of the decision taken. Administrative lawyers make a fundamental distinction between judicial review – which forms part of the inherent jurisdiction of superior courts of record but is limited to an assessment of legality – and appeals – which may only be created by statute and which may permit an appellate body to interfere on the merits.[1] Where judicial review requires a consideration of the merits of an administrative decision, as in situations where the rationality of the decision is under attack, judges need to be careful to respect the autonomy of the decision-maker chosen by the legislature.

Accordingly, the terminology of rationality review is designed to remind judges of their institutional role. Lord Greene M.R.’s formulation of Wednesbury unreasonableness is a classic example: “something so absurd that no sensible person could ever dream that it lay within the powers of the authority”;[2] courts could intervene only “if a decision on a competent matter is so unreasonable that no reasonable authority could ever have come to it…”, but this would require the applicant to demonstrate something “overwhelming”.[3] Other colourful formulations can be found in the cases. Lord Diplock described this ground of review as requiring “a decision which is so outrageous in its defiance of logic or of accepted moral standards that no sensible person who had applied his mind to the question to be decided could have arrived at it”.[4] An Irish judge saw the judicial role as “considering whether the impugned decision plainly and unambiguously flies in the face of plain reason and common sense”.[5] These formulations are salutary reminders of the appropriate limits of judicial oversight of the administration.

Language has played a similar role over the course of recent Canadian administrative law history, since the emergence of the ‘patent unreasonableness’ standard of review of expert administrative decision-makers in the 1970s. As Justice LeBel observed in CUPE (2003), the definitions of ‘patent unreasonableness’ offered by the Supreme Court of Canada were “intended to indicate the high degree of deference inherent in this standard of review”.[6] He offered as examples the definitions suggested by Justice Dickson – “cannot be rationally supported by the relevant legislation”[7] – Justice Cory – a “very strict test…[satisfied when a decision is]…clearly irrational, that is to say evidently not in accordance with reason”[8] – and Justice Iacobucci – a defect characterized by “immediacy or obviousness”[9] leading to a decision “so flawed that no amount of curial deference can justify letting it stand”[10].

However, language has its limits. Justice LeBel’s reasons in CUPE (2003) represent a thoughtful engagement with the limits of language in substantive review. At the time, the Supreme Court of Canada’s framework for substantive judicial review of administrative action was under attack. Commentators decried the “epistemological” confusion[11] created by an unwieldy four-factor analysis[12] combined with three standards of review, a non-deferential standard of correctness that permitted a reviewing court to substitute its view for that of the decision-maker and two deferential standards: reasonableness simpliciter and patent unreasonableness.

In a memorable passage, one first-instance judge wrote:

In attempting to follow the court’s distinctions between “patently unreasonable”, “reasonable” and “correct”, one feels at times as though one is watching a juggler juggle three transparent objects. Depending on the way the light falls, sometimes one thinks one can see the objects. Other times one cannot and, indeed, wonders whether there are really three distinct objects there at all.[13]

Justice LeBel began by noting that the judiciary must ensure a “predictable, workable and coherent jurisprudence” for use by lawyers, litigants and lower courts,[14] a duty they had failed in because the prevailing law “does not currently provide sufficiently clear parameters”.[15] The limits of terminological exactitude could be most clearly observed in judicial attempts to distinguish reasonableness simpliciter from patent unreasonableness. As Justice LeBel explained, the two standards shared an underlying analytical basis – that there will be sometimes, perhaps often, be more than one possible reasonable interpretation of a statutory provision, in which case a reviewing court should defer to the wisdom of the front-line decision-maker: “Because patent unreasonableness and reasonableness simpliciter are both rooted in this guiding principle, it has been difficult to frame the standards as analytically, rather than merely semantically, distinct”.[16]

Indeed, the gallant efforts to distinguish the indistinguishable came to resemble judicial linguistic gymnastics as judges fell over one another in their attempts to achieve a perfect verbal distillation of the two deferential standards of unreasonableness. Cory J.’s invocation of the adjective “clearly” added little more than “rhetorical effect”.[17] LeBel J. also cited a first-instance judge who had commented: “If the unreasonableness of a decision is not clear or obvious, I do not see how that decision can be said to be unreasonable”.[18] As long as the judicial inquiry was to whether a decision was rationally supported by the reasoning and evidence offered by the decision-maker, a meaningful distinction “attempting to qualify degrees of irrationality”[19] would be impossible: “There seems to me to be no qualitative basis on which to differentiate effectively between these various characterizations of a rationality analysis; how, for instance, would a decision that is not “tenably supported” (and is thus “merely” unreasonable) differ from a decision that is not “rationally supported” (and is thus patently unreasonable)?”[20] The same was true of a “somewhat probing examination”[21] of a decision, difficult to distinguish from one that is “simply probing, or…less than probing”;[22] whether it was designed to identify an immediately obvious defect or one that is obvious once it has been discovered.[23] And if a “merely” unreasonable decision should be allowed to stand because it was not “clearly” or “patently” unreasonable, or a defect overlooked because it required more than a somewhat probing examination to uncover it, the law would be according irrationality a “legitimate presence in the legal system”,[24] a most unsavoury prospect.[25]

In 2003, Justice LeBel saw “no easy way out of this conundrum”,[26] but his preferred solution was relatively clear, as he signed off by asking “Should courts move to a two standard system of judicial review, correctness and a revised unified standard of reasonableness”.[27] Sure enough, five years later, Justice LeBel and Justice Bastarache wrote the set of reasons in Dunsmuir v. New Brunswick that abolished the two deferential standards and replaced them with one:

Reasonableness is a deferential standard animated by the principle that underlies the development of the two previous standards of reasonableness: certain questions that come before administrative tribunals do not lend themselves to one specific, particular result. Instead, they may give rise to a number of possible, reasonable conclusions. Tribunals have a margin of appreciation within the range of acceptable and rational solutions. A court conducting a review for reasonableness inquires into the qualities that make a decision reasonable, referring both to the process of articulating the reasons and to outcomes. In judicial review, reasonableness is concerned mostly with the existence of justification, transparency and intelligibility within the decision-making process. But it is also concerned with whether the decision falls within a range of possible, acceptable outcomes which are defensible in respect of the facts and law.[28]

But this definition too has difficulties. What exactly do “justification”, “transparency” and “intelligibility” mean? Just as it added little to “irrational” to add “clearly”, so it added little to “reasonableness” to add these abstract hallmarks of best administrative practice. Indeed, some courts took this aspirational language as a departure from the policy of deferential self-restraint and a signal to intervene more readily in the administrative process.[29] The Court soon replaced its aspirational language with a more restrictive and more precise formulation.[30] As Justice LeBel put it in a subsequent case, the question is whether the reviewing court can “clearly understand” the thought process of the decision-maker.[31] Even here, language alone cannot do all the work – for to apply this test a reviewing court must appreciate the role of reasons and the record in permitting it to perform its task of ensuring that the decision under review was reasonable[32] – but it sets out a comprehensible criterion for judicial judgement.

Being misled by linguistic formulations is not a pitfall unique to the Canadian legal landscape. The United States has its ‘hard look’ test for determining whether an administrative decision was arbitrary or capricious. Does this mean that reviewing courts must literally take a hard look at the material in the record, reading it twice or three times, or perhaps with the aid of a magnifying glass? But shouldn’t they always take a hard look?[33] Similar questions might be posed of England’s ‘anxious scrutiny’ standard: shouldn’t a judge’s scrutiny always be anxious? Is ‘anxious scrutiny’ to be juxtaposed against ‘relaxed scrutiny’, performed perhaps under the influence of barbiturates? In both cases the answer is no. Indeed, the ‘hard look’ and ‘anxious scrutiny’ labels are dangerously misleading. Arbitrary and capricious review is in fact similar to – perhaps indistinguishable from – the grounds of abuse of discretion well known to the common law;[34] and anxious scrutiny imposes additional process obligations on decision-makers in sensitive cases.[35] In neither case does the language of hardness or anxiety do any analytical work.

The difficulty in all of these cases lies in using descriptive language as a substitute for doctrinal precision. As Lord Greene M.R. observed in Wednesbury, reasonableness “has frequently been used and is frequently used as a general description of the things that must not be done”.[36] Yet what really matters is whether judicial intervention is justifiable. This is purely a normative question.[37] Thinking that descriptive language can provide the necessary normative basis for judicial intervention (or, indeed, judicial restraint) is a serious error. A judge must demonstrate why there is a basis for intervention, something that requires engagement with the underlying “substantive principles”[38] or “reason and structure”[39] of unreasonableness or irrationality.[40] Otherwise, the reviewing court risks overstepping the limits of the judicial role by substituting judgement on the merits of the question at issue; at the very least, it “renders doctrinal analysis otiose”.[41]

Sometimes the error of substituting descriptive terms for normative analysis is compounded by the introduction of a hypothetical reasonable observer. Consider these observations of an Irish judge:

[T]he kind of error that produces invalidity is one which no rational or sane decision maker, no matter how misguided could essay to be reviewably irrational. It is not sufficient that a decision maker goes wrong or even hopelessly or fundamentally wrong: he must have gone completely and inexplicably mad: taken leave of his senses and come to an absurd conclusion. It is only when this last situation arises or something akin to it that a court will review the decision for irrationality.[42]

Not only does this passage reveal the standard error of those who search in vain for terminological exactitude – what is the difference between the “wrong” and the “hopelessly or fundamentally wrong”? – but it also conjures up a mythical decision-maker who has been afflicted by madness, as if the judge’s role were to imagine himself as a delirious official asked to take an administrative decision. But judicial review does not require such leaps of imagination. It requires only that judges ask whether a decision should stand or fall and apply appropriate criteria in answering this question.

[1] See generally Paul Daly, « Les appels administratifs au Canada » (2015), 93 Revue du Barreau canadien 1.

[2] Associated Provincial Picture Houses v. Wednesbury, [1948] 1 K.B. 223, at p. 229.

[3] Wednesbury, at p. 230.

[4] Council of Civil Service Unions v. Minister for the Civil Service, [1985] A.C. 374, at p. 410. Strikingly, his Lordship also commented that to “eliminate confusion”, the term “reasonableness” is best avoided in public law. Ibid., at p. 409.

[5] State (Keegan) v. Stardust Victims’ Compensation Tribunal, [1986] I.R. 642, at p. 658.

[6] CUPE (2003), at para. 78.

[7] CUPE, at p. 237.

[8] Canada (Attorney General) v. Public Service Alliance of Canada, [1993] 1 S.C.R. 941, at pp. 963-964.

[9] Canada (Director of Investigation and Research) v. Southam Inc., [1997] 1 S.C.R. 748, at para. 57.

[10] Law Society of New Brunswick v. Ryan, [2003] 1 S.C.R. 247, at para. 52.

[11] D. J. Mullan, “Recent Developments in Standard of Review”, in Canadian Bar Association (Ontario), Taking the Tribunal to Court:  A Practical Guide for Administrative Law Practitioners (2000), at p. 26.

[12] For the uninitiated: statutory language (the presence or absence of a privative clause and right of appeal); the nature of the question (legal or factual); the relative expertise of the decision-maker and the reviewing court; and the purpose of the provision at issue (policy-making choice or the application of an objective legal norm to the facts as found).

[13] Miller v. Workers’ Compensation Commission (Nfld.) (1997), 154 Nfld. & P.E.I.R. 52 (Nfld. S.C.T.D.), at para. 27, per Barry J. (as he then was). I note that Barry J.A. (as he now is) remains interested in these issues: see the excellent discussion in Workplace Health, Safety and Compensation Commission v. Allen, (2014), 379 DLR (4th) 271.

[14] CUPE (2003), at para. 64.

[15] CUPE (2003), at para. 66.

[16] CUPE (2003), at para. 103.

[17] D. J. Mullan, “Recent Developments in Standard of Review”, in Canadian Bar Association (Ontario), Taking the Tribunal to Court:  A Practical Guide for Administrative Law Practitioners (2000), at p. 24.

[18] Hao v. Canada (Minister of Citizenship and Immigration) (2000), 184 F.T.R. 246, at para. 9, per Reed J.

[19] CUPE (2003), at para. 111.

[20] CUPE (2003), at para. 107.

[21] Southam, at para. 56.

[22] D. W. Elliott, “Suresh and the Common Borders of Administrative Law:  Time for the Tailor?” (2002), 65 Sask. L. Rev. 469, at pp. 486-87, cited in CUPE (2003), at para. 112.

[23] CUPE (2003), at paras. 116-120.

[24] CUPE (2003), at para. 132.

[25] CUPE (2003), at paras. 109, 125.

[26] CUPE (2003), at para. 121.

[27] CUPE (2003), at para. 134.

[28] Dunsmuir, at para. 47.

[29]See e.g. Driver Iron Inc. v. International Association of Bridge, Structural Ornamental and Reinforcing Ironworkers, Local Union No. 720 (2011), 502 A.R. 229, rev’d [2012] 3 S.C.R. 405; Public Service Alliance of Canada v. Canada Post Corporation, [2011] 2 FCR 221, rev’d [2011] 3 S.C.R. 572.

[30] Newfoundland and Labrador Nurses’ Union v. Newfoundland and Labrador (Treasury Board), [2011] 3 S.C.R. 708, at para. 14: “the reasons must be read together with the outcome and serve the purpose of showing whether the result falls within a range of possible outcomes”.

[31] Agraira v. Canada (Public Safety and Emergency Preparedness), [2013] 2 S.C.R. 559, at para. 89.

[32] See e.g. Leahy v. Canada (Citizenship and Immigration), 2012 FCA 227, [2014] 1 FCR 766, at para. 122: “Any reviewing court upholding a decision whose bases cannot be discerned would blindly accept the decision, abdicating its responsibility to ensure that it is consistent with the rule of law”. See also Wall v. Office of the Independent Police Review Director, 2014 ONCA 884.

[33] In practice the look might not be very hard at all. See Jacob E. Gersen and Adrian Vermeule, “Thin Rationality Review”, Unpublished, August 4, 2015 (

[34] Motor Vehicle Manufacturers Association v. State Farm Mutual Automobile Insurance (1983) 463 U.S. 29, at p. 43:

Normally, an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.

[35] Paul Craig, “Judicial Review and Anxious Scrutiny: Foundations, Evolution and Application”, [2015] PL 60.

[36] Wednesbury, at p. 229.

[37] See e.g. Delios v. Canada (Attorney General), 2015 FCA 117, at para. 26:

Bearing in mind the margin of appreciation that the administrator should be given – a margin that can be narrow, moderate or wide according to the circumstances – we examine the administrator’s decision in light of the evidentiary record and the law, to examine whether the decision is acceptable and defensible on the facts and the law.

[38] Anthony Lester and Jeffrey Jowell, “Beyond Wednesbury: Substantive Principles of Administrative Law”, [1987] Public Law 368

[39] Paul Daly, “Wednesbury’s Reason and Structure”, [2011] Public Law 237.

[40] See also Paul Craig, “The Nature of Reasonableness Review” (2013), Current Legal Problems 1, though for all that Craig is persuasive in demonstrating that reasonableness review requires consideration by judges of the weight given to relevant factors by decision-makers, he does not provide much guidance to judges charged with assessing reasonableness in a particular case.

[41] Wade & Forsyth, Administrative Law, 11th ed. (Oxford University Press, Oxford, 2014), p. 304, fn. 124.

[42] Aer Rianta v. Commissioner for Aviation Regulation, unreported, High Court, January 16, 2003 (O’Sullivan J.).

This content has been updated on September 9, 2015 at 12:34.