The Search for a Simpler Test: Dunsmuir and Categories (Andrew Green)

Andrew Green is an Associate Professor at the Faculty of Law, University of Toronto

In Dunsmuir v. New Brunswick[i], the Supreme Court of Canada attempted to re-structure administrative law to more fully recognize legislative supremacy, maintain the court’s role in ensuring legality and, at the same time, reduce the costs of making administrative law decisions.  It was a tall order.  It is impressive that the Court has been as successful as it has so far at least on the substantive review side.  It hasn’t made the process simple but that was never going to happen – the hope was for a “simpler test”.[ii] We aren’t there yet but Dunsmuir gives us some keys to the path forward.

In terms of substantive review, Dunsmuir established a choice between two standards of review – correctness and reasonableness.  To structure that choice, the Court combined a categorical approach with the ability to tailor the categories according to factors that essentially embody the former pragmatic and functional approach.  Categories can make the choice of the appropriate standard of review simpler, reducing the costs of arguing about which standard applies and decreasing uncertainty for litigants.   Further, the simplicity may lessen the risk of judges making mistakes in choosing the standard or manipulating the standard to achieve a particular result (choosing a correctness standard solely to overturn a particular decision).  Simple categories can be monitored by higher courts on appeal, by the legislature in revising administrative frameworks or, more tenuously, by the public.[iii] The result could potentially be increased access to justice and greater certainty for litigants and for legislators.

Categories may of course be problematic where they are too vague or are over or under-inclusive.[iv]  Vague categories increase litigation costs and uncertainty.[v]  Over or under-inclusive categories may allow for some decisions to be reviewed on a correctness basis when the standard should be reasonableness and vice versa.  A key concern comes from the trade-off of expertise and independence.[vi]  If a court chooses correctness instead of reasonableness, there is the potential for errors if the decision requires expertise that the judge does not have or if the decision involves making value choices that the executive decision-maker is more legitimately established to make.  Litigants whose views do not align with the court’s may be less likely to challenge decisions, decision-makers may be influenced towards the court’s view to avoid being over-turned on review and legislature’s view of policy may not be vindicated.  On the other hand, if reasonableness is chosen instead of correctness, the court may be less able to police executive decision-makers where they are mistaken in their views or are following their own self-interested view of the right policy (as opposed to that of the legislature), litigants who lose in front of a decision-maker will be less likely to challenge the decision and decision-makers may make more aggressive interpretations that do not accord with legislative goals.  Categories then structure the type and probability of certain errors.

A contextual, factor-based test may be better able to tailor the review to the underlying decision.  Such factors could explicitly take into account factors we care about in making decisions such as what the legislature actually said they wanted, where the relative expertise lies, or whether there is value to political accountability relative to independence.  The difficulty of course is that case-by-case analysis raises the costs of decisions for the litigants, increases the risk of error and manipulation by lower court judges, makes it harder for appellate judges to monitor lower court decisions, and makes life less certain for litigants and legislators.[vii]

The question is – how do you get the value from the simplicity of categories with the reduced error costs of tailoring through the use of factors?  It turns out the Court has used the Dunsmuir framework to come up with a nice answer – what has amounted to a broad presumption of reasonableness with judges required to use principled reasoning to move to a correctness standard.  A broad presumption of reasonableness lowers the cost of choosing the standard in the majority of cases.  It also increases the certainty for parties choosing whether to launch a judicial review and for regulators deciding in the shadow of judicial review.  Further, a presumption of reasonableness ties into the value of deference.   It recognizes both that the legislature has set up expert bodies for a reason and that judges are limited in their ability to make certain decisions, particularly in the case of complex and uncertain policy choices.[viii]

Judges have to consider factors to move to the correctness category but that’s a feature not a bug.  As Daly has pointed out, one of the values of factors is that they force judges to think about and justify their choices in a way that categories do not.[ix]  In order to move off of the presumption of reasonableness, judges have to think about the underlying reasons why they are better positioned to make the decision. Judges may make better choices and there is a clearer basis for appellate judges to monitor lower court judges for mistakes or manipulation and for the legislature to make its wishes known about who should be making a decision.

Such a move only reduces monitoring costs, however, where there is a set of understood bases for such a move to correctness.  Yet this is something that the courts do all the time – it is the common law.  Just as Stratas J.A. has pointed out that the court can point to “badges of unreasonableness” for assessing a decision’s reasonableness, courts can over time develop ‘badges” of when a correctness standard should be applied.[x] Some current considerations should probably not be used such as the notion of jurisdictional error.  Others should become central such as clear statements of legislative intent or actual consideration of relative expertise.  Clarifying these factors, rather than trying to modify the correctness categories, is where the real benefits lie.

Of course everything could be done under a single reasonableness standard.[xi]  The court already has to calibrate reasonableness to take account of the diversity of decisions and decision-makers, so why not just incorporate what were formerly correctness reviews under that umbrella.  Judges could then calibrate reasonableness to different contexts, including ones where there are essentially no degrees of freedom possible for the executive decision-maker. So what would be lost?  Most obviously, as Binnie J. warned, you may be merely shifting the cost to what was formerly the second stage.[xii]  The costs and uncertainty would be higher unless there was some greater specification of the calibration than currently exists, whether it is with Stratas J.A.’s badges of unreasonableness or something else. However, more would be lost.

Loss of the standard of correctness gives rise to the potential loss of the ‘judicial self-discipline’ advocated by Justice Iacobucci[xiii] – eliminating the separate category of correctness may actually weaken judges’ mental focus on deference under reasonableness.  Further, a single standard of reasonableness relies heavily on an assumption that the executive decision-maker has relative expertise with the result being the risk of errors where judges have relative expertise.[xiv] More context is needed in terms of the decision about who has expertise, not less – it is a key factor that has been lost.[xv] In addition, a correctness standard avoids the disingenuous claim that a judge is applying reasonableness when in fact she is undertaking correctness analysis.[xvi] Finally, and perhaps most importantly, in some cases, it seems clear that the legislature wants the court to make the decision, for reasons of expertise or relative independence, and has said so.[xvii]

So moving from the presumption of reasonableness started under Dunsmuir to a single reasonableness standard risks bringing back increased cost and uncertainty. At very least the Dunsmuir framework as it has been developed does not appear correlated with greater discussion of the choice of standard, increased use of correctness or higher rates of overturning decisions.[xviii]  In fact the opposite appears true.  The mistake of Dunsmuir may have been to provide categories for correctness instead of pointing to a contextualized move from reasonableness based on factors. The search for a simpler test is clearly not over but the lessons of simplicity and signaling from Dunsmuir should not be lost.

[i] [2008] 1 S.C.R. 190.

[ii] Dunsmuir, at para. 43.

[iii] Andrew Green, “Can There Be Too Much Context in Administrative Law: Setting the Standard of Review in Canadian Administrative Law” (2014) 47 U.B.C.L.R. 443.

[iv] Adrian Vermeule, Judging Under Uncertainty: An Institutional Theory of Legal Interpretation (Cambridge, MA: Harvard University Press, 2006) and Paul Daly, “The Unfortunate Triumph of Form Over Substance in Canadian Administrative Law” (2012) 50(2) Osgoode Hall Law Journal 317.

[v] Jacob E. Gerson and Adrian Vermeule, “Chevron as Voting Rule” (2007) 116(4) Yale LJ 676.

[vi] See, for example, Vermeule (2006) and Matthew C. Stephenson, “Statutory Interpretation by Agencies” in Danial Farber and Anne Joseph O’Connell, eds., Research Handbook on Public Choice and Public Law (Cheltenham, UK: Edward Elgar, 2010).

[vii] On the trade-off of decision costs and errors, see Vermuele (2006) and Cass Sunstein, “Chevron Step Zero” (2006) 92(2) Virginia Law Review 187.

[viii] Adrian Vermeule, Law’s Abnegation: From Law’s Empire to the Administrative State (Cambridge, MA: Harvard University Press, 2016).

[ix] Daly (2012).

[x] Delios v. Canada (Attorney General), 2015 FCA 117, at para. 27.  Stratas J.A. has pointed to a range of factors that may help provide the basis for a more “nuanced” approach to the choice of standard of review as well as strong arguments against an overall presumption of reasonableness: David Stratas, “The Canadian Law of Judicial Review: Some Doctrine and Cases” (October 2017), available at SSRN: For a discussion of the need to maintain the presumption but rely on the contextual factors to recognize different types of decisions, see the dissent by Cote and Brown JJ in Edmonton (City) v. Edmonton East (Capilano) Shopping Centres Ltd.. [2016] 2 S.C.R. 293.

[xi] Wilson v. Atomic Energy of Canada Ltd., 2016 SCC 29, per Abella J.

[xii] Dunsmuir, para. 139 (per Binnie J.).

[xiii] Law Society of New Brunswick v. Ryan, [2003] 1 S.C.R. 247, at para 46 (per Iacobucci J.).

[xiv] See, for example, Edmonton (City) v. Edmonton East (Capilano) Shopping Centres Ltd.. [2016] 2 S.C.R. 293 where Karakatsanis J. for the majority, notes that expertise “inheres in a tribunal itself as an institution” (at para. 33).

[xv] For a start at bringing a more contextual analysis of expertise back, see the dissenting opinion by Cote and Brown JJ. in Edmonton (City) v. Edmonton East (Capilano) Shopping Centres Ltd.. [2016] 2 S.C.R. 293, paras. 81-90.  See also Martin Olszinski, “Dunsmuir is Dead – Long Live Dunsmuir! An Argument for a Presumption of Correctness” (, arguing for a refinement of the concept of expertise.

[xvi] On “disguised correctness”, see David Mullan, “Unresolved Issues on Standard of Review in Canadian Judicial Review of Administrative Action – The Top Fifteen” (2013), 42 Adv. Q. 1.

[xvii] See the disagreement between the majority and dissent in Edmonton (City) v. Edmonton East (Capilano) Shopping Centres Ltd.. [2016] 2 S.C.R. 293, with the dissent arguing that recognizing the statutory right of appeal as pointing towards correctness is not a new “category” but a principled application of the contextual approach.  See also Stratas (2017).

[xviii] Robert Danay, “Quantifying Dunsmuir: An Empirical Analysis of the Supreme Court of Canada’s Jurisprudence on Standard of Review” (2016) 66 UTLJ 555 and Diana Ginn, William Lahey, David Constantine and Nicholas Hooper, “How Has Dunsmuir Worked? A Legal-Empirical Analysis of Substantive Review of Administrative Decisions after Dunsmuir v. New Brunswick: Findings from the Courts of Nova Scotia, Quebec, Ontario and Alberta” (2017) 30 Can. J. Admin. Law and Practice 317.

This content has been updated on February 13, 2018 at 09:52.