Horrocks: What Happens to Agraira?

One of the more nerdy judicial review questions is the standard of review applicable from an appellate court to a lower court in judicial review cases. That is, how do appellate courts deal with lower court decisions that, either through a right of appeal or by application for judicial review, review administrative decisions?

The current orthodox position is outlined in Agraira. There, the Supreme Court held that when an appellate court reviews a lower court decision disposing of a judicial review application, the appellate court “steps into the shoes” of the lower court (Agraira, at para 46). In effect, as Stratas JA points out in Sharif, at para 5 this turns out to be pure de novo review: the appellate court reconducts the standard of review analysis, in both selection of the relevant standard of review and its application.

Post-Vavilov, Stratas JA questions how long Agraira-review will last (see pg 60 of his masterful work here). Luckily, we won’t have to wait too long for the answer to this question. The Supreme Court recently granted leave in Horrocks, a Manitoba case that raises this question directly. The Supreme Court, in describing the main question in the case, says the following: “What is appropriate standard of appellate review, as between levels of court sitting in review of decision of administrative tribunal?”

This is an open question. In this post, I assess some of the options for dealing with this issue. I first describe the holding of the Manitoba Court of Appeal in Horrocks, then I outline the potential options that are in front of the Court in assessing this question. I ultimately conclude that, on balance, there are good reasons to favour the application of the Housen v Nikolaisen standard of review across the board when an appellate court reviews a lower court’s review of an administrative decision. Put differently, we should rid ourselves of Agraira.

**

Horrocks involves, in addition to the standard of review issue, a tricky issue of “jurisdictional lines” between two or more tribunals. While I do not address this issue in this post, some description of the basic conundrum is necessary for context.

Horrocks was a health care aide at a personal care home operated by the Northern Regional Health Authority in Manitoba (NRHA). NRHA is a party to a collective agreement with CUPE. Horrocks, during work hours, was observed as intoxicated. After refusing to sign a document that would have forced Horrocks to abstain from alcohol completely, her employment was terminated. CUPE grieved under the collective agreement. Prior to a scheduled arbitration, another document was negotiated between the parties, which again asked Horrocks to abstain from alcohol, and contained other revisions, including that any breach of the document would constituted just cause for termination, and that Horrocks could challenge any employment decision by NRHA under the collective agreement’s grievance and arbitration procedures.

Horrocks was, again, found to be intoxicated outside work. NRHA again terminated her employment. Horrocks then filed a discrimination claim with the Manitoba Human Rights Commission (MHRC). The core question was whether the MHRC had jurisdiction over the dispute, or whether the collective bargaining/arbitration provisions governed. At the Manitoba Court of Queen’s Bench, the court found that the MHRC lacked jurisdiction. The Manitoba Court of Appeal allowed the appeal, concluding, interalia, that there was a “modest” place for human rights adjudication; and so the lower court judged erred in overturning the MHRC decision that there was jurisdiction.

In analyzing the issue, the MBCA also commented on standard of review. The Court held that “both the identification and the application of the appropriate standard of review by a superior court judge conducting a judicial review is a question of law under the standard of review framework as set out in Housen [v Nikolaisen]” (Horrocks, MBCA at para 47). Therefore, the Court purported to apply the framework set out in Housen for appellate review (see Horrocks MBCA at para 38), meaning correctness review on any extricable legal questions. However, the Court, relying on Stewart, also said that the basic question on judicial review was whether “there is a principled reason to afforded deference here…” (Stewart, at para 19; Horrocks MBCA at para 49). The Court concluded that there was no such reason, because:

If one returns to the basic question discussed in Stewart (see para 19) as to whether there is a principled reason to afford deference here, I am satisfied that there is not. The record before the reviewing judge was the same that was before the Chief Adjudicator. He was not required to make any original findings of fact or exercises of discretion. Additionally, there are no limitations on the Commission’s right of appeal of the reviewing judge’s decision pursuant to section 89 of The Court of Queen’s Bench Act, CCSM c C280, such as a requirement that the decision being appealed must have wider significance beyond the parties such that leave to appeal must first be obtained. Taken together, these circumstances make it difficult to justify, on a principled basis, that a margin of appreciation should be afforded to the reviewing judge’s decision when he was not required by Dunsmuir (at
para 61) to afford deference to the decision of the Chief Adjudicator on the same issue (Horrocks MBCA at para 49).

Seemingly on the basis of Stewart, the Court therefore concluded that the administrative decision must be reviewed on a correctness standard, in both application and selection of the standard of review.

**

The question: what should the Court do with Agraira in Horrocks?

One option might be the reliance on Stewart as a comprehensive statement of the relationship between appellate courts and lower courts on judicial review. But I do not read Stewart as the MBCA does. That is, I do not see Stewart standing for a broad-based approach to the relationship between appellate courts and lower courts. But on principle, a standard phrased like the one in Stewart is too broad to be of much use on a comprehensive basis. It raises the prospect of standard-less doctrine that does not help reviewing judges to know the proper basis of intervention.

More seriously, there are two other basic approaches that are operating in tandem, to my mind. The first, the Agraira approach, views both the selection and application of the standard of review as questions of law. A lower court choosing and applying a standard of review, then, is owed no deference by an appellate court, even on questions of mixed fact and law or fact. Again, as noted above, this approach results in basically de novo review, a “re-do” on the merits by the appellate court, with the focus on the administrative decision.

This approach has its benefits and drawbacks. In principle, it is aligned with the focus of judicial review: the administrative decision. It ensures that administrative decisions (and errors) are not sheltered by deference doctrine applied by the appellate court. But there is a price to pay for fidelity to principle: as Stratas JA notes in Sharif, at para 5, it is an open question whether this “re-do” is consistent with principles of judicial economy and access to justice, heralded by cases like Hryniak v Maulden.

The second approach is to adopt Housen v Nikolaisen. That is, the lower court decision would be viewed as a pure appeal, notwithstanding the fact that it reviews an administrative decision. This means that, while selecting the relevant standard of review will be a question of law, there will be room for some deference on a palpable and overriding error standard on the application of the standard of review, and the law, to the facts (see, for nuance on this point, John Evans, “The Role of Appellate Courts in Administrative Law” (2007) 20 Can J Admin L & Prac”). A good example of this approach is contained in Hupacasath. There, Stratas JA noted that while he was applying the Agraira standard, that approach does not allow for the substitution of factual (or mixed fact and law) findings made by a lower court (Hupacasath, at para 75). Stratas JA goes on to say:

In my view, as is the case in all areas of appellate review, absent some extricable legal principle, we are to defer to findings that are heavily suffused by the first instance court’s appreciation of the evidence, not second-guess them. Only palpable and overriding error can vitiate such findings. In the context of the existence of Aboriginal title, the Supreme Court held to similar effect in Tsilhqot’in Nation, supra at paragraph 52 (Hupacasath, at para 76; see also Long Plain First Nation, at para 86.

This approach, of course, solves some of the problem of “re-doing” the standard of review analysis that characterizes the other approach. And, at least nominally, it ensures that the selection of the relevant standard of review remains a question of law that is reviewable on a correctness standard under Housen. But it leaves the door open to deference on findings “heavily suffused by the first instance court’s appreciation of the evidence.”

The main difference between approaches, then, is in the application stage, where Housen would not counsel a pure de novo review. People will different priors will favour one approach over another, but a complicating factor in the choice is the decision in Vavilov. While there is much in Vavilov that bears on this question, the main distinction that affects the choice between approaches is the strong distinction now drawn between cases going up to a higher court on a statutory right of appeal and cases proceeding by an application for judicial review. This distinction is driven, in the Vavilov Court’s mind, by an expression of Parliament: “…legislative intent can only be given effect in this framework if statutory appeal mechanisms, as clear signals of legislative intent with respect to the applicable standard of review, are given effect through the application of appellate standards by reviewing courts” (Vavilov, at para 49). So, on this account, there is a distinct difference when Parliament legislates a right of appeal—in effect, it legislates the appellate standards of review (see Vavilov, at para 36).

This suggests that there might be a reason to favour one standard of review framework over another, depending on the context. However, in my view, there are reasons internal and external to Vavilov to favour the Housen standard of review as a comprehensive standard for appellate review.

1) The appellate standard can no longer be Agraira because of reasons internal to Vavilov. This is because of Vavilov‘s holding on statutory rights of appeal. Now, an administrative decision taken to a court via a statutory right of appeal invites the application of the Housen standard of review (Vavilov, at paras 36 et seq). This, in other words, is a legislative signal that courts should treat the lower administrative decision as a decision from which to be appealed. There is no principled reason to differ the approach when the lower court decision is subsequently reviewed. The relationship between the appeal court and the lower court is similarly governed by statute; and the fact that the first instance decision-maker is an administrative decision-maker does not change the statutory relationship going “all the way up” to the appeal court. Agraira, then, is inconsistent with Vavilov on this score.

2) When the case involves an application for judicial review, the question is trickier. Vavilov does not speak directly to the issue. We now have the choice between Agraira and Housen presented directly. Based on reasons external to Vavilov, in my view, there is good reason to ditch Agraira and move to Housen for these cases, whether they involve a reasonableness or correctness standard on pure questions of law. That is because, when viewing the decision of a lower court that deals with an administrative decision, the appellate court’s role is not to directly review the administrative decision. In the hierarchy of courts, the appellate court role is to correct errors of lower courts; it is not to re-review the merits of administrative action. That is, primarily, the task of the superior courts and courts of first instance with judicial review jurisdiction. Based on this overriding first principle, the Agraira rule had three main flaws:

a) In principle, it corrupted the relationship between appellate courts, superior courts, and administrative decision-makers. Appellate courts do not review administrative decisions at first instance; they review decisions of lower courts as a matter of appeal. The appellate standards should, therefore, apply.

b) Whether the relevant standard is reasonableness or correctness, Agraira saps any deference to a lower court’s appreciation of the facts and evidence from the analysis. While the selection of the standard of review is a question of law to be decided by the court (see Monsanto, at para 6), its application to the facts may involve mixed questions of fact and law or assessment of evidence. There are good functional reasons to favour a first instance court’s appreciation of these issues. Agraira, for no good reason, did not account for this issue of first instance courts.

c) As Peter Hogg noted, there is no good reason to favour review of any kind, given that review necessarily involves some duplication of effort. Sometimes, review is a legal necessity, and we bite the bullet on duplication (ie) judicial review in the first instance is a constitutional necessity, and a legal good. But there is no principled reason to favour duplication of effort by an appellate court that did not get a first-hand glimpse of the record. There is no constitutional or legal good served by this sort of review that would vitiate these concerns.

A comprehensive standard for appellate review, then, emerges: it is the Housen v Nikolaisen standard. Adopting that standard in all instances of appellate review of a judicial review decision has two added benefits. First, it creates a comprehensive standard across judicial review contexts for appellate review. Whether the case involves a statutory right of appeal or an application for judicial review, an appellate court’s posture remains the same. While there are different reasons to favour the Housen standard in rights of appeal versus judicial reviews, the bottom line is the same: the posture of the appellate court is focused on reviewing potential errors in a lower court decision. The appellate court, at a basic level, is not a court of original judicial review jurisdiction, and was always a mistake to transform appellate review into first-instance judicial review.

But secondly, on a practical level, appellate courts are familiar and comfortable with Housen review. We now have extensive guidance on how to apply Housen review, including on the tricky issue of what constitutes “palpable and overriding error” (see South Yukon per Stratas JA, at para 46; Mahjoub per Stratas JA at para 61 et seq). There is a practical benefit that supports the in-principle reasons for favouring Housen review on these matters.

Overall, while this issue might appear to be a niche issue for administrative law lawyers, it is actually a fundamental issue. It goes to the relationship between the judiciary and the administrative state, an issue that should captivate all public lawyers. Hopefully, Horrocks helps to clear up some of the confusion that characterizes the current status of Agraira, on this front.