This post is written by Marc-Antoine Gervais, and a larger paper on the subject will appear in the McGill Law Journal (vol. 66).
Canada’s model of judicial review of legislation is unusual. On the one hand, it is “diffuse” in that all courts of law (and many administrative tribunals) may decide constitutional questions. On the other hand, declaratory judgments of unconstitutionality issued by superior courts under s. 52 of the Constitution Act, 1982 “establish the general invalidity of a legislative provision for all future cases.”[1] As Professor Daly’s comment on R v. Sullivan reveals, there is still much uncertainty regarding the true nature and scope of s. 52 declarations rendered by provincial superior courts.
Professor Daly contends that “[t]he effect of a declaration of unconstitutionality, just as with a declaration of invalidity of a regulation, is that it no longer exists.” He further suggests that “res judicata has no application here […]: the matter has been decided but not because the requirements of res judicata are met, rather because a declaration with erga omnes force has been issued by a superior court.” With respect, it is submitted that the judiciary cannot literally invalidate or nullify laws, and that the general or erga omnes effect of s. 52 declarations may instead be explained by the doctrine of res judicata.
Courts cannot literally invalidate laws
Judges and commentators commonly assert that superior courts “strike down”, “nullify”, or “invalidate” unconstitutional laws.[2] These statements, however, should not be interpreted literally. Nothing in the Constitution of Canada empowers courts to exercise such a “negative act of legislation”, as Kelsen described it.[3] In fact, the power to annul laws has only been vested in the Queen in Council, although it has fallen into desuetude.[4]
Unconstitutional laws are “of no force or effect” by virtue of s. 52 of the Constitution Act, 1982. Courts merely declare, but cannot render laws invalid, as Justice Gonthier stressed in Martin: “the invalidity of a legislative provision inconsistent with the Charter does not arise from the fact of its being declared unconstitutional by a court, but from the operation of s. 52(1).”[5] Authoring a powerful dissent espoused in subsequent Supreme Court jurisprudence,[6] Justice McLachlin, as she was then, left no doubt as to the judicial nature of declarations of unconstitutionality in Cooper:
It is common to speak of courts or tribunals “striking down” or invalidating laws […]. This view of the Charter is, with respect, inaccurate. The Charter confers no power on judges or tribunals to strike down laws. The Constitution Act, 1982, however, provides that all laws are invalid to the extent that they are inconsistent with the Charter. Laws are struck down not by judicial fiat, but by operation of the Charter and s. 52 of the Constitution Act, 1982.[7]
Justice La Forest’s opinion in Kourtessis further confirms that declaratory judgments are no different in constitutional matters: “the declaration [of unconstitutionality] by its nature merely states the law without changing anything” — that is, it does not literally extinguish the provision deemed unconstitutional.[8] Proponents of the judicial nullification theory will have a hard time justifying the application by the Supreme Court of Canada of provisions that had been declared unconstitutional in prior cases.[9]
Res judicata and the general effect of s. 52 declarations of unconstitutionality
How, then, can a declaration of unconstitutionality generate erga omnes effects? The Supreme Court of Canada gave a hint in Ravndahl by describing s. 52 declarations as “in rem remed[ies]”.[10] A judgment in rem determines the legal status of a person or a thing independently from the context; it has erga omnes effects because it is “binding not only upon the parties but as against the whole world.”[11] The in rem determination is protected by the doctrine of res judicata, which estops anyone from challenging that legal status in other procedures.
Usually, judgments have inter partes effects because they only bind the parties to the litigation. However, in public law, it has been accepted for a long time that some types of determinations — such as declarations of nullity of regulations — are binding in rem.[12] In fact, the first constitutional law judgment in rem dates back to the famous 1637 Ship Money case, in which the court held that the tax levied by King Charles I was legal despite Parliament’s opposition.[13] Shortly thereafter, another litigant was estopped from challenging the same tax on the basis that the determination on the legality of the tax was no longer up for debate, even by non-parties to the Ship Money case.[14]
In modern Canadian constitutional law, the doctrine of judgment in rem may explain the general effect of s. 52 declarations issued by superior courts, as the apex court suggested in Ravndahl. Nonetheless, some difficulties remain. What is the territorial scope of declarations of unconstitutionality concerning federal legislation? In principle, judgments in rem bind the whole world, but this position is inapposite in the constitutional law context. Mark Mancini convincingly argues that it cannot be reconciled with the principle of federalism:
It would be an affront to the principled federalism balance established by the Constitution Act, 1867 to argue that section 52 declarations should extend throughout the country when issued by one judge in a province. The fact that there may be different findings between one province and another is a feature, not a bug, of the federalist system.
Moreover, a nationwide declaration of unconstitutionality issued by a single provincial superior court would have the very unfortunate effect of circumventing constitutional notice requirements to the attorneys general of other provinces.[15] It is thus submitted that judgments in rem should not run across provincial lines in constitutional matters until the Supreme Court — the only national court in Canada[16] — decides the issue.
In practice, superior courts have refused to automatically give effect to s. 52 declarations pronounced in other provinces.[17] In most cases, constitutional decisions are consistent throughout the country, but not always.[18] For example, the mandatory minimum sentence under s. 153 of the Criminal Code was declared unconstitutional by the Nova Scotia Court of Appeal,[19] but subsequently upheld by the Court of Appeal of Alberta.[20] The Supreme Court of Canada dismissed the application for leave to appeal the Alberta decision.
Not the final word on the constitutional matter
The doctrine of res judicata is flexible in common law, and it should not bar the reopening of a constitutional debate under certain exceptional circumstances.[21] Indeed, when the preconditions to the operation of the estoppel are established, the court possesses a discretionary power to refuse to apply it.[22]
Inconsistent superior court pronouncements on the constitutionality of a law in different provinces engender untenable jurisprudential uncertainty. In such circumstances, the controversy should be allowed to make its way to the Supreme Court of Canada, thereby allowing the “percolation” of diverse lower-court perspectives.[23] It must be stressed that this discretion to refuse the application of the estoppelshould only be exercised in order to foster the integrity and coherence of constitutional law, one of the central goals of the doctrine.[24]
Res judicata is thus distinct from, but operates in tandem with, stare decisis. Within a province, a mere disagreement with the underlying reasoning is not a sufficient ground for a court of appeal to neuter the estoppel barring the parties to challenge a judgment of unconstitutionality in rem issued by a superior in a prior case. Nonetheless, if the court elects to allow relitigation of the constitutional question, stare decisis applies as usual — either in its horizontal or vertical form, depending on the context.
Since courts cannot annul or invalidate legislation, it follows that the law once declared unconstitutional remains capable of “reviving” in the future, unless it is repealed. For example, the minimum sentence under s. 153 of the Criminal Code would “revive” in Nova Scotia if the Supreme Court of Canada holds that the provision passes constitutional muster in a subsequent case. The apex court has repudiated multiple declarations of unconstitutionality issued in earlier decisions,[25] and it would be inappropriate to blindly follow such controverted s. 52 declarations. Canadian[26] and American[27] courts have rightly “revived” laws once declared unconstitutional. Res judicata should not prevent the judiciary from revisiting declarations of unconstitutionality that rest on precarious legal grounds.
In the final analysis, courts cannot literally “strike down” laws or “remove [them] from [their] proper place among statutes.”[28] Rather, declarations of unconstitutionality are in rem determinations that generally establish the judicial ineffectiveness of provisions within the court’s territorial jurisdiction, subject to rare exceptions. The main qualification to the general binding effect of s. 52 declarations is that res judicata should not be applied where the estoppel entrenches inconsistencies in the judicial interpretation and application of the Constitution. The rule of law depends on it.
[1] Nova Scotia (Workers’ Compensation Board) v Martin, 2003 SCC 54 at para 31 [Martin].
[2] See Schachter v Canada, [1992] 2 SCR 679 at 715; Ontario (Attorney General) v G, 2020 SCC 38 at paras 114, 237; Vancouver (City) v Ward, 2010 SCC 27 at 1.
[3] Hans Kelsen, “Judicial Review of Legislation: A Comparative Study of the Austrian and the American Constitution” (1942) 4:2 J Politics 183 at 187.
[4] The Constitution Act, 1867 (UK), 30 & 31 Victoria, c 3, s 56.
[5] Martin, supra note 1 at para 28. See also R v Ferguson, 2008 SCC 6 at 35.
[6] See ibid at para 29; Paul v British Columbia (Forest Appeals Commission), 2003 SCC 55 at para 36; R v Conway, 2010 SCC 22 at paras 20, 64–82.
[7] Cooper v Canada (Human Rights Commission), [1996] 3 SCR 854 at para 83.
[8] Kourtessis v MNR, [1993] 2 SCR 53 at 86.
[9] See R v Malmo‑Levine, 2003 SCC 74 and R v Loewen, 2011 SCC 21. The marijuana prohibition under s 4 of the Controlled Drugs and Substances Act had been declared unconstitutional by the Ontario Court of Appeal in R v Parker (2000), 49 OR (3d) 481), and the government adopted a new regulation to solve the constitutional lacuna. However, no legislative amendment was made. See also R v J-LJ, 2000 SCC 51 and R v Lamy, 2002 SCC 25. The Quebec Court of Appeal had declared s 159 of the Criminal Code unconstitutional in R c Roy, supra note 17, but its constitutionality was not challenged in both Supreme Court cases.
[10] Ravndahl v Saskatchewan, 2009 SCC 7at para 27.
[11] See Barry L Strayer, The Canadian Constitution and the Courts, 3rd ed (Toronto: Butterworths, 1988) at 193.
[12] See Dilworth et al v Bala (Town) et al, [1955] SCR 284 at 289 [Dilworth]; Corporation du Village de Deschênes v Loveys, [1936] SCR 351; Emms v The Queen et al, [1979] 2 SCR 1148 at 1158–62 [Emms];
[13] Rex v Hampden (1637), 3 How St Tr 826.
[14] Lord Say’s Case (1638), Cro Car 524.
[15] See Strayer, supra note 11 at 193–95.
[16] See TA Cromwell, “Aspects of Constitutional Judicial Review in Canada” (1995) 46:5 SCL Rev 1027 at 1042.
[17] See R v Pete (1998), 119 BCAC 161 (BC CA); Reference re Same-Sex Marriage, 2004 SCC 79 at 70; R v EJB, 2018 ABCA 239 at paras 72–75; Parent c Guimond, 2016 QCCA 159 at paras 11–18; R c Roy, [1998] RJQ 1043 (Qc CA); R v Scofield, 2019 BCCA 3 at paras 75–89; R v Boutilier, 2016 BCCA 24 at para 45 (Nielsen J); R v Graham and Parks, 2003 BCPC 369 at paras 12–16; R v Nicholls, 2003 BCPC 132 at 74–76.
[18] See infra notes 19–20; s 151(a) of the Criminal Code was deemed constitutional in British Columbia twice in 2015 and 2017, but declared unconstitutional in multiple other provinces. In 2019, the Court of Appeal for British Columbia eliminated this inconsistency by overruling prior cases in the province and declaring the provision unconstitutional. See R v Scofield, 2019 BCCA 3 at paras 11–12, 75–88.
[19] See R v Hood, 2018 NSCA 18.
[20] See R v EJB, supra note 17 (leave to appeal dismissed).
[21] See Emms, supra note 12; Dilworth, supra note 12 at 289–90.
[22] See Danyluk v Ainsworth Technologies Inc, 2001 SCC 44 at para 33.
[23] See Han-Ru Zhou, “Erga Omnes or Inter Partes? The Legal Effects of Federal Courts’ Constitutional Judments” (2019) 97 R du B Can 276 at 296–98.
[24] See R v Mahalingan, 2008 SCC 63 at 38.
[25] See eg R v Turpin, [1989] 1 SCR 1296 at 1333–34; Libman v Quebec (Attorney General), [1997] 3 SCR 569 at paras 55, 79.
[26] See Saltspring Island Water Preservation Society v Rockliffe, [1993] 4 WWR 601 (BCCA).
[27] See Legal Tender Cases (1870), 79 US (12 Wall) 457, in which the US Supreme Court upheld a law that it had declared unconstitutional two years earlier in Hepburn v Griswold (1868), 75 US (8 Wall) 603. See also West Coast Hotel Co v Parrish (1937), 300 US 379, rejecting Adkins v Children’s Hosp (1923). The law’s validity was later confirmed in Jawish v Morlet (1952), 86 A2d 96 at p 97 (DC Mun Ct App). See also Pait v Ford Motor Co (1987), 500 So 2d 743 (Fla Dist Ct App); Pullum v Cincinnati, Inc (1985), 476 So 2d 657 (Fla Sup Ct) at pp 659–60; State ex rel Gillespie v County of Bay (1933), 112 Fla 687 at p 722 (Fla Sup Ct); Pierce v Pierce (1874), 46 Ind 86 at p 95.
[28] Allison v Corker (1902), 67 NJL 596 at p 601.
Double Aspect 
It is great that Professor Gervais is promoting this conversation. There are all kinds of areas where confusion about what a “declaration of invalidity” really is. I also reacted to Professor Daly’s suggestion that federally-appointed provincial superior courts are “national” courts in the sense of their geographic jurisdiction. I was a little surprised that such a sophisticated public law scholar would put out something that seems to me to be such a basic error.
That being said, let me add a note of disagreement. I think there is much more to be said for the role of res judicata and issue estoppel in limiting the relitigation of the constitutionality of statutes, particularly on the basis that the composition of the apex court has changed. It is unfortunate that this wasn’t considered in Bedford (where it would not have applied because the Prostitution Reference was a reference) and in Carter (where the key issue would be whether the in rem nature of the declaratory relief sought in Rodriguez, and denied, had preclusive effect). I tend to see relitigation based on changes in court composition as fundamentally dangerous in terms of the ultimate legitimacy of the judicial system as more than a partisan/ideological tool (see America, United States of). By contrast, as long as we are not talking about the very same statutes, the courts should wear horizontal stare decisis about the analytic framework of constitutional law more lightly.
Maybe more fundamental is that I do think there is, realistically, more going on in a declaration of invalidity of a statute than a purely declaratory order as we would traditionally think about it – whatever La Forest J may have said in Kourtessis. In Ontario v. G, Karakatsanis J makes the point that a declaration of invalidity is more than a “declaration of inconsistency”, as in New Zealand or the United Kingdom. It goes further and attempts to *remedy* the inconsistency. I wish she had developed this insight in response to some of the arguments made by the three concurring/dissenting justices.
In practice, when a provision of a law is inconsistent with the Constitution, either because it is overbroad or underinclusive, a court *both* declares the nature of the inconsistency (in the G case, that NCR and criminally responsible offenders were treated differently) and then goes on to say what the implications are in terms of what, precisely, is “of no force and effect” (and, crucially, when). I would say this makes sense, since the law is only “of no force and effect” *to the extent of the inconsistency* (language that opponents of suspended declarations and other flexible remedies always leave out). Section 52 descends from s. 2 of the Colonial Laws Validity Act – which was quite clearly as much about validating colonial legislation to the extent it was consistent with imperial legislation as about invalidating it when it wasn’t. If she had gone down this route, then she would have had a textualist/originalist response to the dissenting justices that her four “remedial principles” were not just plucked out of the air, but are really part of s. 52 – it is just as much a court’s duty not to interfere more with legislative choice than necessary as it is to vindicate the constitutional text. In my view, the majority in the G case did this much better than the dissent (since it gave the challenger individual relief and gave the legislature a reasonable amount of time to come up with its own preferred solution to a complex problem).
All of which is to say that a declaration of invalidity contains both a declaration of inconsistency and a remedy for that inconsistency. This latter looks much more like a positive remedy of “striking down.” If, as has happened, the legislature re-enacts virtually the same provision, then stare decisis but not res judicata are in play (because it is not an injunction). But clearly until the declaration is set aside, the law cannot have any legal effects in the jurisdiction in which it has been declared.
Leonid raised this on twitter, but I would also quibble about the claim that the Supreme Court of Canada is the only national court. As you probably know, there is a bubbling issue about whether the Federal Courts can give in rem/erga omnes declarations of invalidity. In Windsor, Justice Karakatsanis put this in doubt, but the Federal Court itself has responded by saying the Federal Court Act provides for this. So if someone wants a “nationwide” declaration of invalidity of a federal law, they may be able to get it through the federal court system, depending on how the Supreme Court of Canada ultimately resolves this.
The not-really-declaratory nature of constitutional declarations can also be seen for “declarations of title” in the aboriginal law context. In Tsilhqot’in, it operates much more like a prospective vesting order than like a declaration. There are all kinds of practical reasons for this, although it seems in tension with the text, and it would have been nice for the SCC to engage with the issue more. One approach would have been to rethink what “existing” is doing in section 35, reading it more realistically as a direction to balance reliance interests as a result of the history of settler colonialism with the imperative of overcoming that history, rather than as a legalistic tautology.