Tsvi Kahana' UNDERSTANDING THE NOTWITHSTANDING
MECHANISM1
I Introduction
In recent years, Canada has gained international recognition as a constitutional empire. Commentators overseas often refer to the Canadian Charter of Rights and Freedoms1 as an exemplary rights protection document,2 and judges across the globe, from China3 to South Africa4 to Israel;' cite Canadian constitutional cases with comfort. This phenomenon evokes pride in Canadian lawyers, especially given the American hegemony in constitutional theory. In light of this pride in Canadian distinctiveness, it is surprising that so little has been written about the most unique feature of the Charter, the Notwithstanding Mechanism (NM)'' provided under s. 33.7 This dearth of literature is glaring, especially
* Executive Director, Centre for Constitutional Studies, University of Alberta.
f This paper is based in part on my doctoral work at. the University of Toronto Faculty of Law and on research I conducted as a Visiting Scholar at Yale Law School. I am profoundly grateful to ray doctoral supervisor, Professor Lorraine Weinrib, for her unwavering support and invaluable guidance, and to Professor Owen Fiss for his insights in developing the manuscript. I wish also to thank the members of my dissertation committee, Professors Carl Baar, Alan Brudner, David Schneidenrtan, andJohnWhyte. For their valuable comments while researching and writing this manuscript, 1 would also like to thank Professors David Beatty, David Dyzenbaus, Victor Ferreres, and Roberto Saba as well as Chris Moore, RobertBlitt, Karen Gauvine, Pippa Reyburn, and Matthew Woodley. I dedicate this paper to the memory of my mother, Esther Golda Kahana.
1 Canadian Charter of Rights and Freedoms, Part I of the Constitution Act, 1982, being Schedule B to the Canada Act 1982 (U.K.), 1982, c. 11 [hereinafter the Charter].
2 See, e.g., Z. Segal, 'Israel Ushers in A Constitutional Revolution: The Israeli Experience, The Canadian Impact' (1995) 6 Const.Foruin 43.
3 See 'CaiiadianLegalWisdomaHolComrhodity Abroad' The Globe and Mail (28 August 2000) Al.
4 See, e.g., State v. Makuianyane (1995), 3 S.A. 391 (C.C.) at. paras. 60-2.
5 See, e.g., United.Mhrahi Bank Ltd., e.t al. v. Migdal Cooperative Village, d al (1995), 49 P.D. 221 at para. 98.
6 Reference is often made to a 'notwithstanding clause.' This term is confusing because it has three different meanings in the literature in which it is discussed. It is alternatively referred to as (a) the text of s. 33, (b) the mechanism that, the section creates, or (c) a clause in an act stating that the act shall operate notwithstanding a Charter provision. I use the term 'notwithstanding clause' exclusively to refer to the text of s. 33, Similarly, the 'limitation clause' denotes the text of s. I of the Charter. In this paper, the mechanism that the notwithstanding clause establishes is termed the 'notwithstanding mechanism' or NM. For the purpose of further clarification, a declaration in an act that.
(2002), 52 UNIVERSITY OF TORONTO LAW JOURNAL
222 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 223
given the reams of scholarly treatment generated by analysis of the other distinctive feature of the Charter, s. l's limitation clause.8 The purpose of this article is to help fill this gap in Canadian constitutional theory by exploring the idea of the NM.
Section 33 provides Canadian legislatures with the ability to displace most Charter rights.9 A Canadian legislature's ability to ignore rights remains constrained in two ways: first, the legislature must expressly declare that, the legislation shall apply notwithstanding certain Charter provisions, thereby risking public condemnation; second, after a period of Five years, if the legislature wants the rights-violating measure to continue, it is required to re-enact the measure, possibly in the face of renewed public outcry. Ultimately, however, if a legislature is determined to preserve a law that violates most fundamental Charter-protected rights, it can do so.-10
the act or a provision thereof operates notwithstanding the Charter is called 'a notwithstanding declaration,' and an act in which a notwithstanding declaration appears is 'a notwithstanding act.' Many observers refer to the NM as the 'legislative override.1 Unlike the term 'notwithstanding,' the term 'override' does not appear in s. 33, and therefore I prefer to use the former.
7 S. 33 reads as follows:
Exception where express declaration
33. (1) Parliament or the legislature of a province may expressly declare in an Act of Parliament or of the legislature, as the case may be, that the Act or a provision thereof shall operate notwithstanding a provision included in section 2 or sections 7 to 15 of this Charter.
Operation of exception
(2) An Act or a provision of an Act in respect of which a declaration made under this section is in effect shall have such operation as it would have but for the. provision of this Charter referred to in the declaration.
Five year limitation
(3) A declaration made under subsection (1) shall cease to have effect five years after it comes into force or on such earlier date as may be specified in the declaration. Re-enactment
(4) Parliament or a legislature of a province may re-enact, a declaration made under subsection (1).
Five year limitation
(5) Subsection (3) applies in respect of a re-enactment made under subsection (4).
8 See, e.g., the abundance of literature referred to in R.F. Devlin, 'The Charier and Anglophone Legal Theory' (1997) 4 Rev.Const.Stud. 19.
9 S. 33 can be applied to fundamental freedoms (s. 2), including freedom of religion and conscience, freedom of expression and assembly, and freedom of association; legal rights (ss. 7-14); and equality rights (s. 15). In contrast, s. 33 cannot be invoked with respect to democratic rights (ss. 3-5), mobility rights (s. 6), rights regarding the official languages of Canada (ss. 16-22), minority language education rights (s. 23), or gender equality rights (s. 28).
10 For an argument that evens. 33 does not allow for extremely severe violations of rights, see B, Slattery, 'Override Clauses under S. 33 - Whether Subject to Review under s. 1' (1983) 61 Can.Bar.Rev. 391; D, Arbess, 'Limitations on Legislative Override under the
This situation exists because no consensus regarding the issue of rights existed when the Charter was adopted. Prime Minister Pierre Trudeau introduced the notwithstanding clause in the final stages of the negotiations surrounding the patriation of the Canadian Constitution." Tru-deau's patriation sought to establish final Canadian independence from Britain, to affix an amendment procedure to the Constitution, and to entrench a Charter of Rights, enumerating the political rights of Canadians.12 However, confronted by wide differences between the opponents and the supporters of entrenchment, the parties settled on the compromise of an entrenched Charter subject to a NM that would allow legislatures to deviate from the rule of rights protection. Pragmatically speaking, there is no doubt that the compromise reflected a clever choice. While the courts have found many laws unconstitutional due to Charter infringement, the legislature has rarely invoked the NM.1S In other words, even if Canada does not enjoy a full rights protection regime de jure, it clearly enjoys such a regime de facto. Although the wisdom of this compromise and the success of the NM in practice are significant issues, they are not the focus of this article.14 Rather, I am concerned here with the theoretical underpinnings of the NM.
Only three Canadian writers - Paul Weiler, Brian Slattery, and Lorraine Weinrib - have offered a theoretical foundation for the NM.15 Weiler sees the mechanism as a tool for correctingjudicial errors; Slavery's view is that
Canadian Charter of Rights and Freedoms; A Matter of Balancing Values' (1983) 21 Osgoode Hall L.J. 113.
11 For a history of the NM, see H. Leeson, 'Section 33, the Notwithstanding Clause: A Paper Tiger' (2000) 6:4 Choices 1 at 6-14 [hereinafter 'Paper Tiger'].
12 For a history of the 1982 patriaüon, see R. Romanow, J. Whyte, & H. Leeson, Canada Notwithstanding (Toronto: Carswell/Methuen, 1984).
13 For an analysis of the practice of the NM in Canada see T. Kahana, 'The Notwithstanding Mechanism and Public Discussion: Lessons from the Ignored Practice of Section 33 of the Charter' (2001) 44 Can.Pub Admin. 255 [hereinafter 'Notwithstanding Mechanism']. The only blatant abuse of the NM seems to be Quebec's infamous Act Respecting the Constitution Act, 1982, S.Q. 1982, c. 21. This Act repealed all of Quebec's legislation and then re-enacted it with newly affixed notwithstanding declarations. Significantly, the use of the NM in this instance was not aimed at protecting a specific piece of legislation but, rather, was an act of political protest against the fact that the Charter was entrenched without the Quebec government's consent. For a critical discussion of Quebec's Act Respecting the Constitution Act and the Supreme Court of Canada's decision to uphold it in Ford v. Quebec. (A.G.), [1988] 2 S.C.R. 712, see L. Weinrib, 'Learning to Live with dre Override' (1990) 35 Mc.Gill L.J. 541 [hereinafter 'Learning to Live'].
14 I address these questions in T. Kahana, The Partnership Model of the Canadian Notwithstanding Mechanism: Failure and Hope (S J.D. Thesis, Faculty of Law, University of Toronto 2000) [unpublished] atch. 5 [hereinafter Partnership Model].
15 Other commentators have expressed support for the NM, but all of them have followed Weiler, even if not always acknowledging this fact. Some of Weiler's followers are mentioned in note 26 infra.
224 UNIVERSITY OF TORONTO LAW JOURNAL
the mechanism is aimed at preventing such errors; and Weinrib reasons that the mechanism enables the legislature to create exceptions to rights.
According to Weiler, while judicial review of the constitutionality of legislation is an important aspect of constitutional democracy, the practice should not be determinative, since courts are not immune from making mistakes. If the court strikes down legislation based on an erroneous interpretation or application of the Charter, the NM allows the legislature to correct this mistake by re-enacting the impugned legislation.16 Slattery believes that the NM can be used to prevent the court from making such mistakes. According to Slattery, if the legislature is confident that an act is constitutional - to such an extent that it believes any judicial decision to strike it down would be erroneous - it can simply invoke the NM pre-emptively and thus prevent judicial review altogether.17 Both Weiler and Slattery advocate the legislature's use of the NM only to protect legislation believed by the legislature to be constitutional, and only in circumstances where the court's striking down of this legislation must necessarily be erroneous. To state the position differently, both believe that while invoking the mechanism, the legislature acts as a 'super-court.'18 in that it makes its own judgment on the constitutionality of the relevant legislation. It is this construction that Weinrib challenges in her approach to the NM. According to Weinrib, the legislature does not declare the constitutionality of an act by invoking the NM, Rather, it concedes that the legislation it is enacting might be unconstitutional, but nevertheless ought to be enacted as an 'exception' to the rights guaranteed by the Charter.™ Creating such exceptions, Weinrib argues, makes the legislature not a 'super-court' but, rather, a 'super-legislature.'50
These three approaches to the NM form the starting point for this article's analysis. Part II of this analysis examines whether these three theories fit within the text of s. 33, concluding that of the three approaches, Weinrib's exceptions approach best fits the text of the Charter because it conceptualizes the NM as a legislative rather than a quasi-judicial mechanism. Part III moves from constitutional law to constitutional theory. It concludes that none of the three approaches survives theoretical scrutiny. Both Slattery's prevention of judicial error approach
16 P.C. Weiler, 'Of Judges and Rights, or Should Canada Have a Constitutional BUS of Rights?' (1980/81) 60 Dal.Rev. 205 [hereinafter Judges and Rights (1980)']; P.C. Weiler, 'Rights and judges in a Democracy: A New Canadian Version' (1984) IS U.Mich.J.Law Reform 51 [hereinafter 'Rights andjudges (1984)'].
17 B. Slattery, 'A Theory of the Charter' (1987) 25 Osgoode HallLJ. 701. Slattery does not mention the notion of prevention of judicial error. This notion encompasses my understanding of Slattery's approach. See text accompanying note 30 infra.
18 This is Weinrib's term. Weinrib, 'Learning to Live,' supra note 13 at 569.
19 Ibid.
20 Ibid.
understanding the notwithstanding mechanism 225
and Weinrib's exceptions to rights approach fail because they allow the legislature to use the nm pre-emptively, thereby silencing the court altogether and obstructing individuals whose rights are infringed from airing their claims. Weiler's check on judicial error approach, does not suffer from this flaw, since his theory requires that the nm be invoked only if the courts err in striking down an act. His approach, however, suffers from a different fault: it supports the idea that a non-judicial body can correct judicial mistakes. Such an idea contradicts the fundamental constitutional notion that courts are better suited to engage in constitutional interpretation than legislatures, since it allows the less able legislature to supervise the more able court.
After rejecting the theories of Weiler, Slattery, and Weinrib, the article moves on to posit an alternative approach to the nm. This alternative, the 'deliberative disagreement approach,' holds that the role of judicial review under the Charter is not to supervise the legislature but, rather, to deliberate about the Charter's meanings. The legislature must permit the court to lead national deliberation about these meanings, and therefore ought not be allowed to use the nm until the Supreme Court has deliberated on tire constitutionality of a given piece of legislation. Once the court delivers its ruling, the legislature is free to reject it and to re-enact the impugned legislation. This approach is similar to Weiler's, since it holds that the nm should be invoked only in response to a Supreme Court decision. However, there is a significant difference between Weiler's check on judicial error approach and the one I shall propose. Weiler's theory holds that the Charter envisages mutual supervision between courts and legislatures: courts supervise legislatures through judicial review, whereas legislatures supervise courts through the nm. This mutual supervision dynamic is the source of the theoretical inconsistency in Weiler's work, as it suggests that the legislature may invade the powers of the judiciary by supervising its decision making. In contrast, this difficulty does not arise under the deliberative disagreement approach because courts and legislatures do not supervise one another. Rather, courts deliberate through judicial decisions and legislatures decide whether to accept the courts' conclusions.
Part rv discusses the idea of a partnership between courts and legislatures. It suggests that when invoking the nm, the legislature should not view the court as its enemy but, rather, should work cooperatively with the court. It posits that in order to successfully foster a meaningful partnership between the courts and the legislatures, the legislatures must adhere to three notions. I refer to these notions as the 'partnership of respect,' the 'partnership of benefit,' and the 'partnership of last resort.' A partnership of respect implies that the legislature invokes the nm out of respect for the constitutional text and for the court. Respect for the constitutional text means that the decision to invoke the nm is based on the
226 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 227
legislature's reading of the Constitution and not on the legislature's political preferences; respect for the court, means that the legislature's decision to re-enact the legislation comes only after it has become conversant with the court's decision. The notion of a partnership of benefit requires that the legislature not use the NM until the country's highest court has ruled on the matter. It is only after the highest court has issued its decision that the legislature and the polity can benefit from a fully developed judicial voice. Finally, the notion of a partnership of last resort means that if the legislature has at its disposal other means with which to achieve its goal, such as enacting new legislation, it should not use the NM.
II Constitutional law: Understanding s. 33
A. THREE APPROACHES TO S. 33
This part of the article presents three approaches to s. 33 and examines them in light of Dworkin's 'fit' test.21 Weiler was the first Canadian academic to suggest that the NM represents an advantageous constitutional evolution and not merely a compromise. In a 1979 lecture, Weiler expressed the view that Canada should adopt a constitutional Bill of Rights with a 'non obstante provision.''2 Weiler reaffirmed his support
21 See R. Dworkin, Law's Empire. (Cambridge: Harvard University Press, 1986) at ch. 7 ('Integrity in Law').
22 'Judges and Rights,' supra note 16 at 232. Far a citation of this paper delivered as a public lecture delivered in 1979, see Weiler.'Rights and Judges,' supra note 16 at 80 note 96.
Weiler saw the idea of including a NM in the Charter as his 'own modest proposal' ('judges and Rights (1980),' supra note 16 at 231). In his 1984 paper, Weiler writes that in late 1981, during constitutional negotiations, he spoke with the premiers of Ontario and Saskatchewan, as well as with officials from Ottawa and British Columbia, about the idea of a NM. Following these discussions, the premiers came to the November 1981 meetings '[e]quipped with this basic familiarity with the non obstante notion within Canadian taw, and with one extended scholarly defense of its virtues in the constitutional context, [and] reached for this formula for their Accord in the early hours of the morning of November 3,1981' ('Rights and Judges (1984),' supra note 16 at 80 note 97). Samuel LaSelva also writes that the fact that the Charter includes a notwithstanding clause may be attributed to Weiler. See S. LaSelva, 'Only in Canada: Reflections on the Charters Notwithstanding Clause' (1983) 63 Daihousie Rev. 383 at 383 [hereinafter 'Only in Canada']. Former Alberta Premier Peter Lougheed has a different account of who introduced the idea of a NM and how the premiers came to learn about it. Lougheed attributes the inclusion of the NM in the Charter to former Alberta Attorney General Merv Leitch. Based on Leitch's advice, Lougheed proposed it to the premiers in late 1980 at a First Ministers' Conference in Ottawa. See P. Lougheed, 'Why aNotwithstanding Clause?' (Inaugural Merv Leitch Memorial Lecture, University of Calgary, 20 November 1991) (1998) G Points of View 1 at 1-2. Janet
for such a provision in a 1984 piece following the enactment of the Charter and s. 33.s3 The crux of Weiler's account of the Charter and the NM is found in the following excerpt:
The premise of the Charter is that the. optimal arrangement for Canada is a new partnership between court and legislature. Under this approach judges will be on the front lines; they will possess both the responsibility and the legal clout necessary to tackle 'rights' issues as they regularly arise. At the same time, however, the Charter reserves for the legislature a final say to be used sparingly in the exceptional case where the judiciary has gone awry.2'1
Weiler states that a NM represents a beneficial check on the court's power in cases where the judiciary has 'gone awry' and has erred in interpreting the Charter. As F.L. Morton puts it, 'fj]tist as judicial review serves as a check on a certain kind of legislative mistake, so "legislative review" serves
Hiebert also writes that 'Alberta Premier Peter Lougheed had first proposed the inclusion of a legislative override in 1979': j.L. Hiebert, Limiting Rights: - The Dilemma of Judicial Review (Montreal 8c Kingston; McGill-Queens University Press, 1996) [hereinafter Limiting Rights] at 163 note 43 (Hiebert does not offer a reference for this statement). Interestingly, Howard Leeson cites a reference to a notwithstanding clause from briefing notes of the Saskatchewan Continuing Committee of Ministers on the Constitution written in 1978, and Ann Bayefsky traces the first version of a notwithstanding clause hack to a 1969 draft constitutional bill of rights, See Leeson, 'Paper Tiger,' supra note 11 at 10, andA.F. Bayefsky, 'The Judicial Function under the Canadian Charter of Rights and Freedoms (1987) 32 McGill L.J. 791 at 813 note 75. Since this work does not focus on the origins or his wry of the Charier or the NM, this question is not one of central importance. Nevertheless, this contradictory evidence is perhaps indicative of the many other controversies surrounding the history of the 1982 patriation.
23 'Rights and Judges (1984),' supra note 16. Both of Weiler's articles trace a path from democracy to judicial review to the problem of judicial finality. The main innovation presented in 'Rights and Judges' is his interpretation of the NM as a partnership between courts and legislatures. See ibid, at 84, 86. This notion, which is not fully developed by Weiler, is the subject of Part IV infra. Aside from this addition in 1984, the distinctions between the theoretical arguments contained in Weiler's two articles arc not significant. Two points are nonetheless worth mentioning. First, in 1980 Weiler suggests that in order to pass a notwithstanding act, the legislature has to pass the legislation twice - once before the next election for that legislature, and then again after the election-while in 1984 he refers to the check that: was added to s. 33 (i.e., the five-year time limit). See 'Judges and Rights (1980),' supra note 16 at 234;'Rights and Judges (1984),' supra note 16 at 81. I address this point in note 36 infra. Secondly, whereas in Judges and Rights' Weiler proposes to grant the notwithstanding power to the Canadian Parliament but not to the provinces, by 1984 he seems to accept without any misgivings the fact that this power had been bestowed upon the provincial legislatures as well. See'Judges and Rights (1980)' at 234-5; 'Rights andjudges (1984) at 85-6. Since this work deals with separation of powers and not with the federal division of powers, I shall not elaborate on this point here.
24 'Rights andjudges (1984),' supra note 16 at 84.
228 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NO'lWITHSTANDING MECHANISM 229
as a check on judicial error.'251, label this approach 'the check on judicial error approach.
Weiler's analysis of the Charter focuses uniquely on s. 33. In contrast, the second approach to the NM, outlined by Slattery, is part: of a broader theory of the Charter. Slattery's theory, which he cails the 'Coordinate Model,' distinguishes between first- and second-order duties of different institutions under the Charter.27 A first-order duty is the Charter's direct imperative to a given institution to do something or to refrain from doing something. A second-order duty is the Charter's imperative to a given institution to review the action (or failure to act) of another institution while the latter is exercising its first-order responsibilities.28 Slattery's account of the NM is based on the distinction between first- and second-order duties:
25 F.L. Morton, 'The Political Impact of the Canadian Charter of Rights and Freedoms' (1987) 20 Can.J.PoS.Sci. 31 at 54 [hereinafter 'Political Impact']. The expression 'legislative review' in the quotation is taken from P. Russell, 'The Effect of a Charter of Rights on die Policy-Making Role of the Canadian Courts' (1982) 25 Can.Pub.Admin. 1 at 32 [hereinafter 'Effect of a Charter'],
26 Weiler's use of the term 'gone awry,' in the passage quoted in the text accompanying note 24 supra, is objective, implying that the NM is aimed at situations in which the court actually made a mistake, as opposed to situations in which the legislature disagrees with the court. Similarly, Weiler refers to situations in which 'the judiciary has miscarried' ('Rights and Judges (1984),' supra note 16 at83) and also suggests that the. 'assumption' thatjudges are 'infallible' is a 'fallacy' ('Judges and Rights,' supranote 16 at 222). Importantly, this objective conceptualization of error is not the only one Weiier mentions. Fie also discusses a situation in which the government is 'strongly persuaded that the Court [has] erred' (ibid, at 233-4), implying a legislative belief that the court has erred even where no objective error occurred. Beyond these two concepts of error, Weiler also talks of a 'disagreement' between the court and the legislature without mentioning the idea of error at all - either in terms of the objective description of the court's action or in terms of what the legislature thinks about the court's action. Thus, Weiler writes of situations its which the judicial 'say' differs from the legislative 'say': ibid, at 222. Similarly, he notes that the NM is there for legislators in cases when they have 'disagreed with the courts': 'Rights and judges (1984),' supra note 16at82. While Weiler does not elaborate on the exact concept of error he has in mind, it is clear that his work at least includes a reference to errors in an objective fashion, and therefore it is fair to talk about a 'check on judicial error approach.' In the works of some of those who followed Weiler's path, the language of objective error is even more clear. See Morton, 'Political Impact,' supra note 25 at 53; P.H. Russell, 'Standing Up for Notwithstanding' (1991) 29 Alta.L.Rev. 293 at 295 [hereinafter 'Standing. Up']; L.G. MacDonald, 'Promoting Social Equality through the Legislative Override' (1994) 4 N.J.C.L. 1 at 23 [hereinafter 'Social Equality']; CP. Manfredi, Judicial Power and the. Charter: Canada and the Paradox of Liberal Constitutionalism, 2d ed. (Toronto: Oxford University Press, 2001) at 170-1 [hereinafter Judicial Pouter].
27 Slattery, 'Theory of the Charter,' supra note 17 at 707.
28 Ibid, at 707-8.
[A] legislature is always under a first-order duty to comply with Charter standards in enacting a statute, even when the statute contains a notwithstanding clause. The effect of a valid notwithstanding clause is to curtail or eliminate judicial review, not to release a legislature from its constitutional responsibilities under the Charter. So when a government drafts a bill with a notwithstanding clause and a legislature considers it, they both have the duty to ensure that in their judgment the bill does not unjustifiably infringe any Charter rights, including those covered by the notwithstanding clause.'1''1
By examining Slattery's terminology, we can easily discern a clear disagreement with Weiler. For Weiler, the legislature's exercise of the NM represents a third-order duty: it is a legislative third-order review of the court's second-order review of the legislature's first-order duty to respect the Constitution when enacting legislation. In contrast, Slattery views the legislature's use of the NM as a first-order power aimed at immunizing legislation from the second-order judicial review power.
Weiler's story responds to two questions with one answer. First, what is the essence of the NM? Second, why have a NM? Weiler's answer is that the NM is a mechanism for correcting judicial errors. This is its essence, and the reason Canada has need of it. In contrast, Slattery answers only one of these questions: what is the essence of the NM? Slattery says that the NM is a mechanism through which the legislature can pre-emptively immunize an act from judicial review in cases where it is certain that the act is constitutional. But this does not explain the need for a NM. Why, given the fact that the Charter, as a general rule, envisions judicial review of the constitutionality of legislation, should the legislature have the power to prevent such review? It is possible to answer this question by viewing Slattery's approach as an extreme version of Weiler's. According to this understanding, the NM is aimed at situations in which the legislature is very interested in a certain act, is convinced of the act's constitutionality, and suspects that the court will mistakenly find the act unconstitutional and therefore strike it down. To avoid the cost and time of litigation, the legislature invokes the NM in advance, at the time of the initial enactment of the legislation. I shall name this approach 'the prevention of judicial error approach' and attribute it to Slattery, although he himself does not explain the exact meaning of his approach.30
29 Ibid, at 739-40 [original emphasis].
30 In a conversation with the author, Professor Peter Hogg has singled out a certain type of legislation that legislatures might be especially interested in immunizing from judicial review. For Hogg, where the court lacks expertise, such as in cases dealing with complicated social science issues, the court should not be given a chance to make a mistake and the legislature should not have to undergo the political hardship of invoking die NM in order to correct po ten tiaijudicia! errors. In such cases, Hoggsuggests, the legislature could legitimately use the NM even in advance of any conflicting judicial decision.
230 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 231
The third approach to s. 33 represents a significant departure from those of Weiler and Slattery. Notably, both Wciler and Slattery expect the legislature to engage in Charter interpretation, Weiler believes that when invoking the NM, the legislature replaces the court's erroneous constitutional interpretation with its own corrected version. Slattery states that, if the legislature enacts a notwithstanding act, it must first ascertain that the act does not violate the legislature's own understanding of the Charter. In contrast, Weinrib's justification for the NM is not grounded in constitutional interpretation at all. Rather, she views the NM as a means for creating constitutional 'exceptions' based on 'majoritarian or representational values.'31 In other words, a legislature needs a NM to somehow legitimate a violation of rights. The legislature does not override the judicial interpretation of the constitutional text (Weiler); nor does it assert its own interpretation of the Charter (Slattery). Simply stated, Weinrib's model holds that the legislature is aware of the fact that it is deviating from the Constitution by nullifying rights but is acting out of the belief that this is the right thing to do.
This idea of creating exceptions to rights leaves unanswered the question of why such exceptions cannot be considered by the courts as part of their rights and limits analysis. If a genuine need to violate rights exists, the court will justify the violation under the limitation clause. Weinrib's response is that judges are constrained:
The courts are nowhere empowered to go beyond the traditional bounds of the judicial function into the legislative waters of balancing or maximizing utility -that function is left to the legislatures, under s. 33.12
Here, the legislature will not simply revisit the impugned legislation using considerations identical to the courts' and correct the courts where they went wrong. Rather, it will, when necessary, bring political considerations into the equation, considerations that are beyond the court's purview. Such an arrangement, as Weinrib describes it, is one where 'courts can be courts and legislatures, legislatures.'33
All three accounts of the NM inevitably raise questions of constitutional theory. These questions will be addressed in the next part of this article. But first we will examine the fit between these three s. 33 theories and the text of the Charter.
B. FITTING S. 33
First, let us outline the various aspects of the section. Four features in particular are relevant to our discussion - two of them concern what is included in s. 33, while the other two concern what is absent.
31 Weinrib, 'Learning to Live,' supra note 13 at 568.
32 Ibid, at 567.
33 Ibid, at 568.
The first two features of s. 33 are the express declaration requirement and the sunset mechanism. These two features embody the intermediate constitutional status that s. 33 confers upon most Charter rights: on the one hand, it is possible to override a right and deviate from the Charter; on the other hand, the enactment of a notwithstanding act is more complex than the enactment of an ordinary act and has a different legal effect. The first feature, the express declaration requirement, demonstrates that the legislature is aware of its action. Moreover, it serves to inform the public about the deviation and thereby encourages public debate and perhaps public criticism.34 As a result, it is more difficult politically to pass a notwithstanding act than to pass an ordinary act. The second feature, the sunset mechanism, does not make it more difficult for the legislature to pass notwithstanding legislation, but it contributes to the intermediate constitutional status of Charter rights in a different way. Unlike a normal legislative provision, which remains in force as long as it is not repealed, a notwithstanding declaration expires after five years. Thus, a notwithstanding act may be struck down by the courts after this fixed period unless the notwithstanding declaration in the act. is renewed.
Moreover, the combination of the express declaration requirement and the sunset mechanism imposes a further check on the notwithstanding power, since, in the event that the legislature opts to renew the legislation, it must repeat the express declaration and thus confront a possible renewed wave of public outcry and political risk. The five-year limitation period also guarantees that an election will occur before the renewal of a notwithstanding declaration, so that the deviation from Charter protection may potentially become an election issue.3n
In addition to the express declaration requirement and the sunset mechanism, which are included in s. 33, there are two aspects of s. 33 that concern things to which it does not apply. The third feature of s. 33 is that it applies only to most Charter rights, not to all of them. Thus, s. 33 does not include a mechanism to override democratic rights (ss. 3-5), mobility rights (s. 6), or language rights (s. 16-23). Finally, the section makes no mention of courts or judicial decisions, but instead refers to the enactment of legislation.
All four features must be dealt with in any attempted theoretical explanation of s. 33. The following table illustrates the interaction between these features and the three NM theories:
34 See Weiler, 'Rights and Judges (1984),' supra note 16 at 81-2; M, Tusbnet, 'Policy Distortion and Democratic Debilitation: Comparative Illumination of the Countermajoritarian Difficulty' (1995) 94 Mich.L.Rev. 245 at 280 [hereinafter 'Policy Distortion'].
35 Weiler, 'Rights and judges (1984),' supra note 16 at 81-2. See also Manfredi, Jwlicial Power, supra note 26 at 193.
232 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 233
	'A Express declaration	B Sunset mechanism	C Some rights are excluded	D Courts are not mentioned
Correction of judicial error (Weiler)	Fit	No fit.	Fit	No fit
Prevention of judicial error {Slattery)	Fit	Some fit	Fit.	Fit
Exceptions to rights (Weinrib)	Fit	Fit	Fit	Fit
The three theories demonstrate similarities with respect to the features of express declaration and the protection of certain Charter rights from subordination under s. 33 (columns A and C). In the first instance (column A), whether the role of the NM is to correct judicial errors, to prevent such errors, or to create legislative exceptions to rights, the mechanism should not be invoked without adequate legislative and public scrutiny. To this end, the express declaration requirement singles out notwithstanding legislation and - although not guaranteeing such an occurrence — increases the likelihood of such scrutiny.
Column C demonstrates that all three theories also fit with the fact that not all Charter rights are susceptible to the NM. The explanation for this limitation might be the legislative bias associated with the Charter rights not affected by s. 33. These protected rights include democratic rights, mobility rights, and language rights. Given that the democratic rights deal directly with legislatures, an immediate apprehension of bias would arise if legislatures were permitted to violate them. Similarly, since mobility and language rights embody the essence of Canadian federalism, they too cannot be left for the provincial legislatures or federal Parliament, as these institutions are biased in terms of their own political agendas within the federalist structure. This is true regardless of whether the notwithstanding power is grounded in the correction of judicial errors (Weiler), the prevention of such errors (Slattery), or the need to provide for legislative exceptions to rights (Weinrib).
The two features that distinguish the theories in terms of explanatory power are the sunset mechanism and the absence in s. 33 of any mention of courts or judicial decisions. These two features of s. 33 cause Weiler's check on judicial error account to fail short.
First, the check on judicial error approach makes no provision for a sunset mechanism. According to Weiler, it is the NM that saves the system from judicial mistakes, But if this is the case, why should the legislative correction come with a fixed expiration date? Why permit the erroneous
judicial decision to regain effect if the legislature, for some reason, does not renew the notwithstanding declaration?3''
The fit of Weiler's approach is further attenuated when we consider that s. 33 is not limited to remedial operations and, in fact, does not mention the courts, a judgment, or a re-enactment at all. If the NM serves as a 'legislative review of judicial review,'" then the scope of s. 33 should have been limited to the re-enactment of laws that have been struck down. In other words, if s. 33 were about the correction of judicial errors, s. 33(1) would have included an opening statement such as 'After a court renders a final judicial decision declaring an Act of Parliament or of the legislature of a province to be of no force and effect1 before its current opening words, 'Parliament or the legislature of a province may expressly declare.,..'38 In response to this criticism, Weiler might invoke Donna Greschner and Ken Norman's argument that s. 33 should be read as requiring a prior judicial striking down of an act:
In order to give the 'notwithstanding' declaration any legal effect, the law must be in conflict with the Charter provision. If the law was not inconsistent with the
36 One might respond that since neither courts nor legislatures are infallible, the legislature must, re-examine every five years its view that the court erred. This response, however, is unsatisfactory. To explains. 33 in terms of the judicial error argument, one must assume that the relevant judicial decision was indeed wrong. Of course, it is possible to suggest that rather than correcting judicial mistakes, the NM is away for the legislature to disagree with the court, even if the court's interpretation'is not wrong. I develop this type of argument in Part III, s, C, below. The judicial error argument, however, does not choose this path.
Irs this regard, it is worth mentioning that Weiler's initial pie-Charter piece, in which he introduced the idea of a NM, makes no mention of a sunset mechanism. Instead, Weiler suggests that the use of the NM be subordinated to two enactments - one before and one after the election of the relevant legislative body ('Judges and Rights (1980),' supra note 16 at 239), After a notwithstanding act is enacted through such a process, it does not expire and remains in force until cancelled by the legislature. Such an approach indeed fits the judicial error argument, since it requires a higher degree of legislative and public conviction that the court indeed erred in its decision to strike down the impugned legislation, With a legislative determination of judicial mistake established, there is no sense in keeping the erroneousjudicial decision in force. Thus Weiler's check onjudicia! error approach fits his proposed NM with the two enactments requirement. It does not however, fit the design of s. 33, which lacks a dual enactment requirement but instead includes a sunset mechanism.
37 Russell, 'Effect of a Charter,' supra note 25.
38 The federal government, indeed, proposed amendments to the text of s, 33 in. this direction. See Canada, Shaping Canada's Future. Together: Proposals (Ottawa: Supply and Sendees Canada, 1991) at 4. In the same vein see Lougheed, 'Why a Notwithstanding Clause?' supra note 22 at 17-8, and Manfredi, Judicial Power, supra note 26. This issue, however, was notadvanced in the Charlottetown conferences, Canada, Consensus Refiort On the Constitution Charlottetown, 28 August, 1992:FmalText (Ottawa, 1992) at 20. As long as this amendment remains unadopted, it is clear that.the current text of s. 33 does not Fit the check on judicial error approach.
234 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 235
Charter right or freedom, the notwithstanding declaration would be unnecessary, a meaningless appendage. In other words, when a legislative assembly says dial a law operates in spite of a Charter provision, there must be a conflict between the law and the provision if the statement is to have any meaning. The institution charged with rendering an authoritative ruling with respect to incompatibility, before s. 33 is invoked, is the court. Hence the override power is contingent on a prior judicial determination that a law violates a Charter right or freedom.3"
This, however, is a weak defence for several reasons. First, the word 'notwithstanding' in s. 33 is not descriptive, suggesting that acts which include notwithstanding declarations do operate notwithstanding a Charter provision. Rather, the word 'notwithstanding' is prescriptive, instructing that the act 'shall operate notwithstanding' the provision, regardless of whether that provision commands that the act is invalid. Second, s. 33 does not say that the notwithstanding act operates notwithstanding an inconsistency between it and the Charter provision; rather, it operates notwithstanding the provision itself, whether such an inconsistency exists or not. Therefore, it is inaccurate to claim that without a prior judicial decision the word 'notwithstanding' in a notwithstanding declaration is 'meaningless.' Such a declaration shields the act from any prospective judicial scrutiny. Since litigation can take a heavy toll on political capital and time, invoking a pre-emptive notwithstanding declaration to avoid judicial entanglements represents a significant gain that ought not to be dismissed as 'unnecessary.'
Furthermore, Greschner's and Norman's textual argument not only fails to support Weiler's judicial error argument but actually provides the best textual argument against it. Greschner and Norman argue that in order for the word 'notwithstanding' to be meaningful, notwithstanding acts must actually violate the Charter. However, according to the judicial error argument, notwithstanding acts never contradict the Charter, since the raison d'etre of the NM is to enable re-enactment of constitutional acts that were mistakenly struck down by the court. In other words, Greschner and Norman's textual argument and the judicial error argument make contradictory stipulations for the use of the nm. Greschner and Norman's argument stipulates that the nm is invoked only if a notwithstanding act. indeed violates the Charter, Weiler's judicial error argument stipulates that the nm is invoked only if the court was wrong in striking down an act, that is, only when a notwithstanding act does not violate the Charter.
39 D. Greschner & K. Norman, "The Courts and Section 33' (1987) 12 Queen's-L.J. 155 ■ at 188 [hereinafter 'Courts and a. 33']. Note that Greschner and Norman admit that this textual basis for requiring a prior judicial decision to invoke the NM is not. 'the strongest basis' (at 189).
Finally, even if we assume that the term 'notwithstanding' is more meaningful if the nm is limited to post-judicial decisions, the problem of fit that the Greschner-Norrnan argument creates is, in my view, greater than the fit advantage their approach generates. The advantage is that the term 'notwithstanding' might be more meaningful, if we required a judicial decision prior to its use. However, the disadvantage stemming from the fact that the court is not mentioned in s. 33 is more significant. The. argument that s. 33 requires a prior judicial decision can be sustained only by reading into the section an institution (the court) and a process (prior judicial decision) neither of which are mentioned in the text. Such a reading hardly fits the section. Notably, the Charter does refer to both the court and judicial decisions in s. 24, suggesting that Charter claims can be petitioned to 'a court of competent jurisdiction' which is in turn authorized to provide 'such remedy as the court considers appropriate.' If the purpose of s. 33 was to create a means of overriding the product of judicial review, s. 33 would have been either included in s. 24 or classified as an independent section under the 'Enforcement' heading of the Charter. Instead, it is an independent section located in the 'Application' part of the Charter (ss. 32-33). This indicates that s. 33 should be seen as operating in relation to the Charter itself, not. in relation to its enforcement by the judiciary.
Slattery's approach fits s. 33 better. Since he does not believe that the NM should be invoked only subsequent to a judicial decision,40 his theory accounts for the fact that neither the court nor judicial decisions are mentioned in s. 33. However, his theory is unable to account fully for the sunset, mechanism. According to Slattery, a decision to invoke the NM implies that the legislature is convinced that a certain act is constitutional, If this is the case, why would the legislature need to reconsider its finding of constitutionality every five years? Admittedly, interpretation is subject to change over time. This.is especially true of constitutional interpretation, given its heavy reliance upon social science evidence, which is iri a state of constant transformation. Yet despite this ready answer, Slattery's account only partially fits the sunset, mechanism. As noted, Slattery reasons that the legislature acts as a 'High Court of Parliament' when it invokes the NM.43 However, engaging in a process of periodic review is an activity one associates with legislatures, not with
40 Slattery does not discuss the question of pre-emptive and remedial uses of the NM, leaving it to future 'full and serious consideration.' He does imply, however, that the purpose of the NM is to 'protect {an act] from judicial scrutiny' and 'curtail or eliminate judicial review' and thereby makes it clear that, in his opinion, the NM can be used pre-emptively. 'Theory of the Charter,' supra note 17 at 742, 741, and 739 respectively.
41 See Slattery, 'Theory of the Charter,' supra note 17 at 743.
236 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 237
courts. Therefore, while the revisiting of notwithstanding acts makes sense for a legislature, it remains incompatible with the idea of the legislature as a 'super-court.'
Of the three theories examined here, Weinrib's approach fits the text of s. 33 best. Weinrib argues that the NM is aimed at creating legislative exceptions to rights. Consequently, both the sunset mechanism and the lack of any reference to courts in s. 33 are appropriate. The need for an exception has nothing to do with Charter interpretation. Rather, it has to do with representational or majoritarian considerations, which are external to the Charter and are based solely on shifting societal preferences, appropriately invoked by legislatures - not courts. Periodic revisions of legislation reflect an integral part of the legislative process. Thus, unlike Weiler and Slattery, Weinrib sees the invocation of the NM not as a quasi-judicial activity but., rather, as a purely legislative one. She realizes that if s. 33 focuses on legislatures and not on courts, its story must be one about legislatures, not courts.4"
The need for an account of s. 33 that focuses on the legislature and not on the court is the crux of the analysis undertaken in this section. Weiler's account fails to fits. 33 because he connects the notwithstanding power to judicial decisions. By suggesting that such a judicial decision precedes the use of a NM, Weiler turns the legislature into a veritable court of appeal, a 'super-court.' In contrast, Weinrib's approach fits s. 33 because exceptions to rights are disassociated from the courts. Slattery's approach partially fits with s. 33, since his view falls between Weiler's and Weinrib's on the court-legislature spectrum. Like Weiler, Slattery believes that while invoking the NM, the legislature acts as a 'High Court.'43 Yet he concludes, as Weinrib does, that exercise of the notwithstanding power need not occur only in response to a judicial decision.
42 Indeed, Weinrib's approach best fits s. 33 also in terms of the title of s. 33, which reads 'Exception where express declaration.' The point is not that Weinrib uses the same terminalogyas that of the title but that Weiler's and Slattery's approaches cannot explain the merits of this title, The idea of an exception to a constitutional right implies that the invocation of s. 33 results in a deviation from rights protection. However, according to Weiler and Slattery, this is not. the case. Weiler believes that (.he NM corrects a judicial error, such that the NM is not an exception to the Charter but, rather, its correct application. Similarly, Slattery believes that the legislature can iiwoke the NM only if it is convinced that the act in question is constitutional and therefore not an exception to rights protection. Weinrib, in contrast, believes that the invocation of the NM is based not on legislative interpretation of the constitution but on legislative interest in injecting majoritarian and representational considerations into die constitutional world. Since such considerations are antithetical to rights protection, they are appropriately labelled by s. 33's title as 'exceptions.'
43 See text accompanying note 4.1 supra.
What Weiler, Slattery, and Weinrib have in common is the argument that the NM can be explained in terms of constitutionalism; but all three approaches suggest modifications to traditional constitutionalism. Weiler envisages a system of judicial review and legislative finality, as opposed to the traditional model of final judicial review. Slattery supports the legislative power to exempt acts from judicial review, thus contradicting the courts' traditional power to scrutinize impugned legislation. Finally, Weinrib's vision of a system where legislatures sometimes act in deviation from the constitution stands in opposition to the traditional model of strict adherence to the constitution. As a result, all three authors describe new divisions of labour between courts and legislatures.
All three versions of this new institutional paradigm raise serious challenges to our understanding of constitutionalism. Weiler's vision of legislatures overriding judicial errors and Slattery's model of legislatures predicting such judicial mistakes raise the following question: Does the idea of the NM as a tool to correct or prevent judicial errors not contradict the very foundation of judicial review, which is that the court is the most appropriate institution for constitutional interpretation and for rights protection? Weinrib's approach does not encounter this problem, since she reasons that the legislature's only function under the NM is to create legislative exceptions to rights. Still, her approach raises a different, question and therefore also requires further clarification: Why should exceptions to rights be part of the Charter when the very basis for a Charter of Rights is to avoid such exceptions altogether? These questions are not limited to the text of s. 33 and, in fact, are not unique to the Canadian context. The issues they raise point at more general and universal themes that underpin ideas of judicial review and a legislative notwithstanding power. It is to these theoretical questions that we now turn.
Ill Constitutional theory: Understanding the notwithstanding mechanism
A. A CRITIQUE OF SLATTERY'S AND WEINRIB'S APPROACHES To understand how the approaches of Weiler, Slattery, and Weinrib explain the NM in light of principles of constitutionalism, consider two questions. The first is process centred and concerns itself with the degree to which legislative displacement of judicial review is permissible. Specifically, this question asks whether the NM can be invoked only in response to judicial review or whether it can be used to avoid judicial review altogether. The second examines whether the legislature must believe it is acting in accordance with the constitution when invoking the NM or whether it is at liberty to consider extra-constitutional considerations. This is a substantive question, as it pertains to the values underpinning-invocations of the NM. The table below provides the answers that each of the three approaches to s. 33 would give to these two questions:
238 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 239
	Can the legislature avoid judicial review?	Can the legislature consider extra-constitutional considerations while invoking the NM?
Check on judicial error		
(Weiler)	No	No
Prevention of judicial error		
(Slattery)	Yes	No
Legislative exceptions to		
rights (Weinrib)	Yes	Yes
From the table, it is easy to see that it is Weiler's approach that deviates least from traditional constitutional rights protection. Since he believes that the role of the NM is to correct judicial mistakes, he does not allow for the legislature's use of the NM until the judiciary has fulfilled its constitutionally mandated task of judicial review. Similarly, Weiler's judicial error argument mandates that the legislature act only to correct a judicial error in Charter interpretation, as opposed to re-enacting an act that the legislature knows is unconstitutional. If the legislature is not convinced of an act's constitutionality, then it cannot use the NM, regardless of the act's merit as a legislative endeavour. Conversely, Weinrib's approach is the farthest from traditional constitutional rights protection. Since she believes that the role of the NM is to create exceptions to rights independently of judicial decisions, there is no need to wait for a court ruling before invoking the NM,44 Moreover, the very essence of the NM is an authorization for the legislature to create exceptions to rights based on 'majoritarian or representational values'43 - both non-normative considerations. Finally, Slattery's prevention of judicial error approach reflects an intermediate position. On the one hand, since the NM is
44 Weinrib herself does not make Lhis argument explicitly but only hints at it. She argues that a legislature that invokes the NM should do so in a specific fashion, as opposed to the omnibus manner demonstrated by Quebec in their use of An Act Respecting the Constitution Ad, 1982 (supra note 13). In the Ford case (supra note 13), the Supreme Court of Canada did not. require such specificity and upheld Quebec's omnibus use of the NM. Criticizing that decision, Weinrib writes, 'It is true that by requiring specificity the Court might eventually find itself called upon to enunciate the requisite standard of specificity. Would reference to a specific right be enough? Would there be an evidentiary requirement (for example, an opinion, a court decision, a legislative debate) demonstrating that tike legislature had considered the Charter impact of the policy?' 'Learning to Live,' supra note 13 at 555. Weinrib does not answer these questions. She merely states that 'wherever one might draw tile line on specificity, the comprehensive override policy would clearly fall on the impermissible side' (ibid.). In other words, Weinrib offers the requirement of judicial decision prior to the use of the NM not as mandatory but merely as optional, alongside other means that could demonstrate that the legislature addressed Charier issues specifically.
45 Ibid, at 568.
aimed at preventing or curtailing judicial review, the legislature is not required to wait for a judicial decision before approving its use. On the other hand, the invocation of the NM is based on legislative pre-ruling about the meaning of the constitution, and non-constitutional considerations are excluded.4(1
Because of their significant deviation from the principles of constitutionalism, neither Slattery's nor Weinrib's approach survives critical examination. First, both approaches condone the legislature's preemptive use of the NM to avoid judicial review altogether. Although neither Slattery nor Weinrib provides an explicit justification for this preemptive power,4' given their accounts of the NM, it is possible to infer their respective rationales for it. Slattery trusts the legislature to make its own pre-ruling about the constitutionality of any given act; therefore, awaiting the process of judicial review represents a waste of time and money. According to Weinrib, the NM is aimed at creating legislative exceptions to rights. Since these exceptions are disassociated from judicial decisions, the legislature need not wait for the court's ruling before invoking the nm.
However, allowing the legislature to immunize its legislation from judicial scrutiny amounts to giving it absolute legislative power, effectively silencing individuals who are injured by the legislation.48 It obstructs the opportunity to air grievances and bring forth arguments that the legislation is antithetical to rights protection. It creates a risk that the use of the NM will go unnoticed.4''*
46 Tellingly, while Weinrib's approach bestfits s. 33, it is furthest removed from traditional constitutionalism, and while Weiler's approach least fits s. 33, it remains closest to traditional constitutionalism. This discrepancy reflects the fact thats, 33 was devised as a compromise between no rights protection and full rights protection. Therefore, a theory that best explains the section (Weinrib) is destined to be problematic in terms of rights protection, while a theory that accounts for rights protection (Weiler) cannot fit the text of the section.
47 See supra notes 40, 44.
48 Manfredi seems to be making the same argument using different language. He opposes pre-emptive use of the NM because' [t]he doctrine of constitutional supremacy includes an important review function for the courts.' Judicial Power, supra note 26 at 192. However, Manfredi's emphasis on the court's 'important function' misses the point. The fact that a few individual laws invoke the NM.pre-emptively and thus prevent, judicial review does nothing to the court's 'important review function,' since, in the vast majority of cases, the NM is not invoked and the court may perform a review. In contrast, if we look at the matter from the individual's perspective, pre-emptive uses of the NM block access to the courts and the individual therefore loses the chance to air her grievances: it does not matter that in most other cases individuals have access to justice.
49 Indeed, most of the uses of the NM have been ignored by the public and the media. In a previous work, I argued that this was partly because these uses were pre-emptive. Kahana, 'Notwithstanding Mechanism,' supra note 13 at 272-80.
240 university of toronto law journal
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 241
Moreover, expanding the exercise of the NM to pre-emptive situations will probably result in an increase in its use. It is reasonable to assume that in certain cases the legislature will fail to predict accurately whether or not the court will strike down a particular act. To remain on the safe side, the legislature will therefore enact the law with a notwithstanding declaration prior to any judicial review. By taking this pre-emptive step, the legislature ensures that the possibility that the court would have upheld the act absent the declaration goes untested. In other words, under this scenario, the legislature unnecessarily increases the use of the NM.:,<>
Weinrib's approach is incompatible with rights protection for an additional reason. Unlike Weiler and Slattery, Weinrib believes that the invocation of the NM has nothing to do with legislative interpretation of the constitution. She holds that the NM is directed at enabling legislatures to create 'exceptions' to rights where social interests necessitate their displacement."1 These exceptions stem from two categories: 'ma-joritarian or representational values'52 and 'balancing or maximizing utility.,ss Weinrib argues that since such considerations should never be entertained by court's, it is the legislature that should consider them and effectuate them through the notwithstanding power.54
However, majoritarian and utilitarian considerations collide directly with the idea of constitutional rights protection, the precise aim of which
50 A more problematic argument that ties limiting the NM power to remedial situations to decreasing its usage focuses on accountability. This argument suggests that the legislature would exercise greater caution using the NM remedially rather than preemptively because, while a pre-emptive use of the NM would be perceived by the public as an action only against the constitution, remedial uses might be perceived as actions against both the constitution and the courts. This argument is problematic, since use of the NM to counter a court's decision does not guarantee that the legislature will be perceived as acting against both the constitution and the court. Rather, it might, very well be perceived as acting against the court but not against the constitution. Such a perception might arise because where the legislature uses the NM in the wake of a judicial decision, it. is saying 'we override the court's interpretation of constitutional rights' and not 'we override the constitution.' In other words, it might actually be easier, politically, to use the NM remedially than to use pre-emptively. This argument is even stronger when the act has been struck down by a majority only and not by a unanimous court. In such a case, the judicial process actually provides the legislature with an explicit stamp of legitimacy that it would lack in the case of a pre-emptive exercise of the NM. In such circumstances, the legislature could say, 'while some judges think the law is unconstitutional, others agree with us that it is constitutional. As a result, this use of the NM is byno means unacceptable.' The legislature could even invoke the reasons of the dissenting judges who found the act to be constitutional. Obviously, this argument only gains strength when the dissenting judge or judges are sitting on the bench of the Supreme Court.
51 See text accompanying notes 31-33 supra.
52 Weinrib,'Learning to Live,'supra note 13 at 568.
53 Ibid, at 567.
54 See text accompanying notes 32-33 supra.
is to secure rights beyond the reach of majorities and their representatives. They should not be part of a rights protection system, nor should they be weighed by either legislatures or courts. The goal of such a system would be defeated by allowing majorities to reassert their preferences or, worse, by immunizing legislation from judicial scrutiny on the basis that the legislation is sustained by majoritarian or utilitarian considerations.
As an alternative to seeing Weinrib's account of the nm as a failed attempt to conform with constitutionalism, we can interpret her defence of exceptions as an attempt to 'learn to live with the override.'"'" From this perspective, while a Charter without a nm is preferable, a Charter with a nm has the advantage of reducing the need.for judicial consideration of, and deference to, majoritarian preference. Although a court operating properly should not concern itself with majoritarian preferences, the nm acknowledges that not all courts operate properly. Therefore, it suggests to courts that concern about utilitarian and majoritarian preferences can be safely excluded from the judicial process, since the majority can ultimately exercise its will by using the nm, regardless of the judicial decision reached.
This argument may be taken one step further to argue that a Charter with a nm is actually better than a Charter without one. The argument in the previous paragraph assumes that the existence of the nm neither increases nor decreases the prevalence of majoritarian values in the system - it. simply shifts the task of determining and applying them from the courts to the legislatures. However, it is possible to argue that the nm could decrease the influence of majoritarian considerations. Arguably, in an effort to be responsive to the majority, judges might become overly cautious and defer to majoritarian preferences too often. In contrast, if judges knew that the will of the majority could be expressed byway of the nm, they would never need to reflect upon such considerations. Since the majority tends to express its preferences through the mechanism only rarefy, the number of cases in which majoritarian considerations appear would decrease.
There are two counter-arguments to this reasoning. First, it implies that the legislature will use the NM to accommodate majoritarian and utilitarian considerations in fewer cases than would a court under a Charter without a nm. While this assumption is reasonable, so too is its opposite. Furthermore, with the nm in place, there is an ancillary danger that the court may perceive the legislature's use of the nm as injurious to the court's credibility. As a consequence, it might seek to prevent an invocation of the nm by appeasing the legislature and including majoritarian and utilitarian considerations injudicial decisions in more cases than it would have in the absence of a nm. Given this possibility, it is feasible
55 As the title of her paper, 'Learning to Live with the Override' (supra note 13), implies.
242 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 243
that majoritarian concerns might occur more frequently under a NM bearing Charter than the optimistic scenario suggests.
Even if we do accept that the existence of the NM is likely to decrease the overall presence of non-normative considerations in the system, the argument in favour of a NM based on the notion of exceptions remains inconsistent with constitutionalism. What this argument suggests is that it might be better for a constitution to accommodate some illegitimate rights infringements in order to decrease the overall prevalence of such rights infringements in political decision making. Clearly, this type of pragmatic argument is more strongly pro-NM than the argument according to which the NM represents a sound compromise between no Charter and a Charter that gives final power to the judiciary,51' for it suggests that a nation ought to prefer, even as a starting point, a constitution with an NM, since it would contribute to a decrease in illegitimate rights infringements within the system. However, the fact that the pragmatic argument may be a stronger one does not repair its theoretical flaws: it still accepts the proposition that an invocation of the NM may he at odds with the goal of rights protection.
Weiler's approach has more potential for compatibility with constitutionalism than Slattery's or Weinrib's. Only Weiler's model rejects the consideration of non-normative factors and prohibits silencing the court or obstructing individuals from airing their constitutional claims. Yet, for all this potential, W'eiler's approach still suffers from theoretical inconsistency, as we shall now see.
B. A CRITIQUE OF WEILER'S APPROACH
Weiler's support of both constitutionalism and judicial review is very clear. His belief injudicial review is grounded in the classical justification that its role is to provide constitutional rights protection. Simply put, this argument suggests that courts are better motivated and better able to protect rights than are legislatures. They are better motivated because they are not accountable to the majority.57 They are better able to deal with matters of principle because the judicial process revolves around parties, arguments, concrete disputes, and reasoning."8 Weiler's reservations are limited to the notion of judicial finality:
The argument [in favour of judicial review] presumes that when judges say what our fundamental rights arc, they are right about that. This is a natural assumption to make, especially for the legal mind ... That is a fallacy nonetheless,
56 This argument underlies Weiler's theory. See note 70 infra and accompanying text.
57 'Judges and Rights (1980),' supra note 16 at.214-5; 'Rights and Judges (1984),' supra note 16 at 68-70.
58 Weiler, Judges and Rights (1980),' supra note 16 at 222; 'Rights and judges (1984); supra note 16 at. 71-2.
nowhere better expressed than by Mr. Justice Jackson of the United States Supreme Court, when he. confessed that 'we are not final because we are infallible, we are only infallible because we are final!>K)
Since, contrary to Mr. Justice Jackson's statements, judges are indeed fallible, the final constitutional word should be legislative.
The First problem with Weiler's judicial error argument is its contention that judicial finality is based on the 'assumption' that judges are 'right' about fundamental rights. In fact, judicial finality is not based on such an assumption, and by suggesting this connection, Weiler actually repeats Jackson J.'s flawed reasoning. The correct understanding of judicial finality is neither that courts are final because they are infallible, nor that they are infallible because they are final, but, instead, that they are final even though they are fallible. The justification for judicial finality is based on the argument that, because of their ability and motivation, courts are more capable than legislatures of protecting rights well, even if they sometimes make mistakes. Thus, a system of traditional constitutional rights protection assigns to courts the final power over fundamental rights with the knowledge that the courts may be wrong.
Now, if we believe - as Weiler does - that the court is the institution best situated to deal with rights issues, how can one trust the legislature to correct the court's mistakes? The problem with this scenario is not in detecting the error: a student can often find a mistake in the way her professor marked her exam, despite the fact that the professor's mastery of the material is undoubtedly superior to her own. The problem with using the judicial error argument to promote legislative finality is that this argument assigns to the legislature supervisory power over the judiciary in all cases because in some cases judges make mistakes. This is like granting students the power to change all marks given by professors because, in some cases, professors make mistakes in marking.1'0
59 'judges and Rights (1980),' supra note 16 at 222.
60 There is, in effect, a much stronger way to state this argument. Instead of comparing the judiciary-legislature situation to the professor-student scenario, we can compare it to two Htigants in a private case. Should the party that loses a case be granted the power Jo override the court's decision because judges sometimes make mistakes? The answer is obviously no. Likewise, it seems illogical to grant to the legislature the power to override a court's decision in a case where it represents the losing side, This way of thinking about the judicial error argument is actually more accurate than the student-professor analogy, since the legislature that can override the judicial decision has, unlike the student, lost its case to an individual rights holder. The reason I still prefer the student-professor analogy is that the judicial error argument could deny the fact that constitutional litigation is similar to private litigation in this regard. With respect to the judicial error argument, there is, according to this approach, not only a rivalry between the two parties to the case but, unlike in private litigation, a power struggle between the court and one of the parties, namely the legislature. The student-professor analogy is nevertheless appropriate because advocates of the judicial error argument
244 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 245
More importantly, in speaking of a scenario in which judicial error is corrected by the legislature, Weiler creates a system of'review that starts with the lower courts and ends with the legislature, just as an upper court might find a lower court's ruling wrong and overrule its decision, so might the legislature find a decision of the Supreme Court to be wrong and override it. The legislative action (i.e., the act) becomes the resting place of judicial decisions. Although Weiler is not as explicit about this point as is Slattery, who expressly suggests that in invoking the NM the legislature 'convert.[s] itself to a High Court of Parliament,'01 the judicial error argument effectively converts the legislature into a 'super-court.'1'2 Taking this conversion seriously entails four curious results that Weiler never discusses and I suspect would not accept. First, it means that legislatures could decide cases without parties, without any judicial process, and without reasoning. This result would be intolerable, especially in criminal cases. Second, the legislative 'judgment' would apply retroactively to the parties before the court. For example, if the court struck down an act imposing a form of criminal liability, re-enacting this act as a notwithstanding act could send the acquitted accused to jail. Third, as is the case with appeals, the NM would operate only within a prescribed period of time following the striking down of an act by a final judicial decision. In order to ensure stability, after this prescribed period elapses, the legislature would no longer be empowered to use the NM. Fourth, future adjudication would have to consider the legislative judgment' as precedent, at least to some degree, in cases similar to the one-struck down by the court and 'upheld' by the legislature.
Weiler could respond that the 'judgment' he is talking about is a different kind of judgment.' Since it does not apply to the parties and is not retroactive, there is no need for parties, process, or reasoning. Similarly, given that it does not apply to specific parties, the 'judgment' can be rendered whenever the legislature desires, and not only within a prescribed period following the final judicial decision. Finally, this different kind of judgment' is also limited to the specific act struck down by the court, and implies that future adjudication will not have to take it into account. This explanation effectively strips the notwithstanding act of its judicial' nature and reassigns it a legislative one. Therefore, it is more accurate to say that the legislature does not remedy a judicial error but, rather, re-enacts an act despite its nullification by the court. Consequently, calling such an action a judicial error is inaccurate.>A
concede that it is appropriate that the court supervise tile legislature, even though in some cases the court is less objective an arbitrator than it is in private litigation.
61 Slattery, 'Theory of the Charter,' supra note 1.7 at 742.
62 Weinrib, 'Learning to Live,* supra note 13.
63 As part of their support for the NM, Russell and MacDonald, two of Weiler's followers, make a special effort to point out judicial errors in several cases. See Russell, 'Standing
Weiler's second argument in support of the NM addresses not what is wrong with courts but what is right with legislatures. The obvious fear generated by the NM is that the legislature will use it not. to correct judicial errors in rights protection but out of a disregard for rights. Weiler deals with this spectre by suggesting that legislatures are trustworthy and not prone to abusing the NM. This 'no abuse argument' is composed of two sub-arguments, the first involving reliability and the second highlighting accountability. Regarding reliability, Weiler reasons that Canadian legislatures are not rights violators. Support for this argument comes from Weiler's conclusion that rights were not less protected in Canada than in the United States prior to 1982.1,4 Although he concedes that Canada has a record of rights violations,1''' he goes on to say,
The Parliament of Canada can, and has, done injustice to individuals and minority groups. So has the Supreme Court of Canada. We must eventually choose between these only too frail human institutions, locating the final authority in that forum where we anticipate it will do the most good and the least damage.
In addition to reliability, Weiler suggests that accountability can prevent abuses of the NM, since a government using the NM would have to 'be prepared to take the flak' from the people.1'7 Mark Tushnet adds that
Up,' supra note 26 at302-3,307; MacDonald, 'SocialEquality,' supra note 26 at 15-26. Elsewhere, I show that their attempts to establish judicial error were unsuccessful and that, in reality, the source of these 'errors' was Russell and MacDonald's own disagreement with the courts. See Kahana, Partnership Model, supra note 14 at 235-40, 244-52.
hi addition to the judicial error argument, Weiler introduces two related arguments in support of legislative finality. First, he recalls that the Charter's ambiguous language has resulted in questions in Charter interpretation 'rarely [having] clear-cut answers': 'Rights and judges (1984),' supra note 16 at 54. Second, he argues that Charter adjudication requires social science expertise and that the court lacks an advantage over the legislature in this area. See 'judges and Rights (1980),' supra note 16 at. 223-4; 'Rights andjudges (1984),' supra note 16 at 74-7. For this argument see also Manfredi, Judicial Power, supra note 26 at ch. 5, These two arguments are not conceptually different from the judicial error argument; rather, they represent more specific versions of it. The argument regarding the ambiguity of Charter provisions is that the Charter text is more difficult to interpret, as a result of which the chance of judicial error increases. (The argument is not that Charter questions do not have correct answers at ail; if that were the case, there would be no point injudicial review.) The argument concemingsocial science is that the court's institutional deficiencies make it susceptible to mistakes when dealing with policy issues. Since these arguments are ultimately about judicial error, my analysis of the judicial error argument is applicable to thorn as well. See Rah ana, Partnership Model, supra note 14 at 65-7, .93-106.
64 'Rights andjudges (1984),' supra note 16 at 53.
65 'Judges and Rights (1980),' supra note 16 at 208. See also 'Rights andjudges (1984),' supra note 16 at 68.
66 'judges and Rights (1980),' supra note 16 at 235.
67 'judges and Rights (1980),' supra note 16 at 234.
246 UNIVERSITY OF TORONTO LAW JOURNAL
the sunset mechanism makes it possible for the public to choose between the court's interpretation of the Charter and that adopted by the legislature.68
Weiler's prediction has clearly proven to be correct. In the two decades since the Charter was adopted, there has not been a single case where the legislature used the NM to legitimize oppressive or tyrannical legislation,09 But neither the judicial error argument nor the no abuse argument renders Weiler's account theoretically consistent, since both arguments are at odds with the very justification Weiler introduces for judicial review. The judicial error argument dictates that although judicial review is based on the assumption that the court is better suited for rights protection than the legislature, it is sometimes prone to error. The no abuse argument suggests that although judicial review is based on a reluctance to place confidence in the legislature, sometimes the legislature is more trustworthy than the court. On this account, legislative finality constitutes an obtrusive exception to the principle of judicial review, since it is at odds with the idea that underlies this principle. It is nevertheless adopted because of the belief that a compromise between no judicial review and final judicial review is the best pragmatic arrangement for rights protection. I call this justification for the NM an 'inconsistent justification,'
An inconsistent justification of the NM is neither innovative nor difficult to devise. The two extremes that such an inconsistent justification seeks to balance, namely final judicial review and the absence of judicial review, have been extensively discussed. Both extremes have their respective advantages and disadvantages. One merely needs to point to the pros and cons of judicial finality, as compared with the pros and cons of the absence of judicial review, and argue that the middle ground of judicial review with legislative finality strikes a better balance between advantages and disadvantages.
68 Tushnet, 'Policy Distortion,' supra note 34 at 280.
69 Even the one use of the NM that provoked public outrage - namely, the Quebec sign law - did not involve an attack on the core of a Charter right. This NM invocation was a response to the Ford decision (supra note 13), where the Supreme Court ruled that a provision forcing the use of solely (as opposed to mainly) French signs unjustifiably infringed freedom of expression. Using s. 33, the Quebec National Assembly re-enacted the French-only stipulation for exterior signs and allowed interior bilingual signs: An Act to Amend the Charter of the French Language, S.Q.. 1988, c. 54, s. 10. Since the sign law dealt with commercial speech and not political speech, it could hardly be said that overriding it amounted to tyranny. Another allegedly oppressive use of the NM, namely Alberta's proposed bill to cap the compensation that could be awarded to victims of forced sterilization, provoked so much public anger that the Alberta government had to withdraw it from the legislature. See Kahana, 'Notwithstanding Mechanism,' supra note 13 at. 271-2.
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 247
This is what Weiler is ultimately doing.70 Another way to suggest an inconsistent justification for the NM is by examining results: one could demonstrate that a system of judicial review with legislative finality produces better rights protection than either a system with final judicial review or a system with no judicial review. From the perspective of an inconsistent justification for the NM, this mechanism is not an innovative project in terms of theory, although a new balance or compromise between two existing ideas is undoubtedly important.71
By way of contrast, a consistent justification for the NM would have to unify the notions of judicial review and legislative finality. This would require finding a justification for judicial review that does not. justify judicial finality, even at the starting point, thus eliminating the tension between judicial review and legislative finality. I attempt to devise such a justification in the next section.
70 It is not clear whether Weiler himself saw his account as an exercise in constitutional theory. On the one hand, he writes that the dilemma of the NM cannot be solved by 'a priori reasoning about rights and democracy' but must be solved by 'practical judgment about the relative competence of two imperfect institutions in the context of a particular nation': 'Rights and Judges (1984),' supra note 16 at 83-4. On the other hand, when he makes the same point two pages later, he says that the merits of the NM 'are appraised here from the point of view of constitutional theory': ibid, at 86.
71 Another indication that Weiler's account is inconsistent with constitutionalism is demonstrated by his suggestion that invoking the NM can be, in effect, an easier way to amend the constitution. One of the traditional features of constitutional rights protection is that the amendment process for altering the protection of rights is significantly more difficult than the amendment of ordinary legislation. Having this fact in mind, Weiler says,
It is rare to find constitutional guarantees, no matter how inalienable they may seem, which are entirely beyond legal amendment, The question is where we should locate the authority to approve such an amendment, I am prepared to trust Parliament with that role, at least as long as it does so in accordance with a procedure which clearly focuses political responsibility for such action, ('Judgesand Rights (1980),' supranote 16 at 234)
Weiler's language of iegal amendment' is very telling. This term distinguishes between a constitutional amendment, which is the common tool for altering constitutional guarantees, and a 'legal amendment,' created by the NM, which is 'more of a surgical instrument' ('Right and Judges (1984),' supranote 16at83note 101) butstill operates to change the guarantees. If these guarantees can be changed by the legislature, then the Charter is not a constitutional document. It is indeed a compromise between having no rights protecting documents and having a constitutional document.
The exact relationship between the NM and the constitutional amendment mechanism deserves separate treatment. It should be noted that this easy amendment idea is not at all a necessary part of Weiler's judicial error account. In fact, it is inconsistent with that argument. The judicial error argument holds that, while invoking the NM, the legislature is asserting its own reading of the constitution in the face of the court's faulty reading. The idea of legal amendment implies that when the legislature invokes the NM, it is not offering a reading of the constitution but is altering the actual text.
248 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 249
C. DELIBERATIVE DISAGREEMENT: TOWARDS A CONSISTENT ACCOUNT OF JUDICIAL REVIEW AND LEGISIjVTIVE FINALITY
The central purpose of my approach to the NM is a desire to have justifications for judicial review and legislative finality unified under one umbrella argument.72 This argument, reasons that the role of judicial review is not to check legislative action but to deliberate on the meaning of the Charter. I call this approach, the 'deliberative disagreement approach.' According to this model, the legislature could abide by a court's decision or choose to reject it and exercise the nm instead. The novel aspect of this approach is that it circumvents the inconsistency created by the check on judicial error model. This inconsistency arises because the role of judicial review under the check on judicial error approach is to check legislative activity. A legislative re-check of this check is inconsistent with the idea that it is the role of the court to protect the constitution. Under the alternative approach suggested here, the role of judicial review is not to block or protect anything but, rather, to deliberate. Since the legislature is free to accept or reject the result of this deliberation, legislative finality therefore becomes consistent with judicial review.
The deliberative disagreement account of judicial review consists of three arguments:
1 It is not the purpose of judicial review to check legislative power. If that were indeed the purpose, legislative finality would be theoretically inconsistent. The purpose of judicial review is to deliberate about the meaning of the constitution through sophisticated and carefully reasoned opinions in concrete cases. Once the legislature is exposed to the product of judicial review, it can decide whether it agrees with the court's interpretation.
2 The probability that the legislature and the public will attach weight to judicial deliberations and seriously discuss the relevant constitutional matters increases if the court enjoys an independent status and the power to strike down legislation. Although the legislature, as a repre-
72 I am not ignoring the fact that in the real world, neither judicial finality nor legislative finality is truly final. The American literature in this area is replete with works showing that the legislative brandh has many ways to retaliate against the judiciary, and vice versa. See, e.g„ G.N. Rozenberg, The Hollow Hope.: Can Courts Bring About Sorial Chang?*' (Chicago: University of Chicago Press, 1991). In Canada, Peter Hogg and Allison Bushel! have recently demonstrated that in most cases, legislatures have been able to achieve their goals even after Charter review. See P.W. Hogg & A.A. Bushel], 'The Charter Dialogue Between Courts and Legislatures (Or Perhaps the Charter of Rights Isn'tSuchaBad Thing After All)' (1997) 35 OsgoodeHallLJ. 75 [hereinafter'Charter Dialogue']. Still, in terms of theory, a constitution can have only one final interpreter, and, therefore, the discussion about finality makes sense. See j. Harrison, 'The Constitutional Origins and Implications of Judicial Review* (1998) 84 Va.L.Rev. 333 at 357-8.
sentative of the people, need not accept the court's opinion, it must re-enact the impugned act if it is to demonstrate that it has seriously considered the court's opinion. 3 There is no danger that such a re-enactment would result in a grave violation of rights, since modern legislatures are equally motivated to respect the constitution and because reversing a Supreme Court decision carries a considerable political price.
1 shall elaborate on these three arguments below.
1. The role of the, court is not to check legislative power but to deliberate on constitutional matters
This argument resolves the theoretical difficulty created by the judicial error argument. That difficulty is the tension that arises from the judicial error argument's premise: that both the court and the legislature are expected to take on the task of constitutional decision making. If the court is charged with deciding constitutional matters for the people, then legislative finality is an abnormality. Conversely, if the court's function is to deliberate on constitutional matters and the legislature is responsible for deciding how this deliberation might affect legislation, then legislative finality is not the exception but part of the rule.
The suggestion that the purpose of judicial review is to deliberate and not to decide rests on a claim that, although courts generally reason and deliberate for the purpose of decision making, it is possible to distinguish between the court's deliberation and its decision. Deliberation is an act of practical reasoning, whereas decision is an act of will giving effect to the result of deliberation. Under the model suggested here, the Constitution gives the courts the power to deliberate on whether a measure violates the Charter, but it does not give them the power to will the outcome of their deliberations into effect. That power belongs to the legislature, which can choose either to effectuate the court's reasoning or to leave it ineffectual, according to whether it agrees with it or not. That power belongs to the legislature, which can choose either to respect the court's ruling and do nothing or to render it ineffectual by legislating against it. Accordingly, since the court's role is to deliberate and not to decide, there is no need for the legislature to find a judicial error. The legislature can simply disagree with the result of the court's deliberation by re-enacting the impugned act.73
73 By structuring the process as one of deliberation through decision by courts followed by decision through legislation by legislatures, I am by no means suggesting that legislatures should not deliberate or should deliberate less. In fact.-because under the deliberative disagreement approach the legislature must take the judicial deliberation seriously, it is encouraged to deliberate. Indeed, Tushnet shows how the combination of judicial review and a NM could, in theory, contribute to public discussion and solve
250 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 251
At the heart of this account stands a distinction between the legislature's ability to interpret the constitution and its motivation to do so.74 This model suggests that the legislature is as motivated to interpret the constitution as the court is. If both bodies are driven by equal motivation, it should be the accountable legislature and not the unaccountable court that ultimately interprets the constitution. That said, in addition to a consideration of motivation, an assessment of the respective interpretive abilities of each body must also be undertaken. While the court is accustomed to interpreting texts, specifying ideas, and offering legal reasoning, the legislature is not. This specialized capacity suggests that the court's analysis of a constitutional matter should be antecedent to any legislative action. The legislature can benefit from the judicial deliberation by reading the decision, becoming informed by it, and conducting its deliberation in terms of the decision. Once the court has published its reasons, it ceases to possess anything that will make its final decision superior to that of the legislature.
2. Although the purpose of judicial review is deliberation and not decision, judicial review should nonetheless be binding
The obvious question raised by the deliberative disagreement approach is this: If the role of judicial review is to deliberate, why should judicial review be operative as opposed to advisory? Why should the legislature have to re-enact a law disapproved by the court rather than simply let it stand? There are four answers to this question.
First, to ensure that the legislators are actually considering the court's deliberation, it is helpful that the court's decisions be binding. This stipulation is based on the assumption that if the legislature can achieve a certain goal through inaction, it has less incentive to act. The goal in our case is to have a certain act in force despite the judicial finding that such an act violates the constitution. In a system with non-binding
the problem of 'democratic debilitation,' where the people and their representatives are discouraged from deliberating constitutional issues because they leave this job for the court. Tushnet, 'Policy Distortion,' supra note 34 at 277-84, 292-9. 74 It seems to me that this distinction underlies Frank Mkheiman's idea of a 'Council of Revision1 that would have the power to override US Supreme Court, decisions. Michelman's council would be accountable - elected and replaceable - but its only occupation would be reviewing judicial decisions. Consequently, this council would have motivation similar to that of a legislature (since it is accountable) but ability similar to that of a court (since it lias the expertise of a court). Michehnan argues that nothing in the concept of law precludes such a mechanism, which creates, as he puts it, judicial leadership withoutjudicial finality,' F. Michehnan, Judicial Supremacy, the Concept of Law, and the Sanctity of Life' in A. Sarat & T.R. Kearns, eds., Justice and injustice in Law and Legal Theory (Ann Arbor: University of Michigan Press, 1996) 139 at 145, 153 elxaj.
judicial review, the legislature can achieve this goal by doing nothing, and therefore it has less incentive to take into account the court's opinion, In contrast, in a system with binding judicial review, the legislature has to re-enact the impugned act to achieve its goal. Therefore, there is a greater likelihood (although definitely not a guarantee) that prior to the re-enactment, the legislature will consider the court's deliberation. A second justification for the binding character of judicial review, despite the court's status as a deliberator, focuses not on the legislature but on the court. Like every institution, the court has an incentive to increase its power, and therefore it will work harder and better if it knows its decisions will be binding. A third reason is efficiency. In most cases, the legislature will accept the court's deliberation. It would be simply inefficient to require that the legislature approve each of the court's decisions rather than rejecting only the few decisions to which it opposed.
The final justification for a binding form of judicial review is concerned with preserving the image of the court as an independent branch of government. If judicial review is merely advisory and judicial decisions become law only if the legislature voluntarily adopts them, the court might be perceived as an organ of the legislature. This is because the mechanism of striking down an act would involve not only a judicial decision stating that the act is unconstitutional but also a legislative decision adopting the court's decision. Under such a system, what gives the judicial decision its power is the legislative adoption. In contrast, if the court has the power to actually strike down legislation, then it is perceived as being an independent branch whose deliberation is not for legislative consideration but for public discussion. While the legislature still has the final word, the court's ability to strike down an act makes it clear that the court remains an independent and powerful branch of government that, must be taken seriously.
3. Legislative finality is not dangerous
The third argument supporting the deliberative disagreement approach is that the legislature is unlikely to abuse its power to reject the court's deliberation. This argument is actually identical to the no abuse argument,75 presented by Weiler as part of his check on judicial error approach. In fact, the no abuse argument not only fits both Weiler's approach and the deliberative disagreement approach but actually renders his judicial error argument unnecessary. Note that Weiler's account of the NM 'rests on the assumption that the chief threat to rights
75 See text accompanying notes 64-68 supra.
252 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 253
in Canada comes from legislative thoughtlessness'7" and not from 'outright legislative oppression.'77 The role of the court, according to Weiler, is to force the legislature to think by 'airing the issue of principles.'78 As Samuel LaSelva puts it, the constitution is supreme 'not because it prevents manifest violations of the constitution by the legislature, but because even a legislature which respects the constitution may violate it through inadvertence.'79 However, if the court's role is to provoke more meticulous drafting on the part of the legislature rather than to prevent legislative oppression, it is unclear why Weiler needs to invoke the judicial error argument. The need for a judicial error in order to justify a legislative trump of judicial decisions would make sense where the role of judicial review is to check legislative power. Such a supervisory function for the court demands that its decisions be final. Under this account, legislative finality could be justified only if a valid reason for rejecting the court's decision, such as a court error, existed. If, however, the court's role is simply to deliberate and to promote new legislative and public discussion, and the final pronouncement on the matter remained with the legislature, the legislature could simply say, 'We rethought the matter, and we disagree with the court's decision.'8(1
It seems that Weiler's invocation of the judicial error argument.stems from his failure to explore fully his own reading of the role of judicial review under the Charter. Once he suggested that the purpose of judicial review was to provoke thought and not to obstruct legislative oppression, he could have abandoned the judicial error argument along with the theoretical inconsistency it involved.
It might be helpful to sketch the ways in which the deliberative disagreement approach compares with the approaches suggested by Weiler, Slattery, and Weinrib. I will do so by reintroducing a table brought forth earlier, this time with the addition of the deliberative disagreement approach as well as that of a third parameter of inquiry:
76 'Rights and Judges (1984),' supra note 16 at 84.
77 Ibid.
78 Ibid.
79 LaSelva, 'Only in Canada,' supra note 22 at390-1. Indeed, Hogg and Bushell's research revealed that, in most cases, the effect of Chattel' review is to send the legislature back to the drafting table to produce better legislation. See Hogg and Bushel!, 'Charter Dialogue,' supra note 72, esp. at 96-8.
80 It should be emphasized that: I neither subscribe to nor reject the no abuse argument Weiler may be right that Canadian legislatures are trustworthy; but then again, he may be wrong. In any event, I am proposing that if he is right, the deliberative disagreement approach is a better way to understand the NM than the judicial error approach.
	Can the legislature avoid judicial review?	Can the legislature consider extra-constitutional considerations while invoking the NM?	While operating the NM, the legislature ...
Check on judicial error (Weiler)	No	No	believes that the court was wrong
Prevention of judicial error (Slattery)	Yes	No	suspects that the court will be wrong
Legislative exceptions to rights (Weinrib)	Yes	Yes	does not think about the court
Deliberative disagreement (proposed approach)	No	No	disagrees with the court
The deliberative disagreement approach suggests a significantly different role for the court in constitutional matters. The court's role is not to engage in dispute resolution but, rather, to engage in what might be called 'dispute deliberation.' Such a role for the court is certainly not foreign to Canadian political culture. Canadian courts have been providing deliberation in the form of advisory opinions to Canadian legislatures for many years. This 'reference' function allows the Canadian Parliament to refer 'important questions of law or fact' to the Supreme Court of Canada.81 Provincial governments can similarly refer questions to the provincial Courts of Appeal.82 The rich and established practice of using this mechanism in the context of constitutional questions in Canada83 reveals the vitality of the dynamic in which the court deliberates through sophisticated reasoning and governments make the final decision. Canadian legislatures seem to have consistently adopted the result of the courts' deliberations.84 Moreover, important reference decisions, such as the Quebec
81 Supreme Court Act, R.S.C 1985, c. S-26, s. 53.
82 See P.W. Hogg, Constitutional Law of Canada, looseleaf (Toronto: Carswell, 1992) at vol. 1, 8.6(b), 8-15 [hereinafter Constitutional Law].
83 See L, Huffman & M. Saathoff, 'Advisory Opinions and Canadian Constitutional Development: The Supreme Court's Reference jurisdiction' (1990) 74 Minn.L.Rev. 1251.
84 Hogg, Constitutional Law, supra note 82 at 8.6(d), 8-17.
254 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 255
Secession Reference^' and the Patriation Reference*" were sources of much public deliberation, even though they were not binding in nature.87
It is important to note the assumptions upon which the deliberative disagreement approach relies, some of which are not uncontested. With regard to courts, this approach assumes that judicial review is good for public discussion and that the court's deliberations can and do educate the public. With regard to legislatures, the deliberative disagreement approach assumes that they are capable of interpreting the constitution, as opposed to simply pursuing political preferences, and that they would not abuse the NM. Examining the validity of these assumptions must involve some difficult theoretical and empirical inquiries that are well beyond the scope of this article. However, once one accepts these assumptions - and it: seems that many constitutional lawyers and political scientists do88 ~ the deliberative disagreement approach has the
85 Re Secession of 'Quebec, [1998] 2 S.C.R. 217.
86 Re Resolution to Amend the Cmutiluiicm, [1981] 1 S.C.R. 753.
87 One might object that, in a system featuring a specific 'reference' mechanism, it is illogical to view the entire purpose of judicial review as deliberative. According to this objection, the fact that, an avenue of deliberation exists for the court: under the auspices of the reference mechanism implies that when the court is asked to work under its general power of review, its review should be directed at supervision and decision, not deliberation. This objection misses the point of the deliberative disagreement approach. This approach does not hold that the essence of judicial review is one of deliberation rather than decision, It holds dial: the purpose of judicial review is deliberation and not decision. But in order for this purpose to be achieved, the process of judicial review itself must include judicial power to decide that an act is unconstitutional Moreover, a reference process, by definition, can be set in motion only by the government, while individuals can initiate judicial review as well, Therefore, even if one believes that the purpose of judicial review is to deliberate and not to decide, there is, in principle, room both for a reference mechanism and for judicial review. While workmgunder its reference capacity, the court's power is limited to giving advice. While reviewing the constitutionality of legislation, the court has the added ability to strike down legislation, and thereby wields the power to trigger a more serious legislative and public debate about the constitutionality of a given act.
88 The assumption that judicial review is good for public discussion and that the court's deliberations can and do educate the public is adopted both by the advocates of the check unjudicial error approach and by other critics of the Supreme Court, of Canada. See, e.g., Weiler, Judges and Rights (1.980),' supra note 16 at 223; Weiler, 'Rights and Judges (1984),'supra note 16 at 76-7; Russell, 'Standing Up,' supra note 26 at 300-1; J.L. Hiebert, 'Why Must a Bill of Rights Ve a Contest of Political and judicial Wills? "f lie Canadian Alternative' (1999) 10 Pub.L.Rev. 22 at 33 [hereinafter 'Why Must'3; Greschner & Norman, (Courts and s. 33,' supra note 39 at 191-2. Further, the assumption that legislatures are capable of engaging in constitutional interpretation is not marie only by the deliberative disagreement approach; it is actually-crucial to both Weiler's check on judicial error approach and Slattery's prevention of judicial error approach. For Weiler, the legislature must decide whether a certain judicial decision is erroneous according its own reading of the constitution. Similarly, for Slattery, the legislature can invoke the NM in order to prevent judicial error only if it believes the
potential to provide a consistent account, of judicial review with legislative finality.
The deliberative disagreement approach gives the legislature the opportunity to be constitutionally engaged rather than constitutionally uninterested. What principles should the legislature follow, when invoking the NM, to both make this constitutional engagement meaningful and, at the same time, preserve the central role of the court in the deliberation of constitutional issues? This is the topic of the next part of the article.
HI Between constitutional law and constitutional theory: The NM and the judicial-legislative partnership
The first part of this article dealt specifically with constitutional law and was therefore veiy concrete. It sought to determine the best interpretation of a text, namely s. 33 of the Charter. The second part dealt with constitutional theory, and therefore was more abstract. It did not address s. 33 or any specific constitutional text at all; rather, it addressed the idea of a judicial review with legislative finality. The third part of the article involves both constitutional law and constitutional theory. It proposes a model for understanding the relationship between courts and legislatures under a NM-bearing Charter. This model requires that the legislature treat the court as its partner rather than its adversary when invoking the NM.
Clearly, the text of s. 33 does not fit my suggested partnership model. As we shall see, a central element of this model is that the judiciary should be allowed to conclude its job before the NM is invoked. However, s. 33 allows for pre-emptive use of the mechanism.89 Indeed, I am not arguing that the partnership model is positive law in Canada, nor should my analysis in this section be read as an attempt to suggest a fourth reading of s. 33 (in addition to the three readings I discussed in the first part of the article). That said, I will demonstrate how the idea of partnership fits well with Canadian constitutional culture and therefore ought to
notwithstanding act to be constitutional according to its reading of the constitution, For the question of what it means for a legislature to interpret the constitution, see T. Kaliana, 'The Future of Law: an Enhanced Role for Legislatures?' (Law Forum 2001, Universite Laval, Quebec, 30 May 2001) [unpublished].
The idea that we should treat legislatures with more respect than traditional constitutionalism does.is gaining increasing support in Constitutional theory, best represented in the works of Mark Tnshnet and Jeremy Waldron. See M.V. Tushnet, Taking the Constitution Away From the Courts (Princeton: Princeton University Press, 1999); J. Waldron, Law and Disagreement (Oxford: Oxford University Press, 1999) at.pt. I ('A Jurisprudence of Legislation'); J. Waldron, The Dignity of legislation (Cambridge: Cambridge University Press, 1999), 89 See text accompanying notes 37-40 supra.
256 university of Toronto lawjournal
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 257
be considered as a model by legislatures interested in invoking the NM in a manner respectful to both the Canadian constitutional setting and the courts. Before delineating this model, I will first expand on the notion of partnership.
A. THE IDEA OF PARTNERSHIP BETWEEN COURTS AND LEGISLATURES Both Weiler and Weinrib describe the NM as a constitutional innovation and explain the innovation in terms of a 'partnership' between courts and legislatures. Weiler suggests that '[t] he premise of the Charter is that the optimal arrangement for Canada is a new partnership between court and legislature.'90 However, Weiler's presentation of the partnership notion as part of his judicial error account trivializes it. If the idea of final judicial review is referred to as judicial supremacy,' and if the idea of no judicial review is labelled 'legislative supremacy,' Weiler implies that a system of judicial review in combination with a NM amounts to a 'partnership.' This concept of partnership has neither practical implications nor conceptual importance. Even if the NM is based on the remnants of legislative supremacy, it can still be called a partnership. While there is no doubt that. Weiler's 'new partnership' is novel, it remains unclear how much of a 'partnership' it truly embodies. The interaction between courts and legislatures envisaged by Weiler is still about supervision 21s opposed to cooperation. Yet, rather than judges simply supervising legislatures, legislatures are assigned the additional role of supervising the judges responsible for supervising the legislatures. It is a partnership only inasmuch as legislatures are also part of the process; it is not a partnership in terms of how the process is conducted.01
Weinrib's invocation of the term 'partnership' is more meaningful. She believes that the nm is aimed at creating exceptions to rights. Further, she stresses that since these exceptions are created in order to promote majoritarian preferences, they cannot and should not be created by judges."2 In this context, Weinrib suggests that the nm brings with it 'not rule by supercourts at the expense of legislatures, but a complex partnership through institutional dialogue between supercourts and superlegis-latures.'"'3 Unlike Weiler's idea of partnership, where the partners supervise each other, Weinrib suggests a theoretical concept of partnership in which the partners cooperate by doing different things: courts deal with
90 'Rights and judges (1984),' supra note 16 at 84; also at 86: 'For better or for worse, Canada did stumble on this distinctive constitutional partnership between court and legislature for the protection of fundamental rights.'
91 Weiler's reference to 'partnership' was adopted by two of his followers, Russel! and MacDonald. See Russell, 'Standing Up,' supra note 26 at 297, and MacDonald, 'Social Equality,' supra note 26 at 23, Like Weiler himself, they do not explain what they mean by 'partnership.'
92 See text accompanying note 32 supra.
93 'Learning to-Live,' supra note 13 at 564—5,
law by protecting rights based on normative considerations, and legislatures deal with politics by creating exceptions to these rights based on majoritarian considerations.94 The partnership between courts and legislatures created by the Charter leaves room for both institutions to focus on what they are best suited for while continuing to interact under the Charter.
While Weinrib's concept of partnership is certainly innovative, it is still anchored in the notion of legislative exceptions to rights - an exception that is illegitimate in a constitutional democracy. As I have explained above, under a genuine constitutional democracy, majoritarian preferences should never be part of the Charter game and should be considered neither by courts nor by legislatures.95 Therefore, while it is true that assigning different types of Charter considerations to courts and legislatures makes the idea of partnership more meaningful, this approach must also be rejected because it allows considerations that are antithetical to a rights protection system.
My approach to partnership focuses not on substantive considerations taken into account by the legislature in deciding to invoke the NM but, rather, on fine-tuning the process of NM invocation, I term this idea 'partnership as a practice.' This notion occupies an intermediate position between two other approaches that address how legislatures should respond to unfavourable judicial decisions: the 'clashes approach,' which holds that the legislature should use all the power at its disposal to retaliate against the court's decision, and the 'acceptance approach,' according to which the legislature must accept judicial rulings.
Traditional American scholarship, adhering to the clashes approach, sometimes sees the three branches - separated by virtue of the principle of separation of powers - as supervising one another by means of checks and balances, creating a dynamic of 'tension and competition.'90 Each branch should do everything it can to promote its agenda, for it is the constant conflict among branches that secures the system from tyranny.97 Following this account, the NM is yet another check on judicial power,
94 See text accompanying note 33 supra.
95 See text accompanying notes 51-55 supra.
96 M.E. Magill, 'The Real Separation in Separation ol Powers Law* (2000) Va.L.Rev. 1127 at 1130, See also H.C. Mansfield, 'Separation of Powers in the American Constitution' in B.P. Wilson &P.W. Schramm, eds., Separation of Powers and Good Government (Lanham, MD: Rowmart & LittSefield, 1994) 3 at 3 [hereinafter 'Separation of Powers']. It should be noted, however, that the clashes approach is not the only one found in American discourse. There is also talk about coordination. See, e.g., M.O. Fiss, 'A Life Lived Twice' (1991) 100 Yale L.J. 1117 at 1120.
97 James Madison wrote that the accumulation of governmental power in one branch is 'the very definition of tyranny.' J. Madison, 'Federalist No. 47' in J. Madison, A. Hamilton, &j.Jay, The. Federalist, ed. Benjamin Fletcher Wright (Cambridge: Harvard University Press, 1961) 336 at 336.
258 UNIVERSITY OF TORONTO I AW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 259
qualitatively similar to the constitutional amendment, the appointment process, or any other means through which the legislature can respond to constitutional decision making by judges.
An example of the clashes conception is reflected in the work of American political scientist John Agresto, who supports legislative finality. Reasoning that a NM could function as a legislative tool to retaliate against the court, Agresto believes that such a mechanism fits the American system of checks and balances.'18 This proposal for an American NM emerges out of Agresto's discussion of other existing checks on judicial power, all of which he rejects.'1'' The check on judicial power that Agresto
98 J. Agresto, The Supreme Court and Constitutional Democracy (Ithaca, NY: Cornel! University Press, 1984) at 134 {hereinafter Supreme Court]. Agresto does not mention the actual Canadian NM, and his proposed NM is different from the Canadian one for several reasons. First, he does not limit his proposal to rights cases. Moreover, he does not talk about the re-enactment of legislation that has been struck down, referring instead to striking down the striking down. In addition, his NM does not produce temporary legislation. Finally, he suggests a supermajority requirement. Still, Agresto does propose a system of judicial review and legislative finality, and as such his analysis is of value to us.'
99 Agresto dismisses constitutional amendments as loo difficult to accomplish and still subject to subsequent judicial interpretation, Moreover, he argues that viewing the process of consti tu tional amendment as the only way to check the court wrongly assumes that the court's interpretation of the constitution iscorrect: ibid, at 107-11. He similarly rejects calls for judicial self-restraint, since the court is unlikely to heed such calls and, furthermore, self-restraint would negate the importance.of the court as a national thinker: ibid, at 112-6. Other checks on judicial power that Agresto discounts are impeachment, 'court packing,' and using the Congress's power under Article III of the USConstitulion to create exceptions to the Supreme Court'sjurisdiction: ibid, at 116-25,
The Article III power deserves special attention, since it is sometimes mentioned as the US equivalent of the NM. See, e.g., GR. Massey, 'The Locus of Sovereignty: Judicial Review, Legislative Supremacy, and Federalism in the Constitutional Traditions of Canada and the United States' (1990) Duke L.j. 1229 at. 1285-98; Tushnet, 'Policy Distortion,' supra note 34 at 285-7. Nonetheless, the NM power is not the Canadian equivalent of the Article III power; Manfredi, Judicial Power, supra note 26 at 175-6. Canada's equivalent to Article III is s. 101 of the Constitution Ad, 1867 (U.K.), reproduced in R.S.C. 1985, Appendix II, no. 5. This provision authorizes Parliament 'to provide for the Constitution, Maintenance, and Organization of a General Court of Appeal for Canada.' Correspondingly, Parliament established the Supreme Court in 1875. See the Sufrreme, Court Art, supra note 81, ivhich replaced the Supreme and Exc.kequer CourhAct, 1875, S.C. 1875, c. 11. Since the Supreme Court of Canada is not created by the Constitution but by the Supreme Court Act, the Parliament of Canada has the power to regulate the jurisdiction of the Supreme Court, which is the equivalent of Congress's power under Article III. The following table demonstrates this point:
General power to regulate Specific power to override _Supreme Court'sjurisdiction judicial decisions
Canada s. 101 of the Constitution Ad, 1867 s. 33 of the Charter
US _Article III of the Constitution No equivalent
does endorse is the 'power of the Congress literally to force reconsideration.'10" Congress can do this by re-passing legislation under an 'alternative constitutional guise,'by rewriting it, or by using 'the ability to circumscribe the holding of any decision in an attempt to delimit its effects.'502
Although Agresto notes the possibility of rewriting legislation to accommodate the court's decision and acknowledges that the introduction of new legislation 'argues for a differing constitutional view,'5,w what he really describes is a clash or struggle between the court and the legislature in which the legislature does whatever it can to get its way. Significantly, when Agresto moves to his suggestion of a NM, he does not emphasize the legislature's role in constitutional interpretation; rather, he stresses the importance of symmetry between the branches:
In many ways the perfect constitutional solution to the problem of interpretive finality ... would have been for the judiciary to possess the same legislative relationship to Congress as that which governs the executive. Just as Congress, by special majority, can override a presidential veto, a similar process could from the outset have been established to review judicial objections.1(14
Not surprisingly, Agresto's lone reference to the term 'partnership' simply describes a negation of judicial finality. He writes that 'the principle of checks and balances is a partnership and not simply a donation of power to the Court.,m This is a trivial use of the term, since it encompasses only the idea of checks and balances and not different types of interaction between courts and legislatures.
A very different approach to the interaction between courts and legislatures under the constitution is depicted in Peter Hogg and Allison Bushell's work. Their study shows that, as a general rule, Canadian legislatures have been able to achieve the goals of legislation struck down by courts, through the introduction of new legislation that adheres to court-
In other words, while overriding judicial decisions could perhaps be achieved ill the United States through Article III, the fact remains that s. 33 does not have an American equivalent. The American scholarly need to show that there is a 'U.S. analog to Section 33' (Tushnet, 'Policy Distortion,' supra note 34 at 285) is curious. Difficult as it might be for American scholars to realize, the NM is one of the few things that Canada has and the United States does not.
100 Agresto, Supreme Court, supra note 98 at 126.
101 Ibid.
102 Ibid, at 130. For the same idea see L. Fisher, Constitutional Dialogues: Interpretation as Political Process (Princeton: Princeton University Press, 1988) at 251-2.
103 Ibid, at 131 [emphasis added],
104 Ibid, at 134,
105 Ibid, at: 166.
260 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 261
prescribed standards.101'Furthermore, they show that the court's decision influences the legislative response even when the court does not strike down an act.107 .Although Agresto's as well as Hogg and Bushell's work discuss both the reintroductton of old legislation and new legislation, the stories they tell are different. Agresto describes a legislature that does not accept the court's view and acts out of disagreement with the court, whereas Hogg and Bushel! describe a legislature that accepts the court's decision.103
The notion of partnership as a practice suggests a middle ground between-the approach advocated by Agresto and that described by Hogg and Bushel!.109 It calls on the legislature to do what Agresto would have it do, but to do it in the manner described by Hogg and Bushell. Following Agresto, such a notion of partnership authorizes the legislature to reject a judicial reading of the constitution. However, in line with Hogg and Bushell's model, it requires that if the legislature rejects a judicial reading, it must do so while being informed by the judicial decision. The notion of partnership as a practice stands in contrast, then, to both the notion that the role of the legislature is to clash with the court and the suggestion that the duty of the legislature is to accept the court's view. It
106 Hogg & Bushell, 'Charter Dialogue,' supra note 72 esp. at 96-8. For critique and response see, respectively, C. Manfredi & J. Kelly, 'Six Degrees of Dialogue: A Response to Hogg and Bushell' (1999) 37 Osgoode Hall L.J. 513, and P. Hogg & A. Thornton, 'Reply to "Six Degrees of Dialogue'" (1999) 37 Osgoode Hall LJ. 529.
107 Hogg & Bushell, 'Charter Dialogue,' supra note 72 at 101-5.
108 Notably, the term 'dialogue' is employed in both Agresto's book and Hogg and Bushell's paper, although their approaches are very different. Hogg and Bushell entitle their article 'The Charter Dialogue Between Courts and Legislatures' and elaborate on this conception throughout the entire paper. Similarly, Agresto talks about a 'dialogue' between the branches (Agresto, Supreme Court supra note 98 at 97), Although the meaning of the idea of'dialogue' is outside the scope of this work, my impression is that this almost contradictory usage of the term reflects the fact that 'dialogue' has come, in recent years, to express many different ideas with regard to the interaction between courts and legislatures, between legislatures and the people, and between the courts and the people.
Hiebert speaks of the NM as a means of 'conversation' between the court and the legislature. Hiebert, 'Why Must,' supra note 88 at 31-4, Unlike the approach suggested here, Hiebert seems to be in line with Agresto and organizes her thoughts about 'conversation' in terms of checks and balances rather than partnership. See ibid, at 25.
109 I use the word 'described' in the context of Hogg and Bushell because, unlike Agresto's, the purpose of their work is not to suggest, a new argument, in favour of judicial review or to call on the legislature to use a certain type of check on judicial power but, rather, to show that, in most cases, the accountable legislature does achieve its social goals and thus 'any concern about the legitimacy of judicial review is greatly diminished': 'Charter Dialogue,' supra note 72 at 80. It remains unclear whether Hogg and Bushell support or object to the use of the NM. They do say, however, that the use of the NM in Canada is 'relatively unimportant' and that when the NM is used, there is no dialogue between the court and the legislature. Ibid, at 83.
asks the court and the legislature to see themselves as working together harmoniously in the explication of the principles of the constitution.
It is necessary to explain why the notion of partnership as a practice is preferable to both the acceptance and the clashes models. The acceptance model is inferior because the motivation for having a NM stems either from the fear of judicial error (as Weiler suggests110) or the desire to provide the legislature with the power to disagree with the court (as my alternative approach suggests111). If the legislature is to accept the court's view, no need for the NM arises. But what about the clashes model? If it is beneficial for courts and legislatures to play an active role in constitutional decision making, should they not be encouraged to clash with each other as much and as vehemently as possible? For the answer to this question, we must return to the Canadian context.
In the United States, fear of tyranny of one branch of government animates the principle of checks and balances.112 Analogizing to another field of law, the branches are not permitted to work together to create an anti-trust limitation, since the consumer - or, in the constitutional situation, the individual - might then have to pay higher prices for inferior products. However, under this system, the political discourse becomes more power-oriented than reason-oriented, since every branch tries to play its cards right rather spend time trying to convince others that its view is correct. Assuming that a reasoned debate is superior to a power struggle, there is some loss arising out of a clashes system. Still, the system Is willing to absorb this flaw in order to protect itself from tyranny. It is exactly here that the Canadian context differs. Given the absence of an entrenched fear of tyranny in Canada, no justification exists for paying the price exacted by a clashes system.113 Notably, Weiler's no abuse argument114 regains significance in this respect, since it trusts legislatures to respect rights. It implies that there is no fear of legislative oppression, and therefore the argument in favour of the clashes approach, namely the need to prevent legislative tyranny, is rendered invalid.
Furthermore, even after the adoption of the Charter, the principle of separation of powers - the central principle behind the idea of checks
110 See Pan.Ill, s. B, above.
111 See Part III, s. C, above. I explain why the partnership model cannot be applied to Siatiery's and Weinrib's models in note 121 infra.
112 Mansfield, 'Separation of Powers,' supra note 96.
113 Note that I am talking about an 'entrenched' fear of tyranny. Clearly, the framers of the Canadian Constitution did envision the possibility of tyranny and in order to block it devised a Senate and a disallowance power in s. 56 of the Constitution Ad, 1867, supra note 99. See P. Mackiem, 'Constitutional Ideologies' (1988) Ottawa L.Rev. 117 at 129-33. However, in contrast to the US context, this fear is not 'entrenched' in the sense that the fear of tyranny is not an inherent element of the political cukure.
11.4 See text accompanying notes 64-69 supra.
262 university of toronto law journal
UNDERSTANDING THE NOTWITHSTANDING'MECHANISM 263
and balances and the idea of clashing branches - is not a formal part of Canadian constitutionalism. Rather, Canada's system of responsible government provides for a fusion of the executive and legislative branches.115 Additionally, partnership is a concept appropriate to the Canadian context, given the constitution's 'reference' system, through which the Supreme Court delivers advisory opinions to Parliament."0 In the United States, such judicial activity exists only at the state level and is forbidden at the federal level as contradicting the court's role as a separate branch that should not cooperate with the legislative branch lest it lose its independence.'17 In a system where the branches are not as fundamentally separate from each other, the expectation of clashes is a priori-lower and the branches' ability to work in a partnership is a priori higher.
To put it differently, the Canadian NM should not be examined from the American perspective of checks and balances, as the latter is based on a fundamental mistrust of both courts and legislatures.118 Not surprisingly, when Michael Perry contemplated the idea of importing the nm to the United States, he suggested that such a mechanism was based on skepticism 'both shout the capacity of ordinary politics to specify constitutional indeterminacy and about the capacity of many of our judges and justices to do so.'"9 However, the Canadian view of the interaction between court and legislature is based not on a lack of public trust in either institution but on trust in both institutions. While it is clear that each institution makes its unique contribution to the interpretation, and enforcement of the constitution, an underlying assumption in Canadian constitutionalism is trust and respect between the branches, not suspicion and confrontation.120
Before expanding on the idea of partnership as a practice, it will be helpful to note that the partnership argument fits well with both the proposed deliberative disagreement approach and Weiler's check on judicial error approach. The partnership notion is an argument about the way in which the NM should operate. Whether it invokes the NM out of a convic-'
115 See Hogg, QmstitutumdlLata, supra note 82 at 7.3(a), 7-24-7-26.
116 See text accompanying notes 81-87 supra.
117 See G. Gunther, Constitutional Law, 12th ed. (Westbury, NY: Foundation Press, 1991) at 1593-6.
118 In conversation with the author, Professor Owen Fiss raised the question of what is the cause and what is the result: Are Americans less trusting of legislatures because of judicial finality, or is judicial finality a result of this American lack of trust? Needless to say, this question is beyond the scope of this article.
119 M. Perry, The Constitution in the Courts: Law or Politics (New York: Oxford University Press, 1994) at 197 [original emphasis],
120 One is tempted to extend this argument beyond the context of constitutionalism and suggest that generally speaking, the role of conflict arid suspicion is more central in American politics and culture than in Canada. This article is clearly not the appropriate forum for such an argument.
don that the court has erred (the check on judicial error approach) or simply because it disagrees with the court (the deliberative disagreement approach), the legislature must respond in a way that reflects a dynamic of partnership and not a dynamic of legislative supremacy. In either case, the idea of partnership requires that the legislature first discuss the court's decision and only then decide to use the NM in a manner respectful of the court, acting out of cooperation and not retaliation.121
B. THE PARTNERSHIP IDEA EXPLICATED 1, A partnership of respect
If the clashes model is rejected in favour of the alternative approach, competing agendas can be replaced with constructive deliberation. By implication, then, courts and legislatures must show respect both for the constitution and for each other. Respect for the constitution entails that the partners act with the understanding that the partnership is about interpreting the constitution. If a decision to invoke the NM is not connected to the constitutional text but, rather, is made because the legislature favours a certain substantive result, such a decision constitutes reassertion of legislative supremacy and not legislative participation in the interpretation of the constitution.
The notion of mutual respect between the partners is not intended to imply that the very fact that the legislature has acted is sufficient reason for the court to accept that action. Nor does it suggest that once the court strikes down a legislative act, this alone should serve as a reason for the legislature not to re-enact the legislation using the NM. Respect, in this context, means that the court and the legislature listen to each other, present each other's views fairly, and, in responding, employ arguments rather than rhetoric.
As a way of ensuring that the court listens to the legislature before striking down, a law, the current mechanism of litigation is sufficient.m The law is presented in front of the court. The government and other interveners have a chance to justify its enactment, to argue for its constitutionality, and to respond to questions posed by the court. In addition, the government has an opportunity to respond formally to arguments put forth by those who claim that the act is unconstitutional. If necessary,
121 While the partnership approach fits well with both Weiler's check on judicial error approach and my proposed deliberative disagreement approach, it does not fit well with Slattery's and Weinrib's approaches to the NM. Whereas the partnership approach requires that the court speak first, Slattery's and Weinrib's approaches allow the NM to be used pre-emptively.
122 I obviously am not implying that the judicial process is a perfect process. What I am saying is that, in terms of the idea of respect for the constitution and for the legislature, this process is sufficient,
264 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 265
all parties can submit social science evidence and expert opinions. The arguments that the impugned legal measures are constitutional are heard, analyzed, and responded to by judicial reasoning.
While the legislature need not engage in ajudicial process in responding to ajudicial decision (indeed, I reject the judicial error argument precisely because it puts the legislature in the position of a super-court523), its decision to re-enact legislation that has been struck down must be informed by the judicial decision. This is required according to both the check on judicial error approach and the deliberative disagreement approach. For the check on judicial error approach, this requirement is obvious, given that before the legislature can view a judicial decision as being wrong, it must know what the decision says. According to the alternative approach that I have suggested, which sees the court as national deliberator and legislative finality as a means of indicating popular acceptance or rejection of the court's decision, there is a need for the court's deliberation to be heard. According to both accounts, then, the legislators must consider the overridden judicial decision in its entirety, and not only in its result. Moreover, while discussing the decision, legislators must attempt to address the merits of the reasons and engage in a principled discussion, rather than in a simple assertion of
124
power.
These requirements are not difficult to institutionalize, since many legislative procedures aimed at enhancing the earnestness of legislative debate already exist. Traditionally, these methods have included absolute majorities or supermajorities, quorums, multiple readings, and express declarations. Indeed, express declaration is already provided for under s. 33, and some people have proposed including in this section the requirement of an absolute majority or supermajority.12'' Beyond traditional requirements, other alternatives include the periodic expiration set out in s. 33 and the dual enactment requirement originally proposed by Weiler.'20 Another possibility is the requirement of a preamble declaration in the overriding act setting out the legislature's reasons for overriding the court's decision. Other types of requirements can also be envisioned, such as mandatory public hearings held by committees of the legislature.
Obviously, the legislature cannot declare that if the Court strikes down a certain law, it will use the NM to re-enact it. In the framework of part-
123 See text accompanying notes 61-63 supra.
124 For an examination of the legislative debates thatsurrounded the only invocation of the NM to re-enact legislation struck down by ajudicial decision to date, namely Quebec's response to the Ford case (supra note 13), see Kahana, Partnership Model, supra note 3.4 at 207-14.
125 See Shaping Canada's Future Together, supra note 38; Manfredi, judicial Power, supra note 35 at 193; Lougheed, 'Why a Notwithstanding Clause?' supra note 22 at .17-8.
126 See note 23 supra.
nership, the court's decision signals the beginning of the decisionmaking process; therefore, a pre-emptive override would show disrespect for the court and render the judicial effort pointless. Moreover, such a declaration would generate an atmosphere of intimidation, forcing the court to rule in a certain way or else have its decision overridden, abridging judicial independence.127 The legislature must also respect the court in the way in which it presents the court's decisions. When legislators respond to ajudicial decision, the representation of the decision in their own language should'be fair, reflecting the fact that the court is seen not as an enemy to attack but as a partner with whom to reason.iiH
2. A partnership of benefit
Should the legislature delay its use of the NM until the highest court invalidates an act, or could it be invoked once a lower court nullifies the impugned act?1KI At the root of this dilemma is the requirement - under both Weiler's check on judicial error approach and my deliberative disagreement approach - that the court speak first. This requirement is in place to prevent the legislature from immunizing an act from judicial scrutiny before the court has had a chance to speak. However, once a court's voice is heard, must the legislature wait until the highest court speaks?
Greschner and Norman write that limiting the use of the NM to the overriding of decisions of the highest court is necessary to the legislature's showing 'respect for the judicial process."3" However, respect for
127 An example of a lack of respect for the judiciary can be found, ironically, in a newspaper article co-authored by Weiler himself. This article supports the use of the NM in response to the Ford decision (supra note 13) and also mentions National Citizens' Coalition Inc. v. A.G. Canada, [1984] 5W.W.R. 436 (Aita. Q.B.), asadecision in response to which the NM could have been invoked. In this decision, the Alberta Court of Queen's Bench struck down the Actio amend the Canadian Elections Act (No. 3), S.G 1980-81-82-83, c. 164,s. 15, which prohibited private interest groups from funding candidates or parties during election campaigns. Weiler and his co-author, Russell, said that they 'would strongly endorse the use of Canada's override procedure if our Supreme Court were so to interpret "freedom of expression" in the Charter.' P.H. Russell & P.C. Weiler, 'Don't Scrap Override Clause - It's a Very Canadian Solution' The 'Toronto Star (4 June 1989) B3. Russell and Weiler's threat that if the court does not obey their interpretive command they would support the use of the NM is extremely disrespectful towards the judiciary. In fact, the decision of the Alberta Court of Queen's Bench was never appealed.
128 Unfortunately, while discussing variousjudicial decisions as part of their support of the NM, some of the advocates of die NM failed to present judicial decisions in a fair way. See Kahana, Partnership Model, supra note 14 at 236-40, 245-52, 265-79.
129 By 'the highest court,' I mean the highest court possible. In cases where the parties do not have a right to appeal to the Supreme Court but can appeal only to one higher court (or in cases where such a right of appeal is discretionary and has been denied to the government), then 'the highest court' would mean the highest court possible.
130 Greschner & Norman, 'Courts and s. 33,' supra note 39 at 193.
266 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 267
the judicial process might be exactly the reason why the legislature should not wait for the highest court's decision. If the legislature plans to invoke the NM anyway, it may in fact be more respectful towards the courts to avoid having the higher courts waste their time. A better argument in favour of waiting until the highest court issues its decision is the notion of mutual benefit, which seeks to make the best use of the legislative-judicial partnership.
The partnership approach's marked interest in the judicial component of constitutional interpretation acknowledges that there is virtue in leaving the court alone and allowing 'courts {to] be courts, and legislatures, legislatures.'13' The courts 'should be left free to engage in their task' and 'go about their business of interpretation' as usual.18'2 For the individual court, part of what it. means 'to be a court' and to do 'business as usual' is to be surrounded and supported by the judicial system in its normal function. This normal function assumes a dynamic of progress and conversation within the judiciary. Cases start their journey at one court, move to another, and so on. They may also be sent back to a lower-court, or to a different court, as the dynamic of progress and conversation continues.1SS Ultimately,, several different hands shape the interpretation of the relevant provisions, and only when a case reaches its final stop within the judicial system can it be said that the judiciary's contribution has been exhausted. Because a premise of the partnership notion requires that the legislature listen to the judicial voice before deciding how to deal with a constitutional issue, it ought to wait until the internal judicial process is complete.
This argument is supported not only by the internal needs of the judicial system but also by the court-legislature relationship. The argument for using the NM immediately, rather than waiting for the highest court's decision, assumes that the higher court will either approve or reject the lower court's decision, and therefore it is more efficient simply to use the NM immediately: if the higher court approves the decision, then the NM will be used to override it, and if the higher court overrules the lower court's decision, then there will he no need to invoke the NM. In either case, the final result will be that the lower court's decision does not stand. However, part of the judiciary's internal dynamic presents the higher court with a third option: that of modifying the lower court's decision. Under this alternative, the legislature may be satisfied with the change, such that invocation of the NM would be unnecessary.
131 Weinrib, 'Learning to Live,' supra note 13 at 568.
132 Greschner & Norman, 'Courts and s. 33,' supra note 39 at 192.
133 See A. Reichman, 'A Theory of Practice: Constitutional Adjudication as Professional Discourse' {paper delivered at the Center for Ethics and the Professions, Harvard University, 2001) [unpublished]; on file with the author.
Consider, for example, the Vriend decision.134 In that case, the Supreme Court of Canada found the failure of Alberta's Human Rights, Citizenship and Multiculturalism Act135 to include sexual orientation as a prohibited ground of discrimination in the private sector to be unconstitutional. The Court read sexual orientation into the Act, and the Alberta government reacted negatively to this decision.13(1 Suppose now that this decision had been handed down by the Alberta Court of Appeal and not by the Supreme Court. Suppose further that Alberta's Legislative Assembly had considered using the NM in advance of any Supreme Court ruling, since it remained opposed to making sexual orientation a prohibited ground for discrimination. Such a legislative undertaking would fail to consider the possibility that the Supreme Court could, for example, strike down the act's impugned provision without 'reading in' sexual orientation. Under this scenario, the Supreme Court neither adopts nor rejects the Court of Appeal's decision; rather, it alters the lower court's decision in a manner that eliminates the need to invoke the NM.
Moreover, the judiciary's internal dynamic at higher levels can also result in novel reasoning that, while reaching the same result as the lower courts, presents a previously unconsidered perspective. Such reasoning may persuade the legislature that use of the NM is unnecessary. This is true from the viewpoint of both the check on judicial error approach and the deliberative disagreement approach. In terms of the judicial error argument, it is possible that the higher court would articulate the decision in language that corrects the error of a lower court. In terms of the deliberative disagreement approach, the decisions of the highest court would supply tire legislature with an additional source for reflection and deliberation.
The final justification for awaiting the highest court's decision before invoking the NM relates to the public. First, since the court's role is to serve as a deliberator for the people, the public deserves the court's most-considered product, which is likely to emerge only after the judicial system as a whole has had the opportunity to act.1S7 This would create not only a plurality of judicial voices but also an opportunity for a longer and gradual public discussion.138 Second, Canada is a federal country: for an
134 Vriend v. Alberta, [1998] 1 S.C.R. 493.
135 Human Rights, Citizenship and Multicultumlism Act, R.S,A. 1980, c. H-11.7,
136 See Kahana, Partnership Model, supra note 14 at 221-7. The NM was not used after Vriend, although the Alberta government contemplated its use. Ibid.
137 See Greschner & Norman, supra note 39 at 191; Hiebert, supra note 22 at 142-4; Hiebert, 'Why Must,' supra note 88 at 31-3. Hiebert's argument, like Greschner and Norman's, focuses on the need for public debate prior to the use of the NM and the potential for ajudiciai decision to improve such a debate. Their analysts does not focus on the role of the Supreme Court: in such a discussion.
138 For how the passage of time can influence people's and legislators' preferences, see Tushnet, 'Policy Distortion,' supra note 34 at 293-4,
268 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 269
event to be exceptional, it would be conducive that it be considered of national, rather than provincial, importance. If a provincial legislature invokes the NM in response to a judicial decision in that province without waiting for the Supreme Court, the discussion has a greater chance of being considered a provincial issue, and it may stay within the province. If, on the other hand, the NM is used only after the Supreme Court of Canada has ruled on the matter, there is a greater chance that the entire country will discuss the issue. Even though the NM was created partly to ensure provincial control over some constitutional issues within each province, the exceptional nature of the NM demands a country-wide discussion of the matter prior to and in response to this decision. Third, the no abuse argument suggests that if the legislature does not restrain itself from abusing the NM, the public may respond negatively. For the public to be able to evaluate a given use of the NM, it must receive the most thorough judicial deliberation available. Fourth, overriding the highest court's decision is likely to be more hazardous politically than overriding a lower court decision, given the greater prestige of the highest court and the heightened national attention surrounding the issue at stake. If the legislature wishes to use the NM, it should be ready to pay the political price associated with not merely overriding a court decision but overriding the decision of the nation's most prestigious court.181'
While the partnership of benefit favours awaiting the highest court's ruling, it may be appropriate for the legislature to exercise the NM in response to a lower court decision in exceptional circumstances, such as where this decision risks generating irreparable damage to social interests. Consider the example of legislation prescribing a maximum delay of one year between the filing of a criminal charge and the ensuing trial. The court in this example might deem the waiting period to be excessive in light of s. 11(b) of the Charter, which provides for a trial to be held 'within a reasonable time.' Suppose further that in making its ruling, the court reduces the waiting period to eight months, thereby forcing the crown to discontinue prosecutions of numerous detainees.1111 Even if the
139 This argument is especially strong in a federal system in which a court in jurisdiction A may strike down an act that was upheld in jurisdiction B. In such a case, the legislature that invokes the NM can support its position by resorting to the ruling from jurisdiction B, which found the act to be constitutional. After the Supreme Court rules that the act is unconstitutional, this line of argument is weakened (although it still exists, since the act was upheld by at least one court). It should be noted, however, that a Supreme Court decision may actually make it. politically easier for the legislature to invoke the NM in cases where a dissenting opinion upholds the constitutionality of the impugned act. See note 50 supra.
140 This example is based on R. v. Askav, [1990] 2 S.C.R. 1199. In Askov, however, the eight-month formula was prescribed by the court not in the context of striking *dowh legislation but, rather, as an interpretation of s. 11 (b) of the Charter.
highest court later overrules the decision and restores the one-year maximum, the negative effects of the lower court's ruling cannot be repaired retroactively. In such exceptional situations, society should not be forced to endure the resulting irreparable damage.
Arguably, this consideration can be extended further - from permitting the use of the NM following a lower court ruling to allowing it prior to any judicial consideration at all. Suppose, for example, that in the wake of a foreign crisis, tens of thousands of refugees sail towards Canadian waters. Suppose that no case law exists regarding the right of a refugee to a hearing.141 Fearing great expense and the paralysis of the Canadian immigration department, Parliament enacts new legislation that disallows such refugees any right to a hearing. However, given that the judiciary might misconstrue the extent of the problem, the constitutional validity of the new legislation is left uncertain. Consequently, if Parliament is forced to await a judicial decision affirming the new law, the refugees might already be on Canadian soil, and such legislation would create a public outcry. Why should Parliament be prohibited from using the NM in advance of a court ruling in such a case? The answer is that, under the partnership model, while the need for an issue to be examined by the judiciary as a whole is not absolute, the need for the issue to be examined by a court is. In the present example, the government should request a court ruling on an urgent basis to quickly ascertain the lower court's view before invoking the NM. The advantages that the judicial process has over the legislative process - arguments, parties, evidence, independence - all produce qualitative differences between the ways in which legislatures and courts make decisions. If an act that potentially violates rights is immunized completely from judicial scrutiny, this judicial quality will be absent from the system of rights protection. By contrast, if an act is examined by a court but is thereafter shielded from further judicial scrutiny, what is absent from rights protection is simply additionaljudicial consideration.142 Note, however, that this is acceptable only in exceptional situations. In all other circumstances, the rule created by the idea of partnership of benefit requires that the highest court speak before the NM is used.
3, A partnership of last resort-Should the legislature be authorized to use the NM when it can instead enact legislation that does not invoke the mechanism? Consider these three examples:
141 Compare Singh v. Canada (Minr. ofEplt. andlmm.), [1985] 1 S.C.R. 177.
142 Indeed, the examination of social evidence, which is a crucial component of rulings on the justification of limits on rights, is normally done by the lower courts; most lower court decisions are upheld by higher courts.
270 UNIVERSITY OF TORONTO LAWJOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 271
1 The highest court strikes down an act. The legislature wishes to enact legislation very similar to that struck down by the court, knowing that the new act is likely unconstitutional based on the court's decision. Should it wait until the new act is reviewed by the court - now we can say by the highest court - only to re-enact it with a NM, or should it enact it immediately with a notwithstanding declaration?
2 The highest court strikes down an act, ruling that the social science evidence presented to the court does not satisfy the Oakes requirements. K1 Some years later, new social science evidence becomes available, and the government believes that, based on this evidence, the act does satisfy the Oakes requirements. Should the legislature enact this measure as an ordinary act, without a notwithstanding declaration, making it possible for the court to examine the new evidence, or should it re-enact the measure with a notwithstanding declaration to save time and money on litigation?
3 The highest, court creates a common law rule, based on a certain interpretation of the Charter. The legislature wants to enact a law that clearly contradicts this common law rule. Should it enact the measure as an ordinary act and then, if the highest court strikes it down, re-enact it with a notwithstanding declaration? Or should it enact the measure with a notwithstanding declaration without waiting to see whether the court will strike it down?
The last scenario is exemplified in the Daviault case.144 There, a majority of the Supreme Court concluded that the common law rule rejecting intoxication as a defence to general intent crimes violated sections 7 and 11(d) of the Charter. The majority reasoned that even crimes of general intent require a minimal mental element and that substituting self-induced intoxication for this requirement, infringed upon the s. 7 right to life, liberty, and security in a way that is not 'in accordance with the principles of fundamental justice.',4r' Further, the Court found that it infringed upon s. ll(d)'s presumption of innocence. Since these violations were not justifiable under s. 1 of the Charter,14" the Court altered the common law rule and allowed extreme drunkenness leading to automatism to be a defence to the general intent crime of sexual assault.147 The dissent in Daviault upheld the common law rule and suggested that the accused's drunkenness resulted from his 'own blameworthy con-
143 R. v. Oalii's, [1986] 1 S.C.R. 103 at 138.
144 R. v. Daviault, [1994] 3 S.C.R. 63 at 79-93.
] 45 S. 7 of the Charter guarantees that one's life, liberty, and security of person will not be infringed in a way that is not 'in accordance with the principles of fundamentaljustice.'
146 Ibid, at 93-103.
147 Ibid, at 103-4.
duct.'148 In the wake of public, outcry,14'1 Parliament amended the Criminal Code to reflect the Daviault dissent, essentially converting the impugned common law rule into a statutory one. The legislation states that '[i]t is not a defence ... that the accused, by reason of self-induced intoxication, lacked the general intent or the voluntariness required to commit the offence.'150
For the purpose of this discussion, assume that Daviault was indeed decided wrongly. Did Parliament err by not invoking the NM to prevent judicial review of this Criminal Code amendment? The partnership formula set out in the previous section does not provide a full answer to this question. On the one hand, this formula requires that the court speak first. No court has ever ruled on the validity of the Criminal Code amendment overriding Daviault. Therefore, the legislature should first enact the new law as an ordinary act, and if the highest court strikes it down, the legislature should ponder the court's reasoning and only then decide whether to invoke the NM. On the other hand, in overruling Daviault, what Parliament would want to say is not: 'we do not like the result of Daviault,' but rather 'we think that the common law rule that self-induced intoxication is not a defence to general intent crimes does not violate section 7.' Regarding this finding, the Supreme Court has already ruled. In other words, since this legislative reaction to the court was, as Kent Roach put it, an 'in-your-face reply,' the government should have simply invoked the NM to protect the legislation though it was never actually struck down.15! Using the NM, as others put it, would be 'straightforward,',:,a 'symbolic,'1''3 and 'honest.'154
The resolution of this problem is that under the partnership approach the legislature is not a super-court. Respecting the role of the court implies that the legislature lacks the power to simply overrule judicial decisions whenever it believes that the court's interpretation is wrong. Rather, the legislature's power under the partnership model is limited to those situations in which the legislature disagrees with the court's ruling and the court's decision stops the legislature from achieving its goal. This
148 Ibid, at 11.5-32, quotation at 121.
149 See K, Roach, Due Process and Victims' Rights: The New Law and Polities of Criminal Justice (Toronto: University of Toronto Press, 1999) at 178-80.
150 Criminal Code, R.S.C, 1985, c. C-46, s. 33.1 (1), as am by A nAci to Amend the Criminal Code (self-induced inloxirMitm), S.C. 1995, e, 32, s, 1.
151 K. Roach, 'Editorial: When Should the Section 33 Override be Used?' (1999) 42 C.L.Q. 1 at 2; K. Roach, 'Constitutional and Common Law Dialogues between the Supreme Court and Canadian Legislatures' (2001) 80 Can.Bar.Rev. 481 at 525.
152 R. Litkowski, 'Roundtable III' in J. Cameron, ed,, The Charter's Impact on the Criminal justice System (Scarborough, ON: Carswell, 1996) 293 at 309.
153 T. Williams, ibid.
154 A. Young, ibid.
272 UNIVERSITY OF TORONTO LAW JOURNAL
UNDERSTANDING THE NOTWITHSTANDING MECHANISM 273
power is therefore exercised through the re-enactment of laws and not by issuing declarations overruling judicial decisions. In other words, for the operation of the NM, it is not enough that the highest court has ruled; its ruling must involve the striking down of an act, such that the NM is used as a means of re-enactment and not as a means of overruling.
Furthermore, supporting the use of the NM in response to a judicial decision that did not strike down an act assumes that once the newly enacted measure reaches the courts, it will be struck down. But this is not necessarily the case. In fact, the British Columbia Supreme Court, upheld the amendment to the Criminal Code created by the legislation enacted in response to Daviault, thus contradicting the majority in that decision.155 Similarly, the Supreme Court decision in R. v. Mills15'' upheld legislation that clearly contradicted its decision in R. v, O'Connor*-'1 by practically adopting the dissent in that decision.158
The conclusion that the NM ought to be used only as a response to the striking down of an act can be reached via another line of reasoning. The partnership approach sees the invocation of the NM as an exceptional event that should attract exceptional public attention. The less it is used, the more it is conceived as exceptional. To decrease the use of the NM, the rule for the partnership approach should be that court decisions are adhered to, thus avoiding use of the mechanism unless necessary.
Requiring that the NM be used only if the legislature has no other choice has two other implications. First, where the legislature could achieve the goals of a law struck down by the court by enacting new legislation harmonious with the court's decision, the legislature should avoid using the NM. As Hogg and Bushel] have suggested, this option is frequently available to legislatures.15" The other implication pertains to the exception to the requirement that the NM be used only subsequent to the highest court's ruling - an exception I alluded to in the last section. I suggested that in cases where exceptional and irreparable damage might be inflicted by requiring the legislature to await a higher court ruling, the legislature ought to be able to override a lower court's decision.
155 See /{. v. Vickberg (1998), 16 C.R. (5th) 164 (B.C. S.C.).
156 [1999] 3 S.C.R. 668.
157 [1995] 4S.C.R. 411.
158 The argument I am making here is slightly different from the one offered by Greschner and Norman against the pre-emptive use of the NM when a similar act is struck down. They suggest that'[t] t is difficult to imagine a case in which every Charter issue is on all fours with a case already decided ... Truly identical cases can be anticipated to be so rare that [they are] justified in enunciating ... that the statute first be subject to judicial review': Greschner 8c Norman, 'Courts ands. 33,' supra note 39 at 193-4! My point here is that even if itis clear that two cases are 'trulyidentical,' the legislature should not use the NM prior to a judicial decision, since the NM should be used only if there is no other choice,
159 Hogg & Bushel!, 'Charter Dialogue,' supra note 72.
However, based on the discussion in this section, before the legislature can take such a step, it must appeal the lower court decision and request a stay of the court's order until the appeal is heard. Only in the event that this request is refused can the legislature justify invoking the NM.'(,(I
V Conclusion: A nexu era of constitutionalism?
The idea of a legislative override of the judiciary in constitutional cases marks an important development in constitutional theory, one whose impact extends far beyond Canadian borders. In a world in which it seems that 'everything useful that could possibly be said about the legitimacy of judicial review has now been said,'1" this constitutional innovation has attracted the attention of constitutional law scholars worldwide, from the United States11'2 to the United Kingdom1"3 to Spain.164 Reflecting this interest, this article has tried to achieve two main goals: first, to explain the NM within the structure of Canadian constitutionalism; and, second, to explore various models purporting to explain and justify the NM - models applicable in many constitutional democracies.
Weiler, Slattery, and Weinrib have produced a body of academic literature that is of tremendous importance to the understanding of the NM. Weiler's formative work laid the foundation for approaching the NM as a beneficial development rather than a necessary evil. Slattery's work took the discussion a step further by explaining the NM in light of a general theory of the Charter. Finally, Weinrib's analysis added the distinction between 'super-court' and 'super-legislature,' thereby emphasizing the need to explain the NM in light of the division of labour between courts and legislatures.
Notwithstanding their importance, these works fail to render the NM consistent with general principles of constitutionalism. Weiler's idea of the legislature as a 'super-court' and Slattery's concept of the legislature as a 'High Court of Parliament' contradict the principles underlying the idea of judicial review. Weinrib's account does not suffer this flaw, yet it allows the legislature to violate rights for the sake of majoritarian preferences, a notion antithetical to the idea of rights.
Building on the work of these scholars, this article offers an approach for resolving the inconsistency between the NM and constitutionalism.
160 See Greschner & Norman, 'Courts and s. 33,' supra note 39 at 193.
161 Hogg & Bushell, 'Charter Dialogue,' supra note 72 at 79.
162 See, e.g., Tushnet, 'Policy Distortion,' supra note 34; M.A. Glendon, Rights Talk: The Impoverishment of Political Discourse (New York: Free Press, 1991) at 39,
163 See, e.g., G. Marshall, 'Taking Rights for an Override: Free Expression and Commercial Expression' (1989) Pub.L.Rev. 4 at 5.
164 See V. Ferreres-Comeila, judicial Review of Legislation in a Parliamentary Democracy: The Case of Spam' (JSD Thesis, Yale Law School 2000) [unpublished] at 51.
274 UNIVERSITY OF TORONTO LAW JOURNAL
This can be achieved by rethinking not only the legislature's role when invoking the NM but also the court's role when reviewing legislation. The deliberative disagreement approach holds that, since the purpose of judicial review is to deliberate on constitutional issues and not to block legislative decisions, it is appropriate for the legislature to have the final word but to utter it only after having considered the court's reasons. This paradigm makes judicial review and legislative finality theoretically consistent. In addition to the theoretical unification of judicial review and legislative finality, I suggest a model for the operation of the NM. The NM should operate within the framework of a partnership between the branches, and not as a tool of one branch in its struggle with another,
The model offered here presupposes trust in legislatures not to abuse the NM. Weiler likewise trusts legislatures to exercise self-restraint, relying partly on their fear of public condemnation. His no abuse argument provides the necessary departure point for both the deliberative disagreement approach and the partnership model. Since legislatures are trustworthy, the argument goes, it is possible to suggest that the purpose of judicial review is not to supervise legislative activity but to infuse the public domain with meaningful deliberation. Moreover, the trustworthiness of legislatures makes our confidence in their adherence to the partnership model realistic.
The question of whether Canadian legislatures - and legislatures generally - are indeed worthy of the trust Weiler reposes in them is a difficult one. While Canadian'legislatures have arguably not yet used the NM to enact any repugnant laws, this history does not guarantee that abuses will not occur in the future. This article takes no position on this question, however, because it has neither advocated nor opposed the existence of a NM, Rather, it has sought to uncover the theoretical framework within which a NM can be reconciled with constitutional rights protection. The theoretical consistency of the NM is one thing, its workability another.
If indeed modern legislatures are trustworthy, then the constitutional world is entering a new era of constitutionalism, one in which fear of legislative oppression is reduced and trust is increased, resulting in a corresponding rise in the recognition that legislatures need to be constitutionally engaged as 'super-legislatures.' If this is the case, the NM could very well serve as the perfect symbol to usher in this, new era.