Fraser v. Canada (Attorney General), 2020 SCC 28, [2020] 3 SCR 113

    

 

SUPREME COURT OF CANADA

 

Citation: Fraser v. Canada (Attorney General), 2020 SCC 28, [2020] 3 S.C.R. 113

Appeal Heard: December 12, 2019 Judgment Rendered: October 16, 2020 Docket: 38505

 

Between:

Joanne Fraser, Allison Pilgrim and Colleen Fox

Appellants

 

and

 

Attorney General of Canada

Respondent

 

- and -

 

Attorney General of Ontario, Attorney General of Quebec, Women’s Legal Education and Action Fund Inc., Public Service Alliance of Canada and National Police Federation

Interveners

 

 

 

Coram: Wagner C.J. and Abella, Moldaver, Karakatsanis, Côté, Brown, Rowe, Martin and Kasirer JJ.

 

Reasons for Judgment: (paras. 1 to 139)   Joint Dissenting Reasons: (paras. 140 to 230)   Dissenting Reasons: (paras. 231 to 256)

Abella J. (Wagner C.J. and Moldaver, Karakatsanis, Martin and Kasirer JJ. concurring)   Brown and Rowe JJ.     Côté J.

 

 

 

fraser v. canada (attorney general)

Joanne Fraser,

Allison Pilgrim and

Colleen Fox                                                                                                    Appellants

v.

Attorney General of Canada                                                                      Respondent

and

Attorney General of Ontario,

Attorney General of Quebec,

Women’s Legal Education and Action Fund Inc.,

Public Service Alliance of Canada and

National Police Federation                                                                          Interveners

Indexed as: Fraser v. Canada (Attorney General)

2020 SCC 28

File No.: 38505.

2019: December 12; 2020: October 16.

Present: Wagner C.J. and Abella, Moldaver, Karakatsanis, Côté, Brown, Rowe, Martin and Kasirer JJ.

on appeal from the federal court of appeal

                    Constitutional law — Charter of Rights  — Right to equality — Discrimination based on sex — Adverse impact discrimination — Systemic discrimination — RCMP allowing members to job‑share — Job‑sharing members not allowed under pension plan to buy back pension credits — Job‑sharers are mostly women — Retired members claiming that pension consequences of job‑sharing have discriminatory impact on women and violate their constitutional right to equality — Whether limitation on job‑sharers’ ability to buy back pension credits discriminates on basis of sex — If so, whether infringement justified — Canadian Charter of Rights and Freedoms, ss. 1 , 15(1)  — Royal Canadian Mounted Police Superannuation Act ,  R.S.C. 1985, c. R‑11  — Royal Canadian Mounted Police Superannuation Regulations, C.R.C., c. 1393.

                    The claimants are three retired members of the RCMP who took maternity leave in the early‑to‑mid 1990s. Upon returning to full‑time service, they experienced difficulties combining their work obligations with their childcare responsibilities. At the time, the RCMP did not permit regular members to work part‑time. In December 1997, the RCMP introduced a job‑sharing program in which members could split the duties and responsibilities of one full‑time position. The three claimants enrolled in the job‑sharing program; they and most of the other RCMP members who job‑shared were women with children. Pursuant to the Royal Canadian Mounted Police Superannuation Act , and the associated Royal Canadian Mounted Police Superannuation Regulations (“pension plan”), RCMP members can treat certain gaps in full‑time service, such as leave without pay, as fully pensionable. The claimants expected that job‑sharing would be eligible for full pension credits. However, they were later informed that they would not be able to purchase full‑time pension credit for their job‑sharing service.

                    The claimants initiated an application arguing that the pension consequences of job‑sharing have a discriminatory impact on women contrary to s. 15(1)  of the Charter . Their claim failed at the Federal Court. The application judge found that job‑sharing is part‑time work for which participants cannot obtain full‑time pension credit and that this outcome did not violate s. 15(1). The application judge held that there was insufficient evidence that job‑sharing was disadvantageous compared to leave without pay. The Federal Court of Appeal dismissed the claimants’ appeal.

                    Held (Côté, Brown and Rowe JJ. dissenting): The appeal should be allowed.

                    Per Wagner C.J. and Abella, Moldaver, Karakatsanis, Martin and Kasirer JJ.: Full‑time RCMP members who job‑share must sacrifice pension benefits because of a temporary reduction in working hours. This arrangement has a disproportionate impact on women and perpetuates their historical disadvantage. It is a clear violation of their right to equality under s. 15(1)  of the Charter .

                    To prove a prima facie violation of s. 15(1), a claimant must demonstrate that the impugned law or state action, on its face or in its impact, creates a distinction based on enumerated or analogous grounds, and imposes burdens or denies a benefit in a manner that has the effect of reinforcing, perpetuating, or exacerbating disadvantage. The claimants contend that the negative pension consequences of job‑sharing infringe s. 15(1) because they have an adverse impact on women. Resolving their claim requires considering how adverse impact discrimination is applied.

                    Adverse impact discrimination occurs when a seemingly neutral law has a disproportionate impact on members of groups protected on the basis of an enumerated or analogous ground. There is no doubt that adverse impact discrimination violates the norm of substantive equality which underpins the Court’s equality jurisprudence. Substantive equality requires attention to the full context of the claimant group’s situation, to the actual impact of the law on that situation, and to the persistent systemic disadvantages that have operated to limit the opportunities available to that group’s members. At the heart of substantive equality is the recognition that identical or facially neutral treatment may frequently produce serious inequality. This is precisely what happens when seemingly neutral laws ignore the true characteristics of a group which act as headwinds to the enjoyment of society’s benefits.

                    The same two‑step approach to s. 15(1) applies regardless of whether the discrimination alleged is direct or indirect. At the first step, in order for a law to create a distinction based on prohibited grounds through its effects, it must have a disproportionate impact on members of a protected group. A law, for example, may include seemingly neutral rules, restrictions or criteria that operate in practice as “built‑in headwinds” for members of protected groups. In other cases, the problem is not “headwinds” built into a law, but the absence of accommodation for members of protected groups.

                    Two types of evidence will be especially helpful in proving that a law has a disproportionate impact on members of a protected group. The first is evidence about the situation of the claimant group. Courts will benefit from evidence about the physical, social, cultural or other barriers which provide the full context of the claimant group’s situation. The goal of such evidence is to show that membership in the claimant group is associated with certain characteristics that have disadvantaged members of the group. These links may reveal that seemingly neutral policies are designed well for some and not for others.

                    Courts will also benefit from evidence about the outcomes that the impugned law or policy has produced in practice. This evidence may provide concrete proof that members of protected groups are being disproportionately impacted. The evidence may include statistics, especially if the pool of people adversely affected by a criterion or standard includes both members of a protected group and members of more advantaged groups. The goal of statistical evidence is to establish a disparate pattern of exclusion or harm that is statistically significant and not simply the result of chance. The weight given to statistics will depend on, among other things, their quality and methodology.

                    Ideally, claims of adverse effects discrimination should be supported by evidence about the circumstances of the claimant group and the results produced by the challenged law. However, both kinds of evidence are not always required. In some cases, evidence about a group will show such a strong association with certain traits that the disproportionate impact on members of that group will be apparent and immediate. Similarly, clear and consistent statistical disparities can show a disproportionate impact on members of protected groups, even if the precise reason for that impact is unknown. In such cases, the statistical evidence is itself a compelling sign that the law has not been structured in a way that takes into account the group’s circumstances.

                    In sum, both evidence of statistical disparity and of broader group disadvantage may demonstrate disproportionate impact, but neither is mandatory and their significance will vary depending on the case. Whether the legislature intended to create a disproportionate impact is irrelevant. Proof of discriminatory intent has never been required to establish a claim under s. 15(1), and an ameliorative purpose is not sufficient to shield legislation from s. 15(1) scrutiny.

                    If claimants successfully demonstrate that a law has a disproportionate impact on members of a protected group, they need not also prove that the protected characteristic “caused” the disproportionate impact. It is also unnecessary for them to prove that the law itself was responsible for creating the background social or physical barriers which made a particular rule, requirement or criterion disadvantageous. In addition, claimants need not show that the impugned law affects all members of a protected group in the same way. The fact that discrimination is only partial does not convert it into non‑discrimination, and differential treatment can occur on the basis of an enumerated ground despite the fact that not all persons belonging to the relevant group are mistreated.

                    The second step of the s. 15 test — whether the law has the effect of reinforcing, perpetuating, or exacerbating disadvantage — will usually proceed similarly in cases of direct and indirect discrimination. The goal is to examine the impact of the harm caused to the affected group, which must be viewed in light of any systemic or historical disadvantages faced by the claimant group. The presence of social prejudices or stereotyping are not necessary factors in the s. 15(1) inquiry, and the perpetuation of disadvantage does not become less serious under s. 15(1) simply because it was relevant to a legitimate state objective. The test for a prima facie breach of s. 15(1) is concerned with the discriminatory impact of legislation on disadvantaged groups, not with whether the distinction is justified, an inquiry properly left to s. 1. Similarly, there is no burden on a claimant to prove that the distinction is arbitrary to prove a prima facie breach of s. 15(1). It is for the government to demonstrate that the law is not arbitrary in its justificatory submissions under s. 1.

                    Full‑time RCMP members who work regular hours, who are suspended, or who go on unpaid leave can obtain full pension credit for those periods of service under the pension plan, but full‑time members who temporarily reduce their hours under a job‑sharing agreement are classified as part‑time workers under the Regulations and are unable to acquire full‑time pension credit for their service. The question is whether this arrangement has a disproportionate impact on women.

                    In relying on the claimants’ “choice” to job‑share as grounds for dismissing their claim, the Federal Court and Court of Appeal misapprehended the Court’s s. 15(1) jurisprudence. The Court has consistently held that differential treatment can be discriminatory even if it is based on choices made by the affected individual or group. The Federal Court and Court of Appeal also engaged in a formalistic comparison between the remuneration offered under job‑sharing and leave without pay, even though s. 15(1) guarantees the claimants and others in the job‑sharing program the right to substantive equality with respect to full‑time RCMP workers.

                    Under a proper assessment, the s. 15(1) claim succeeds. The use of an RCMP member’s temporary reduction in working hours as a basis to impose less favourable pension consequences plainly has a disproportionate impact on women. The relevant evidence showed that RCMP members who worked reduced hours in the job‑sharing program were predominantly women with young children. These statistics were bolstered by compelling evidence about the disadvantages women face as a group in balancing professional and domestic work. This evidence shows the clear association between gender and fewer or less stable working hours, and demonstrates that the RCMP’s use of a temporary reduction in working hours as a basis for imposing less favourable pension consequences has an adverse impact on women.

                    This adverse impact perpetuates a long‑standing source of disadvantage to women: gender biases within pension plans, which have historically been designed for middle and upper‑income full‑time employees with long service, typically male. Because the RCMP’s pension design perpetuates a long‑standing source of economic disadvantage for women, there is a prima facie breach of s. 15 based on the enumerated ground of sex.

                    Section 1 allows the state to justify a limit on a Charter  right as demonstrably justified in a free and democratic society. To start, the state must identify a pressing and substantial objective for limiting the Charter  right. The Attorney General has identified no pressing and substantial policy concern, purpose or principle that explains why job‑sharers should not be granted full‑time pension credit for their service. On the contrary, this limitation is entirely detached from the purposes of both the job‑sharing scheme and the buy‑back provisions. Job‑sharing was clearly intended as a substitute for leave without pay for those members who could not take such leave due to personal or family circumstances. It is unclear, then, what purpose is served by treating the two forms of work reduction differently when extending pension buy‑back rights. The government has not offered a compelling objective for this differential treatment.

                    Since the prima facie breach cannot be justified under s. 1, it is a violation of s. 15(1) to preclude the claimants and their colleagues from buying back their pension credits. The appropriate remedy is a declaration that there has been a breach of the s. 15(1) rights of full‑time RCMP members who temporarily reduced their working hours under a job‑sharing agreement, because of the inability of those members to buy back full pension credit for that service.

                    Per Brown and Rowe JJ. (dissenting): The RCMP pension plan does not violate s. 15  of the Charter  in its application to members who job‑share by denying them the right to accrue full‑time pension credit for periods when they job‑shared for childcare reasons. While the pension plan does create a distinction that, in its impact, is based on sex, its effect cannot be to hinder government efforts to address pre‑existing inequality. Any disadvantage the claimants face is caused not by the impugned provisions or any government action, but by the unequal division of household and family responsibilities and social circumstances such as the availability of quality childcare. Substantive equality has become almost infinitely malleable, allowing judges to invoke it as rhetorical cover for their own policy preferences in deciding a given case. This discretion does not accord with, but rather departs from, the rule of law.

                    Understanding the pension plan and the underlying legislative scheme in its entirety is key to adjudicating the s. 15 claim. It is incumbent on the Court to understand and account for how the scheme operates as a whole. The only employment statuses provided for under the pension plan are full‑time, part‑time, and leave without pay. There are no specific provisions relating to job‑sharing but the relevant policies define job‑sharing as a form of part‑time work. Job‑sharers are treated as working part‑time during the period in which they job‑share. As they work part‑time hours, they receive part‑time pension benefits for the period they job‑share.

                    The s. 15 test, as it is currently framed, can address claims of adverse‑impact discrimination by its inquiry into whether a law creates a distinction in its impact. At step one of the s. 15 test, it is not necessary to demonstrate that the impugned law or state action has the effect of contributing to an existing disadvantage. As a search for impact is a search for causation, establishing causation is critical. This is particularly so in instances where the state acts in order to address systemic discrimination given that the inquiry at step one is into whether the gap in outcomes is fully explained by pre‑existing disadvantage or whether state conduct has contributed to it. Section 15 is concerned with state conduct that contributes to pre‑existing disadvantage. The state does not have a freestanding positive obligation to remedy social inequalities and it can act incrementally, by putting in place policies that narrow a gap without closing it.

                    Ultimately, the onus is on the claimant to establish causation between the impugned law and the disadvantage. The analysis should not assume that correlation is the function of causation, where it might be the function of independent factors — correlation itself is not proof of causation. Where a law is enacted to incrementally narrow a pre‑existing systemic disadvantage without eradicating it, an element of disparity will necessarily remain. In such cases, it is not enough to refer to a statistical disparity and a broader group disadvantage.

                    The focus at step one on identifying a distinction is consistent with the comparative nature of equality. The two ways in which a distinction can be framed on the enumerated ground of sex in this case are by comparison to full‑time members and by comparison to members who take leave without pay. The comparison to members who take leave without pay is a distinction that is not based on sex because there is no evidence that members taking leave without pay are less likely to be women than members participating in the job‑sharing program. However, the distinction by comparison to full‑time members is a distinction based on sex because members of the job‑sharing program are disproportionately women, whereas uninterrupted full‑time employment is a male pattern of employment. Therefore, the pension plan creates a distinction that, in its impact, is based on sex.

                    Step two of the s. 15 analysis asks whether that distinction is discriminatory in that it fails to respond to the actual capacities and needs of the group and instead imposes burdens or denies a benefit in a manner that has the effect of reinforcing, perpetuating, or exacerbating the disadvantage of women. The analysis at this step must consider whether the unequal impact corresponds with a group’s actual circumstances or needs or whether it is in any other sense substantively discriminatory.

                    To establish substantive discrimination, an element of arbitrariness or unfairness has always been required in the s. 15 analysis. Factors relating to arbitrariness or unfairness must not be confined to the s. 1 analysis. The element of arbitrariness or unfairness has been most often expressed as a failure to respond to individuals’ capacities, needs and circumstances but it does not need to take the form of promoting negative attitudes. It has never been confused with a discriminatory purpose, which is not required to establish substantive discrimination. Substantive discrimination cannot be reduced to historical disadvantage. In some circumstances, laws can maintain significant disadvantage while treating individuals equally and without discrimination. Substantive equality has become so vague that it is impossible for claimants or legislatures to anticipate its demands in advance. Legislatures are instead effectively expected to hit a moving target.

                    It is not arbitrary or unfair and therefore not discriminatory for an employer to prorate compensation, including benefits, according to hours worked when this responds to employees’ actual capacities and circumstances. Employers must be able to compensate employees based on hours worked and offering pension benefits that are prorated to hours worked is not substantive discrimination. In accordance with the contextual analysis of the broader scheme, the provisions on leave without pay remain an important consideration at step two even though the distinction based on members who job‑share compared to those who take leave without pay is not based on sex. Offering pension benefits that are prorated to hours worked does not become substantive discrimination where members who take leave without pay have the right to buy back hours of pension benefits, because the focus of the contextual analysis must be on the actual impact of the law in its full context and must consider each benefit program in full.

                    In the case at bar, the record does not suggest that the lines drawn are inappropriate, having regard to all the circumstances. The pension plan does not represent a source of ongoing systemic disadvantage as it does not contribute to women’s systemic disadvantage; nor does it reinforce, perpetuate, or exacerbate the pre‑existing disadvantage of women in the workplace which arises in part from unequal distribution of parental responsibilities. The impugned provisions of the pension plan represent an example of a government acting incrementally to address inequities that exist in society, using provisions that do not have a discriminatory impact. The legislation is ameliorative in both intent and effect.

                    Given that s. 15(1)  of the Charter  is not infringed, there is no need to consider s. 1

                    Per Côté J. (dissenting): The claim fails at step one of the s. 15(1) analysis because the impugned provisions of the pension plan do not create a distinction on the basis of the enumerated ground of sex. The effect of the impugned provisions of the pension plan is to create a distinction not on the basis of being a woman, that is, sex simpliciter, but on the basis of caregiving responsibilities alone or as a result of a combination of sex with caregiver status.

                    The impugned provisions of the pension plan that discriminate against those with caregiving responsibilities do not necessarily discriminate against women. There is disagreement that, in effect, discrimination on the basis of childcare is tantamount to discrimination on the basis of sex due to their historical association with one another because caregiving status can be separated from sex; rather, same‑sex couples with children, as well as those individuals with elderly caregiving responsibilities will all be disproportionately affected.

                    Disproportionate impact alone is not sufficient to meet step one of the s. 15(1) analysis. Ultimately, in cases of adverse effect discrimination, the question under step one is whether the law, while facially neutral, creates an adverse distinction based on an enumerated or analogous ground. Step one includes a requirement of causation, nexus or tether between the impugned provisions and their effect. It cannot be satisfied in the absence of such a nexus between the impugned law and the disproportionate impact. In cases of adverse effect discrimination, step one is a step at which claimants have more work to do, unlike at s. 1, where the burden is placed on the government. If disproportionate impact alone were sufficient, this would invite statistics‑based litigation which would not be desirable, in part because statistics are constantly shifting.

                    In the present case, the claim is on behalf of women with children, and not simply women. It is critical that the claimants had caregiving responsibilities that made them decide to job‑share. The statistical disparity in results showing that women are disproportionately affected — given that the majority of job‑sharers are women with children — is insufficient to say that step one has been met. There is no reason why job‑sharing is a singularly sex‑based issue: rather, it is a caregiving status issue because job‑sharing is a solution for all members with caregiving responsibilities, not just a solution for those of a certain sex who have children.

                    In light of the conclusion that any distinction depends not on sex but on caregiving responsibility and that the Court has not recognized caregiving, parental, or family status as an analogous ground, in this case, the claimants’ contention must fail at step one of the s. 15(1) analysis. To be sure, the impugned provisions may very well not be rational — there may indeed be no logical reason to deprive job‑sharers of full pension benefits that are guaranteed to full‑time members and members on leave without pay. But it is not the Court’s role to constitutionalize normative judgments to this effect; that is the role of the electorate, and in turn, the legislature. It therefore falls to the legislature, not the courts, to remedy any under‑inclusiveness in this legislation, which purportedly was meant to assist with caregiving responsibilities in the first place.

                    As no distinction can be made out on the basis of sex, there is no need to proceed to the second step of the analysis.

Cases Cited

By Abella J.

                    Referred to: Quebec (Attorney General) v. A, 2013 SCC 5, [2013] 1 S.C.R. 61; Kahkewistahaw First Nation v. Taypotat, 2015 SCC 30, [2015] 2 S.C.R. 548; Quebec (Attorney General) v. Alliance du personnel professionnel et technique de la santé et des services sociaux, 2018 SCC 17, [2018] 1 S.C.R. 464; Centrale des syndicats du Québec v. Quebec (Attorney General), 2018 SCC 18, [2018] 1 S.C.R. 522; Withler v. Canada (Attorney General), 2011 SCC 12, [2011] 1 S.C.R. 396; Griggs v. Duke Power Co., 401 U.S. 424 (1971); British Columbia (Public Service Employee Relations Commission) v. BCGSEU, [1999] 3 S.C.R. 3; Canada (Human Rights Commission) v. Taylor, [1990] 3 S.C.R. 892; Ontario Human Rights Commission v. Simpsons‑Sears Ltd., [1985] 2 S.C.R. 536; Canadian National Railway Co. v. Canada (Canadian Human Rights Commission), [1987] 1 S.C.R. 1114; Andrews v. Law Society of British Columbia, [1989] 1 S.C.R. 143, rev’g (1986), 2 B.C.L.R. (2d) 305; R. v. Kapp, 2008 SCC 41, [2008] 2 S.C.R. 483; Eldridge v. British Columbia (Attorney General), [1997] 3 S.C.R. 624; Vriend v. Alberta, [1998] 1 S.C.R. 493; McKinney v. University of Guelph, [1990] 3 S.C.R. 229; Tétreault‑Gadoury v. Canada (Employment and Immigration Commission), [1991] 2 S.C.R. 22; Law v. Canada (Minister of Employment and Immigration), [1999] 1 S.C.R. 497; Eaton v. Brant County Board of Education, [1997] 1 S.C.R. 241; British Columbia (Superintendent of Motor Vehicles) v. British Columbia (Council of Human Rights), [1999] 3 S.C.R. 868; Moore v. British Columbia (Education), 2012 SCC 61, [2012] 3 S.C.R. 360; Egan v. Canada, [1995] 2 S.C.R. 513; Central Alberta Dairy Pool v. Alberta (Human Rights Commission), [1990] 2 S.C.R. 489; R. v. Spence, 2005 SCC 71, [2005] 3 S.C.R. 458; Homer v. Chief Constable of West Yorkshire Police, [2012] UKSC 15, [2012] 3 All E.R. 1287; Vancouver Area Network of Drug Users v. Downtown Vancouver Business Improvement Association, 2018 BCCA 132, 10 B.C.L.R. (6th) 175; Gaz métropolitain inc. v. Commission des droits de la personne et des droits de la jeunesse, 2011 QCCA 1201; Oršuš v. Croatia, No. 15766/03, ECHR 2010‑II; Essop v. Home Office (U.K. Border Agency), [2017] UKSC 27, [2017] 3 All E.R. 551; O’Connor v. Bar Standards Board, [2017] UKSC 78, [2018] 2 All E.R. 779; D.H. v. the Czech Republic, No. 57325/00, ECHR 2007‑IV; Horváth and Kiss v. Hungary, [2013] E.L.R. 102; R. v. Turpin, [1989] 1 S.C.R. 1296; Brooks v. Canada Safeway Ltd., [1989] 1 S.C.R. 1219; Janzen v. Platy Enterprises Ltd., [1989] 1 S.C.R. 1252; Nova Scotia (Workers’ Compensation Board) v. Martin, 2003 SCC 54, [2003] 2 S.C.R. 504; Miron v. Trudel, [1995] 2 S.C.R. 418; Lavoie v. Canada, 2002 SCC 23, [2002] 1 S.C.R. 769; Nova Scotia (Attorney General) v. Walsh, 2002 SCC 83, [2002] 4 S.C.R. 325; Moge v. Moge, [1992] 3 S.C.R. 813; New Brunswick (Minister of Health and Community Services) v. G. (J.), [1999] 3 S.C.R. 46; Symes v. Canada, [1993] 4 S.C.R. 695; Young v. Young, [1993] 4 S.C.R. 3; Jenkins v. Kingsgate (Clothing Productions) Ltd, C‑96/80, [1981] E.C.R. I‑911; Bilka‑Kaufhaus GmbH v. Weber von Hartz, C‑170/84, [1986] E.C.R. I‑1607; Rinner‑Kühn v. FWW Spezial‑Gebäudereinigung GmbH, C‑171/88, [1989] E.C.R. I‑2743; Vroege v. NCIV Instituut voor Volkshuisvesting BV, C‑57/93, [1994] E.C.R. I‑4541; Schönheit v. Stadt Frankfurt am Main, C‑4/02 and C‑5/02, [2003] E.C.R. I‑12575; Reg. v. Secretary of State for Employment, Ex parte Equal Opportunities Commission, [1995] 1 A.C. 1; London Underground Ltd. v. Edwards (No. 2), [1999] I.C.R. 494; Phillips v. Martin Marietta Corp., 400 U.S. 542 (1971); Bostock v. Clayton County, Georgia, 140 S. Ct. 1731 (2020); Campbell River & North Island Transition Society v. Health Sciences Assn. of British Columbia, 2004 BCCA 260, 28 B.C.L.R. (4th) 292; Brown v. Department of National Revenue (1993), 93 CLLC ¶17,013; Canada (Attorney General) v. Johnstone, 2014 FCA 110, [2015] 2 F.C.R. 595; Misetich v. Value Village Stores Inc., 2016 HRTO 1229, 39 C.C.E.L. (4th) 129; Thibaudeau v. Canada, [1995] 2 S.C.R. 627; Canada (Attorney General) v. Lesiuk, 2003 FCA 3, [2003] 2 F.C. 697; R. v. Oakes, [1986] 1 S.C.R. 103; Vancouver (City) v. Ward, 2010 SCC 27, [2010] 2 S.C.R. 28; Doucet‑Boudreau v. Nova Scotia (Minister of Education), 2003 SCC 62, [2003] 3 S.C.R. 3.

By Brown and Rowe JJ. (dissenting)

                    Quebec (Attorney General) v. Canada (Attorney General), 2015 SCC 14, [2015] 1 S.C.R. 693; Quebec (Attorney General) v. Alliance du personnel professionnel et technique de la santé et des services sociaux, 2018 SCC 17, [2018] 1 S.C.R. 464; Withler v. Canada (Attorney General), 2011 SCC 12, [2011] 1 S.C.R. 396; Vriend v. Alberta, [1998] 1 S.C.R. 493; RWDSU v. Dolphin Delivery Ltd., [1986] 2 S.C.R. 573; Centrale des syndicats du Québec v. Quebec (Attorney General), 2018 SCC 18, [2018] 1 S.C.R. 522; Public Service Alliance of Canada v. Canada (Department of National Defence), [1996] 3 F.C. 789; Newfoundland (Treasury Board) v. N.A.P.E., 2004 SCC 66, [2004] 3 S.C.R. 381; Kahkewistahaw First Nation v. Taypotat, 2015 SCC 30, [2015] 2 S.C.R. 548; Andrews v. Law Society of British Columbia, [1989] 1 S.C.R. 143; R. v. Kapp, 2008 SCC 41, [2008] 2 S.C.R. 483; Winko v. British Columbia (Forensic Psychiatric Institute), [1999] 2 S.C.R. 625; Quebec (Attorney General) v. A, 2013 SCC 5, [2013] 1 S.C.R. 61; Snell v. Farrell, [1990] 2 S.C.R. 311; Blatch v. Archer (1774), 1 Cowp. 63, 98 E.R. 969; Auton (Guardian ad litem of) v. British Columbia (Attorney General), 2004 SCC 78, [2004] 3 S.C.R. 657; Gosselin v. Quebec (Attorney General), 2002 SCC 84, [2002] 4 S.C.R. 429; Corbiere v. Canada (Minister of Indian and Northern Affairs), [1999] 2 S.C.R. 203; Hodge v. Canada (Minister of Human Resources Development), 2004 SCC 65, [2004] 3 S.C.R. 357; Law v. Canada (Minister of Employment and Immigration), [1999] 1 S.C.R. 497; IBM Canada Limited v. Waterman, 2013 SCC 70, [2013] 3 S.C.R. 985; Parry v. Cleaver, [1970] A.C. 1; Egan v. Canada, [1995] 2 S.C.R. 513; R. v. Edwards Books and Art Ltd., [1986] 2 S.C.R. 713; Williamson v. Lee Optical of Oklahoma, 348 U.S. 483 (1955); McKinney v. University of Guelph, [1990] 3 S.C.R. 229; Schachter v. Canada, [1992] 2 S.C.R. 679; Alberta (Aboriginal Affairs and Northern Development) v. Cunningham, 2011 SCC 37, [2011] 2 S.C.R. 670; Thibaudeau v. Canada, [1995] 2 S.C.R. 627; Lovelace v. Ontario, 2000 SCC 37, [2000] 1 S.C.R. 950; Ferrel v. Ontario (Attorney General) (1998), 42 O.R. (3d) 97; British Columbia (Public Service Employee Relations Commission) v. BCGSEU, [1999] 3 S.C.R. 3; Baldwin v. Missouri, 281 U.S. 586 (1930); R. v. Nova Scotia Pharmaceutical Society, [1992] 2 S.C.R. 606; Irwin Toy Ltd. v. Quebec (Attorney General), [1989] 1 S.C.R. 927; Law Society of British Columbia v. Trinity Western University, 2018 SCC 32, [2018] 2 S.C.R. 293.

By Côté J. (dissenting)

                    Public Service Alliance of Canada v. Canada (Department of National Defence), [1996] 3 F.C. 789; Quebec (Attorney General) v. Alliance du personnel professionnel et technique de la santé et des services sociaux, 2018 SCC 17, [2018] 1 S.C.R. 464; Gosselin v. Quebec (Attorney General), 2002 SCC 84, [2002] 4 S.C.R. 429; Andrews v. Law Society of British Columbia, [1989] 1 S.C.R. 143; R. v. Kapp, 2008 SCC 41, [2008] 2 S.C.R. 483; Withler v. Canada (Attorney General), 2011 SCC 12, [2011] 1 S.C.R. 396; Quebec (Attorney General) v. A, 2013 SCC 5, [2013] 1 S.C.R. 61; Kahkewistahaw First Nation v. Taypotat, 2015 SCC 30, [2015] 2 S.C.R. 548; Centrale des syndicats du Québec v. Quebec (Attorney General), 2018 SCC 18, [2018] 1 S.C.R. 522; Brooks v. Canada Safeway Ltd., [1989] 1 S.C.R. 1219; D.H. v. the Czech Republic, No. 57325/00, ECHR 2007‑IV; Griggs v. Duke Power Co., 401 U.S. 424 (1971); Washington, Mayor of Washington, D.C. v. Davis, 426 U.S. 229 (1976); Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988); Texas Department of Housing and Community Affairs v. Inclusive Communities Project, Inc., 135 S. Ct. 2507 (2015); Ricci v. DeStefano, 557 U.S. 557 (2009); Smith v. City of Jackson, Mississippi, 544 U.S. 228 (2005); Symes v. Canada, [1993] 4 S.C.R. 695; Miceli‑Riggins v. Canada (Attorney General), 2013 FCA 158, [2014] 4 F.C.R. 709; Grenon v. Minister of National Revenue, 2016 FCA 4, 482 N.R. 310; Canada (Attorney General) v. Lesiuk, 2003 FCA 3, [2003] 2 F.C. 697; Begum v. Canada (Citizenship and Immigration), 2018 FCA 181, [2019] 2 F.C.R. 488.

Statutes and Regulations Cited

Canadian Charter of Rights and Freedoms , ss. 1 , 3 , 14 , 15 , 20 , 23 , 32 .

Income Tax Act , R.S.C. 1985, c. 1 (5th Supp .).

Income Tax Regulations, C.R.C., c. 945.

Royal Canadian Mounted Police Superannuation Act , R.S.C. 1985, c. R‑11, ss. 5 , 6 , 7(1) , 29.2 .

Royal Canadian Mounted Police Superannuation Regulations, C.R.C., c. 1393, ss. 2.1, 5.2(1), 5.4, 10(1), (4), 10.1, 10.8.

Treaties and Other International Instruments

Beijing Declaration and Platform for Action, U.N. Doc. A/CONF.177/20, October 17, 1995.

Authors Cited

Bingham, Tom. The Rule of Law. London: Allen Lane, 2010.

Bisom‑Rapp, Susan. “What We Know About Equal Employment Opportunity Law After Fifty Years of Trying” (2018), 22 Employee Rts. & Employment Pol’y J. 337.

Bisom‑Rapp, Susan, and Malcolm Sargeant. “It’s Complicated: Age, Gender, and Lifetime Discrimination Against Working Women — The United States and the U.K. as Examples” (2014), 22 Elder L.J. 1.

Braun, Evelyn. “Adverse Effect Discrimination: Proving the Prima Facie Case” (2005), 11 Rev. Const. Stud. 119.

British Columbia Law Institute. Human Rights and Family Responsibilities: Family Status Discrimination under Human Rights Law in British Columbia and Canada. Vancouver, 2012.

Canada. Commission of Inquiry into Part‑time Work. Part‑time Work in Canada: Report of the Commission of Inquiry into Part‑time Work. Ottawa: Labour Canada, 1983.

Canada. House of Commons. House of Commons Debates, vol. VI, 3rd Sess., 34th Parl., February 24, 1992, p. 7487.

Canada. Royal Commission on Equality in Employment. Report of the Commission on Equality in Employment. Ottawa: Supply and Services Canada, 1984.

Canada. Royal Commission on the Status of Women in Canada. Report of the Royal Commission on the Status of Women in Canada. Ottawa, 1970.

Canada. Statistics Canada. Table 11-10-0055-01 — High income tax filers in Canada (online: https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1110005501; archived version: https://www.scc-csc.ca/cso-dce/2020SCC-CSC28_1_eng.pdf).

Canada. Statistics Canada. Women in Canada: A Gender‑based Statistical Report, 7th ed. by Melissa Moyser. Ottawa, 2017.

Canadian Institute for Health Information. Nursing in Canada, 2019: A Lens on Supply and Workforce. Ottawa, 2020.

Chaykowski, Richard P., and Lisa M. Powell. “Women and the Labour Market: Recent Trends and Policy Issues” (1999), 25 Can. Pub. Pol’y S1.

Collins, Hugh, and Tarunabh Khaitan. “Indirect Discrimination Law: Controversies and Critical Questions”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law. Portland, Or.: Hart Publishing, 2018, 1.

Craig, Elaine. “Family as Status in Doe v. Canada: Constituting Family Under Section 15 of the Charter ” (2007), 20 N.J.C.L. 197.

Demers, Diane L. “La discrimination systémique: variation sur un concept unique” (1993), 8 C.J.L.S. 83.

Eberts, Mary, and Kim Stanton. “The Disappearance of the Four Equality Rights and Systemic Discrimination from Canadian Equality Jurisprudence” (2018), 38 N.J.C.L. 89.

Eisen, Jessica. “Grounding Equality in Social Relations: Suspect Classification, Analogous Grounds and Relational Theory” (2017), 42 Queen’s L.J. 41.

Faraday, Fay. “One Step Forward, Two Steps Back? Substantive Equality, Systemic Discrimination and Pay Equity at the Supreme Court of Canada” (2020), 94 S.C.L.R. (2d) 301.

Fredman, Sandra. “Direct and Indirect Discrimination: Is There Still a Divide?”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law. Portland, Or.: Hart Publishing, 2018, 31.

Fredman, Sandra. Discrimination Law, 2nd ed. New York: Oxford University Press, 2011.

Fredman, Sandra. “The Reason Why: Unravelling Indirect Discrimination” (2016), 45 Indus. L.J. 231.

Gopnik, Adam. “The illiberal imagination: Are liberals on the wrong side of history?”, The New Yorker, March 20, 2017, 88.

Hogg, Peter W. Constitutional Law of Canada, vol. 2, 5th ed. Supp. Toronto: Thomson Reuters, 2019 (loose‑leaf updated 2019, release 1).

Hughes, Patricia. “Supreme Court of Canada Equality Jurisprudence and ‘Everyday Life’” (2012), 58 S.C.L.R. (2d) 245.

International Labour Organization. Rights, jobs and social security: New visions for older women and men. Geneva, 2008.

Kanee, Lyle, and Adam Cembrowski. “Family Status Discrimination and the Obligation to Self‑Accommodate” (2018), 14 J.L. & Equality 61.

Kaplan, Ari, and Mitch Frazer. Pension Law, 2nd ed. Toronto: Irwin Law, 2013.

Khaitan, Tarunabh. A Theory of Discrimination Law. New York: Oxford University Press, 2015.

Khaitan, Tarunabh, and Sandy Steel. “Wrongs, Group Disadvantage and the Legitimacy of Indirect Discrimination Law”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law. Portland, Or.: Hart Publishing, 2018, 197.

Koshan, Jennifer, and Jonnette Watson Hamilton. “Meaningless Mantra: Substantive Equality after Withler” (2011), 16 Rev. Const. Stud. 31.

Law Commission of Ontario. Vulnerable Workers and Precarious Work. Toronto, 2012.

Lawrence, Sonia. “Choice, Equality and Tales of Racial Discrimination: Reading the Supreme Court on Section 15”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms. Markham, Ont.: LexisNexis, 2006, 115.

Macpherson, Suzi. “Reconciling employment and family care‑giving: a gender analysis of current challenges and future directions for UK policy”, in Nicole Busby and Grace James, eds., Families, Care‑giving and Paid Work: Challenging Labour Law in the 21st Century. Northampton, Mass.: Edward Elgar Publishing, 2011, 13.

Majury, Diana. “Women Are Themselves to Blame: Choice as a Justification for Unequal Treatment”, in Fay Faraday, Margaret Denike and M. Kate Stephenson, eds., Making Equality Rights Real: Securing Substantive Equality under the Charter. Toronto: Irwin Law, 2006, 209.

McIntyre, Sheila. “Deference and Dominance: Equality Without Substance”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms. Markham, Ont.: LexisNexis, 2006, 95.

Miller, Bradley W. “Justification and Rights Limitations”, in Grant Huscroft, ed., Expounding the Constitution: Essays in Constitutional Theory. New York: Cambridge University Press, 2010, 93.

Miller, Bradley W. “Majoritarianism and Pathologies of Judicial Review”, in Grégoire Webber et al., Legislated Rights: Securing Human Rights through Legislation. New York: Cambridge University Press, 2018, 181.

Mize, Selene. “Indirect Discrimination Reconsidered” (2007), N.Z.L. Rev. 27.

Monahan, Patrick J., Byron Shaw and Padraic Ryan. Constitutional Law, 5th ed. Toronto: Irwin Law, 2017.

Moran, Mayo. “Protesting Too Much: Rational Basis Review Under Canada’s Equality Guarantee”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms. Markham, Ont.: LexisNexis, 2006, 71.

Moreau, Sophia. “The Moral Seriousness of Indirect Discrimination”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law. Portland, Or.: Hart Publishing, 2018, 123.

Moreau, Sophia. “What Is Discrimination?” (2010), 38 Philosophy & Public Affairs 143.

National Action Committee on the Status of Women. Brief on Equality for Women in Pensions, Taxation and Federal Benefits to Parents. Toronto, 1985.

Ontario. Gender Wage Gap Strategy Steering Committee. Final Report and Recommendations of the Gender Wage Gap Strategy Steering Committee. Toronto, 2016.

Ontario. Ontario Human Rights Commission. The Cost of Caring: Report on the Consultation on Discrimination on the Basis of Family Status. Toronto, 2006.

Ontario. Royal Commission on the Status of Pensions in Ontario. Report of the Royal Commission on the Status of Pensions in Ontario. Toronto, 1980.

Osborne‑Brown, Sheila. “Discrimination and Family Status: The Test, the Continuing Debate, and the Accommodation Conversation” (2018), 14 J.L. & Equality 87.

Philipps, Lisa, and Margot Young. “Sex, Tax and the Charter : A Review of Thibaudeau v. Canada” (1995), 2 Rev. Const. Stud. 221.

Pothier, Dianne. “Connecting Grounds of Discrimination to Real People’s Real Experiences” (2001), 13 C.J.W.L. 37.

Pothier, Dianne. “M’Aider, Mayday: Section 15  of the Charter  in Distress” (1996), 6 N.J.C.L. 295.

Pothier, Dianne. “Tackling Disability Discrimination at Work: Toward a Systemic Approach” (2010), 4 McGill J.L. & Health 17.

Puchta, Alicja. “Quebec v A and Taypotat: Unpacking the Supreme Court’s Latest Decisions on Section 15  of the Charter ” (2018), 55 Osgoode Hall L.J. 665.

Réaume, Denise G. “Harm and Fault in Discrimination Law: The Transition from Intentional to Adverse Effect Discrimination” (2001), 2 Theor. Inq. L. 349.

Rivet, Michèle, et Anne‑Marie Santorineos. “Juger à l’ère des droits fondamentaux” (2012), 42 R.D.U.S. 363.

Sangiuliano, Anthony Robert. “Substantive Equality As Equal Recognition: A New Theory of Section 15 of the Charter ” (2015), 52 Osgoode Hall L.J. 601.

Sealy‑Harrington, Joshua. “Assessing Analogous Grounds: The Doctrinal and Normative Superiority of a Multi‑Variable Approach” (2013), 10 J.L. & Equality 37.

Sharpe, Robert J. Good Judgment: Making Judicial Decisions. Toronto: University of Toronto Press, 2018.

Sheppard, Colleen. “Grounds of Discrimination: Towards an Inclusive and Contextual Approach” (2001), 80 Can. Bar Rev. 893.

Sheppard, Colleen. Inclusive Equality: The Relational Dimensions of Systemic Discrimination in Canada. Montréal: McGill‑Queen’s University Press, 2010.

Sheppard, Colleen. “Mapping anti‑discrimination law onto inequality at work: Expanding the meaning of equality in international labour law” (2012), 151 Int’l Lab. Rev. 1.

Sheppard, Colleen. “Of Forest Fires and Systemic Discrimination: A Review of British Columbia (Public Service Employee Relations Commission) v. B.C.G.S.E.U.” (2001), 46 McGill L.J. 533.

Shilton, Elizabeth. “Family Status Discrimination: ‘Disruption and Great Mischief’ or Bridge over the Work‑Family Divide?” (2018), 14 J.L. & Equality 33.

Shilton, Elizabeth. “Gender Risk and Employment Pension Plans in Canada” (2013), 17 C.L.E.L.J. 101.

Vizkelety, Béatrice. Proving Discrimination in Canada. Toronto: Carswell, 1987.

Watson Hamilton, Jonnette, and Jennifer Koshan. “Adverse Impact: The Supreme Court’s Approach to Adverse Effects Discrimination under Section 15  of the Charter ” (2015), 19 Rev. Const. Stud. 191.

Watson Hamilton, Jonnette, and Jennifer Koshan. “Equality Rights and Pay Equity: Déjà Vu in the Supreme Court of Canada” (2019), 15 J.L. & Equality 1.

Watson Hamilton, Jonnette, and Jennifer Koshan. “Kahkewistahaw First Nation v. Taypotat: An Arbitrary Approach to Discrimination” (2016), 76 S.C.L.R. (2d) 243.

Young, Claire. “Pensions, Privatization, and Poverty: The Gendered Impact” (2011), 23 C.J.W.L. 661.

Young, Margot. “Blissed Out: Section 15 at Twenty”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms. Markham, Ont.: LexisNexis, 2006, 45.

Young, Margot. “Unequal to the Task: ‘Kapp’ing the Substantive Potential of Section 15” (2010), 50 S.C.L.R. (2d) 183.

Yowell, Paul. Constitutional Rights and Constitutional Design: Moral and Empirical Reasoning in Judicial Review. Portland, Or.: Hart Publishing, 2018.

                    APPEAL from a judgment of the Federal Court of Appeal (Gauthier, Gleason and Woods JJ.A.), 2018 FCA 223, [2019] 2 F.C.R. 541, 426 C.R.R. (2d) 190, 44 C.C.P.B. (2nd) 167, 2019 CLLC ¶230‑018, [2018] F.C.J. No. 1228 (QL), 2018 CarswellNat 7614 (WL Can.), affirming a decision of Kane J., 2017 FC 557, [2017] F.C.J. No. 609 (QL), 2017 CarswellNat 2726 (WL Can.). Appeal allowed, Côté, Brown and Rowe JJ. dissenting.

                    Paul Champ and Bijon Roy, for the appellants.

                    Christopher M. Rupar, Zoe Oxaal and Gregory Tzemenakis, for the respondent.

                    Rochelle S. Fox and Yashoda Ranganathan, for the intervener the Attorney General of Ontario.

                    Amélie Pelletier Desrosiers and Catheryne Bélanger, for the intervener the Attorney General of Quebec.

                    Danielle Bisnar, Kate Hughes and Janet Borowy, for the intervener the Women’s Legal Education and Action Fund Inc.

                    Andrew Astritis and Morgan Rowe, for the intervener the Public Service Alliance of Canada.

                    Christopher Rootham and Andrew Montague‑Reinholdt, for the intervener the National Police Federation.

The judgment of Wagner C.J. and Abella, Moldaver, Karakatsanis, Martin and Kasirer JJ. was delivered by

[1]                             Abella J. — In 1970, the Royal Commission on the Status of Women in Canada set out a galvanic blueprint for redressing the legal, economic, social and political barriers to full and fair participation faced by Canadian women for generations. Many of the inequities it identified have been spectacularly reversed, and the result has been enormous progressive change for women in this country. But despite the sweep of legislative initiatives and the positive realignment of many social expectations, the long reach of entrenched assumptions about the role of women in a family continues to leave its mark on what happens in the workplace.

[2]                             One of the ways it does so is in how women are remunerated generally; the corollary is how they are remunerated when they seek to combine work with family responsibilities by working part‑time. As the Royal Commission noted, “ways must be found to provide [part-time] employees with pay and working conditions no less equitable than those provided for the full-time worker” (Report of the Royal Commission on the Status of Women in Canada, at p. 105). Fifty years later, this appeal raises that very issue.

[3]                             Members of the Royal Canadian Mounted Police (“RCMP”) receive benefits upon retirement from a pension plan. Greater benefits are provided to members with a record of high pay and long, uninterrupted full‑time service. Certain gaps in a member’s record of service — such as being suspended or time spent on unpaid leave — can be filled in through a “buy back” process, leaving the member’s pension benefits unaffected. No such choice is available to full‑time members who temporarily reduced their working hours under a job-sharing agreement. Nearly all of the participants in the job-sharing program are women and most of them reduced their hours of work because of child care.

[4]                             Three retired members of the RCMP claim that the pension consequences of job‑sharing have a discriminatory impact on women and violate s. 15(1)  of the Canadian Charter of Rights and Freedoms . Their claim failed at the Federal Court (2017 FC 557). The application judge concluded that job-sharing is not disadvantageous when compared to unpaid leave and, even if it is, that any such disadvantage is the result of an individual employee’s choice to job-share, not her gender or family status. The Federal Court of Appeal upheld the application judge’s decision ([2019] 2 F.C.R. 541).

[5]                             I would allow the appeal. Full-time RCMP members who job‑share must sacrifice pension benefits because of a temporary reduction in working hours. This arrangement has a disproportionate impact on women and perpetuates their historical disadvantage. It is a clear violation of their right to equality under s. 15(1)  of the Charter .

Background

[6]                             Ms. Fraser, Ms. Pilgrim and Ms. Fox served as police officers in the RCMP for over 25 years. Ms. Fraser was posted to Fort Saskatchewan, Alberta, where she worked rotating 10‑hour shifts, 7 days a week. Ms. Pilgrim worked in the Commercial Crime Unit in Quebec City. Ms. Fox began her policing work in Toronto before being transferred to a small community in central Newfoundland.

[7]                             Ms. Fraser, Ms. Pilgrim and Ms. Fox took maternity leaves in the early‑to‑mid 1990s. Upon returning to full‑time service, they experienced difficulties combining their work obligations with caring for their children. Ms. Fraser described feeling “overwhelmed” as she tried to balance work and family; Ms. Pilgrim felt like she was “on a treadmill”; and Ms. Fox described the experience as “hell on earth”. These difficulties caused Ms. Fox to retire from the RCMP in 1994 and resulted in Ms. Fraser taking unpaid leave in 1997. At the time, the RCMP did not permit regular members to work part-time.

[8]                             In December 1997, the RCMP introduced a job-sharing program to provide members with an alternative to taking leave without pay. Under the job-sharing program, two or three RCMP members could split the duties and responsibilities of one full-time position, which allowed each participant to work fewer hours than a full‑time employee. Parties to a job‑sharing agreement could be asked, on one month’s notice, to resume full-time work based on administrative or operational needs.

[9]                             Job-sharing was meant to be “mutually beneficial” for the RCMP and participating members. Participants were able to “remain operationally connected to the Force while having a work schedule that better accommodated their individual circumstances” (A.R., vol. V, at p. 810). The RCMP benefitted from the participants’ services, which helped, among other things, in addressing staff shortages in smaller communities and in emergency situations.

[10]                         Ms. Fraser, Ms. Fox and Ms. Pilgrim enrolled in the job‑sharing program along with 137 other RCMP members between 1997 and 2011. Most participants were women with children. From 2010 to 2014, all RCMP members who job-shared were women, and most of them cited childcare as their reason for joining the program.

[11]                         After enrolling in the job‑sharing program, Ms. Fraser, Ms. Fox and Ms. Pilgrim became aware that their participation would have consequences for their pensions. Understanding those consequences requires a brief review of the RCMP’s pension plan.

[12]                         All RCMP members engaged to work at least 12 hours a week must enrol in and contribute to a statutory pension plan.[1] Upon retirement, members receive benefits based on, among other things, their years of service.[2] One year of work counts as one year of pensionable service. More years of pensionable service lead to higher pension benefits.

[13]                         Years of full-time work and part‑time work are treated differently when pension benefits are calculated. Part-time work is pro‑rated to reflect the lower number of hours worked. It is therefore less valuable than full‑time work in the formula used to calculate pension benefits.

[14]                         RCMP members acquire full‑time pension credit for periods of service in which they were engaged to work 40 hours a week. Members can also treat certain gaps in full-time service, such as leave without pay, as fully pensionable. Upon returning from unpaid leave, a member can “buy back” the service they missed by making the contributions that both she and the RCMP would have made had she been actively employed. This increases the member’s years of full‑time pensionable service, which results in a more valuable pension.

[15]                         Ms. Fraser, Ms. Fox and Ms. Pilgrim expected that job‑sharing, like leave without pay, would be eligible for full pension credit. Both situations, they noted, involve a temporary interruption in regular service for full‑time members — a decrease to between 12 and less than 40 hours of work a week when job-sharing, and to 0 hours a week when on unpaid leave. It was logical, in their view, that members in both situations would be allowed to “buy back” their lost service and associated pension benefits.

[16]                         The RCMP initially accepted this position in communications with Ms. Pilgrim. Subsequently, however, the RCMP took the position that job-sharing was part‑time work for which participants could not receive full-time pension credit.

[17]                         When informed that they would not be able to purchase full‑time pension credit for their job‑sharing service, participants in the program raised concerns with senior management. In a memo to the RCMP’s commissioner, 14 female members from across Canada explained why they considered the pension consequences of job‑sharing to be illogical and unfair:

Members returning to full‑time work from maternity leave, LWOP [Leave without Pay], SFLWOP [Self-funded Leave without Pay], and Disciplinary Actions (Suspended without Pay) are given the opportunity to buy back their pension benefits. Members returning to work from extended periods of ODS [Off Duty Sick] and Disciplinary Actions (Suspended with Pay) maintain their pension benefits, despite not working. Members who have departed the Force and are subsequently re‑engaged are able to buy back their pension. . . . Thus, it would seem logical that members returning to full‑time work from job‑share arrangements would be entitled to the same opportunity to purchase pension benefits.

 

. . .

 

Job‑sharing is a progressive, proactive and innovative step for the RCMP. It is time to support members who choose to job-share, rather than penalizing them for choosing an option the Force has made available. It is important for management to remember that it is not only the member who benefits from job-sharing, but also the RCMP. Job-sharing allows the Force to retain its investment in human resources; members with training, skills and seniority. It provides a pool of trained people who can be called on in emergency situations. Members who job-share stay current with changing technology, legislation and training, among other things, because they are still working. Why is the RCMP penalizing those who choose to job-share when it stands to benefit from the arrangement?

[18]                         The RCMP’s then-assistant commissioner, G. J. Loeppky, responded to the memo and acknowledged that there “may be an element of unfairness” to the RCMP’s approach. He presented the matter to the RCMP Pension Advisory Committee, which retained an actuary to provide advice on available options. The actuary acknowledged that the RCMP’s pension plan could be amended under the Income Tax Act  and Income Tax Regulations[3] to extend pension buy-back rights to participants in the job-sharing program. The actuary noted that the flexibility under the Income Tax Regulations “is particularly useful in responding to employee requests for reduced work-hours at various stages of their family life or career” (p. 459).

[19]                         While this process was ongoing, three female RCMP members filed internal grievances challenging the denial of their requests to buy back full‑time pension credit for their job-sharing service. The RCMP External Review Committee found in their favour. The Committee saw no legal barriers to the RCMP’s defining job‑sharing as a combination of hours worked and a period of leave without pay. The Committee cited a similar Treasury Board policy about the working hours of certain public service employees on the verge of retirement:

. . . there was a precedent for such a categorization. In 1999, the Treasury Board instituted a program of pre‑retirement transition leave by introducing the Pre‑retirement Transition Leave Policy. This policy allowed certain Public Service employees close to retirement to reduce their hours of work by up to 40%. Their pay was reduced accordingly, but the hours not worked were treated as LWOP with respect to pay, deductions, allowances, other leave, benefits, and pensions.

[20]                         The RCMP’s Acting Commissioner, William Sweeney, did not follow the External Review Committee’s recommendations and dismissed the grievances. In his view, it was not legally possible for job‑sharing to be defined as a combination of full‑time work and leave without pay. Although “immensely sympathetic” to the grievances, he concluded that the classification of job‑sharing as part-time work was not discriminatory.

[21]                         After the Commissioner’s decision, Ms. Fraser, Ms. Fox and Ms. Pilgrim started this Charter  application. They advanced two submissions. First, they argued that the pension plan, properly interpreted, allows participants in the job‑sharing program to acquire full pension credit. Second, if this was not possible, they argued that the pension plan violates s. 15(1)  of the Charter  because it prevents women with children — the majority of participants in the job-sharing program — from contributing to their pensions in the same way as members who work full-time or take leave without pay. In support of their application, Ms. Fraser, Ms. Fox and Ms. Pilgrim filed expert evidence and other material addressing the disadvantages women with children face in the labour force.

[22]                         The application judge found that job-sharing is part‑time work for which participants cannot obtain full‑time pension credit. This outcome, in her view, did not violate s. 15(1) because there was insufficient evidence that job-sharing was disadvantageous compared to unpaid leave. Even assuming that there were negative consequences to job-sharing, these outcomes were the result of a participant’s choice to job-share. The Charter application was therefore dismissed.

[23]                         An appeal to the Federal Court of Appeal was unsuccessful. The court held that job-sharing RCMP members did not receive inferior compensation to members on leave without pay, and that any adverse impact on job-sharing participants flowed from their choice to work part-time, not from the pension plan.

[24]                         For the reasons that follow, I would allow the appeal.

Analysis

[25]                         Unlike full‑time members who work regular hours,[4] who are suspended or who take unpaid leave, full‑time RCMP members who job-share are classified as part‑time workers under the Regulations and cannot, under the terms of the pension plan, obtain full‑time pension credit for their service. Ms. Fraser and her colleagues submit that this limitation violates s. 15(1)  of the Charter  on the basis of sex and, alternatively, on the basis of family/parental status.

[26]                         Section 15(1)  of the Charter  states:

15. (1) Every individual is equal before and under the law and has the right to the equal protection and equal benefit of the law without discrimination and, in particular, without discrimination based on race, national or ethnic origin, colour, religion, sex, age or mental or physical disability.

[27]                          Section 15(1) reflects a profound commitment to promote equality and prevent discrimination against disadvantaged groups (Quebec (Attorney General) v. A, [2013] 1 S.C.R. 61, at para. 332; Kahkewistahaw First Nation v. Taypotat, [2015] 2 S.C.R. 548, at paras. 19‑20). To prove a prima facie violation of s. 15(1), a claimant must demonstrate that the impugned law or state action:

•            on its face or in its impact, creates a distinction based on enumerated or analogous grounds; and

•            imposes burdens or denies a benefit in a manner that has the effect of reinforcing, perpetuating, or exacerbating disadvantage.

(Quebec (Attorney General) v. Alliance du personnel professionnel et technique de la santé et des services sociaux, [2018] 1 S.C.R. 464, at para. 25; Centrale des syndicats du Québec v. Quebec (Attorney General), [2018] 1 S.C.R. 522, at para. 22.)

[28]                         Ms. Fraser does not suggest that the negative pension consequences of job‑sharing are explicitly based on sex. Rather, she claims that they have an adverse impact on women with children.

[29]                          How adverse impact or systemic discrimination is applied has received extensive academic consideration (see, for example, Colleen Sheppard, Inclusive Equality: The Relational Dimensions of Systemic Discrimination in Canada (2010), at pp. 19‑21; Evelyn Braun, “Adverse Effect Discrimination: Proving the Prima Facie Case” (2005), 11 Rev. Const. Stud. 119; Jonnette Watson Hamilton and Jennifer Koshan, “Adverse Impact: The Supreme Court’s Approach to Adverse Effects Discrimination under Section 15  of the Charter ” (2015), 19 Rev. Const. Stud. 191; Michèle Rivet and Anne‑Marie Santorineos, “Juger à l’ère des droits fondamentaux” (2012), 42 R.D.U.S. 363, at p. 374; Diane L. Demers, “La discrimination systémique: variation sur un concept unique” (1993), 8 C.J.L.S. 83; Lisa Philipps and Margot Young, “Sex, Tax and the Charter : A Review of Thibaudeau v. Canada” (1995), 2 Rev. Const. Stud. 221). As Prof. Colleen Sheppard notes:

Why is it so critical to expand on our understanding of adverse effect discrimination? If we do not, there is a significant risk that discrimination embedded in apparently neutral institutional policies, rules, or procedures will not be recognized as discriminatory. This risk is accentuated by the necessity in anti‑discrimination law to connect the experience of exclusion, harm, prejudice, or disadvantage to a recognized ground of discrimination. . . . We need a sophisticated and coherent theory of adverse effect discrimination to assist claimants, lawyers, and adjudicators with the complexities of the manifestations of systemic discrimination.

(“Of Forest Fires and Systemic Discrimination: A Review of British Columbia (Public Service Employee Relations Commission) v. B.C.G.S.E.U.” (2001), 46 McGill L.J. 533, at p. 542; see also Braun, at p. 122)

[30]                          It is helpful to start by defining the concept. Adverse impact discrimination occurs when a seemingly neutral law has a disproportionate impact on members of groups protected on the basis of an enumerated or analogous ground (see Watson Hamilton and Koshan (2015), at p. 196; Sheppard (2001), at p. 549; see also Withler v. Canada (Attorney General), [2011] 1 S.C.R. 396, at para. 64; Taypotat, at para. 22). Instead of explicitly singling out those who are in the protected groups for differential treatment, the law indirectly places them at a disadvantage (Sophia Moreau, “What Is Discrimination?” (2010), 38 Philosophy & Public Affairs 143, at p. 155).

[31]                          Increased awareness of adverse impact discrimination has been a “central trend in the development of discrimination law”, marking a shift away from a fault‑based conception of discrimination towards an effects‑based model which critically examines systems, structures, and their impact on disadvantaged groups (Denise G. Réaume, “Harm and Fault in Discrimination Law: The Transition from Intentional to Adverse Effect Discrimination” (2001), 2 Theor. Inq. L. 349, at pp. 350‑51; see also Béatrice Vizkelety, Proving Discrimination in Canada (1987), at p. 18; Sheppard (2010), at pp. 19‑20). Accompanying this shift was the recognition that discrimination is “frequently a product of continuing to do things ‘the way they have always been done’”, and that governments must be “particularly vigilant about the effects of their own policies” on members of disadvantaged groups (Fay Faraday, “One Step Forward, Two Steps Back? Substantive Equality, Systemic Discrimination and Pay Equity at the Supreme Court of Canada” (2020), 94 S.C.L.R. (2d) 301, at p. 310; Sophia Moreau, “The Moral Seriousness of Indirect Discrimination”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law (2018), 123, at p. 145).

[32]                         Griggs v. Duke Power Co., 401 U.S. 424 (1971) was one of the first cases to apply this concept and is a classic example of adverse impact discrimination. An employer required employees to have a high school diploma and pass standardized tests to work in certain departments at a power plant. Neither requirement was significantly related to successful job performance; both, however, had the effect of disqualifying African Americans at a substantially higher rate than white applicants.

[33]                         The United States Supreme Court held that the education and testing requirements infringed Title VII of the Civil Rights Act of 1964, Pub. L. 88-352, 78 Stat. 241 (1964). The court emphasized that the Act  prohibits “practices that are fair in form, but discriminatory in operation”:

Congress has now provided that tests or criteria for employment or promotion may not provide equality of opportunity merely in the sense of the fabled offer of milk to the stork and the fox. On the contrary, Congress has now required that the posture and condition of the job seeker be taken into account. It has — to resort again to the fable — provided that the vessel in which the milk is proffered be one all seekers can use. The Act  proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation. . . . Good intent or absence of discriminatory intent does not redeem employment procedures or testing mechanisms that operate as “built-in headwinds” for minority groups and are unrelated to measuring job capability. [Emphasis added; pp. 431‑32.]

[34]                          Griggs explains that the application of “neutral” rules may not produce equality in substance for disadvantaged groups. Membership in such groups often brings with it a unique constellation of physical, economic and social barriers. Laws which distribute benefits or burdens without accounting for those differences — without accounting for the “posture and condition of the job seeker”, as in Griggs — are the prime targets of indirect discrimination claims. I agree with Profs. Lisa Philipps and Margot Young that

we are not always conscious of the ways in which the distinctions we draw . . . will implicate group identities and single out specific groups for distinctive treatment. This is because the constellations of factors or characteristics that go into the construction of identities often masquerade as unconnected, purely individual traits, behaviours, choices, or situations. Yet, in social reality they may be tightly linked to one group or another. So the law has had to recognize that state action may be discriminatory even though on its face and in terms of the intentions informing it there is no obvious evidence that such discrimination is occurring. [p. 258]

(See also Sandra Fredman, Discrimination Law (2nd ed. 2011), at pp. 38 and 108.)

[35]                          Addressing adverse impact discrimination can be among the “most powerful legal measures available to disadvantaged groups in society to assert their claims to justice” (Hugh Collins and Tarunabh Khaitan, “Indirect Discrimination Law: Controversies and Critical Questions”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law (2018), 1, at p. 30). Not only is such discrimination “much more prevalent than the cruder brand of openly direct discrimination”,[5] it often poses a greater threat to the equality aspirations of disadvantaged groups:

. . . even more common are situations where the discrimination occurs in a context like an employment relationship, government program or statute, or educational setting, and there is no single identifiable “villain”, no single action identifiable as “discriminatory”, and the outward appearance of a neutral set of rules or practices being applied across the board. This invisible structure, with its accompanying set of practices, is a powerful limit on the equality aspirations of many who must deal within that structure but have characteristics that do not match those of persons intended to benefit from the structure.

(Mary Eberts and Kim Stanton, “The Disappearance of the Four Equality Rights and Systemic Discrimination from Canadian Equality Jurisprudence” (2018), 38 N.J.C.L. 89, at p. 92)

[36]                         By recognizing the exclusionary impact of such discrimination, courts can better address “discrimination in its diverse forms”, including at “the systemic or institutional level” (Vizkelety, at p. viii; see also Colleen Sheppard, “Mapping anti‑discrimination law onto inequality at work: Expanding the meaning of equality in international labour law” (2012), 151 Int’l Lab. Rev. 1, at p. 8; Faraday, at p. 319). Remedying adverse effects discrimination allows courts

[to] go to the heart of the equality question, to the goal of transformation, to an examination of the way institutions and relations must be changed in order to make them available, accessible, meaningful and rewarding for the many diverse groups of which our society is composed.

(Meiorin, at para. 41, quoting Shelagh Day and Gwen Brodsky, “The Duty to Accommodate: Who Will Benefit?” (1996), 75 Can. Bar Rev. 433, at p. 462)

[37]                         This Court first dealt with adverse impact discrimination in Ontario Human Rights Commission v. Simpsons‑Sears Ltd., [1985] 2 S.C.R. 536. Employees at a department store were periodically required to work on Friday evenings and Saturdays. Theresa O’Malley, an employee of the store and a member of the Seventh‑Day Adventist Church, was required by her faith to observe the Sabbath from sundown Friday until sundown Saturday. She brought a complaint against the store under the Ontario Human Rights Code, R.S.O. 1980, c. 340, claiming that the rule requiring her to work on Saturdays discriminated against her on the basis of religion.

[38]                         Writing for a unanimous Court, McIntyre J. agreed. He stressed that the Ontario Human Rights Code was meant to provide protection against the “result or the effect” of discriminatory conduct (p. 547). Citing Griggs and several Canadian decisions, McIntyre J. concluded that the Act  prohibited adverse effects discrimination, which he distinguished from direct discrimination as follows:

A distinction must be made between what I would describe as direct discrimination and the concept already referred to as adverse effect discrimination in connection with employment. Direct discrimination occurs in this connection where an employer adopts a practice or rule which on its face discriminates on a prohibited ground. For example, “No Catholics or no women or no blacks employed here.” There is, of course, no disagreement in the case at bar that direct discrimination of that nature would contravene the Act . On the other hand, there is the concept of adverse effect discrimination. It arises where an employer for genuine business reasons adopts a rule or standard which is on its face neutral, and which will apply equally to all employees, but which has a discriminatory effect upon a prohibited ground on one employee or group of employees in that it imposes, because of some special characteristic of the employee or group, obligations, penalties, or restrictive conditions not imposed on other members of the work force. [p. 551]

[39]                         Simpsons‑Sears was the first of several human rights decisions where this Court grappled with adverse effects discrimination. In Canadian National Railway Co. v. Canada (Canadian Human Rights Commission), [1987] 1 S.C.R. 1114 (“Action Travail”), Dickson C.J. upheld a discrimination claim against an employer whose hiring and promotion practices led to women being drastically under‑represented in certain jobs. Some of these practices were neutral on their face; Dickson C.J., however, highlighted the importance of looking “at the results of a system”:

A thorough study of “systemic discrimination” in Canada is to be found in the Abella Report on equality in employment.

 

. . .

 

     Discrimination . . . means practices or attitudes that have, whether by design or impact, the effect of limiting an individual’s or a group’s right to the opportunities generally available because of attributed rather than actual characteristics . . . .

 

It is not a question of whether this discrimination is motivated by an intentional desire to obstruct someone’s potential, or whether it is the accidental by‑product of innocently motivated practices or systems. If the barrier is affecting certain groups in a disproportionately negative way, it is a signal that the practices that lead to this adverse impact may be discriminatory.

 

This is why it is important to look at the results of a system . . . . [Emphasis added; pp. 1138-39.]

[40]                          These principles were soon imported into the Court’s s. 15 jurisprudence. In Andrews v. Law Society of British Columbia, [1989] 1 S.C.R. 143, the Court rejected a “sameness” or formal theory of equality, instead identifying substantive equality as the philosophical premise of s. 15 and outlining a theory of equality centred on “the impact of the law on the individual or the group concerned” (p. 165). In developing this theory, McIntyre J. emphatically rejected the approach to s. 15 adopted by the British Columbia Court of Appeal, which had defined the “essential meaning” of equality as ensuring that the “similarly situated be similarly treated” ((1986), 2 B.C.L.R. (2d) 305, at p. 311, quoting Joseph Tussman and Jacobus tenBroek, “The Equal Protection of the Laws” (1949), 37 Cal. L. Rev. 341, at p. 344). Justice McIntyre described this approach as “seriously deficient”, on the basis that “mere equality of application to similarly situated groups or individuals does not afford a realistic test for a violation of equality rights” (pp. 165‑67).

[41]                          Drawing on the Court’s human rights jurisprudence, while recognizing that “not all distinctions and differentiations created by law are discriminatory” (at p. 182), McIntyre J. endorsed an approach to equality and discrimination that was centred on the actual effects, rather than the purpose or facial neutrality of a law on a claimant group:

I would say then that discrimination may be described as a distinction, whether intentional or not but based on grounds relating to personal characteristics of the individual or group, which has the effect of imposing burdens, obligations, or disadvantages on such individual or group not imposed upon others, or which withholds or limits access to opportunities, benefits, and advantages available to other members of society. [p. 174]

Andrews provided a robust template for substantive equality that subsequent decisions “enriched but never abandoned” (R. v. Kapp, [2008] 2 S.C.R. 483, at para. 14). It was a remedy for exclusion and a recipe for inclusion.

[42]                         Our subsequent decisions left no doubt that substantive equality is the “animating norm” of the s. 15 framework (Withler, at para. 2; see also Kapp, at paras. 15‑16; Alliance, at para. 25); and that substantive equality requires attention to the “full context of the claimant group’s situation”, to the “actual impact of the law on that situation”, and to the “persistent systemic disadvantages [that] have operated to limit the opportunities available” to that group’s members (Withler, at para. 43; Taypotat, at para. 17; see also Quebec v. A, at paras. 327‑32; Alliance, at para. 28; Centrale, at para. 35).

[43]                         The Court, applying these principles, has acknowledged the existence of adverse impact discrimination under s. 15(1). In Eldridge v. British Columbia (Attorney General), [1997] 3 S.C.R. 624, the Court recognized that there is a disparate impact on persons with hearing loss in a health‑care system in which they are unable to access interpreters. The Court confirmed that a s. 15(1) violation could arise through “the adverse effects of rules of general application” (para. 77).

[44]                         Similarly, in Vriend v. Alberta, [1998] 1 S.C.R. 493, the Court declared unconstitutional an Alberta human rights statute which did not include sexual orientation as a prohibited ground of discrimination, because of its “disproportionate impact” on members of the LGBTQ+ community:

. . . there is, on the surface, a measure of formal equality: gay or lesbian individuals have the same access as heterosexual individuals to the protection of the [Act ] in the sense that they could complain to the Commission about an incident of discrimination on the basis of any of the grounds currently included. However, the exclusion of the ground of sexual orientation, considered in the context of the social reality of discrimination against gays and lesbians, clearly has a disproportionate impact on them as opposed to heterosexuals. Therefore the [Act ] in its underinclusive state denies substantive equality to the former group. [Emphasis added; para. 82.]

[45]                          Several other decisions of this Court have confirmed that “not only does the Charter  protect from direct or intentional discrimination, it also protects from adverse impact discrimination” (McKinney v. University of Guelph, [1990] 3 S.C.R. 229, at p. 279; Tétreault‑Gadoury v. Canada (Employment and Immigration Commission), [1991] 2 S.C.R. 22, at p. 41; see also Law v. Canada (Minister of Employment and Immigration), [1999] 1 S.C.R. 497, at para. 36; Withler, at para. 64; Taypotat, at para. 23).

[46]                          The Court most recently addressed this issue in Taypotat. While concluding that there was no discrimination demonstrated on the facts of the case, the Court acknowledged that “facially neutral qualifications like education requirements” can be a breach of s. 15(1) because of their “disproportionate effect” on protected groups (paras. 15 and 22).

[47]                          There is no doubt, therefore, that adverse impact discrimination “violate[s] the norm of substantive equality” which underpins this Court’s equality jurisprudence (Withler, at para. 2). At the heart of substantive equality is the recognition that identical or facially neutral treatment may “frequently produce serious inequality” (Andrews, at p. 164). This is precisely what happens when “neutral” laws ignore the “true characteristics of [a] group which act as headwinds to the enjoyment of society’s benefits” (Eaton v. Brant County Board of Education, [1997] 1 S.C.R. 241, at para. 67; Eldridge, at para. 65).

[48]                         The “animating norm” of the current s. 15 framework guaranteeing substantive equality is also the core value engaged in cases of adverse effects discrimination (Withler, at para. 2; Watson Hamilton and Koshan (2015), at pp. 192 and 197). This Court has never suggested that cases of adverse impact discrimination should be resolved under a different approach (see, for example, Andrews, at pp. 173‑74; Eldridge, at paras. 59‑60; Vriend, at paras. 81‑82 and 87‑89; Law, at paras. 36‑39; Taypotat, at paras. 19‑22; Alliance, at para. 25). On the contrary, we have clarified that the same approach applies regardless of whether the discrimination alleged is direct or indirect. Withler leaves little doubt on this point:

The substantive equality analysis under s. 15(1), as discussed earlier, proceeds in two stages . . . . The role of comparison at the first step is to establish a “distinction”.

 

. . .

 

In some cases, identifying the distinction will be relatively straightforward, because a law will, on its face, make a distinction on the basis of an enumerated or analogous ground (direct discrimination) . . . . In other cases, establishing the distinction will be more difficult, because what is alleged is indirect discrimination: that although the law purports to treat everyone the same, it has a disproportionately negative impact on a group or individual that can be identified by factors relating to enumerated or analogous grounds . . . . In that kind of case, the claimant will have more work to do at the first step. Historical or sociological disadvantage may assist in demonstrating that the law imposes a burden or denies a benefit to the claimant that is not imposed on or denied to others. The focus will be on the effect of the law and the situation of the claimant group. [Emphasis added; paras. 61‑62 and 64.]

[49]                         In the human rights context, the Court has not used different legal tests for direct and indirect discrimination since Meiorin (paras. 50‑54; see also British Columbia (Superintendent of Motor Vehicles) v. British Columbia (Council of Human Rights), [1999] 3 S.C.R. 868, at paras. 18‑19; Moore v. British Columbia (Education), [2012] 3 S.C.R. 360, at para. 61). A unified approach, in my view, is equally justified under the Charter .

[50]                         To prove discrimination under s. 15(1), claimants must show that a law or policy creates a distinction based on a protected ground, and that the law perpetuates, reinforces or exacerbates disadvantage. These requirements do not require revision in adverse effects cases. What is needed, however, is a clear account of how to identify adverse effects discrimination, because the impugned law will not, on its face, include any distinctions based on prohibited grounds (Withler, at para. 64). Any such distinctions must be discerned by examining the impact of the law (Alliance, at para. 25).

[51]                         This inquiry has frequently been described as a search for a “disproportionate” impact on members of protected groups (see Vriend, at para. 82; Withler, at para. 64; Taypotat, at paras. 21‑23; Action Travail, at p. 1139; Egan v. Canada, [1995] 2 S.C.R. 513, at para. 138, per Cory and Iacobucci JJ. dissenting; Moreau (2010), at p. 154; Braun, at pp. 124‑25; Vizkelety, at p. 176; Watson Hamilton and Koshan (2015), at p. 196; Collins and Khaitan, at pp. 3‑4; Dianne Pothier, “M’Aider, Mayday: Section 15  of the Charter  in Distress” (1996), 6 N.J.C.L. 295, at p. 322).

[52]                         In other words, in order for a law to create a distinction based on prohibited grounds through its effects, it must have a disproportionate impact on members of a protected group. If so, the first stage of the s. 15 test will be met.

[53]                         How does this work in practice? Instead of asking whether a law explicitly targets a protected group for differential treatment, a court must explore whether it does so indirectly through its impact on members of that group (see Eldridge, at paras. 60‑62; Vriend, at para. 82). A law, for example, may include seemingly neutral rules, restrictions or criteria that operate in practice as “built‑in headwinds” for members of protected groups. The testing requirement in Griggs is the paradigmatic example; other examples include the aerobic fitness requirement in Meiorin, and the policy requiring employees to work on Saturdays in Simpsons‑Sears (see also Central Alberta Dairy Pool v. Alberta (Human Rights Commission), [1990] 2 S.C.R. 489). To assess the adverse impact of these policies, courts looked beyond the facially neutral criteria on which they were based, and examined whether they had the effect of placing members of protected groups at a disadvantage (Moreau (2018), at p. 125).

[54]                         In other cases, the problem is not “headwinds” built into a law, but the absence of accommodation for members of protected groups (Tarunabh Khaitan, A Theory of Discrimination Law (2015), at p. 77; Dianne Pothier, “Tackling Disability Discrimination at Work: Toward a Systemic Approach” (2010), 4 McGill J.L. & Health 17, at pp. 23‑24). Eldridge is a good example. Under the health care scheme in that case, all patients lacked access to sign language interpreters — but this lack of access had a disproportionate impact on those who had hearing loss and required interpreters to meaningfully communicate with health care providers (paras. 69, 71 and 83).

[55]                         Disproportionate impact can be proven in different ways. In Eldridge, it was established because “the quality of care received by [those with hearing loss] was inferior to that available to hearing persons” (para. 83 (emphasis added)). In Griggs and Meiorin, by contrast, the relevant impact was the higher rate at which African Americans and women were disqualified from employment. Both are examples of how a law or policy can have a disproportionate impact on members of a protected group. Griggs, Meiorin, and other leading cases leave no doubt that disproportionate impact can be established if members of protected groups are denied benefits or forced to take on burdens more frequently than others. A difference in “quality” of treatment, as in Eldridge, may strengthen a claim of disproportionate impact, but it is not a necessary element (Philipps and Young, at pp. 244‑45; see also Pothier (1996), at p. 322; Selene Mize, “Indirect Discrimination Reconsidered” (2007), N.Z.L. Rev. 27, at p. 39).

[56]                          Two types of evidence will be especially helpful in proving that a law has a disproportionate impact on members of a protected group. The first is evidence about the situation of the claimant group. The second is evidence about the results of the law.

[57]                          Courts will benefit from evidence about the physical, social, cultural or other barriers which provide the “full context of the claimant group’s situation” (Withler, at para. 43; see also para. 64). This evidence may come from the claimant, from expert witnesses, or through judicial notice (see R. v. Spence, [2005] 3 S.C.R. 458). The goal of such evidence is to show that membership in the claimant group is associated with certain characteristics that have disadvantaged members of the group, such as an inability to work on Saturdays or lower aerobic capacity (Homer v. Chief Constable of West Yorkshire Police, [2012] 3 All E.R. 1287 (S.C.), at para. 14; Simpsons‑Sears; Meiorin, at para. 11). These links may reveal that seemingly neutral policies are “designed well for some and not for others” (Meiorin, at para. 41). When evaluating evidence about the group, courts should be mindful of the fact that issues which predominantly affect certain populations may be under‑documented. These claimants may have to rely more heavily on their own evidence or evidence from other members of their group, rather than on government reports, academic studies or expert testimony.

[58]                         Courts will also benefit from evidence about the outcomes that the impugned law or policy (or a substantially similar one) has produced in practice. Evidence about the “results of a system” may provide concrete proof that members of protected groups are being disproportionately impacted (Action Travail, at p. 1139; Vizkelety, at pp. 170‑74). This evidence may include statistics, especially if the pool of people adversely affected by a criterion or standard includes both members of a protected group and members of more advantaged groups (Sheppard (2001), at pp. 545‑46; Braun, at pp. 120‑21).

[59]                         There is no universal measure for what level of statistical disparity is necessary to demonstrate that there is a disproportionate impact, and the Court should not, in my view, craft rigid rules on this issue. The goal of statistical evidence, ultimately, is to establish “a disparate pattern of exclusion or harm that is statistically significant and not simply the result of chance” (Sheppard (2001), at p. 546; see also Vizkelety, at p. 175; Fredman (2011), at pp. 186‑87). The weight given to statistics will depend on, among other things, their quality and methodology (Vizkelety, at pp. 178‑84).

[60]                         Ideally, claims of adverse effects discrimination should be supported by evidence about the circumstances of the claimant group and about the results produced by the challenged law. Evidence about the claimant group’s situation, on its own, may amount to merely a “web of instinct” if too far removed from the situation in the actual workplace, community or institution subject to the discrimination claim (Taypotat, at para. 34). Evidence of statistical disparity, on its own, may have significant shortcomings that leave open the possibility of unreliable results. The weaknesses with each type of evidence can be overcome if they are both present (Braun, at p. 135; Vizkelety, at p. 192; Vancouver Area Network of Drug Users v. Downtown Vancouver Business Improvement Association (2018), 10 B.C.L.R. (6th) 175 (C.A.), at para. 98). Prof. Colleen Sheppard (2001) recognizes this possibility:

While in some cases the overwhelming correspondence between certain categories and the gender or racial composition of the category makes the sex or race discrimination claims relatively easy to substantiate, in other cases the statistical preponderance may be less marked. In such cases it may also be important to consider the qualitative components of the harm that constitutes discrimination. [p. 548]

[61]                         This is not to say, of course, that both kinds of evidence are always required. In some cases, evidence about a group will show such a strong association with certain traits — such as pregnancy with gender — that the disproportionate impact on members of that group “will be apparent and immediate” (Taypotat, at para. 33; see also Fredman (2011), at pp. 187‑88; Sheppard (2001), at pp. 544‑45; Gaz métropolitain inc. v. Commission des droits de la personne et des droits de la jeunesse, 2011 QCCA 1201, at paras. 27 and 47 (CanLII); Oršuš v. Croatia, No. 15766/03, ECHR 2010-II, at para. 153).

[62]                         Similarly, clear and consistent statistical disparities can show a disproportionate impact on members of protected groups, even if the precise reason for that impact is unknown. Prof. Sandra Fredman has argued forcefully against requiring claimants to specify “the reason why” they are being disadvantaged by a rule or policy:

To require the complainants to show the ‘reason why’ the PCP [policy, criteria or practice] disadvantages the group as a whole is to fundamentally misunderstand the meaning of indirect discrimination. It is the disparate impact on the group of a PCP itself which constitutes the prima facie discrimination . . . .

(“Direct and Indirect Discrimination: Is There Still a Divide?”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law (2018), 31, at p. 46; see also Sandra Fredman, “The Reason Why: Unravelling Indirect Discrimination” (2016), 45 Indus. L.J. 231.)

[63]                         I agree. If there are clear and consistent statistical disparities in how a law affects a claimant’s group, I see no reason for requiring the claimant to bear the additional burden of explaining why the law has such an effect. In such cases, the statistical evidence is itself a compelling sign that the law has not been structured in a way that takes into account the protected group’s circumstances (see Fredman (2011), at p. 181; Vizkelety, at pp. 174‑76; Action Travail, at p. 1139).

[64]                         The United Kingdom Supreme Court reached a similar conclusion in Essop v. Home Office (U.K. Border Agency), [2017] 3 All E.R. 551 (S.C.). At issue was a core skills assessment that immigration officers had to pass to be promoted. Racial minorities and older candidates were shown to be less likely to pass the assessment, but there was no evidence available to explain why this disparity was occurring (para. 9).

[65]                         The Supreme Court concluded that there was disparate impact. Lady Hale D.P.S.C. explained that a claimant does not need to “establish the reason for the particular disadvantage to which the group is put” (para. 33). She noted that such a requirement made it more difficult to combat “hidden barriers which are not easy to anticipate or to spot” (para. 25). She also recognized that it is “commonplace for the disparate impact, or particular disadvantage, to be established on the basis of statistical evidence” — which would be impossible if claimants had to offer an explanation for why any given statistical imbalances had occurred (para. 28).

[66]                         Essop confirmed a flexible approach to proving disparate impact, under which proof of statistical disparity and broader group disadvantage may each be sufficient to establish a claim, but are not rigid requirements (see also O’Connor v. Bar Standards Board, [2018] 2 All E.R. 779 (S.C.), at para. 43). The European Court of Human Rights has similarly held that “when it comes to assessing the impact of a measure or practice on an individual or group, statistics which appear on critical examination to be reliable and significant will be sufficient to constitute the prima facie evidence the applicant is required to produce”; however, “[t]his does not . . . mean that indirect discrimination cannot be proved without statistical evidence” (D.H. v. the Czech Republic, No. 57325/00, ECHR 2007-IV, at para. 188 (emphasis added); see also Oršuš, at paras. 152‑53; Horváth and Kiss v. Hungary, [2013] E.L.R. 102 (E.C.H.R.), at para. 107).

[67]                         I agree with this approach. Both evidence of statistical disparity and of broader group disadvantage may demonstrate disproportionate impact; but neither is mandatory and their significance will vary depending on the case.

[68]                         Some further observations.

[69]                         First, whether the legislature intended to create a disparate impact is irrelevant (Sheppard (2001), at pp. 543‑44; Watson Hamilton and Koshan (2015), at pp. 196‑97; Faraday, at p. 310). Proof of discriminatory intent has never been required to establish a claim under s. 15(1) (Andrews, at pp. 173‑74; Eldridge, at para. 62; Vriend, at para. 93; Alliance, at para. 28; Centrale, at para. 35). Nor is an ameliorative purpose sufficient to shield legislation from s. 15(1) scrutiny (Centrale, at paras. 8 and 35; Alliance, at paras. 32‑33).

[70]                         Second, if claimants successfully demonstrate that a law has a disproportionate impact on members of a protected group, they need not independently prove that the protected characteristic “caused” the disproportionate impact (Tarunabh Khaitan and Sandy Steel, “Wrongs, Group Disadvantage and the Legitimacy of Indirect Discrimination Law”, in Hugh Collins and Tarunabh Khaitan, eds., Foundations of Indirect Discrimination Law (2018), 197, at pp. 203-4 and 220; Fredman (2018), at p. 46; Braun, at p. 146; Watson Hamilton and Koshan (2015), at p. 197; West Yorkshire Police, at paras. 12‑14; Essop, at paras. 24‑27). Put differently, there was no need for the claimant in Griggs to address whether his exclusion was based on his race or lack of a high school education. The whole point of the adverse impact analysis was to show that the use of a high school education as a criteria for employment had a disproportionate impact on African Americans (Fredman (2011), at p. 189).

[71]                         It is also unnecessary to inquire into whether the law itself was responsible for creating the background social or physical barriers which made a particular rule, requirement or criterion disadvantageous for the claimant group. Returning to Griggs, this would amount to asking whether Duke Power Co. was responsible for lower rates of high school education among African Americans. Plainly, it was not — but this was entirely irrelevant to whether a disproportionate impact had been established. Section 15(1) has always required attention to the systemic disadvantages affecting members of protected groups, even if the state did not create them (Alliance, at para. 41; Centrale, at para. 32; Vriend, at paras. 84 and 97; Eldridge, at paras. 64‑66; Eaton, at para. 67; R. v. Turpin, [1989] 1 S.C.R. 1296, at pp. 1331‑32).

[72]                         Third, claimants need not show that the criteria, characteristics or other factors used in the impugned law affect all members of a protected group in the same way. This Court has long held that “[t]he fact that discrimination is only partial does not convert it into non‑discrimination” (Brooks v. Canada Safeway Ltd., [1989] 1 S.C.R. 1219, at p. 1248, quoting James MacPherson, “Sex Discrimination in Canada: Taking Stock at the Start of a New Decade” (1980), 1 C.H.R.R. C/7, at p. C/11). In Brooks, the Court held that a corporate plan which denied benefits to employees during pregnancy discriminated on the basis of sex. The employer argued that the plan did not deny benefits to “women”, but only to “women who are pregnant” (p. 1248, quoting MacPherson, at p. C/11). Writing for the Court, Dickson C.J. explained that practices amounting to “partial discrimination” are no less discriminatory than those in which all members of a protected group are affected (pp. 1247‑48).

[73]                         The Court reiterated this principle in Janzen v. Platy Enterprises Ltd., [1989] 1 S.C.R. 1252, where it held that the sexual harassment of two female employees was discrimination on the basis of sex. The Court rejected the employer’s argument that sex discrimination had not occurred because only some of the female employees at the store had been sexually harassed. Dickson C.J. reiterated the approach to partial discrimination he had previously set out in Brooks:

If a finding of discrimination required that every individual in the affected group be treated identically, legislative protection against discrimination would be of little or no value. It is rare that a discriminatory action is so bluntly expressed as to treat all members of the relevant group identically. In nearly every instance of discrimination the discriminatory action is composed of various ingredients with the result that some members of the pertinent group are not adversely affected, at least in a direct sense, by the discriminatory action. To deny a finding of discrimination in the circumstances of this appeal is to deny the existence of discrimination in any situation where discriminatory practices are less than perfectly inclusive. It is to argue, for example, that an employer who will only hire a woman if she has twice the qualifications required of a man is not guilty of sex discrimination if, despite this policy, the employer nevertheless manages to hire some women. [Emphasis added; pp. 1288-89.]

[74]                         The Court’s approach in Brooks and Janzen “had obvious implications for claims based on multiple grounds of discrimination” (Dianne Pothier, “Connecting Grounds of Discrimination to Real People’s Real Experiences” (2001), 13 C.J.W.L. 37, at p. 58). As Dianne Pothier has explained:

It is an easy step to move from saying, as in Janzen, that not all women need be affected to constitute sex discrimination to also accepting that different groups of women . . . can be differently affected by, or have different experiences of, sex discrimination. Janzen also meant that a claim based on, for example, both gender and race could not be defeated simply by saying that it could not be sex discrimination because white women were unaffected or that it could not be race discrimination because Black men were unaffected. [p. 58]

[75]                         The Court subsequently confirmed that “heterogeneity within a claimant group does not defeat a claim of discrimination” (Quebec v. A, at para. 354). In Quebec v. A, for example, the Court held that certain provisions of the Civil Code of Québec that distinguished between de facto and legally married spouses for the purposes of support and division, discriminated on the basis of marital status. It reached this conclusion even though there was “a range of need or vulnerability among de facto spouses” (para. 354). Similarly, in Nova Scotia (Workers’ Compensation Board) v. Martin, [2003] 2 S.C.R. 504, the Court held that a provincial compensation scheme that provided lesser benefits to those suffering from chronic pain, discriminated on the basis of disability. The Court confirmed that “differential treatment can occur on the basis of an enumerated ground despite the fact that not all persons belonging to the relevant group are equally mistreated” (para. 76; see also Centrale, at para. 28; Pothier (2010), at pp. 35-36; Watson Hamilton and Koshan (2015), at pp. 197‑98; Braun, at p. 147; Sheppard (2001), at p. 549).

[76]                         This brings us to the second step of the s. 15 test: whether the law has the effect of reinforcing, perpetuating, or exacerbating disadvantage (Alliance, at para. 25). This inquiry will usually proceed similarly in cases of disparate impact and explicit discrimination. There is no “rigid template” of factors relevant to this inquiry (Quebec v. A, at para. 331, quoting Withler, at para. 66). The goal is to examine the impact of the harm caused to the affected group. The harm may include “[e]conomic exclusion or disadvantage, [s]ocial exclusion . . . [p]sychological harms . . . [p]hysical harms . . . [or] [p]olitical exclusion”, and must be viewed in light of any systemic or historical disadvantages faced by the claimant group (Sheppard (2010), at pp. 62‑63 (emphasis deleted)).

[77]                         The purpose of the inquiry is to keep s. 15(1) focussed on the protection of groups that have experienced exclusionary disadvantage based on group characteristics, as well as the protection of those “who are members of more than one socially disadvantaged group in society” (Colleen Sheppard, “Grounds of Discrimination: Towards an Inclusive and Contextual Approach” (2001), 80 Can. Bar Rev. 893, at p. 896; see also Withler, at para. 58). As the Court noted in Quebec v. A when discussing the second stage of the s. 15 test:

The root of s. 15 is our awareness that certain groups have been historically discriminated against, and that the perpetuation of such discrimination should be curtailed. [para. 332]

(See also Taypotat, at para. 20.)

[78]                         Notably, the presence of social prejudices or stereotyping are not necessary factors in the s. 15(1) inquiry. They may assist in showing that a law has negative effects on a particular group, but they “are neither separate elements of the Andrews test, nor categories into which a claim of discrimination must fit” (Quebec v. A, at para. 329), since

[w]e must be careful not to treat Kapp and Withler as establishing an additional requirement on s. 15 claimants to prove that a distinction will perpetuate prejudicial or stereotypical attitudes towards them. Such an approach improperly focuses attention on whether a discriminatory attitude exists, not a discriminatory impact, contrary to Andrews, Kapp and Withler. [Emphasis in original; para. 327.]

(See also paras. 329-31.)

[79]                         The perpetuation of disadvantage, moreover, does not become less serious under s. 15(1) simply because it was relevant to a legitimate state objective. I agree with Dean Mayo Moran that adding relevance to the s. 15(1) test — even as one contextual factor among others — risks reducing the inquiry to a search for a “rational basis” for the impugned law (“Protesting Too Much: Rational Basis Review Under Canada’s Equality Guarantee”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms (2006), 71, at pp. 81‑84; Eberts and Stanton, at pp. 90 and 119‑20; Sheila McIntyre, “Deference and Dominance: Equality Without Substance”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms (2006), 95, at pp. 108‑13). The test for a prima facie breach of s. 15(1) is concerned with the discriminatory impact of legislation on disadvantaged groups, not with whether the distinction is justified, an inquiry properly left to s. 1 (Andrews, at pp. 181‑82; Turpin, at pp. 1325‑26; Miron v. Trudel, [1995] 2 S.C.R. 418, at paras. 129‑30; Eldridge, at paras. 77 and 79; Quebec v. A, at para. 333; Centrale, at para. 35).

[80]                         Similarly, there is no burden on a claimant to prove that the distinction is arbitrary to prove a prima facie breach of s. 15(1). It is for the government to demonstrate that the law is not arbitrary in its justificatory submissions under s. 1 (see Eberts and Stanton, at p. 117; Jonnette Watson Hamilton and Jennifer Koshan, “Kahkewistahaw First Nation v. Taypotat: An Arbitrary Approach to Discrimination” (2016), 76 S.C.L.R. (2d) 243, at pp. 259‑60; Alicja Puchta, “Quebec v A and Taypotat: Unpacking the Supreme Court’s Latest Decisions on Section 15  of the Charter ” (2018), 55 Osgoode Hall L.J. 665, at p. 704).

[81]                         In sum, then, the first stage of the s. 15 test is about establishing that the law imposes differential treatment based on protected grounds, either explicitly or through adverse impact. At the second stage, the Court asks whether it has the effect of reinforcing, perpetuating, or exacerbating disadvantage (Alliance, at para. 25).

[82]                         Where possible, the two inquiries should be kept distinct, but there is clearly potential for overlap in adverse effects cases based on “the impossibility of rigid categorizations” (Sheppard (2010), at p. 21). What matters in the end is that a court asks and answers the necessary questions relevant to the s. 15(1) inquiry, not whether it keeps the two steps of the inquiry in two impermeable silos.

Application

[83]                         Returning to the claim before us in this appeal, as previously noted, full‑time RCMP members who work regular hours, who are suspended, or who go on unpaid leave can obtain full pension credit for those periods of service under the pension plan. Full‑time members who temporarily reduce their hours under a job‑sharing agreement, however, are classified as part‑time workers under the Regulations and are unable to acquire full‑time pension credit for their service.

[84]                         Under the pension scheme, therefore, a full‑time RCMP member’s temporary reduction in working hours results in their losing out on potential pension benefits. The question is whether this arrangement has a disproportionate impact on women.

[85]                         The Federal Court and Court of Appeal acknowledged that the vast majority of members in the job‑sharing program who lose out on pension benefits are women with children. In their view, however, these losses occurred because “the appellants . . . elected to job‑share”, not because of their gender or parental status (C.A. reasons, at para. 53).

[86]                         In relying on Ms. Fraser’s “choice” to job‑share as grounds for dismissing her claim, the Federal Court and Court of Appeal, with respect, misapprehended our s. 15(1) jurisprudence. This Court has consistently held that differential treatment can be discriminatory even if it is based on choices made by the affected individual or group.

[87]                         In Brooks, for example, Dickson C.J. rejected an employer’s argument that providing unequal benefits to pregnant women is not sex discrimination because pregnancy is “voluntary” (pp. 1237‑38). After Brooks, the Court “repeatedly rejected arguments that choice protects a distinction from a finding of discrimination” (Quebec v. A, at para. 336). In Lavoie v. Canada, [2002] 1 S.C.R. 769, for example, the Court held that a statute which gave preferential treatment to Canadian citizens infringed s. 15(1), despite the government’s argument that becoming a Canadian citizen was a choice. Chief Justice McLachlin and Justice L’Heureux‑Dubé, concurring on this issue, made clear that

the fact that a person could avoid discrimination by modifying his or her behaviour does not negate the discriminatory effect. If it were otherwise, an employer who denied women employment in his factory on the ground that he did not wish to establish female changing facilities could contend that the real cause of the discriminatory effect is the woman’s “choice” not to use men’s changing facilities. The very act of forcing some people to make such a choice violates human dignity, and is therefore inherently discriminatory. The law of discrimination thus far has not required applicants to demonstrate that they could not have avoided the discriminatory effect in order to establish a denial of equality under s. 15(1). The Court in Andrews was not deterred by such considerations. On the contrary, La Forest J. specifically noted that acquiring Canadian citizenship could in some cases entail the “serious hardship” of losing an existing citizenship. He left no doubt that this hardship was a cost to be considered in favour of the individual affected by the discrimination. [Emphasis added; citation omitted; para. 5.]

(See also para. 51, per Bastarache J.)

[88]                         Justice L’Heureux‑Dubé had expressed a similar view in her dissenting reasons in Nova Scotia (Attorney General) v. Walsh, [2002] 4 S.C.R. 325, a decision the Court overturned in Quebec v. A, explaining that a choice‑based approach was fundamentally flawed:

In Walsh, the majority’s focus on choice rather than on the impact of the distinction on members of the group also paid insufficient attention to the requirement for a true substantive equality analysis, affirmed in Kapp and Withler. In contrast to formal equality, which assumes an “autonomous, self‑interested and self‑determined” individual, substantive equality looks not only at the choices that are available to individuals, but at “the social and economic environments in which [they] pla[y] out”. [Emphasis added; citation omitted; para. 342.]

[89]                         Several scholars have made this point as well. Prof. Margot Young, for example, points out that

th[e] closure of critical examination by way of characterization of the inequality of which an individual complains as “natural”, “chosen” or “merited” is deeply problematic. Indeed, many of the major steps in the progression towards women’s equality have come precisely from the revelation of the “natural” as “social”, the “chosen” as “coerced” and the “merited” as “undeserved”.

 

. . .

 

Claims of merit, nature and choice are difficult to critically unpack; they so often are the roots of discrimination. This makes these notions deeply functional in the perpetuation and obfuscation of inequality.

(“Blissed Out: Section 15 at Twenty”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms (2006), 45, at pp. 55-56; see also Margot Young, “Unequal to the Task: ‘Kapp’ing the Substantive Potential of Section 15” (2010), 50 S.C.L.R. (2d) 183, at pp. 190‑91 and 196.)

[90]                         Prof. Sonia Lawrence makes the critical point that choices are themselves shaped by systemic inequality:

. . . a contextual account of choice produces a sadly impoverished narrative, in which choices more theoretical than real serve to eliminate the possibility of a finding of discrimination . . . . The result is a jurisprudence which almost mocks a more nuanced version of the what and how of discrimination, through frequent recourse to the idea that any harm to the claimant was actually the result of her choice, or her unwise exercise of her own judicially protected liberty.

 

. . .

 

Any number of structural conditions push people towards their choices, with the result that certain choices may be made more often by people with particular “personal characteristics”. This is a key feature of systemic inequality — it develops not out of direct statutory discrimination, but rather out of the operation of institutions which may seem neutral at first glance. [Emphasis added.]

(“Choice, Equality and Tales of Racial Discrimination: Reading the Supreme Court on Section 15”, in Sheila McIntyre and Sanda Rodgers, eds., Diminishing Returns: Inequality and the Canadian Charter of Rights and Freedoms (2006), 115, at pp. 115‑16 and 124‑25; see also Diana Majury, “Women Are Themselves to Blame: Choice as a Justification for Unequal Treatment”, in Fay Faraday, Margaret Denike and M. Kate Stephenson, eds., Making Equality Rights Real: Securing Substantive Equality under the Charter (2006), 209, at pp. 219‑25.)

[91]                         The case before us highlights the flaws of over‑emphasizing choice in the s. 15 inquiry. For many women, the decision to work on a part‑time basis, far from being an unencumbered choice, “often lies beyond the individual’s effective control” (Miron, at para. 153; Quebec v. A, at para. 316; see also Watson Hamilton and Koshan (2015), at p. 202). Deciding to work part‑time, for many women, is a “choice” between either staying above or below the poverty line. The job-sharing program, moreover, was introduced precisely because some members required access to an alternative to taking leave without pay “due to [their] personal or family circumstances” (A.R., vol. V, at p. 810). Ms. Fox made a similar point in her affidavit:

In my experience, this policy is particularly harmful to women who work in rural or isolated communities. The RCMP regularly posts women members in such communities where there is simply no around‑the‑clock child care available. As such, job‑sharing is often the only child care solution for members with children.

[92]                         By invoking the “choice” to job‑share as a basis for rejecting the s. 15(1) claim, the Federal Court and Court of Appeal removed the “challenged inequality from scrutiny, effectively taking it off the radar screen so as to circumvent examination of the equality issues at stake” (Majury, at p. 219). It is an approach that this Court’s s. 15 jurisprudence eschews.

[93]                         The Federal Court and Court of Appeal also held that the pension plan does not treat those who job‑share less favourably than those who go on unpaid leave. They reached this conclusion based on a formalistic comparison between the remuneration offered under job‑sharing and leave without pay.

[94]                         This is precisely the type of “mirror comparator group” analysis that this Court squarely rejected in Withler (paras. 55‑64; see also Moore, at paras. 28‑31). Section 15(1) guarantees Ms. Fraser and others in the job‑sharing program the right to substantive equality with respect to full‑time RCMP workers, not merely members on leave without pay. A narrow focus on the buy‑back provisions ignores their role within the pension scheme: they are themselves the means by which those who go on unpaid leave “get meaningful access” to the pension benefits available to all full‑time employees (Moore, at para. 28).

[95]                         This aspect of Ms. Fraser’s claim is indistinguishable from Centrale. In that case, Quebec delayed implementation of a pay equity program by up to four years for women employed in workplaces with male comparators, and six years for women employed in workplaces with no male comparators. This Court held that the implementation delay infringed s. 15(1). Rather than comparing the situation of women in different workplaces, the Court explained how the delay in implementing pay equity disadvantaged women relative to men in other workplaces earning full value for their work:

The legislature chose to act to address pay discrimination against women, but denied access by delaying it for a group of women, leaving them, in comparison to male workers, paid less for longer. Whatever the motives behind the decision, this is “discrimination reinforced by law”, which this Court has denounced since Andrews (p. 172). The fact, then, that women in one type of workplace — with male comparators — received a remedy promptly is not an answer to the question of whether women in another type of workplace were also disadvantaged. It is no defence to a claim of discrimination by one group of women to suggest that another group has had its particular discrimination addressed. [Emphasis added; para. 33.]

(See also paras. 29, per Abella J., and 155-56, per McLachlin C.J., concurring on this point.)

[96]                         This leaves the question of whether, under a proper assessment, the s. 15(1) claim should succeed.

[97]                         In my respectful view, the use of an RCMP member’s temporary reduction in working hours as a basis to impose less favourable pension consequences plainly has a disproportionate impact on women. The relevant evidence — the results of the system — showed that:

•            RCMP members who worked reduced hours in the job‑sharing program were predominantly women with young children.

•            From 2010‑2014, 100 percent of members working reduced hours through job‑sharing were women, and most of them cited childcare as their reason for doing so.

[98]                         These statistics were bolstered by compelling evidence about the disadvantages women face as a group in balancing professional and domestic work. Evidence submitted by Ms. Fraser indicated that women have historically borne the overwhelming share of childcare responsibilities, that part‑time workers in Canada are disproportionately women, and that they are far more likely than men to work part‑time due to child care responsibilities. As a result, they experience less stable employment and periods of “scaling back at work”, including within police services.

[99]                         This evidence finds firm support in commission reports, judicial decisions and academic work. The landmark Report of the Royal Commission on the Status of Women in Canada (1970, Florence Bird, chair) acknowledged that a “larger proportion of women work[ed] only part-time” (at p. 61) and warned that the inequitable treatment of part‑time workers would disadvantage women:

We recognize one major problem in the use of part‑time workers: the provision of fringe benefits for those not employed on a regular basis. We nevertheless believe that ways must be found to provide these employees with pay and working conditions no less equitable than those provided for the full‑time worker. [p. 105]

[100]                     In its report, the Commission of Inquiry into Part‑time Work (1983, Joan Wallace, comm.) confirmed that most employees in part-time, lower-paid positions were women (Part-time Work in Canada: Report of the Commission of Inquiry into Part-time Work (1983), at pp. 21‑22, 46 and 151). The Commission also studied the use of job‑sharing programs across Canada. The data it collected suggested that almost all job‑sharing participants were women, and that “[t]he arrival of a new baby was the most common primary reason for initiating job sharing” (pp. 177‑78).

[101]                     The Report of the Commission on Equality in Employment (1984, Rosalie Silberman Abella, comm.) expanded on the link between part‑time work and childcare, observing that

[t]he demand and the need for remedial measures derive from the increasing number of mothers in the workforce. Their children need adequate care. By Canadian law both parents have a duty to care for their children, but by custom this responsibility has consistently fallen to the mother. It is the mother, therefore, who bears any guilt or social disapprobation for joining the workforce. And it is the mother who normally bears the psychological and actual responsibility for making childcare arrangements.

 

. . .

 

From the point of view of mothers, access to childcare and the nature of such care limits employment options. “In balancing the responsibilities of family and career, women more frequently than men must make decisions (such as to withdraw from the labour force to care for young children) of consequence to their career” . . . . Various studies show that a major reason women are over‑represented in part‑time work is that they are combining childcare responsibilities with jobs in the paid labour market. [Emphasis added.]

(pp. 177 and 185-86, quoting Ontario Manpower Commission, The Employment of Women in Ontario: Background Paper (1983), at p. 17)

[102]                     The final report of the Law Commission of Ontario’s 2012 study on vulnerable workers also confirmed that

Canadian studies show that women are more likely to be engaged in precarious work than men. For example, women are over‑represented in part‑time and temporary work.

 

. . .

 

The high numbers of women in precarious work are, in some measure, the result of their traditional social role as caregivers. Under the “gender contract” that typified the 1950s middle class, men were primarily responsible for financial support and women stayed home to care for the family. (Women in many working‑class families have always worked outside the home, caring for other women’s children, cleaning homes and working in factories and shops, for example.) Today, under current social and economic conditions, two incomes are often necessary to support a family and women’s choices and involvement in many spheres of life have expanded. The majority of women have joined the workforce. The family unit is also more varied with increasing numbers of single parents. And yet women continue to bear primary responsibility for care‑giving. In 2010 Canadian women spent an average total of 50 hours per week caring for household children, double that spent by men (24 hours). In 2008, just over 9 percent of women reported working part‑time because of childcare responsibilities as compared to less than 1 percent of men. As a result, the precarity of women’s jobs is partly influenced by public policy on maternity benefits and childcare. [Emphasis added.]

(Vulnerable Workers and Precarious Work, at pp. 19‑20; see also Statistics Canada, Women in Canada: A Gender‑based Statistical Report (7th ed. 2017).)

[103]                     Judgments of this Court have also recognized that women face disadvantages in the workplace because of their largely singular responsibility for domestic work. The Court has acknowledged the sacrifices women make at work “for the sake of domestic considerations” (Moge v. Moge, [1992] 3 S.C.R. 813, at p. 861; New Brunswick (Minister of Health and Community Services) v. G. (J.), [1999] 3 S.C.R. 46, at para. 113, per L’Heureux‑Dubé J., concurring); and that “women bear a disproportionate share of the child care burden in Canada” (Symes v. Canada, [1993] 4 S.C.R. 695, at pp. 762‑63; see also Young v. Young, [1993] 4 S.C.R. 3, at pp. 49‑50, per L’Heureux‑Dubé J., dissenting).

[104]                     Recognizing the reality of gender divisions in domestic labour and their impact on women’s working lives is neither new nor disputable (see Beijing Declaration and Platform for Action, U.N. Doc. A/CONF.177/20, October 17, 1995, at paras. 155‑56 and 158). Elizabeth Shilton has eloquently described the link between the division of unpaid care work and women’s over‑representation in part-time work:

In twenty‑first‑century Canada, the male breadwinner family has largely vanished along with the idea of the “family wage”; women are almost as likely as men to belong to the paid workforce. Two constants remain, however. Employers continue to demand an “unencumbered worker,” along with the right to organize work without regard to workers’ care obligations. And gender roles within families have been slow to change. Care work still needs to be done, and women still bear most of the practical responsibility for doing it. In consequence, women are forced to manage family care without impinging on their work obligations. Their strategies — euphemistically labelled “choices” — often include part‑time and precarious forms of work that typically come with lower wages, fewer benefits, fewer promotional opportunities, and minimal or no retirement pensions. The impact on women’s economic welfare is compounded by stereotypical assumptions that women do not merit or want more responsible, higher‑paying jobs because they will inevitably prioritize family over work. The unequal burden of family care creates and reinforces women’s continuing inequality both inside and outside the workplace. [Emphasis added.]

(“Family Status Discrimination: ‘Disruption and Great Mischief’ or Bridge over the Work‑Family Divide?” (2018), 14 J.L. & Equality 33, at p. 35; see also Sheppard (2010), at p. 26; Richard P. Chaykowski and Lisa M. Powell, “Women and the Labour Market: Recent Trends and Policy Issues” (1999), 25 Can. Pub. Pol’y S1; Braun, at pp. 137‑38; Fredman (2011), at pp. 38, 45 and 181; Rivet and Santorineos, at p. 373; Suzi Macpherson, “Reconciling employment and family care-giving: a gender analysis of current challenges and future directions for UK policy”, in Nicole Busby and Grace James, eds., Families, Care‑giving and Paid Work: Challenging Labour Law in the 21st Century (2011), 13, at pp. 13‑30; Susan Bisom‑Rapp, “What We Know About Equal Employment Opportunity Law After Fifty Years of Trying” (2018), 22 Employee Rts. & Employment Pol’y J. 337, at pp. 348‑49.)

[105]                     Courts, by identifying adverse impact discrimination, have “been particularly effective in dealing with criteria which specifically disadvantage women with childcare responsibilities” (Fredman (2011), at p. 181). The European Court of Justice, for example, has held that providing workers with less favourable benefits based on their working hours can amount to adverse impact discrimination against women (see Jenkins v. Kingsgate (Clothing Productions) Ltd, C-96/80, [1981] E.C.R. I-911; Bilka‑Kaufhaus GmbH v. Weber von Hartz, C-170/84, [1986] E.C.R. I-1607; Rinner‑Kühn v. FWW Spezial‑Gebäudereinigung GmbH, C-171/88, [1989] E.C.R. I‑2743; Vroege v. NCIV Instituut voor Volkshuisvesting BV, C-57/93, [1994] E.C.R. I‑4541; Schönheit v. Stadt Frankfurt am Main, C-4/02 and C-5/02, [2003] E.C.R. I‑12575; see also Reg. v. Secretary of State for Employment, Ex parte Equal Opportunities Commission, [1995] 1 A.C. 1 (H.L.); Braun, at pp. 137‑40).

[106]                     All of these sources — and more — show the clear association between gender and fewer or less stable working hours. They provide powerful support for Ms. Fraser’s core argument: that the RCMP’s use of a temporary reduction in working hours as a basis for imposing less favourable pension consequences has an adverse impact on women. The first part of the s. 15(1) test has therefore been met.

[107]                     This leads me to the second part of the s. 15(1) inquiry: whether this adverse impact reinforces, exacerbates or perpetuates disadvantage.

[108]                     There is no doubt that it does. I agree with Ms. Fraser that the negative pension consequences of job‑sharing perpetuate a long‑standing source of disadvantage to women: gender biases within pension plans, which have historically been designed “for middle and upper‑income full‑time employees with long service, typically male” (Report of the Royal Commission on the Status of Pensions in Ontario (1980), at p. 116).

[109]                     The National Action Committee on the Status of Women (“NAC”) expressed concerns about gender biases within pension plans in a brief to the House of Commons Sub‑Committee on Equality Rights (presented in June 1985 by Louise Dulude and Carole Wallace). In the brief, the NAC described how pension plans treat women unequally:

The differences in impact of pensions on women and men are well known and amply documented. They are the results of the combined effects of the elements which make up the multilayered cake that is Canada’s pension system today.

 

. . .

 

Women are more affected by these inadequacies than men because they have a higher rate of turnover and drop out of the labour force more often than their male counterparts. As a result, the small proportion of female earners who are members of employer pension plans are exceedingly unlikely to ever collect decent pensions from that source. In fact, experts have said that many women who participate in employer‑sponsored pension plans would probably have been better off putting their contributions to them in a bank.

(Brief on Equality for Women in Pensions, Taxation and Federal Benefits to Parents, at pp. 2 and 8-9)

[110]                     Others have echoed these concerns. Elizabeth Shilton notes that although progress has been made in securing equal pension coverage for women, the level of benefits they derive from those pensions remains unequal (“Gender Risk and Employment Pension Plans in Canada” (2013), 17 C.L.E.L.J. 101, at pp. 110‑12). She links the gender biases in pension plans to their preference for “male pattern employment”:

From the beginning, pension plans were calibrated to the career trajectories of skilled workers whose training and experience were particularly valuable to their employers. The reward structures embedded in those plans therefore favoured permanent, full‑time workers with long service and relatively high pay — what has been called “male pattern employment.” Long after explicitly gendered pension plan rules were made illegal, typical benefit structures still forced lower‑paid, temporary or part‑time employees — those in typical “female pattern employment” — to subsidize the benefits of workers with more market power. This is true of all pension plans, although the way in which the gender dynamic works depends on the type of plan. [p. 112]

[111]                     The International Labour Organization has also commented on how increased periods of part‑time work result in lower pension benefits for women:

Throughout their life cycles, women accumulate disadvantages that pile up at older ages. Double or triple discrimination is often amplified as women advance in age. Women are especially vulnerable owing to their high numbers in unpaid, low‑paid, part‑time, frequently interrupted, or informal economy work. As a result they are less often entitled to any contributory pension benefits in their own right. Even if they are, their pensions are often significantly lower than those of men due to lower earnings and shorter contribution periods.

(Rights, jobs and social security: New visions for older women and men (2008))

[112]                     The structural inequality within pension plans has tangible impacts for women upon retirement. This Court has described the “feminization of poverty” as an “entrenched social phenomenon” in Canada (Moge, at p. 853). Claire Young has linked this problem to disparities in pension policies:

. . . when one examines statistics on income security in retirement, women are disproportionately worse off financially than men, with 7.6 percent of women having incomes below the low income cut off (LICO), which is colloquially called the poverty line, compared to 3.6 percent of elderly men . . . . [C]urrent Canadian pension policies are a major contributing factor to this income disparity.

 

. . .

 

. . . over 72 percent of those aged sixty‑five or older living below the poverty line are women. It is also important to note that single elderly women are the poorest of the poor in Canada, with 80 percent of unattached women over the age of sixty‑five living in poverty. [Emphasis added.]

(“Pensions, Privatization, and Poverty: The Gendered Impact” (2011), 23 C.J.W.L. 661, at pp. 663 and 665; see also Shilton (2013), at pp. 102‑3; Commission of Inquiry into Part-time Work, at p. 151; Gender Wage Gap Strategy Steering Committee, Final Report and Recommendations of the Gender Wage Gap Strategy Steering Committee (2016), at pp. 18 and 60‑61; Fredman (2011), at pp. 47‑48; Susan Bisom‑Rapp and Malcolm Sargeant, “It’s Complicated: Age, Gender, and Lifetime Discrimination Against Working Women — The United States and the U.K. as Examples” (2014), 22 Elder L.J. 1, at p. 99.)

[113]                     Pension design choices have, in sum, “far‑reaching normative, political and tangible economic implications for women” (Shilton (2013), at p. 140, quoting Bernd Marin, “Gender Equality, Neutrality, Specificity and Sensitivity — and the Ambivalence of Benevolent Welfare Paternalism”, in Bernd Marin and Eszter Zólyomi, eds., Women’s Work and Pensions: What is Good, What is Best? Designing Gender-Sensitive Arrangements (2010), 203, at p. 210). Because the RCMP’s design perpetuates a long‑standing source of economic disadvantage for women, the second stage of the s. 15(1) test is satisfied and there is a prima facie breach of s. 15 based on the enumerated ground of sex.

[114]                     In light of the conclusion that there is a prima facie breach of s. 15(1) based on sex, it is unnecessary to decide whether Ms. Fraser’s alternative argument that this Court recognize parental/family status as an analogous ground should succeed.[6] Some observations may be helpful, however, for future cases.

[115]                     The Attorney General was prepared to accept that the narrower ground of “parental” status should be recognized as an analogous ground under s. 15(1), but only for these proceedings. I am uncomfortable with this Court accepting a new analogous ground as a one‑off. It is either a sustainable legal principle that this Court should accept or it is not. It should not get a trial run subject to periodic review. Moreover, where it is protected in human rights statutes in Canada, parental status is part of family status, not a distinct category. I would be reluctant to sever them without submissions on what the implications are.

[116]                     In my respectful view, this is not the right case to resolve whether family/parental status should be recognized as an analogous ground under s. 15(1). Not only is recognizing a new analogous ground unnecessary to fully and fairly resolve Ms. Fraser’s discrimination claim, a robust intersectional analysis of gender and parenting — as this case shows — can be carried out under the enumerated ground of sex, by acknowledging that the uneven division of childcare responsibilities is one of the “persistent systemic disadvantages [that] have operated to limit the opportunities available” to women in Canadian society (Taypotat, at para. 17; see also Withler, at para. 43; Quebec v. A, at paras. 327‑32; Alliance, at para. 28; Centrale, at para. 35). Human rights cases in other jurisdictions confirm that claims of parental discrimination can be brought as claims of adverse impact discrimination on the basis of sex (see Fredman (2011), at pp. 181‑82; Shilton (2018), at p. 36; London Underground Ltd. v. Edwards (No. 2), [1999] I.C.R. 494 (E.W.C.A.); Phillips v. Martin Marietta Corp., 400 U.S. 542 (1971); Bostock v. Clayton County, Georgia, 140 S. Ct. 1731 (2020), at p. 1743).

[117]                     There is another more compelling basis for not definitively resolving the issue in this appeal: the record and submissions before us do not provide the necessary assistance in exploring the implications of such a step. There are several complex questions about recognizing family/parental status as an analogous ground that have not been addressed at any stage of these proceedings. There was only a brief discussion of family/parental status in Ms. Fraser’s factum, the issue was largely unaddressed in the submissions of the Attorney General, almost all the interveners and during oral argument,[7] and it was completely absent in the reasons of the Federal Court and Court of Appeal.

[118]                     The parties recognized that family status is a protected ground in most provincial human rights statutes, and that while there is no separate express protection for parental status, family status has been defined or interpreted to include protection for parents (British Columbia Law Institute, Human Rights and Family Responsibilities: Family Status Discrimination under Human Rights Law in British Columbia and Canada (2012), at p. 26). The question of what constitutes a prima facie case of family status discrimination has been the source of considerable “uncertainty and controversy” in the human rights arena (British Columbia Law Institute, at p. 10; see Ontario Human Rights Commission, The Cost of Caring: Report on the Consultation on Discrimination on the Basis of Family Status (2006), at p. 4; Campbell River & North Island Transition Society v. Health Sciences Assn. of British Columbia (2004), 28 B.C.L.R. (4th) 292 (C.A.); Brown v. Department of National Revenue (1993), 93 CLLC ¶17,013 (C.H.R.T.); Canada (Attorney General) v. Johnstone, [2015] 2 F.C.R. 595 (C.A.); Misetich v. Value Village Stores Inc. (2016), 39 C.C.E.L (4th) 129 (Ont. H.R.T.), at paras. 35-48; see also Shilton (2018); Sheila Osborne‑Brown, “Discrimination and Family Status: The Test, the Continuing Debate, and the Accommodation Conversation” (2018), 14 J.L. & Equality 87; Lyle Kanee and Adam Cembrowski, “Family Status Discrimination and the Obligation to Self‑Accommodate” (2018), 14 J.L. & Equality 61).

[119]                     But there were almost no submissions before us about whether or how the unsettled state of the human rights jurisprudence does or should affect the recognition of family/parental status under the Charter , about the definition or possible scope of “family” or “parental” status, or about the possibility of addressing parental or family status discrimination by recognizing other grounds (see Thibaudeau v. Canada, [1995] 2 S.C.R. 627, at pp. 722‑25, per McLachlin J., dissenting (“separated or divorced custodial parent”); Canada (Attorney General) v. Lesiuk, [2003] 2 F.C. 697 (C.A.), at para. 37 (“women in a parental status”)).

[120]                     Nor did we receive any submissions or evidence on how or whether recognition of family/parental status would affect protection for women above and beyond that available under the enumerated ground of sex. The record is similarly silent on the nature of the disadvantages that fathers may have experienced or continue to experience because of parenting responsibilities, or on the possible impact of recognizing a new analogous ground on fathers’ relationships with a co‑parent.

[121]                     And finally, we received no submissions on whether or how these questions are or should be relevant to the test for recognizing a new analogous ground under s. 15(1), a test which has itself been the subject of renewed scholarly attention (see Joshua Sealy‑Harrington, “Assessing Analogous Grounds: The Doctrinal and Normative Superiority of a Multi‑Variable Approach” (2013), 10 J.L. & Equality 37; Jessica Eisen, “Grounding Equality in Social Relations: Suspect Classification, Analogous Grounds and Relational Theory” (2017), 42 Queen’s L.J. 41).

[122]                     These are some of the issues that merit close examination by this Court, as do issues like the “growing and urgent need related to eldercare” (Ontario Human Rights Commission, at p. 12), and the implications of our evolutionary understanding from a conjugal‑centric meaning of “family”, to one more appreciative of the variations in intimate relationships that make up today’s households (see Elaine Craig, “Family as Status in Doe v. Canada: Constituting Family Under Section 15 of the Charter ” (2007), 20 N.J.C.L. 197, at pp. 207‑8). But these issues were barely addressed in this appeal.

[123]                     While recognizing multiple, interactive grounds of discrimination can allow for a fuller appreciation of the discrimination involved in particular cases, the gap in submissions and evidence means that critical questions about the implications of adopting family/parental status as an analogous ground were not explored in the record. That is not to say that this status should not eventually be recognized as an analogous ground, or that we should shy away from recognizing analogous grounds which raise complexities — rarely do enumerated or analogous grounds come neatly packaged — but before we do so, it seems to me to be wiser to have the benefit of sufficient argument and submissions so that the recognition, when it comes, pays full tribute to the breadth of what is at stake.

[124]                     And so to s. 1.

[125]                     Section 1 allows the state to justify a limit on a Charter  right as “demonstrably justified in a free and democratic society”. To start, the state must identify a pressing and substantial objective for limiting the Charter  right (R. v. Oakes, [1986] 1 S.C.R. 103, at pp. 138‑39). The Attorney General bears the burden of showing that classifying full‑time RCMP members who enter job‑sharing as part‑time workers and excluding them from accessing full‑time pension credit, achieves a compelling state objective. As the Court noted in Alliance, it is the limitation on equality rights that must be justified, not the legislative scheme as a whole:

Where a court finds that a specific legislative provision infringes a Charter  right, the state’s burden is to justify that limitation, not the whole legislative scheme. Thus, the “objective relevant to the s. 1 analysis is the objective of the infringing measure, since it is the infringing measure and nothing else which is sought to be justified” (RJR‑MacDonald Inc. v. Canada (Attorney General), [1995] 3 S.C.R. 199, at para. 144; R. v. K.R.J., [2016] 1 S.C.R. 906, at para. 62). [Emphasis in original; para. 45.]

[126]                     The Attorney General, in my respectful view, has identified no pressing and substantial policy concern, purpose or principle that explains why job‑sharers should not be granted full‑time pension credit for their service. On the contrary, this limitation is entirely detached from the purposes of both the job‑sharing scheme and the buy‑back provisions, which were intended to ameliorate the position of female RCMP members who take leave to care for their children. As the Hon. Gilles Loiselle, then President of the Treasury Board, said in support of amendments to the public service superannuation legislation:

I might mention too that this provision, like that for coverage of part‑time employees, would particularly benefit women who continue to be the employees with the greatest need for room to balance family and career commitments. Many women, for example, take advantage of extended periods of leave without pay for the purpose of caring for young children or for elders, and this provision would enhance their ability to return to work without undue financial hardship. [Emphasis added.]

(House of Commons Debates, vol. VI, 3rd Sess., 34th Parl., February 24, 1992, at p. 7487)

[127]                     The job‑sharing program had a similar objective, as confirmed in an affidavit filed as part of the Attorney General’s record for the application hearing:

The job-sharing policy was instituted to facilitate work‑life balance for members of the Force who, due to personal or family circumstances, would benefit from being able to work part‑time instead of taking extended leaves of absence in the form of LWOP. Job‑sharing was thus seen as being mutually beneficial as it enabled members to remain operationally connected to the Force while having a work schedule that better accommodated their individual circumstances. [Emphasis added.]

(A.R., vol. V, at p. 810)

[128]                     Job‑sharing was clearly intended as a substitute for leave without pay for those members who could not take such leave “due to personal or family circumstances”. It is unclear, then, what purpose is served by treating the two forms of work reduction differently when extending pension buy‑back rights. The RCMP’s plan provides buy‑back rights when a full‑time member reduces her hours from 40 to 0 to care for her child, but, inexplicably, withholds such rights if the same member for the same reasons reduces her hours from 40 to 10, 20, 30 or some other number. And this despite the RCMP benefitting from the member’s services in the latter scenario. I see no justification for this limitation, let alone a pressing and substantial one. The distinction becomes even more difficult to understand when considering that buy-back rights are available to members who have been suspended.

[129]                     In my respectful view, therefore, the government has not offered a compelling objective for the limitation on job‑sharing participants wishing to buy back full‑time pension credit.

[130]                     Since the prima facie breach cannot be justified under s. 1, it is a violation of s. 15 to preclude Ms. Fraser and her colleagues from buying back their pension credits.

[131]                     Finally, my colleagues’ reasons call for response.

[132]                     The version of s. 15(1) advanced in my colleagues’ reasons is essentially that advanced in the dissenting reasons in Alliance. They argued then, as they do now, that a finding of a breach would have a “chilling effect” on legislatures;[8] that the impugned legislation was not “the source of the differences in compensation between men and women” (at para. 97);[9] that the Court should not interfere with “incremental” efforts intended to narrow the gap between a group and the rest of society;[10] and that finding a s. 15(1) breach would place legislatures under a freestanding positive obligation “to act in order to obtain specific societal results such as the total and definitive eradication of gender-based pay inequities” (para. 65).[11]

[133]                     All of these propositions were squarely rejected by the majority in Alliance. Nothing, as far as I can see, has happened since Alliance was decided in 2018 to justify discarding its premises. And no one involved in this case argued that we should, except, inferentially, my colleagues, who tug at the strands of a prior decision they disagree with in search of the occasional phrase or paragraph by which they can unravel the precedent. Their arguments are based on conjecture not reality, calling to mind one writer’s wry observation that “setting straw men on fire is not what we mean by illumination”.[12]

[134]                     And, above all, they continue their insistent attack on the foundational premise of this Court’s s. 15 jurisprudence — substantive equality — in favour of a formalistic approach that embraces “a mechanical and sterile categorization process conducted entirely within the four corners of the impugned legislation” (Turpin, at p. 1332). This Court has consistently rejected this “thin and impoverished vision of s. 15(1)” (Eldridge, at para. 73), as have even the scholars cited by my colleagues.[13] It is unfortunate that as the global jurisprudence has increasingly embraced substantive equality, my colleagues continue to endorse an approach which evokes the rejected pre‑Charter  theory whose effect was to deny access to benefits when that access required accommodation based on difference.

[135]                     Whatever my colleagues’ definition of “rule of law” is, it must surely include the assumption that decisions of the Supreme Court will be respected not only by the public, but by members of the Court. And it must surely also include an assurance to those seeking constitutional protections that the ongoing repetition in dissenting reasons of rejected arguments will not require them “with each new case, [to] stand ready to defend the exact gains that have been won multiple times in the past” (Faraday, at p. 330; see also Jonnette Watson Hamilton and Jennifer Koshan, “Equality Rights and Pay Equity: Déjà Vu in the Supreme Court of Canada” (2019), 15 J.L. & Equality 1).

[136]                     For over 30 years, the s. 15 inquiry has involved identifying the presence, persistence and pervasiveness of disadvantage, based on enumerated or analogous grounds. Its mandate is ambitious but not utopian: to address that disadvantage where it is identified so that in the pursuit of equality, inequality can be reduced one case at a time. That is why there is a s. 15(1) breach in this case — not because women continue to have disproportionate responsibility for childcare and less stable working hours than men, but because the pension plan “institutionalize[s] those traits as a basis on which to unequally distribute” pension benefits to job‑sharing participants (see Faraday, at p. 318). This is ‘“discrimination reinforced by law’, which this Court has denounced since Andrews” (Centrale, at para. 33, quoting Andrews, at p. 172). Contrary to the views of my colleagues, there is nothing “extraordinary” about holding, as we do here, that such discrimination violates s. 15(1)  of the Charter . Based on our jurisprudence, it would be extraordinary if we did not.

[137]                     The final question relates to remedy.

[138]                     In my view, the appropriate remedy is a declaration that there has been a breach of the s. 15(1) rights of full‑time RCMP members who temporarily reduced their working hours under a job‑sharing agreement, based on the inability of those members to buy back full pension credit for that service. The methodology for facilitating the buy‑back of pension credit is for the government to develop, but any remedial measures it takes should be in accordance with this Court’s reasons. They should also have retroactive effect in order to give the claimants in this case and others in their position a meaningful remedy (Vancouver (City) v. Ward, [2010] 2 S.C.R. 28, at para. 20; Doucet‑Boudreau v. Nova Scotia (Minister of Education), [2003] 3 S.C.R. 3, at paras. 55‑58).

[139]                     I would allow the appeal with costs throughout.

The following are the reasons delivered by

                    Brown and Rowe JJ. (dissenting) —

I.               Introduction

[140]                     At one level, this appeal presents the simple question: is tying pension benefits to hours worked discriminatory? The Royal Canadian Mounted Police (“RCMP”) allows two or more of its regular members (“members”) to share one full‑time position through the option of “job‑sharing.” The pension benefits of members who job‑share, like all other members, are determined through the Royal Canadian Mounted Police Superannuation Act , R.S.C. 1985, c. R‑11  and the Royal Canadian Mounted Police Superannuation Regulations, C.R.C., c. 1393 (collectively, the “Plan”). Under the Plan, the pension benefits of members who job‑share are prorated to reflect the hours they worked during the job‑sharing period.

[141]                     In addition to job‑sharing, the RCMP has also given its members the option of taking leave without pay (“LWOP”). These members may “buy back” pension benefits when they return to work from an extended period of LWOP. 

[142]                     Through its job‑sharing policy and the LWOP provisions, the RCMP has sought to provide flexible working arrangements in recognition of the burden women face in pursuing a career due to the unequal distribution of childcare responsibilities in society. For members with childcare responsibilities, job‑sharing accommodates those who are able to remain operationally connected to the force and want to keep their policing skills up to date, while LWOP accommodates those who temporarily leave the force by enhancing their ability to return to work without undue financial hardship. The Plan and the RCMP’s policy on job‑sharing are not anathema to the vision of equality that underlies s. 15  of the Canadian Charter of Rights and Freedoms , but instead represent an attempt to accommodate employees in light of their particular circumstances.

[143]                     And yet, our colleague Abella J. finds these aspects of the Plan to be unconstitutional. She describes the historical disadvantages women have faced in the workplace and then concludes, in effect, that the Plan does not do enough to remedy these disadvantages. One may reach this conclusion as a matter of policy, but that is not the question to be decided. Rather, and at its most fundamental level, it is whether, as a matter of law, the Constitution empowers (or even requires) the courts to substitute their views as to how to remedy those disadvantages for those of the legislature and the executive.

[144]                     The circumstances here are extraordinary, in that it is acknowledged that Parliament was not obliged to enact the Plan (see, e.g., transcript, at p. 66), nor is it barred from repealing it (see Quebec (Attorney General) v. Canada (Attorney General), 2015 SCC 14, [2015] 1 S.C.R. 693, at para. 25; Quebec (Attorney General) v. Alliance du personnel professionnel et technique de la santé et des services sociaux, 2018 SCC 17, [2018] 1 S.C.R. 464, at para. 33). But is not the next extension of our colleague’s line of reasoning that governments (federal and provincial) have a positive duty under s. 15(1) to initiate measures that will remove all effects of historic disadvantage, and that they are constitutionally barred from repealing or even amending such measures? These are profoundly complex matters of public policy that no Canadian court is institutionally competent to deal with.

[145]                     Our colleague’s line of reasoning in this case lays the groundwork for just that position. Among other things, it effectively overrules this Court’s recent statement in Alliance that the state may “act incrementally in addressing systemic inequality” (para. 42 (emphasis added)). That the Plan does not eradicate disadvantage should not mean that it should attract censure as “discriminatory.” Rather, considered in its entirety and with proper attention to its object, the Plan is a package of benefits designed to balance the needs of some RCMP employees during their child‑raising years, prorated according to hours worked. On any permutation of s. 15(1), its effect surely cannot be to hinder government efforts to address pre‑existing inequality.

[146]                     This leads to a more fundamental concern presented by this appeal — a concern which, we observe, has been repeatedly made by legal commentators, but which has yet to be taken up by this Court. The gauge of “substantive equality” by which this Court has measured s. 15(1) claims of right, not having been defined (except by reference to what it is not — e.g. “formal equality”), has become an open‑ended and undisciplined rhetorical device by which courts may privilege, without making explicit, their own policy preferences. As we explain below, and with respect, this case is an instance of that inherent malleability being deployed so as to strike down a scheme which was, after all, designed to be ameliorative.

[147]                     The impugned provisions of the Plan are not unconstitutional. For the reasons that follow, we would dismiss the appeal. 

II.            Facts

[148]                     While we agree generally with the facts as recounted by our colleague, we would add some observations regarding the appellants’ employment status. Further, her description of the Plan obscures some key aspects of how it functions. This is not a small matter: understanding the Plan and the underlying legislative scheme in its entirety is key to adjudicating the s. 15 claim. As this Court has explained:

     Where . . . the impugned distinction is the denial of a benefit that is part of a statutory benefit scheme that applies to a large number of people, the discrimination assessment must focus on the object of the measure alleged to be discriminatory in the context of the broader legislative scheme, taking into account the universe of potential beneficiaries. [Emphasis added.]

(Withler v. Canada (Attorney General), 2011 SCC 12, [2011] 1 S.C.R. 396, at para. 3)

A.           The Appellants’ Employment Status

[149]                     Our colleague does not directly engage with the appellants’ employment status. At para. 21, she notes that the appellants argued that “the pension plan, properly interpreted, allows participants in the job‑sharing program to acquire full pension credit.” This passage refers to an issue before the Federal Court: the appellants argued that job‑sharers were “presumptively” full‑time members who agreed to work temporarily reduced hours or that they worked full‑time with periods of LWOP (2017 FC 557, at paras. 43‑44 (CanLII)). This argument was also advanced before this Court. And, having considered this issue in detail, the Federal Court found that the appellants worked part‑time while job‑sharing (para. 47). This finding was based on a thorough review of the record, including the 1997 Bulletin that introduced job‑sharing, the RCMP Administration Manual II.10, and the appellants’ Memoranda of Agreement (paras. 48‑53). We highlight that the Manual explicitly states that “[j]ob sharing is considered as part‑time employment” (A.R., vol. II, at p. 238 (emphasis added)). Further, the Federal Court expressly rejected the arguments that job‑sharers were partly on LWOP (at para. 55) or worked full‑time with temporarily reduced hours (para. 56). The Federal Court of Appeal upheld these findings, noting that the Federal Court’s conclusions on this point were “largely factual” (2018 FCA 223, [2019] 2 F.C.R. 541, at para. 33). Applying a standard of palpable and overriding error, it found no basis to interfere (paras. 33‑36).

[150]                     Despite this, our colleague repeatedly refers to job‑sharers as “full‑time members who temporarily reduced their working hours” (at para. 3; see also paras. 83, 97 and 138) or “full‑time RCMP members who job‑share” (paras. 5 and 25). She therefore seems to overturn this finding without setting out the applicable standard or explaining the error.

[151]                     This finding is key to the rest of our colleague’s decision. By treating job‑sharers as full‑time employees, she is able to say they are entitled to the same pension benefits as full‑time employees. With respect, this approach skates over key aspects of the Plan, most notably how it treats part‑time members and the different situations in which members take LWOP. In other words, our colleague does not account for the context of the entire scheme, which she must do in order to be faithful to Withler.

B.            The Operation of the Plan

[152]                     At the outset, it is important to note that the appellants do not allege that it is unfair in general for members who work part‑time to have their pensions adjusted for periods of part‑time work. Rather, the core of their argument is that they should be able to “buy back” pensionable service in the same way as members who take LWOP, a submission we dismiss below. To properly assess this submission, it is crucial to understand how the Plan operates and, in particular, how it applies in the same manner to full‑time and part‑time members except to make adjustments to account for hours actually worked.

[153]                     The Plan is a “contributory defined benefit pension plan,” which means that contribution rates are based on a percentage of a member’s earnings (A.R., vol. V, at p. 801; see also A. Kaplan and M. Frazer, Pension Law (2nd ed. 2013), at pp. 2‑3). Both members and the RCMP are required to contribute to the Plan (Act , ss. 5  and 29.2 ). A member’s pension is determined based on the years of “pensionable service,” that is, the period of service credited to a member at retirement (Act , s. 6 ). All members accrue pensionable service at the same rate regardless of whether they work part‑ or full‑time. This means that one‑year of part‑time work and one year of full‑time work both count as one year of pensionable service.

[154]                     This important nuance is missing from our colleague’s reasons, which speak only of full‑time pensionable service (see, e.g., para. 14).

[155]                     The only employment statuses provided for under the Plan are full‑time, part‑time, and LWOP. The Regulations define a “full‑time member” as one who is engaged to work 40 hours per week (s. 2.1; see also A.R., vol. V, at p. 803). Meanwhile, a “part‑time member” is described as a member who is engaged to work a minimum of 12 hours per week and is not a full‑time member (ss. 2.1 and 5.2(1)). Importantly, there are no specific provisions relating to job‑sharing in the Plan. This makes sense, given that the relevant policies define job‑sharing as a form of part‑time work. Job‑sharers are therefore treated as working part‑time during the period in which they job‑share.

[156]                     All members contribute to the pension fund at the same rate, set by Treasury Board as a percentage of their salary. Upon retirement, members receive a pension benefit that is proportional to the assigned hours of work: the benefits are prorated to account for any part‑time service. As the respondent’s pension expert explained, “This approach aims to ensure that part‑time members and full‑time members are treated equitably by tying pension benefits to assigned hours of work” (A.R., vol. V, at p. 805).

[157]                     Members’ pension benefits are based on the average annual pay received for the five best consecutive years of highest paid pensionable service. For periods of part‑time pensionable service, the average annual pay is based on the “full‑time equivalent” of the authorized rate of pay and is then prorated to reflect the hours actually worked. The respondent’s pension expert explains that “[t]his method ensures that members will not be penalized based on when in their career the part‑time service occurred” (p. 806).

[158]                     As we will discuss, the core of the appellants’ argument focuses on the “buy‑back” option available to members who take LWOP. It is therefore also important to understand how the “buy‑back” option works. Members who take LWOP have no assigned hours of work and are therefore not remunerated but maintain continuity of employment. For the first three months of an LWOP period, members must contribute to the pension fund the amount they would have contributed had they been working (Act , s. 6 (a)(ii)(A); Regulations, s. 10(1)(a) and (4)). Following that period, members may elect not to contribute for some or all of the balance of the LWOP period (Act , s. 7(1) ). However, members who choose to treat the remaining LWOP period as pensionable must pay two or two and a half times the amount they would have paid had they been working (Act , s. 7(1) ; Regulations, s. 10(1)(b)).[14] This is what is meant by “buying back” pensionable service.

[159]                     Crucially, pension benefits for members who take LWOP are determined based on the status they held immediately before taking LWOP (Regulations, ss. 5.4, 10(4) and 10.1). In other words, members who worked part‑time prior to taking LWOP and buy back that pensionable service will earn a part‑time pension benefit for the period of LWOP. 

[160]                     As the foregoing shows, members who “buy back” pensionable service following a period of LWOP are making contributions for periods in which they did not work. Conceptually, therefore, it makes sense to say that they are “buying back” that time. In contrast, members who job‑share make pension contributions throughout the job‑sharing period. We therefore agree with the respondent that it is inaccurate to speak of “buying back” service when it comes to job‑sharing. The hours worked while job‑sharing are already fully pensionable; there is no remaining time to be “bought back”. Job‑sharers worked part‑time hours and received part‑time pension benefits for the period they job‑shared. The appellants, however, are seeking to obtain a full‑time pension benefit in respect of a period where they have worked part‑time hours. To be clear, no other members are entitled to such a benefit. Even members who take LWOP are limited to the hours they worked prior to taking LWOP (e.g. the part‑time member who takes LWOP is able to buy back part‑time, but not full‑time, pension benefits for their time spent on LWOP). The appellants are, in this sense, asking to be put in a better position than everyone else under the Plan, and, indeed, under any of the other 10 public sector pension plans. Our colleague elides this vital point, which underlies our reasoning and undermines hers.

[161]                     Instead, she draws attention to the fact that members who are suspended can obtain full‑time pension credit (para. 25). The RCMP’s Administration Manual II.8 does state that “[t]he period of time a member is suspended without pay and allowances will count as pensionable service” (A.R., vol. II, at p. 229). But there is almost no evidence before the Court as to how this works. It may be that a member who worked part‑time prior to a suspension is entitled only to part‑time pension credit, which would seem to align with how the Plan otherwise works. We simply do not know. Moreover, during the hearing, the respondent stated this option is no longer available and has not been for some time, noting that the document relied on by the appellants dates back to 2003. It is not appropriate to ascribe any significance whatsoever to the situation of suspended members given this evidentiary void.

[162]                     In short, it is incumbent upon this Court in judging the constitutionality of the Plan to understand and account for how the scheme operates as a whole, rather than compare options that are available to different groups line‑by‑line. And, considering the Plan as a whole, it is clear that it accommodates various stages of a member’s life and career. It is meant to be flexible and meet different needs at different times.

III.         Analysis

[163]                     The appellants say the Plan violates s. 15  of the Charter  in its application to members who job‑share, by denying them the right to accrue full‑time pension credit for periods when they job‑shared for childcare reasons. In our view, it does not.

[164]                     We stress in coming to this conclusion that the Court is not called upon to decide whether the Plan represents good or bad policy on the part of Parliament (in the legislation) and the executive (in the regulations). The task of the Court, rather, is to assess whether the Plan respects the bounds of the constitutional obligations imposed on the state (Vriend v. Alberta, [1998] 1 S.C.R. 493, at para. 136).

[165]                     Section 15 protects individuals against state‑imposed discrimination. Like any other Charter  right, it applies to the actions of the state, and not to private acts of discrimination (s. 32; RWDSU v. Dolphin Delivery Ltd., [1986] 2 S.C.R. 573, at p. 597; P. W. Hogg, Constitutional Law of Canada (5th ed. Supp.), vol. 2, at pp. 55‑12 to 55‑13). In the case at bar, the appellants’ claim is premised on the relationship between their sex, the allocation of childcare responsibilities in society, and the fact they job‑shared. Their argument is that, because primary responsibility for childcare has historically fallen on women, the Plan ought to have allowed them to “buy back” additional pensionable service for the time they job‑shared to fulfill their childcare responsibilities.

[166]                     It is indisputable that women have historically been disadvantaged in the workplace in part by the demands of childcare. This Court has recognized this disadvantage, notably in matters of compensation, in Alliance, at para. 6, and Centrale des syndicats du Québec v. Quebec (Attorney General), 2018 SCC 18, [2018] 1 S.C.R. 522, at paras. 2, 58 and 138. In this case, the application judge was presented with, and accepted, evidence showing that “women continue to assume traditional roles in the home and [that] women are more likely than their male counterparts to scale back at work to respond to ‘role overload’ and work‑life conflict” (para. 72). She also accepted evidence that women continue to make up a larger proportion of the part‑time labour force, particularly between the ages of 25 and 44 when they are most likely to be raising children (para. 168).

[167]                     Further, the Federal Court of Appeal has described systemic discrimination as “a continuing phenomenon which has its roots deep in history and in societal attitudes. [And] cannot be isolated to a single action or statement” (Public Service Alliance of Canada v. Canada (Department of National Defence), [1996] 3 F.C. 789 (C.A.), at para. 16). And it exists within the private and public spheres. Some aspects of government employment policies, for example, have contributed to women’s systemic disadvantage (see, e.g., Newfoundland (Treasury Board) v. N.A.P.E., 2004 SCC 66, [2004] 3 S.C.R. 381). However, many private sources — that is, longstanding phenomena other than government policies — also contribute to women’s ongoing systemic disadvantage. A clear example of this is how parents share and expect each other to share domestic responsibilities, including childcare.

[168]                     In the present case, the Plan represents neither a public nor private source of ongoing systemic disadvantage. It does not contribute to women’s systemic disadvantage; nor does it reinforce, perpetuate, or exacerbate the pre‑existing disadvantage of women in the workplace which arises in part from unequal distribution of parental responsibilities. Rather, it seeks to ameliorate (although without eliminating) the effects of that pre‑existing disadvantage on women’s careers in the RCMP by providing employment options which allow them the flexibility to continue to pursue their careers while raising children. This case therefore raises the question: can a court strike down part of a statutory scheme for simply being insufficiently remedial? In our respectful view, and as we explain below, it cannot.

A.           Step One of the Section 15 Analysis

[169]                     As our colleague explains, this claim alleges adverse‑impact discrimination. We agree that the s. 15 test, as it is framed, can address such claims (paras. 48‑50). That test consists of the following two steps:

1.            Does the law, on its face or in its impact, create a distinction on the basis of an enumerated or analogous ground?

2.            Does the law fail to respond to the actual capacities and needs of the group and instead impose burdens or deny a benefit in a manner that has the effect of reinforcing, perpetuating, or exacerbating their disadvantage?

(Kahkewistahaw First Nation v. Taypotat, 2015 SCC 30, [2015] 2 S.C.R. 548, at paras. 19‑20; Alliance, at para. 25; Centrale, at para. 22.)

[170]                     That the test can account for adverse‑impact discrimination is shown by its inquiry into whether a law creates a distinction in its impact. As this Court recognized in Withler, in adverse‑impact cases, “the claimant will have more work to do at the first step” (para. 64).

[171]                     The concept of a “distinction” has been central to our s. 15 jurisprudence since Andrews v. Law Society of British Columbia, [1989] 1 S.C.R. 143, which described discrimination as “a distinction, whether intentional or not but based on grounds relating to personal characteristics of the individual or group” (p. 174 (emphasis added)). As the Court explained in Withler:

Inherent in the word “distinction” is the idea that the claimant is treated differently than others. Comparison is thus engaged, in that the claimant asserts that he or she is denied a benefit that others are granted or carries a burden that others do not, by reason of a personal characteristic that falls within the enumerated or analogous grounds of s. 15(1). [Emphasis added; para. 62.]

[172]                     As this passage demonstrates, the focus at step one upon identifying a distinction is consistent with the comparative nature of equality. Indeed, and as this Court has repeatedly emphasized, equality is “an inherently comparative concept” (R. v. Kapp, 2008 SCC 41, [2008] 2 S.C.R. 483, at para. 15; see also Andrews, at p. 164; Winko v. British Columbia (Forensic Psychiatric Institute), [1999] 2 S.C.R. 625, at para. 77; Withler, at para. 62). Comparison is relevant throughout both steps of the s. 15 analysis (Withler, at para. 61).

[173]                     Despite the comparative nature of equality, however, this Court has also cautioned that seeking out a “mirror comparator group” is not the proper method to evaluate s. 15 claims (Withler, at paras. 55‑60). As stated in Withler, “[t]he analysis is contextual, not formalistic, grounded in the actual situation of the group and the potential of the impugned law to worsen their [circumstances]” (para. 37; see also Centrale, at para. 135). Nonetheless, the identification of a distinction is crucial.

[174]                     Our colleague identifies a distinction between job‑sharers and full‑time members who work regular hours, who are suspended or who take LWOP (while failing, as noted, to consider individuals who work part‑time prior to taking LWOP) (paras. 25 and 83). Yet, she later rejects (at least facially) a comparison to members who take LWOP, dismissing it as a “formalistic” and “mirror comparator group” analysis (at paras. 93‑94), maintaining that job‑sharers are entitled to substantive equality relative to full‑time members (para. 94). As we explain below, however, it becomes evident upon closer examination that her rejection of a comparison to members who take LWOP is only superficial. The existence of the buy‑back provisions for members who take LWOP is the very reason that she finds a breach; absent that basis for comparison, the alleged breach disappears.

[175]                     As noted, step one of the s. 15 analysis considers whether there is a distinction “in . . . impact” (Centrale, at para. 22 (emphasis added)). A search for impact is a search for causation. The inquiry here is into whether the gap in outcomes is fully explained by pre‑existing disadvantage or whether state conduct has contributed to it. In other words, s. 15 is concerned with state conduct that contributes to — that is, augments — pre‑existing disadvantage (Taypotat, at para. 20, citing Quebec (Attorney General) v. A, 2013 SCC 5, [2013] 1 S.C.R. 61, at para. 332; Vriend, at paras. 75‑76).

[176]                     Further, the analysis is directed to “the effects of the challenged law or action on the claimant group” (Taypotat, at para. 18 (emphasis added), citing Quebec v. A, at para. 331). While disadvantage may persist with or without the impugned law or state action, a demonstration that it has the effect of contributing to that disadvantage is necessary.

[177]                     Establishing causation is particularly critical in instances where the state acts in order to address systemic discrimination. In such cases, policies that narrow a gap may fail to close it. Despite our colleague’s views to the contrary, this Court made clear in Alliance that the state does not have, by virtue of s. 15, a freestanding positive obligation to remedy social inequalities. Moreover, when the state does act with such a purpose, it can do so incrementally:

     The result of finding that Quebec’s amendments breach s. 15 in this case is not, as Quebec suggests, to impose a freestanding positive obligation on the state to enact benefit schemes to redress social inequalities. Nor does it undermine the state’s ability to act incrementally in addressing systemic inequality. But s. 15 does require the state to ensure that whatever actions it does take do not have a discriminatory impact (Vriend; Eldridge v. British Columbia (Attorney General), [1997] 3 S.C.R. 624, at paras. 72‑80). [Emphasis added; para. 42.]

This is consistent with the general focus of s. 15, as explained in Quebec v. A: “If the state conduct widens the gap between the historically disadvantaged group and the rest of society rather than narrowing it, then it is discriminatory” (para. 332 (emphasis added)). These clear and recent statements of this Court should mandate dismissal of this appeal.

[178]                     Our colleague overcomes the requirement of causation, and these statements, by relaxing the claimant’s evidentiary burden to the point of insignificance. She posits that “evidence of statistical disparity and of broader group disadvantage may demonstrate disproportionate impact,” although she adds that “neither is mandatory and their significance will vary depending on the case” (para. 67 (emphasis added)). With respect to “broader group disadvantage,” a single claimant’s “own evidence,” or even judicial notice, is all she requires (paras. 57 and 66‑67). While courts must evaluate evidence in light of “the proof which it was in the power of one side to have produced, and in the power of the other to have contradicted” (Snell v. Farrell, [1990] 2 S.C.R. 311, at p. 328, quoting Blatch v. Archer (1774), 1 Cowp. 63, 98 E.R. 969, at p. 970), ultimately the onus is on the claimant to establish causation. In many contexts, subjective anecdotal evidence is simply incapable of meeting this objective onus (e.g. Taypotat, at paras. 33‑34; Auton (Guardian ad litem of) v. British Columbia (Attorney General), 2004 SCC 78, [2004] 3 S.C.R. 657, at paras. 58 and 62; Gosselin v. Quebec (Attorney General), 2002 SCC 84, [2002] 4 S.C.R. 429, at paras. 8 and 47). Our colleague’s relaxed approach also risks overlooking the interests of the public good (B. W. Miller, “Majoritarianism and Pathologies of Judicial Review”, in G. Webber et al., Legislated Rights: Securing Human Rights through Legislation (2018), 181, at p. 196).

[179]                     With respect to statistical disparity, our colleague contends that “clear and consistent statistical disparities can show a disproportionate impact on members of protected groups, even if the precise reason for that impact is unknown” (para. 62). In this case, then, because there is a statistical disparity between women and men with respect to who has taken advantage of the job‑sharing program, and because there is evidence that women have historically borne a greater part of childcare responsibilities and formed a greater proportion of the part‑time labour force, she finds that the law has a disproportionate impact on women (paras. 97‑106).

[180]                     With respect, this analysis is unsound, since it assumes that correlation between the number of women who have taken advantage of the job‑sharing program and evidence of disproportionate childcare responsibilities falling upon women is the function of causation, whereas it might well be the function of independent factors. Correlation itself is not proof of causation. Indeed, where one is dealing with complex social and economic considerations, like sex and employment, one can readily assume that there are many factors involved, some of which will give rise to causation while others will simply be the result of coincidence (that is, caused by independent factors).

[181]                     But it is causation that must be demonstrated between the law and the disadvantage. It is not enough for our colleague to refer to a statistical disparity and a broader group disadvantage. Indeed, the presence of a statistical disparity is precisely what is to be expected where a law is enacted, as the relevant portions of the Plan were, to incrementally narrow a pre‑existing systemic disadvantage. When the law fails to completely eradicate such disadvantage, an element of disparity will obviously remain. It follows that to accept statistical disparity and broader group disadvantage as sufficient to demonstrate that a law creates a distinction in impact is to do away with this Court’s statement made only two years ago in Alliance that s. 15 neither imposes “a freestanding positive obligation on the state to . . . redress social inequalities” nor “undermine[s] the state’s ability to act incrementally in addressing systemic inequality” (para. 42). It would also represent an undisciplined judicial expansion of the scope of s. 15, which does not apply to private acts of discrimination, because it would render the state responsible for discrimination it has not caused.

[182]                     We turn now to the appellants’ argument that the Plan, on its face or in its impact, draws a distinction based on not just the enumerated ground of sex, but also on “family status,” or “parental status” which they argue should be recognized as an analogous ground.

[183]                     We agree with our colleague that it is inappropriate to recognize an analogous ground solely for the purpose of this litigation (para. 114). Not only is this inappropriate, this approach was squarely rejected in Corbiere v. Canada (Minister of Indian and Northern Affairs), [1999] 2 S.C.R. 203, at paras. 8‑9. We also agree that, because this case can be resolved on the basis of the enumerated ground of sex, it is, substantially for the reasons she gives, unnecessary and unwise to consider parental or family status. The recognition of an analogous ground has significant implications as it opens up the possibility of new lines of Charter  claims; once an analogous ground is recognized, it “always stand[s] as a constant marker of potential legislative discrimination” (Corbiere, at para. 10). Recognition of further analogous grounds should be left for cases where there is sufficient argument and evidence, and where it is necessary to do so. It should not be done on an ad hoc basis.

[184]                     Focusing, then, on the enumerated ground of sex, there are two ways in which a distinction can be framed in this case. Each is based to some extent on comparison, which is to be expected given the inherently comparative nature of s. 15. But neither is based on the “mirror comparator” group approach rejected by this Court, in which the comparator ‘“mirrors the characteristics of the claimant (or claimant group) relevant to the benefit or advantage sought’ except for the personal characteristic on which the claim [is] based” (Withler, at para. 49, quoting Hodge v. Canada (Minister of Human Resources Development), 2004 SCC 65, [2004] 3 S.C.R. 357, at para. 23). In other words, these distinctions do not look for a situation that is in every way identical to the claimant group except for the enumerated ground. Our analysis, rather, is contextual, considering the various facets of the Plan and the evidence that is available on the composition of RCMP membership. 

[185]                     The first distinction can be drawn by looking to full‑time members. Job‑sharers, unlike full‑time members, do not obtain a full 40 hours of weekly pensionable service. The distinction is said to be based on sex because members of the job‑sharing program are disproportionately women, whereas uninterrupted full‑time employment is a male pattern of employment. We agree that a distinction is shown here.

[186]                     A second way of identifying a distinction in this case is by comparison to members who take LWOP. Our colleague expressly rejects this as the relevant comparison (paras. 93‑94). Nonetheless, she begins her analysis by stating that “[u]nlike full‑time members who work regular hours, who are suspended or who take unpaid leave, full‑time RCMP members who job‑share are classified as part‑time workers under the Regulations and cannot . . . obtain full‑time pension credit for their service” (para. 25 (emphasis added; footnote omitted)). In other words, the reason that our colleague finds a distinction is based on a comparison to members who take LWOP. If the option to buy back pensionable service for periods of LWOP did not exist, the distinction would vanish. Further, and as we have already observed, not all members who take LWOP have the right to receive full‑time pension credit for their service. Only members who were working full‑time hours before taking LWOP may do so. Throughout her reasons, our colleague fails to account for this nuance.

[187]                     In any event, the substance of this alleged distinction is that, with respect to pensionable service, job‑sharers are limited to the numbers of hours they work per week, while members who take LWOP are not as they have the option to “buy back” pensionable service. But this distinction is not based on sex: there is no evidence that members taking LWOP are less likely to be women than members participating in the job‑sharing program. As the Court of Appeal noted, “there was very little evidence about the number of RCMP members who have opted to job share or to work part‑time and no evidence about those who have opted to take an unpaid leave of absence” (para. 17 (emphasis added)). Nor was there “any evidence to suggest that more men than women or more childless individuals than those with children had opted to take leaves without pay” (para. 52). Further, the RCMP’s Administration Manual II.5 on leave recognizes various approved reasons for LWOP: education, spousal relocation, care and nurturing of preschool‑aged children, and personal needs. Our colleague cannot simply assume, without evidence, that job‑sharers as a group are more likely to be women than members who take LWOP. Both options are specifically but not exclusively available for women.

[188]                     Therefore, the distinction between job‑sharing and LWOP is not based on an enumerated or analogous ground. The analysis fails at step one. However, bearing in mind that we have found that the comparison between job‑sharers (who do not obtain a full 40 hours of weekly pensionable service) and full‑time members (who do) qualifies at the first step of the s. 15 analysis as a distinction based on sex, we turn to the second step. Here, given the contextual nature of the analysis, taking LWOP into account remains important.

B.            Step Two of the Section 15 Analysis

[189]                     Having shown that the Plan creates a distinction that, in its impact, is based on sex, the second stage of the s. 15(1) analysis asks whether that distinction is discriminatory in that it fails to respond to the actual capacities and needs of the group and instead imposes burdens or denies a benefit in a manner that has the effect of reinforcing, perpetuating, or exacerbating the disadvantage of women (Andrews, at p. 182; Withler, at para. 31; Taypotat, at para. 20).

[190]                     This Court has said that historic disadvantage plays a significant role in identifying substantive discrimination (Law v. Canada (Minister of Employment and Immigration), [1999] 1 S.C.R. 497, at paras. 63‑67; Kapp, at paras. 15‑16 and 35‑36; Taypotat, at para. 20). Substantive discrimination, however, cannot be reduced to historical disadvantage. In some circumstances, laws can maintain significant disadvantage while treating individuals equally and without discrimination. For example, securities and property legislation represent state action that supports a system of private property ownership. Where the distribution of private property between men and women is unequal, this legislation allows that unequal distribution to persist. Nonetheless, the legislation is not discriminatory. Under our colleague’s approach, the second step of the s. 15 test serves only to check if the unequal impact of a law impacts a historically disadvantaged group; there is no analysis of whether the unequal impact corresponds with a group’s actual circumstances or needs or whether it is in any other sense substantively discriminatory.

[191]                     Though substantive discrimination has been described differently over the years, it has always required an element of arbitrariness or unfairness. This element has been most often expressed as a failure to respond to individuals’ actual capacities, needs and circumstances (Andrews, at pp. 174‑75; Law, at para. 70; Withler, at paras. 32 and 65; Taypotat, at para. 20). For a time, perpetuating prejudice and stereotyping were seen as essential features of substantive discrimination (Kapp, at paras. 23‑24; Withler, at paras. 34‑36). More recently, a more contextual analysis has been preferred, because arbitrary discrimination need not take the form of promoting negative attitudes (Quebec v. A, at paras. 327‑31; see also Alliance, at para. 28).

[192]                     Howsoever it has been expressed, this element of arbitrariness or unfairness has never been confused with a discriminatory purpose. A discriminatory purpose can indicate substantive discrimination, but is not required to establish it (Andrews, at p. 174; Quebec v. A, at paras. 325‑30; Centrale, at para. 35). Substantive discrimination is focused on effect rather than intention.

[193]                     The academic literature cited by our colleague supports requiring an element of arbitrariness or unfairness to establish discrimination. It indicates that, while discrimination need not be intentional, it is fundamentally a form of wrongful behaviour (D. G. Réaume, “Harm and Fault in Discrimination Law: The Transition from Intentional to Adverse Effect Discrimination” (2001), 2 Theor. Inq. L. 349, at pp. 351 and 376‑80; H. Collins and T. Khaitan, “Indirect Discrimination Law: Controversies and Critical Questions”, in H. Collins and T. Khaitan, eds., Foundations of Indirect Discrimination Law (2018), 1, at pp. 25‑29; S. Moreau, “What Is Discrimination?” (2010), 38 Philosophy & Public Affairs 143, at p. 146).

[194]                     Our colleague now suggests, however, that the sole focus of the substantive discrimination analysis is historical disadvantage (at para. 77) and that any factors relating to arbitrariness or unfairness must be confined to the s. 1 analysis (paras. 78‑80). This robs the substantive discrimination analysis of its purpose, departing significantly and without acknowledgment or justification from decades of jurisprudence.

[195]                     Our colleague’s removal of considerations of arbitrariness or unfairness from the s. 15 analysis has far‑reaching consequences for this appeal. As we identified above, a distinction arises based on sex when comparing members who job‑share to members with male pattern employment, that is, members who work 40‑hour weeks throughout their careers. Because the pension benefits of members in the job‑sharing program are prorated to reflect the lower number of hours they worked, they receive a lower level of pension benefits than members with a male pattern of employment.

[196]                     Pension benefits are best viewed as part of a whole compensation package, as they are a form of remuneration (IBM Canada Limited v. Waterman, 2013 SCC 70, [2013] 3 S.C.R. 985, at para. 4; Parry v. Cleaver, [1970] A.C. 1 (H.L.), at p. 16). In this case, members who job‑share receive no additional compensation to offset their lower level of pension benefits. On the contrary, they receive lower pay as well as lower pension benefits, because both pay and pension benefits are tied to hours worked. Our colleague suggests that offering a lower level of pension benefits to members in the job‑sharing program “perpetuates a long‑standing source of disadvantage” of women relative to men (para. 108). This disadvantage, she says, cannot be justified on the basis that job‑sharers choose to work fewer hours.

[197]                     We accept that in most contexts, choice cannot protect differential treatment from a finding of discrimination. In this case, however, the logical extension of our colleague’s argument is that if other part‑time members (i.e. those in arrangements other than job‑sharing) are predominantly women (as it has been suggested), they too are entitled to substantive equality relative to full‑time members in matters of compensation. Are all part‑time members therefore entitled to the option of contributing to the Plan as though they were full‑time members?

[198]                     Herein lies the significance of our colleague’s relegation of considerations of arbitrariness or unfairness to the s. 1 analysis. It is not, in general, discriminatory (in an arbitrary or unfair sense) for an employer to prorate benefits according to hours worked (see, e.g., Royal Commission on Equality in Employment, Report of the Commission on Equality in Employment (1984), at pp. 27‑28).[15] Employers are expected to treat employees equally within the context of an employment relationship that is predicated on an exchange of work for compensation. Prorating compensation, including benefits, according to work is not arbitrary or discriminatory when it responds to employees’ actual capacities and circumstances (Taypotat, at para. 20).

[199]                     But even were we to take our colleague’s recasting of the s. 15(1) analysis as legitimate, her open‑ended approach leaves much to be desired as a matter of logic. For example, if, as she says, it is discriminatory towards the appellants to tie pension benefits and other compensation to hours worked, why stop at allowing part‑time and job‑sharing members to “buy back” additional pension benefits? After all, full‑time members do not have to “buy back” their pensions. On our colleague’s logic, if hours worked are not relevant, then part‑time and job‑sharing members should receive a full‑time pension without buying back hours. And if compensation cannot be tied to hours worked, then part‑time and job‑sharing members should receive a full‑time salary as well. Taking our colleague’s argument to its natural conclusion shows the vast implications of her position. One searches in vain for a logical or rational stopping point to either the entitlements that would flow from her line of reasoning, or the scope of judicial intervention to award them.

[200]                     Conversely, we posit simply that employers must be able to compensate employees based on hours worked. This is our central point, which our colleague does not attempt to answer.

[201]                     While prorating pension benefits according to hours worked is not in itself discriminatory, it might be so on this Court’s jurisprudence if some groups of employees were to receive more favourable treatment than others where such treatment reinforces, perpetuates, or exacerbates disadvantage. As we have discussed, the distinction based on members who job‑share compared to those who take LWOP is not based on sex. However, in accordance with the contextual analysis of the broader scheme mandated by Withler, the LWOP provisions remain an important consideration at step two.

[202]                     We accept, as our colleague says (at para. 94), that the proper analysis as it was described in Withler must not devolve into a narrow comparison between members who job‑share and those who take LWOP. There is, however, a degree of inconsistency in our colleague’s approach as she states that members who job‑share are entitled to “substantive equality” to members with a male pattern of employment and yet focuses her comparison on members who take LWOP, who (unlike members with a male pattern of employment) have the option to buy back additional pension benefits. This is made plain by our colleague’s remedy: she does not require that all part‑time members should have the option of buying back the same pension benefits as full‑time members; rather, this option is made available only to those who have temporarily reduced their hours like those on LWOP. It is, therefore, the presence of LWOP, and buy‑back rights for members on LWOP, that render the Plan unconstitutional for our colleague. LWOP is the linchpin of her decision.

[203]                     With respect, this is contrary to the contextual approach mandated by Withler. Job‑sharing and LWOP are options that may be valuable to members at different points in their lives and careers. Job‑sharing allows members to continue to work with reduced hours, while LWOP allows members to maintain continuity of employment without working. Each of these programs represents a package of benefits and responsibilities designed to balance the needs of the employer and of members at various stages of their careers. It is inappropriate to cherry‑pick particular elements out of such packages and compare them in isolation, seeking line‑by‑line parity (Withler, at paras. 73, 76 and 79). Instead, a contextual analysis must consider the full packages and ask “whether the lines drawn are generally appropriate, having regard to the circumstances of the persons impacted and the objects of the scheme. Perfect correspondence between a benefit program and the actual needs and circumstances of the claimant group is not required” (para. 67). In other words, the focus must be on the “actual impact” of the law in its full context (P. J. Monahan, B. Shaw and P. Ryan, Constitutional Law (5th ed. 2017), at p. 469).

[204]                     Each benefit program will have its advantages and drawbacks. While members who take LWOP must pay the employer contribution for any pension benefits they buy back, the employer makes contributions for job‑sharers for each hour they work. While members who take LWOP receive no pay, job‑sharers receive income for the hours they work. The option to buy back pensionable service is an attractive feature of the LWOP package, and it is understandable that members in the job‑sharing program might want it. But on the whole, the record does not suggest that the LWOP package confers improved financial security or pension benefits when taking into account the job‑sharing program as well as continuous full‑time employment. Nor does it suggest that the lines drawn are inappropriate, having regard to all the circumstances.

[205]                     Offering pension benefits that are prorated to hours worked is not substantive discrimination, and it does not become substantive discrimination because members who take LWOP have the right to buy back hours of pension benefits. As a result, s. 15(1)  of the Charter  is not infringed, and there is no need to consider s. 1.

C.            Practical Implications

[206]                     While the foregoing is sufficient to dispose of the matter, we wish to highlight the practical implications of our colleague’s reasons for judgment.

[207]                     Governments must be afforded the latitude to act incrementally when addressing a deeply ingrained, complex and persistent social phenomenon such as inequality. (This assumes that the inequality arises from factors in society; where the government itself has created the inequality, matters are, as we have already indicated, somewhat different.) There are processes by which a government must set its priorities, allocate its budget, and obtain parliamentary approval of its programs. In designing legislation to address a particular equality issue, a government can draw on far more internal and external expertise than we judges can. As a result, it is better positioned than we are to appreciate the consequences of a particular course of law‑making, both upon society and upon public resources. With these practical realities in mind, we must also recognize that, were a government expected to remove all inequalities for all groups on every occasion it acted, it may be disinclined to act, given that any remedial scheme will inevitably be under‑inclusive in some respect. Governments would, understandably, become “reluctant to create any new [remedial] benefit schemes because their limits would depend on an accurate prediction of the outcome of court proceedings under s. 15(1)  of the Charter ” (Egan v. Canada, [1995] 2 S.C.R. 513, at para. 104, per Sopinka J.).

[208]                     To avoid this chilling effect, and to encourage governments to enact remedial legislation addressing pre‑existing disadvantage, this Court has (until now) judiciously accepted that governments may implement reforms “one step at a time, [and] addres[s] [the reforms] to the phase of the problem which seems most acute to the legislative mind” (R. v. Edwards Books and Art Ltd., [1986] 2 S.C.R. 713, at p. 772, quoting Williamson v. Lee Optical of Oklahoma, 348 U.S. 483 (1955), at p. 489). The focal point in assessing remedial legislation is not to ask whether the government has met “the gold standard” (Auton, at para. 62; see also paras. 59‑61), but to recognize that government

should not be obliged to deal with all aspects of a problem at once. It must surely be permitted to take incremental measures. It must be given reasonable leeway to deal with problems one step at a time, to balance possible inequalities under the law against other inequalities resulting from the adoption of a course of action, and to take account of the difficulties, whether social, economic or budgetary, that would arise if it attempted to deal with social and economic problems in their entirety, assuming such problems can ever be perceived in their entirety. [Emphasis added.]

(McKinney v. University of Guelph, [1990] 3 S.C.R. 229, at p. 317, per La Forest J.; see also pp. 318‑19; Schachter v. Canada, [1992] 2 S.C.R. 679, at p. 727, per La Forest J., concurring; Alberta (Aboriginal Affairs and Northern Development) v. Cunningham, 2011 SCC 37, [2011] 2 S.C.R. 670, at para. 41.)

[209]                     At the risk of repeating ourselves, we stress that, as recently as two years ago in Alliance, the Court affirmed this commitment to judicial restraint by permitting the government to address pre‑existing disadvantage incrementally (para. 42). And it also bears repeating that Alliance stated that, first, there is no freestanding positive obligation to remedy social inequities, and secondly, that the state is entitled to act incrementally to address such inequities.

[210]                     That is precisely what the impugned provisions of the Plan represent: an example of a government acting incrementally to address inequities that exist in society, when it has no obligation to do so, using provisions that do not in themselves have a discriminatory impact. Like our colleague, we note that both the LWOP buy‑back provisions and job‑sharing option are ameliorative (para. 126). Under this Court’s statement of the law in Alliance, these provisions should be upheld.

[211]                     Our colleague disagrees. Under her approach legislation must not simply be favourable or beneficial, but sufficiently so to achieve substantive equality: s. 15(1) is breached because the Plan, though part of a remedial scheme, perpetuates (that is, fails to remove) economic disadvantage for women (para. 113). In other words, the Plan is insufficiently remedial. 

[212]                     The practical effect of this decision is to abandon the foundational principles so recently affirmed in Alliance and to discourage governments from offering ameliorative programs (or, as in this case, employment options to its employees) in the future. This is because our colleague has, in effect, imposed a positive obligation on legislatures, where they attempt merely to ameliorate the effects of inequality, to eradicate those effects altogether. Such an obligation exceeds the ambit this Court has given s. 15(1), which, unlike certain other provisions of the Charter  that appear to compel government action (e.g. ss. 3, 14, 20 and 23), “does not impose upon governments the obligation to take positive actions to remedy the symptoms of systemic inequality” (Thibaudeau v. Canada, [1995] 2 S.C.R. 627, at para. 38, per L’Heureux‑Dubé J., dissenting, but not on this point; see also Auton, at paras. 2 and 41; Alliance, at para. 42; Centrale, at para. 33; Andrews, at pp. 163, 171 and 175; McKinney, at p. 318; Lovelace v. Ontario, 2000 SCC 37, [2000] 1 S.C.R. 950, at paras. 90‑92).

[213]                     Requiring that legislation be sufficiently remedial not only changes the scope of s. 15(1), but also pulls courts outside their institutional competence. The judiciary is ill‑equipped to grapple with the public policy and budgetary complexities in legislating benefit plans. “[A]dvancing the cause of human rights . . . . invites a measure of deference for legislative choice” (McKinney, at p. 318). This is because only legislatures have the institutional capacity to conduct the research and study necessary to assess how, and at what pace, its resources should be applied to most effectively address a particular pre‑existing equality issue (and ultimately, to oversee that implementation). Courts are not well placed to define the nature and scope of an obligation to enact sufficiently remedial legislation (Ferrel v. Ontario (Attorney General) (1998), 42 O.R. (3d) 97 (C.A.), at p. 113).

[214]                     Consider this Plan, for example. In light of our colleague’s decision, what is to stop an RCMP employee from working full‑time for a very short time before entering a job‑sharing agreement? That employee would apparently be permitted to buy back full‑time pension benefits, whereas the employee who entered the force as a part‑time member, and has since remained part‑time, will not. Relatedly, what does it mean for employees to “temporarily reduce” their hours? What if an employee entered a job‑sharing agreement to care for her children, but did not return to full‑time work once her children became adults? Of course, we do not know these things, and we cannot know them. Nor are we expected to know them, or even anticipate them. But for that same reason, we are expected to keep to the limits of our institutional competencies, and not fiddle with the complex mechanics of legislative schemes like the Plan.

[215]                     In the case at bar, any disadvantage the appellants face is caused not by the impugned provisions or any government action, but by the unequal division of household and family responsibilities and social circumstances such as the availability of quality childcare. The solution to addressing these underlying matters, which exist outside of the Plan and the purview of courts, is surely not to strike down remedial legislation. In our colleague’s view, however, these true causes of the disadvantage are “entirely irrelevant” (para. 71). She allows for judicial intervention whenever a court is able to identify in the case before it a related social circumstance it wishes to address. Courts are now empowered to engage in “transformation” of the law if they simply believe that “institutions and relations must be changed” (para. 36, citing British Columbia (Public Service Employee Relations Commission) v. BCGSEU, [1999] 3 S.C.R. 3, at para. 41, quoting S. Day and G. Brodsky, “The Duty to Accommodate: Who Will Benefit?” (1996), 75 Can. Bar Rev. 433, at p. 462). Respectfully, we say that this is not our role.

[216]                     A related and final point on the practical implications of our colleague’s decision. For 30 years, this Court has struggled to define the term “substantive equality.” An intelligible and principled definition continues to be elusive. Indeed, this case illustrates the difficulties posed by the slippery quality of “substantive equality” — the core value of our colleague’s decision (at paras. 47‑48) — and its constant shifting in this Court’s jurisprudence.

[217]                     While the Court has stated that substantive equality is not formal equality (Withler, at paras. 2 and 39; Kapp, at para. 15; Centrale, at para. 25; Hodge, at para. 25), it has said little to address what substantive equality is. Scant guidance has been offered beyond describing substantive equality as the “animating norm,” “goal,” “approach,” “engine,” and now the “philosophical premise” of s. 15(1) (Withler, at para. 2; Cunningham, at para. 38; Alliance, at para. 25; Centrale, at para. 25; Taypotat, at para. 17; majority reasons, at paras. 40, 42 and 48). Metaphor and ascription has prevailed over actual definition. Commentators have taken notice:

It is . . . essential [for this Court] to frame a conceptually rigorous understanding of substantive equality as it operates in Canadian equality jurisprudence.

 

. . .

 

. . . there has been no comparable agreement on substantive equality as a state of affairs, i.e., what a situation in which the ideal is instantiated through law would look like, as opposed to the ideal’s methodological dimension. Nor has substantive equality been given a positive definition by the Court. Rather, it has been defined negatively as an approach to section 15(1) contrasting with a formal equality approach. [Emphasis added.]

(A. R. Sangiuliano, “Substantive Equality As Equal Recognition: A New Theory of Section 15 of the Charter ” (2015), 52 Osgoode Hall L.J. 601, at pp. 606‑8)

Others have gone further, for example, noting that despite the efforts of “[m]any scholars . . . to flesh out the precise requirements of substantive equality,” this Court’s elaboration of same has been “sketchy and occasionally contradictory” (J. Eisen, “Grounding Equality in Social Relations: Suspect Classification, Analogous Grounds and Relational Theory” (2017), 42 Queen’s L.J. 41, at pp. 60‑61).

[218]                     To be clear, we do not seek to overturn the jurisprudence that our colleague recounts in her reasons. Rather, we aim to give effect to it. Our disagreement is about the meaning and requirements of substantive equality: we view her approach as lacking in the clarity and guidance necessary to give effect properly to the Charter ’s purposes, notably with regard to legislation that is fundamentally ameliorative. It is for this reason that we highlight the lack of a substantive definition surrounding the norm of “substantive equality” in this Court’s jurisprudence. The concept has not been defined in a manner that renders s. 15 rights, or even the criteria by which they are adjudicated, knowable in advance by claimants and the state, or applicable with any consistency by courts.

[219]                     This lack of definition ex ante is antithetical to any notion of judicial restraint. Where a legal test lacks defined bounds, courts applying it exercise truly arbitrary powers of review. And that is the point at which we have arrived with “substantive equality.” It has become an unbounded, rhetorical vehicle by which the judiciary’s policy preferences and personal ideologies are imposed piecemeal upon individual cases. Consider our colleague’s approach here: legislation that is ameliorative in both intent and effect is judicially reconfigured because it is not ameliorative enough, or more precisely, not ameliorative in ways our colleague would prefer. It is also a prime example of how the goalposts of “substantive equality” are constantly on the move, evidenced most clearly by our colleague’s abandonment of the prudent guidance in Alliance, at para. 42, regarding incremental measures to alleviate systemic inequality. Indeed, her approach in this case lends support to Professor Young’s damning criticism that substantive equality is “an unelaborated, cryptic guidepost pointing to the equality outcome the author prefers” (M. Young, “Unequal to the Task: ‘Kapp’ing the Substantive Potential of Section 15” (2010), 50 S.C.L.R. (2d) 183, at p. 185). Justice Oliver Wendell Holmes’s observation, made in Baldwin v. Missouri, 281 U.S. 586 (1930), at p. 595, regarding the U.S. doctrine of substantive due process, is also apposite:

As the decisions now stand, I see hardly any limit but the sky to the invalidating of those rights if they happen to strike a majority of this Court as for any reason undesirable. I cannot believe that the [Fourteenth] Amendment was intended to give us carte blanche to embody our economic or moral beliefs in its prohibitions. Yet I can think of no narrower reason that seems to me to justify the present and the earlier decisions to which I have referred.

[220]                     The result of all this is corrosive of the rule of law. Our colleague wonders aloud what our definition of the rule of law is (para. 135). We share the views of jurists such as Lord Bingham and Sharpe J.A.: the concept of the rule of law has interlocking components (see T. Bingham, The Rule of Law (2010), at pp. 160‑70; R. J. Sharpe, Good Judgment: Making Judicial Decisions (2018), at pp. 122‑24). One is pertinent here: Canadians should be governed by rules, stated and knowable in advance, that enable them to guide their conduct. As Sharpe J.A. writes:

. . . the [Supreme] [C]ourt has insisted that there must be an intelligible standard, capable of providing “an adequate basis for legal debate . . . as to its meaning by reasoned analysis applying legal criteria.”

 

     In Irwin Toy . . . the majority wrote that . . . . “[w]here there is no intelligible standard” and where the decision‑maker has been “given a plenary discretion to do whatever seems best in a wide set of circumstances,” the essential minimum requirements of the rule of law are not met.

(p. 123, quoting R. v. Nova Scotia Pharmaceutical Society, [1992] 2 S.C.R. 606, at p. 639; Irwin Toy Ltd. v. Quebec (Attorney General), [1989] 1 S.C.R. 927, at p. 983)

[221]                     Our colleague’s appeal to stare decisis, we say with respect, therefore misses the point. The issue is not whether this Court’s s. 15 jurisprudence should be “respected” (para. 135). The issue, rather, is whether that jurisprudence as she has interpreted it states a standard that is practically knowable and reasonably predictable as to results. In our view, the concept of “substantive equality” has become so vague that it is impossible for claimants or legislatures to anticipate its demands in advance. Legislatures are, in effect, expected to hit a moving target, as “the Court continues to revise its analytical approach to section 15(1) without ever overruling or even really disapproving of its earlier judgments on the aspects of those judgments that have now been reversed” (J. Koshan and J. Watson Hamilton, “Meaningless Mantra: Substantive Equality after Withler” (2011), 16 Rev. Const. Stud. 31, at p. 61). This “revisionist approach” — the constant moving of the goal posts of “substantive equality” to suit judicial preference — “will undoubtedly continue to cause further confusion” (p. 61).

[222]                     This suggests another, less normative, but more practical concern: if the demands of substantive equality could be predicted, governments might actually be able to comply with them. Instead,

[i]n the view of many commentators, the equality jurisprudence, despite the Court’s self‑identified efforts to establish clear interpretations, has been muddled and inconsistent . . . . Put simply, it lacks the coherence to offer serious guidance about how to realize substantive equality “on the ground”.

(P. Hughes, “Supreme Court of Canada Equality Jurisprudence and ‘Everyday Life’” (2012), 58 S.C.L.R. (2d) 245, at pp. 254‑55)

In other words, legislatures can comply only with rules they can know well enough to abide by. How can a legislature know what any given court will determine to be sufficiently remedial? The immensity of the obligation that our colleague foists upon legislatures and governments — in both its unpredictability and in the amount of resources it requires — is such that none of them can reasonably be expected to bear it. This wholly disregards La Forest J.’s apt admonition in McKinney that inequality issues cannot even be fully understood, let alone remedied, all at once.

[223]                     The failure to properly define the scope of s. 15(1) also has the practical effect of pushing the bulk of the analysis to s. 1 (e.g. majority reasons, at paras. 79‑80). As a result, courts are not to engage in a substantive analysis of discrimination (where they have a comparative advantage relative to the legislature), but rather in the evaluation of policy (where they do not) (Law Society of British Columbia v. Trinity Western University, 2018 SCC 32, [2018] 2 S.C.R. 293, at paras. 190 and 192, per Rowe J., concurring). This fails to attend to the limits inherent in s. 15. By their very nature, “the demands of equality [cannot] be determined without considering the needs and circumstances of persons and groups in addition to the claimant,” including the practical, moral, economic and social underpinnings of the legislation in question (B. W. Miller, “Justification and Rights Limitations”, in G. Huscroft, ed., Expounding the Constitution: Essays in Constitutional Theory (2010), 93, at p. 106; see also p. 100). Otherwise, we risk allowing “for the intermediate conclusion of a right infringement to trade on the higher prestige and greater strength of a moral right that provides an undefeated reason for action” (at p. 96), watering down the significance of a finding of a s. 15(1) infringement beyond recognition, and requiring the state to justify even its most trivial decisions.

[224]                     In defining substantive equality, courts must bear in mind two considerations. First, s. 15(1) cannot actually guarantee equality, in its broadest sense, throughout society. Systemic disadvantage is just that — systemic, being rooted in social attitudes and institutions. This does not mean that systemic discrimination cannot or should not be addressed; rather, it simply means that s. 15(1) is limited in its capacity to do so. Section 15(1) responds only to state action, and judicial review of state action is ill‑suited to implementing the kind of wide‑ranging institutional or policy changes that are necessary to fully address systemic disadvantage. That this is so becomes apparent when one considers that all of the reports cited by our colleague are directed towards government, not court, action (e.g. Royal Commission on the Status of Women in Canada, Report of the Royal Commission on the Status of Women in Canada (1970), at p. vii). Secondly, bringing analytical discipline to s. 15(1) must start by reaffirming that equality is an inherently comparative notion (Withler, at paras. 61‑67; Koshan and Watson Hamilton, at pp. 45‑46). While mirror comparators have proved unworkable, the analysis must be imbued with a measure of comparison in order to avoid what our colleague now endorses: an unbounded, unpredictable search for “inequality.”

[225]                     To all this, our colleague suggests (at paras. 133‑34) that we — and, by extension, those scholars and judges who have questioned the utility and meaning of the Court’s jurisprudence on “substantive equality” — advance a straw man argument which, at root, is an argument for “formal equality.” But, and again with respect, it is our colleague who marshals a straw man argument. Our reasons apply an approach to substantive equality firmly rooted in this Court’s jurisprudence, including Alliance. The conclusion we reach was shared by all four lower‑court judges who tried to apply this Court’s jurisprudence to the facts of this case.

[226]                     Our colleague, on the other hand, casts aside para. 42 of Alliance as a useless “stran[d]” (at paras. 132‑33) and sidesteps Withler’s call for a contextual approach, which notably arose precisely in the circumstances of a pension scheme. In fact, her reasons are entirely devoid of any consideration of the approach to pension schemes necessitated by Withler.

[227]                     Indeed, more telling than what our colleague does say in response is what she does not say. Notably, she does not explain what “substantive equality” means (save by reference to what it is not — “formal equality”). She fails to define “substantive equality” in terms that allow its meaning to be understood so that the requirements of s. 15 can be practically knowable and reasonably predictable in advance. Thus loosely defined, substantive equality is almost infinitely malleable, allowing judges to invoke it as rhetorical cover for their own policy preferences in deciding a given case. Such vast and little‑bounded discretion does not accord with, but rather departs from, the rule of law.

IV.         Conclusion

[228]                     As we see it, the sole reason the Plan is being judicially reviewed is because Parliament and the government tried to be accommodating in their employment options. If they had not offered pension buy‑back rights for members who take LWOP, there would be no basis for judicial intervention at all. The upshot of our colleague’s reasoning is that the public is now burdened with new financial obligations, simply because Parliament and the executive dared to address pre‑existing inequality incrementally, instead of taking more radical measures to eliminate it. In the future, they may well reason that inaction is the safer route.

[229]                     Similar issues will undoubtedly arise with any other social welfare legislation or government attempts to remedy systemic disadvantage. By reserving the right to arbitrarily second‑guess and undo any legislation that attempts to incrementally address systemic disadvantage, the Court makes it more practically difficult for legislatures and governments to implement policies that promote equality. Put simply, we see restricting the government’s ability to incrementally address disadvantage as a peculiar way to promote equality.

[230]                     We would therefore dismiss the appeal.

The following are the reasons delivered by

                    Côté J. (dissenting) —

I.               Introduction

[231]                     Discrimination on the basis of sex is of the most invidious order. Like race, sex is an innate and immutable characteristic, and bears no relevant relationship to capability. Without question, women have faced a prolonged fight for equal treatment under the law, a fight marked by a society where women have historically been disadvantaged and where they continue to be so today: Public Service Alliance of Canada v. Canada (Department of National Defence), [1996] 3 F.C. 789 (C.A.), at para. 16; Quebec (Attorney General) v. Alliance du personnel professionnel et technique de la santé et des services sociaux, 2018 SCC 17, [2018] 1 S.C.R. 464. But that is not the question before this Court. Nor is the question before this Court whether the impugned legislative provisions[16] are irrational, illogical, or even under-inclusive — that a law is not perfect, or even excludes some, does not make it per se unconstitutional: Gosselin v. Quebec (Attorney General), 2002 SCC 84, [2002] 4. S.C.R. 429, at para. 55; Andrews v. Law Society of British Columbia, [1989] 1 S.C.R. 143, at pp. 168-69. Rather, over the course of three decades, this Court has carefully crafted a test to assess whether a particular form of alleged discrimination is discrimination in fact and runs afoul of the guarantee of equal treatment under the law found in s. 15  of the Canadian   Charter of Rights and Freedoms . A crucial element of s. 15(1)  of the Charter  is that it enumerates the grounds of discrimination that fall within its constitutional prohibition. Here, with respect, I cannot agree with my colleagues that the impugned provisions of the pension plan create a distinction on the basis of the enumerated ground of sex. I proceed to elucidate why.

II.            Analysis

[232]                     My colleagues offer a comprehensive overview of this Court’s s. 15(1) jurisprudence, so I do not purport to do so here. While there exists clear disagreement between them as to certain elements of doctrine, I understand my colleagues and I to agree on the following: to prove discrimination under s. 15(1), including in cases of adverse effect discrimination, a claimant must (i) show that a law creates an adverse distinction based on an enumerated or analogous ground, and (ii) show that the law perpetuates, reinforces, or exacerbates pre-existing disadvantage (majority reasons, at paras. 27 and 50; Brown and Rowe JJ.’s reasons, at para. 169; R. v. Kapp, 2008 SCC 41, [2008] 2 S.C.R. 483, at para. 17; Withler v. Canada (Attorney General), 2011 SCC 12, [2011] 1 S.C.R. 396, at paras. 30 and 61; Quebec (Attorney General) v. A, 2013 SCC 5, [2013] 1 S.C.R. 61, at para. 185; Kahkewistahaw First Nation v. Taypotat, 2015 SCC 30, [2015] 2 S.C.R. 548, at paras. 19-20; Centrale des syndicats du Québec v. Quebec (Attorney General), 2018 SCC 18, [2018] 1 S.C.R. 522, at para. 22).

[233]                     Here, I am of the view that there is no need to proceed to the second step of the analysis, as my colleagues do, because no distinction can be made out on the basis of sex under step one.

[234]                     With the greatest respect, allowing the appeal on the basis of sex simpliciter is an attempt to square a circle. The effect of the impugned provisions of the pension plan is to create a distinction not on the basis of being a woman, but being a woman with children. In other words, a distinction exists not because one is a woman, but because one has caregiving responsibilities. It is telling that the statistics used by my colleague Abella J. to make her point — setting aside for now the question of their validity and even their appropriate role in a s. 15(1) analysis — are all in reference to women with children:

In my respectful view, the use of an RCMP member’s temporary reduction in working hours as a basis to impose less favourable pension consequences plainly has a disproportionate impact on women. The relevant evidence — the results of the system — showed that:

 

•               RCMP members who worked reduced hours in the job‑sharing program were predominantly women with young children.

 

•               From 2010-2014, 100 percent of members working reduced hours through job-sharing were women, and most of them cited childcare as their reason for doing so.

 

. . .

 

         The data [the Commission of Inquiry into Part-time Work] collected suggested that almost all job-sharing participants were women, and that “[t]he arrival of a new baby was the most common primary reason for initiating job sharing” (pp. 177-78).

 

         The Report of the Commission on Equality in Employment (1984, Rosalie Silberman Abella, comm.) expanded on the link between part-time work and childcare . . . . [Emphasis added; paras. 97 and 100-101.]

[235]                     The notion of women with children is inextricable from the evidence presented. It seems to me, then, that caregiving status is the only distinction created by the pension plan. In other words, the distinction in this case exists not on the basis of being a woman, but on the basis of needing to take care of someone: see Taypotat, at para. 21 (holding that under step one a claimant must “demonstrate that the law at issue has a disproportionate effect on the claimant based on his or her membership in an enumerated or analogous group”) (emphasis added); contra Centrale, at para. 28 (finding that step one was satisfied on the basis that “the claimants disproportionately suffer an adverse impact because they are women”) (emphasis in original). One does not job-share because one is a woman; one job-shares because one needs to take care of someone: Appellants’ Affidavits, A.R., vol. II, at pp. 129 et seq., and vol. III, at pp. 327-42; 2017 FC 557, at para. 22 and Annex A (CanLII).

[236]                     Importantly, I wish to draw particular attention to the following illustration. Consider, for instance, same-sex male couples who also have to bear the burden of taking care of their children. Consider further those individuals who bear the burden of taking care of their aging parents or spouse. These individuals, along with women with children, will all be under disproportionate pressure to job-share due to their caregiving responsibilities. Those individuals with no caregiving responsibilities will have no such pressure to job-share. The impugned provisions therefore create a distinction on the basis of caregiving responsibilities, not sex simpliciter: Withler, at para. 62.

[237]                     As I see it, it is essential to consider the nature of the claim before this Court. Indeed, the appellants make their claim on behalf of women with children, and not simply women — nobody has asserted a claim on behalf of women without children. It is critical to the claim, then, that the appellants had caregiving responsibilities in relation to children that made them decide to job-share. This explains why the appellants argued their claim on appeal on the basis of “intersecting” grounds of sex and parental or family status: A.F.; 2018 FCA 223, [2019] 2 F.C.R. 541, at paras. 3 and 42.

[238]                     However, of crucial importance to the disposition of this appeal is the fact that caregiving, parental, or family status is not recognized by this Court as an analogous ground under s. 15(1)  of the Charter , nor would I recognize any of them as such here. Indeed, my colleague Abella J. provides compelling reasons not to (at paras. 119-23), as do my colleagues Brown and Rowe JJ. (at para. 183), and I find myself in agreement that it would be inappropriate to do so. Accordingly, in light of the conclusion that any distinction here depends not on sex but on caregiving responsibilities alone and that this Court cannot recognize caregiving, parental, or family status as an analogous ground in this case, the appellants’ claim must fail at step one of the s. 15(1) analysis.

[239]                     Thus, at least doctrinally, this case is relatively straightforward. Lengthy reasons are elicited only by virtue of the attempt to fit the claim under the enumerated ground of sex simpliciter. However, it is clear that the distinction created by the pension plan manifests itself not as a result of sex alone, but as a result of a combination of sex with caregiver status. If the majority wishes to allow the appeal, then the more doctrinally sound method would be to either recognize intersecting grounds as the appellants urge, or recognize a new analogous ground.[17] Without doing so, however, this Court has only attempted to square a circle. And, as a result, doctrinal uncertainty seeps into this Court’s s. 15(1) jurisprudence and obscures, rather than illuminates, the way forward.

[240]                     This brings me to my next point — responding to the analytical and doctrinal methodology needed to support the conclusion that the pension plan creates a distinction on the basis of sex alone.

[241]                     First, in order to support the finding that the impugned provisions create a distinction on the basis of sex (without recognizing an intersecting or analogous ground of parental, family or caregiving status), my colleague Abella J. says that, in effect, discrimination on the basis of childcare is tantamount to discrimination on the basis of sex due to their historical association with one another: see paras. 98 et seq. This argument follows the rationale of Brooks v. Canada Safeway Ltd., [1989] 1 S.C.R. 1219, where the Court recognized that pregnancy-based discrimination necessarily constituted sex-based discrimination. Indeed, my colleague Abella J. posits that evidence about certain groups “will show such a strong association with certain traits — such as pregnancy with gender — that the disproportionate impact on members of that group ‘will be apparent and immediate’”: para. 61, quoting Taypotat, at para. 33.

[242]                     Setting aside for now the doctrinal proposition that disproportionate impact is sufficient to meet step one, I am of the view that there is a meaningful distinction between pregnancy and sex, on the one hand, and caregiving status and sex, on the other. This case is accordingly unlike Brooks. In Brooks, the Court held that a corporate insurance plan which denied benefits to employees during pregnancy discriminated on the basis of sex. Chief Justice Dickson grounded this conclusion on the fact that pregnancy, by definition, affects only women. Writing for the Court, Dickson C.J. explained that “[w]hile pregnancy-based discrimination only affects part of an identifiable group, it does not affect anyone who is not a member of the group”: p. 1247. The present case is manifestly different. Caregiving status can be separated from sex. In Brooks, an insurance plan that discriminated against pregnant employees necessarily discriminated against women. Here, impugned provisions of the pension plan that discriminate against those with caregiving responsibilities do not necessarily discriminate against women. In other words, caregiving, unlike pregnancy, is not, by definition, associated with sex. Rather, same-sex couples with children and other individuals with caregiving responsibilities will also all be disproportionately affected. The appellants — women with children — are not “denied a benefit that others are granted . . . by reason of a personal characteristic that falls within the enumerated . . . grounds of s. 15(1)”: Withler, at para. 62. The impugned provisions create a distinction on the basis of caregiver status, not sex simpliciter.

[243]                     Accordingly, the only remaining way to support the conclusion that the pension plan discriminates against women — without recognizing an intersecting or analogous ground or relying on the Brooks argument — is to dispose of any requirement of causation, nexus, or tether between the impugned provisions and their effect, and look only to the statistical disparity in results (i.e. women are disproportionately affected). Indeed, my colleague Abella J. takes this doctrinal step and seemingly reduces the step one analysis to a mere search for disproportionate impact evidenced by statistical disparity: she says that “in order for a law to create a distinction based on prohibited grounds through its effects, it must have a disproportionate impact on members of a protected group. If so, the first stage of the s. 15 test will be met” (para. 52 (emphasis added); see also paras. 5, 63, 66-67, 70 and 84). My colleague then expresses “agree[ment]” (at para. 67) with the proposition that statistical disparity “will be sufficient” by itself to demonstrate disproportionate impact (para. 66, quoting D.H. v. the Czech Republic, No. 57325/00, ECHR 2007-IV, at para. 188). This doctrinal development warrants significant caution — and not just because it confounds a necessary condition with a sufficient one.[18]

[244]                     Disproportionate impact alone cannot be sufficient to meet step one of the s. 15(1) analysis. In other words, simply pointing to the fact that the majority of job‑sharers are presently women with children cannot in itself be sufficient to say that step one has been met. Otherwise, for example, a law that regulates the top one percent of income earners in Canada would proceed past the step one analysis simply by virtue of the fact that the top one percent of income earners in Canada are majority male.[19] Analogously, a law regulating the nursing profession would proceed past the step one analysis simply by virtue of the fact that the nursing profession is majority female.[20] Surely, the aforementioned examples are not instances of prima facie discrimination, yet they exemplify how, if disproportionate impact alone were sufficient, step one would become a mere rubber stamp in cases of adverse effect discrimination, rather than a step at which “the claimant will have more work to do”, belying the sage guidance from Withler: para. 64. Worse yet, if statistical disparities alone were sufficient, the s. 15(1) analysis would, in effect, be replaced with a green light to s. 1, where the burden is reversed and placed on the government. My colleagues may very well consider such a state of affairs to be appropriate or desirable, and I cast no judgment in this regard; however, I simply wish to point out that such a doctrinal development is not currently supported by this Court’s jurisprudence — or if it is, then with respect, it requires more justification or clarification.

[245]                     Relatedly, I express further trepidation over the majority’s potential invitation for statistics-based litigation: paras. 58-59, 62-63 and 66-67. In effect, as I see it, parties may now routinely proffer and challenge statistical evidence (my colleague invites claimants to “rely . . . on their own evidence . . . rather than on government reports, academic studies or expert testimony” (at para. 57)), and trial judges must now become arbiters of statistics and their veracity (my colleague says that the “weight given to statistics will depend on, among other things, their quality and methodology” (at para. 59)), thereby bolstering their findings, which are owed deference upon appellate review (and which are findings the judiciary is institutionally ill-equipped to be making (see, e.g., P. Yowell, Constitutional Rights and Constitutional Design: Moral and Empirical Reasoning in Judicial Review (2018), at pp. 70-71)). The nature of statistics means that they are presented to courts frozen at a point in time, yet in reality, they are constantly shifting — it cannot be that the very same law that is constitutional one day is unconstitutional the next based solely on statistical evidence.

[246]                     On this point, my colleague Abella J.’s reliance on Griggs v. Duke Power Co., 401 U.S. 424 (1971) is, respectfully, inapposite. She repeatedly cites Griggs (see paras. 32-34, 38, 53, 55 and 70-71) for the proposition that disproportionate impact is sufficient in itself to demonstrate that a law creates an adverse distinction based on a protected ground and thereby satisfies step one: see paras. 55 and 70-71. I do not think Griggs can be read in such a manner. Griggs was a statutory interpretation case concerned with Title VII of the Civil Rights Act of 1964, Pub. L. 88-352, 78 Stat. 241 (1964) — not a constitutional case. Its result was compelled by the text of the statute rather than by the principles underlying unconstitutional adverse effect discrimination: Washington, Mayor of Washington, D.C. v. Davis, 426 U.S. 229 (1976). Further, as evidenced by its progeny, Griggs is limited in its scope, such that it cannot stand for the proposition that statistical disparity alone is sufficient: see Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988), at p. 994 (holding that “the plaintiff’s burden in establishing a prima facie case goes beyond the need to show that there are statistical disparities” and that “the plaintiff must offer statistical evidence of a kind and degree sufficient to show that the practice in question has caused the exclusion of applicants for jobs or promotions because of their membership in a protected group”) (emphasis added); see also Texas Department of Housing and Community Affairs v. Inclusive Communities Project, Inc., 135 S. Ct. 2507 (2015); Ricci v. DeStefano, 557 U.S. 557 (2009); Smith v. City of Jackson, Mississippi, 544 U.S. 228 (2005). Regardless, in hopes of avoiding a jurisprudential debate on case law sourced from outside this country, at bottom the American tradition of equality is markedly distinct from the Canadian tradition. While Griggs was admittedly a case of adverse effect discrimination, it was also fundamentally a case informed by the unique legacy of slavery and segregated schools endemic to American history: in Griggs, a high school education effectively served as a proxy for race (indeed, the impugned high school education and aptitude test requirements there were imposed on the same day the Civil Rights Act of 1964 came into force). In this way, in Griggs, there was a legitimate nexus established between the imposition of a high school education and aptitude test requirements and racial discrimination, and not simply a nexus that depended on statistical disparity alone (as the majority says is sufficient).

[247]                      While this Court has of course cited Griggs in the past, it has never done so to support the proposition that step one can be satisfied in the absence of a nexus between the impugned law and the disproportionate impact. Since the inception of our jurisprudence interpreting s. 15(1), this Court has cautioned as follows:

If the adverse effects analysis is to be coherent, it must not assume that a statutory provision has an effect which is not proved. We must take care to distinguish between effects which are wholly caused, or are contributed to, by an impugned provision, and those social circumstances which exist independently of such a provision.

(Symes v. Canada, [1993] 4 S.C.R. 695, at pp. 764-65)

This pronouncement continues to be applied in the search for a necessary nexus between the impugned law and its effects in cases of adverse effect discrimination: see Miceli-Riggins v. Canada (Attorney General), 2013 FCA 158, [2014] 4 F.C.R. 709, at para. 76; Grenon v. Minister of National Revenue, 2016 FCA 4, 482 N.R. 310, at paras. 38-39 and 45; see also Canada (Attorney General) v. Lesiuk, 2003 FCA 3, [2003] 2 F.C. 697, at para. 33; Begum v. Canada (Citizenship and Immigration), 2018 FCA 181, [2019] 2 F.C.R. 488, at para. 81.

[248]                     Ultimately, the question under step one is whether the law, while facially neutral, creates an adverse distinction based on an enumerated or analogous ground: see Kapp, at para. 17; Withler, at paras. 30 and 61; Quebec v. A, at para. 185; Taypotat, at para. 19. This means that the claimant “must therefore demonstrate that the law at issue has a disproportionate effect on the claimant based on his or her membership in an enumerated or analogous group”: Taypotat, at para. 21 (emphasis added). Although my colleague Abella J.’s phrasing of the question under step one correctly uses the word creates (at para. 50), she later seems to read this word out by determining that mere disproportionate impact is sufficient: para. 52. This cannot be the case: to proceed past step one and show a prima facie breach, “the evidence must amount to more than a web of instinct” (Taypotat, at para. 34). A nexus between the impugned legislation and the disproportionate impact is required.

[249]                     This is essential to understanding why both the Federal Court and the Federal Court of Appeal held that the claim here fails at step one. Indeed, both courts found that this case “turns on the first step”: C.A. reasons, at para. 40. Applying step one to the facts, they each found that the pension plan creates no distinction based on the enumerated ground of sex. They held the following:

The fact that the vast majority of part-time members and members in a job-sharing arrangement are women, and that these job-sharers do not have the option of contributing to their pension at the full-time rate, is not a consequence of or connected to the provisions of the RCMPSA. The “trigger” is whether the member works part-time. This is not connected to the RCMPSA. Rather, this is based on the decisions the member makes, as difficult as those may be, as a family to balance work and child care, by having one parent, usually the woman, work part-time for a few years.

(F.C. reasons, at para. 137)

. . . the mere fact that women disproportionately take advantage of the job-sharing option does not mean that the pension treatment afforded to those who job-share under the RCMPSA and the Regulation creates a distinction on an enumerated or analogous ground. Likewise, as in Begum, the general expert opinion evidence filed by the appellants fails to establish the requisite nexus between the impugned provisions and a protected ground so as to give rise to a section 15 breach. In sum, the appellants were not denied buy-back rights based on their personal characteristics of being female RCMP members with young children, but rather because they elected to job-share as opposed to taking care and nurturing leave. The requisite nexus to establish a breach of section 15  of the Charter  is therefore absent in this case as the appellants cannot show that the impugned provisions in the RCMPSA and Regulation impact them more negatively than others because of their sex . . . .

(C.A. reasons, at para. 53)

[250]                     My colleague Abella J. rejects this reasoning and its emphasis on “choice” (at paras. 85-92), and my colleagues Brown and Rowe JJ. — by implication of finding that the appellants have satisfied step one (at para. 186) — also seem to reject this reasoning (albeit perhaps for different reasons). I need not express any conclusion in this regard because, in any case, no distinction can be demonstrated here on the basis of sex.

[251]                     As I have established, it is hard to see how the impugned provisions create a distinction on the basis of being a woman alone. I reiterate that one does not job-share because one is a woman (i.e. sex simpliciter); one job-shares because one has caregiving responsibilities (in this case, children to take care of). The claim brought by the appellants — on behalf of women with children — crucially depends on the intersection of sex and parental or family status. The arguments proffered and statistics employed are all in reference to women with children. Like my colleague Abella J., I accept the appellant Ms. Fox’s evidence that “job-sharing is often the only child care solution for members with children” because “there is simply no around-the-clock child care available” in “rural or isolated communities”: para. 91, quoting A.R., vol. III, at p. 334. But there is no reason why this is a singularly sex-based issue: rather, it is a caregiving status issue. For instance, in those same rural or isolated areas, there is likely no around-the-clock elderly care available either, making job-sharing the only solution available for members with caregiving responsibilities. Thus the burden to job-share, and the associated distinction created by the impugned provisions of the pension plan, is faced by those members with caregiving responsibilities and is not limited to those of a certain sex. Indeed, those without such responsibilities do not face any comparable pressure to job-share. But without recognition of an analogous protected ground of parental, family, or caregiver status, or alternatively the intersection of that ground and sex (as in Lesiuk, at para. 37), the appellants’ claim must fail at step one of the s. 15(1) analysis.

[252]                     To be sure, the impugned provisions may very well not be rational — there may indeed be no logical reason to deprive job-sharers of full pension benefits that are guaranteed to full-time members and members on leave without pay. But it is not this Court’s role to constitutionalize normative judgments to this effect; that is the role of the electorate, and in turn, the legislature.

[253]                     In this sense, I wish to add one final point on the role of the judiciary vis‑à‑vis the role of the legislature. The result the majority reaches may certainly be desirable insofar as it guarantees the opportunity for increased pension benefits to RCMP members who job-share. But when the Court reaches this result in such a doctrinally precarious fashion, and when and if the impugned provisions are illogical, irrational, or under‑inclusive, then it is the legislature’s role to rectify — the remedy does not lie in the Constitution, which proscribes particular forms of discrimination, forms that do not exist here. With the greatest respect, then, the remedy sought in the case at bar should be granted by the democratic process.

III.         Conclusion

[254]                     While I agree with Abella J. that sex-based discrimination is particularly insidious and I similarly lament society’s continued history of discrimination against women, I am in respectful disagreement as to whether the appellants have sufficiently made out a s. 15 violation here under this Court’s well-established jurisprudence.

[255]                     Likewise, while I agree with my colleagues Brown and Rowe JJ.’s proposed disposition of the appeal, as well as with their observations regarding the appellants’ employment status and the majority’s unprincipled expansion of the scope of s. 15(1), I write separately to express disagreement with their dismissal of the appeal on the second step of the s. 15(1) analysis rather than the first step, which is the step on which this case turns: C.A. reasons, at para. 40. Further, I write separately in order to highlight the under-inclusive nature of the pension plan, which disproportionately affects all those with caregiving responsibilities, including same-sex couples with children and individuals with caregiving responsibilities for their aging partners or parents. It therefore falls to the legislature, not the courts, to remedy any under-inclusiveness in this legislation, which was purportedly meant to assist with caregiving responsibilities in the first place.

[256]                     For these reasons, and with the greatest respect for my colleagues, I would dismiss the appeal.

                    Appeal allowed with costs throughout, Côté, Brown and Rowe JJ. dissenting.

                    Solicitors for the appellants: Champ & Associates, Ottawa.

                    Solicitor for the respondent: Attorney General of Canada, Ottawa.

                    Solicitor for the intervener the Attorney General of Ontario: Attorney General of Ontario, Toronto.

                    Solicitor for the intervener the Attorney General of Quebec: Attorney General of Quebec, Québec.

                    Solicitors for the intervener the Women’s Legal Education and Action Fund Inc.: Cavalluzzo, Toronto.

                    Solicitors for the intervener the Public Service Alliance of Canada: Raven, Cameron, Ballantyne & Yazbeck, Ottawa.

                    Solicitors for the intervener the National Police Federation: Nelligan O’Brien Payne, Ottawa.