Zurich Insurance Co. v. Ontario (Human Rights Commission), 1992 CanLII 67 (SCC), [1992] 2 SCR 321

minicounsel

Zurich Insurance Co. v. Ontario (Human Rights Commission), [1992] 2 S.C.R. 321

IN THE MATTER OF the Ontario Human Rights

Code, 1981, S.O. 1981, c. 53, as amended;

AND

IN THE MATTER OF the complaint made by

Mr. Michael G. Bates, of Islington, Ontario,

alleging discrimination in the right to

contract and services, goods and facilities

by the Zurich Insurance Company,

188 University Avenue, Toronto, Ontario

Ontario Human Rights Commission Appellant

Zurich Insurance Company Respondent

and

Michael G. Bates Complainant

and

Commission des droits de la personne du Québec

and Alberta Human Rights Commission Interveners

Indexed as: Zurich Insurance Co. v. Ontario (Human Rights Commission)

File No.: 21737.

1991: November 5; 1992: June 25.

Present: Lamer C.J. and La Forest, L'Heureux‑Dubé, Sopinka, Gonthier, McLachlin and Iacobucci JJ.

on appeal from the court of appeal for ontario

Human rights ‑‑ Discrimination on basis of sex, age and marital status ‑‑ Insurance ‑‑ Young unmarried male drivers charged higher car insurance premiums than similarly situated females or married males ‑‑ Insurance industry exempted from Act if grounds reasonable and bona fide ‑‑ Whether or not reasonable and bona fide grounds ‑‑ Human Rights Code, 1981, S.O. 1981, c. 53, ss. 1, 3, 8, 21, 24(1)(b), 46(3).

A Board of Inquiry, appointed pursuant to the Ontario Human Rights Code, 1981, upheld a complaint made in May, 1983, that the respondent insurer engaged in prohibited discriminatory practices in that young, single, male drivers were charged higher car insurance rates than young, single, female drivers or, young, married, male drivers or any drivers 25 years of age or over. The complainant alleged denial of the right to contract on equal terms without discrimination and of the right to equal treatment in services.

Respondent conceded that the classification system upon which its premiums were based constituted a prima facie infringement of the Code but argued that the distinction between classes of drivers was based on reasonable and bona fide grounds and therefore fell within the exception provided by s. 21 of the Code.

The Board of Inquiry found that the Code had been contravened but the Divisional Court and the Court of Appeal held otherwise. At issue here was whether differentiation in automobile insurance rates based upon age, sex and marital status was reasonable and bona fide within the meaning of s. 21 of the Human Rights Code, 1981.

Held (L'Heureux‑Dubé and McLachlin JJ. dissenting): The appeal should be dismissed.

Per Lamer C.J. and La Forest, Sopinka, Gonthier and Iacobucci JJ.: The determination of insurance rates and benefits does not fit easily within traditional human rights concepts. The underlying philosophy of human rights legislation is that an individual has a right to be dealt with on his or her own merits and not on the basis of group characteristics. Exceptions to this legislation should be narrowly construed. Insurance rates, however, are based on statistics relating to the degree of risk associated with a class or group of persons. Although not all persons in the class share the same risk characteristics, it is wholly impractical that each insured be assessed individually. Sometimes the class or group classification chosen will coincide with a prohibited ground of discrimination, bringing the rating scheme into conflict with human rights legislation. The Code, however, exempts an insurer from liability for discrimination if based on reasonable and bona fide grounds.

An important principle of insurance practice is that premiums charged to individual policy holders vary as much as possible in accordance with the degree of risk posed by the policy holder. The degree of risk is necessarily determined on the basis of groups sharing characteristics material to the risk. Inevitably, some will be placed in a group who do not share the average characteristics of that group with the result that the rate discriminates against them. The basic human rights principles must take into account these differences when applied in the context of insurance.

Section 21 provides a defence to a prima facie discriminatory practice if reasonable and bona fide grounds for that practice exist. A discriminatory practice is "reasonable" within the meaning of s. 21 of the Code (a) if it is based on a sound and accepted insurance practice, and (b)if there is no practical alternative. A practice is sound if it is desirable to adopt it for the purpose of achieving the legitimate business objective of charging premiums that are commensurate with risk. The availability of a practical alternative is a question of fact to be determined having regard to all of the facts of the case. The practice, to meet the test of "bona fides", must be adopted honestly, in the interests of sound and accepted business practice and not for the purpose of defeating the rights protected under the Code.

Human rights values cannot be over‑ridden by business expediency alone. To allow "statistically supportable" discrimination would undermine the intent of human rights legislation which attempts to protect individuals from collective fault. It would also perpetuate traditional stereotypes with all of their invidious prejudices. Whether there was an alternative, which in all the circumstances was practicable, must be considered.

Setting premiums on a basis that did not rely on the classifications then in use was not a practical alternative for the industry. It would have been even more impractical for the respondent unilaterally to adopt its own classifications based on non‑discriminatory criteria because actuarially reliable figures, based on only its own data, were impractical or impossible to obtain. The fact that it was theoretically possible to collect data based on other criteria does not establish that reasonable alternatives to the current system exist. In the absence of alternative criteria to set premiums, it was unreasonable to expect Zurich to simply guess at an equitable distribution of premiums among the consumers it insured.

To require an insurer to establish that the very essence of its business would be undermined if it could no longer rely on discriminatory group characteristics for its rate classification system is too high a standard and is not required by s. 21. This standard ascribes too narrow a meaning to what constitutes a practical alternative. An alternative may be impractical even though its adoption would not undermine the very essence of a business. Furthermore, this standard tended to minimize the importance of the statutory framework within which the industry and the respondent were required to operate.

The insurance industry must be allowed time to determine whether it can restructure its classification system in a manner that will eliminate discrimination based on enumerated group characteristics and still reflect the disparate risks of different classes of drivers. While the situation as it existed in 1983 did not provide a reasonable alternative to setting premiums based on age, sex and marital status, the situation today and in the future may be quite different. The insurance industry must strive to avoid setting premiums based on enumerated grounds.

Per L'Heureux‑Dubé J. (dissenting): Legislation prohibiting discrimination must be broadly construed using a purposive approach and defences to discrimination must be narrowly construed. Section 21, which allows discrimination if "reasonable and bona fide" grounds exist, must therefore be narrowly construed. The party seeking to justify discrimination against an individual must show that there is no reasonable or practical alternative. The rule will not be bona fide if one exists.

The test for "reasonable and bona fide grounds" should be based on the test established for employment situations. The terms "reasonable and bona fide" are stronger than the single term "bona fide". The protection of individual rights is equally important in employment and insurance contexts.

A two‑pronged test with subjective and objective components should be adopted. The subjective component requires that a distinction in an insurance contract based on age, sex, marital status, family status or handicap be imposed honestly, in good faith, and in the sincerely held belief that such distinction accurately reflects the cost of the risk insured, rather than for ulterior or extraneous reasons aimed at objectives which could defeat the purpose of the Code. The objective component requires that the distinction be reasonably necessary to assure the proper allocation of risk among insured groups. Any rational connection must causally link distinction and the insured risk and must be more than a statistical correlation or a simple reliance on traditional or accepted insurance practices which may be nonetheless needlessly discriminatory. The distinction must also be a reasonable means of identifying and classifying similar risks.

Respondent, given a prima facie case of discrimination, must demonstrate the "reasonable and bona fide grounds" underlying the discriminatory classification system. The first, subjective component of the applicable test was met as the respondent used its discriminatory classification system in good faith. The objective component, however, was not met. The respondent's conclusion that a rational connection existed since the evidence demonstrated a statistical correlation between young, single males and high risk is dangerously and fundamentally flawed. The mere statistical correlation between a group and higher risk cannot suffice to justify discrimination on prohibited grounds. Such correlation accepts the very stereotyping that is deemed unacceptable by human rights legislation: prohibited grounds of discrimination are used to ascribe the characteristics of the group to all individuals in the class. Discrimination based on statistical correlation is simply discrimination in a more invidious form. Strong statistical proof is required to demonstrate a rational connection between the discriminatory classification and high risk. This proof, however, must not simply be one of correlation, but one of causal connection. Respondent has not established this.

The availability of alternative means of classifying individuals into groups can be considered under the objective branch of the test. The complaint of discrimination is not answered by the fact that the collection of statistics, which would have provided the basis for an alternative classification system, was not required by the Superintendent of Insurance. It was not up to the Superintendent to decide whether or not there was an infringement of the Code.

In keeping with the policy of curial deference to the findings of fact of specialized tribunals, the facts as stated by the Board should not be second‑guessed by this Court in the absence of a clear error or misapprehension on the part of the Board. The Board had the advantage of seeing and hearing the witnesses.

To simply assert that a system has been in place for 50 years is an inadequate response to the argument that there are alternatives to the system. The insurance industry cannot rely on its inaction and its self‑serving claim that the practice "has always been done this way" to defend its discriminatory practices on grounds of a lack of statistical data. If it could, complacency and a history of discrimination would be rewarded at the cost of progress and the recognition of higher societal norms of behaviour.

An alternative to the discriminatory classification system existed at the time of the complaint. The premiums of drivers over the age of 25 were set according to completely non‑discriminatory classifications. Respondent, given this alternative and the absence of evidence that it could not be adopted for unmarried male drivers under age 25, did not establish that the discriminatory system was reasonable.

Per McLachlin J. (dissenting): The law clearly forbids discrimination on the grounds of age, sex and marital status, unless the offender can establish that its intentions were bona fide and that, on an objective test, there was no practical or reasonable alternative to discriminating on those grounds. The insurer, however, did not demonstrate that there was no practical alternative in 1983 to basing its premiums for young men on age, sex and marital status.

An absence of statistics based on grounds other than age, sex and marital status does not establish that there are no risk factors other than age, sex and marital status which would achieve the objective of charging premiums that are commensurate with risk. An absence of statistics does not establish that there is no other alternative. It only establishes that it is not known whether or not other viable alternative bases of risk evaluation exist and that the insurer does not have the means to establish them. As a matter of logic, the absence of alternative statistics is equally consistent with the inference that alternatives exist as with the inference that alternatives do not exist. The insurer must bear the burden of showing no reasonable alternative exists, and, through its own failure to collect the required data, has failed to meet this burden.

The effect of confusing absence of a reasonable alternative with absence of proof of a reasonable alternative is to remove the burden of proof from the shoulders of the person who prima facie violates the Code and to place it on the shoulders of the person who complains of discrimination. This is contrary to human rights tradition and contrary to the rule of evidence that the burden of proof should lie on the person most likely to be in possession of the relevant facts. And on the plane of policy, such an approach would encourage prima facie offenders to continue in their traditional discriminatory ways rather than reforming their practices in accordance with the object of the Code.

The law should not allow difficulties of proving the non‑existence of reasonable alternatives to stand as a defence to a charge under the Code. To the extent that it may take time for industry to convert from ingrained discriminatory practices to non‑discriminatory alternatives (or alternatively to show that none is possible), those difficulties are better dealt with under the rubric of remedies than by saying there is no violation of the Code.

The insurer cannot be excused for its failure to demonstrate reasonable alternatives. First, there is no precedent in human rights cases for excusing a person who is guilty of prima facie discrimination on any ground absent reasonable alternatives. Second, its excuses adduced here were not sufficient. An insurer doing business in a province must comply with all the laws of that province, and failure of the Superintendent of Insurance to require the collection of statistics on bases other than prohibited grounds of discrimination does not amount to satisfying the Code. The size of the firm and its capacity to fix rates unilaterally on non‑discriminatory statistics cannot be accepted as a defence as it would create differing standards for large and small firms. To accept less than requiring the prima facie offender to provide actual justification for what may in the past have been assumed to be justified would constitute an `about‑face' in human rights law which could seriously undermine the effectiveness of human rights codes. Third, the insurer cannot discharge that burden merely by showing that it is difficult to ascertain whether there are reasonable alternatives and if so, what they are. Difficulty alone has never been accepted as an excuse for discriminatory conduct contrary to human rights legislation.

The Code requires the classifications on which premiums are set to be rationally defensible. If they are not, they must be found to violate, even if other classes are negatively affected. Any adverse effect on drivers in other categories arises from the insurer's failure to shoulder the burden which the law imposes of conforming to the Code or demonstrating the absence of reasonable alternatives.

Cases Cited

By Sopinka

Considered: Saskatchewan (Human Rights Commission) v. Saskatoon (City), 1989 CanLII 18 (SCC), [1989] 2 S.C.R. 1297; distinguished: Ontario Human Rights Commission v. Borough of Etobicoke, 1982 CanLII 15 (SCC), [1982] 1 S.C.R. 202; National Corn Growers Assn. v. Canada (Import Tribunal), 1990 CanLII 49 (SCC), [1990] 2 S.C.R. 1324; Lester (W.W.) (1978) Ltd. v. United Association of Journeymen and Apprentices of the Plumbing and Pipefitting Industry, Local 740, 1990 CanLII 22 (SCC), [1990] 3 S.C.R. 644; Gendron v. Supply and Services Union of the Public Service Alliance of Canada, Local 50057, 1990 CanLII 110 (SCC), [1990] 1 S.C.R. 1298; CAIMAW v. Paccar of Canada Ltd., 1989 CanLII 49 (SCC), [1989] 2 S.C.R. 983; referred to: Fletcher v. Manitoba Public Insurance Co., 1990 CanLII 59 (SCC), [1990] 3 S.C.R. 191; Bell Canada v. Canada (Canadian Radio‑television and Telecommunications Commission), 1989 CanLII 67 (SCC), [1989] 1 S.C.R. 1722; Ontario Human Rights Commission v. Simpsons‑Sears Ltd., 1985 CanLII 18 (SCC), [1985] 2 S.C.R. 536; Brossard (Town) v. Quebec (Commission des droits de la personne), 1988 CanLII 7 (SCC), [1988] 2 S.C.R. 279; Bhinder v. Canadian National Railway Co., 1985 CanLII 19 (SCC), [1985] 2 S.C.R. 561; Griggs v. Duke Power Co., 401 U.S. 424 (1971); Central Alberta Dairy Pool v. Alberta (Human Rights Commission), 1990 CanLII 76 (SCC), [1990] 2 S.C.R. 489; Hartford Accident and Indemnity Co. v. Insurance Commissioner of the Commonwealth of Pennsylvania, 482 A.2d 542 (1984); Pennsylvania National Organization for Women v. Commonwealth of Pennsylvania Insurance Department, 551 A.2d 1162 (1988).

By L'Heureux‑Dubé J. (dissenting)

Ontario Human Rights Commission v. Simpsons‑Sears Ltd., 1985 CanLII 18 (SCC), [1985] 2 S.C.R. 536; Insurance Corporation of British Columbia v. Heerspink, 1982 CanLII 27 (SCC), [1982] 2 S.C.R. 145; Brossard (Town) v. Quebec (Commission des droits de la personne), 1988 CanLII 7 (SCC), [1988] 2 S.C.R. 279; Bhinder v. Canadian National Railway Co., 1985 CanLII 19 (SCC), [1985] 2 S.C.R. 561; Caldwell v. Stuart, 1984 CanLII 128 (SCC), [1984] 2 S.C.R. 603; National Corn Growers Assn. v. Canada (Import Tribunal), 1990 CanLII 49 (SCC), [1990] 2 S.C.R. 1324; Lester (W.W.) (1978) Ltd. v. United Association of Journeymen and Apprentices of the Plumbing and Pipefitting Industry, Local 740, 1990 CanLII 22 (SCC), [1990] 3 S.C.R. 644; Bell Canada v. Canada (Canadian Radio‑television and Telecommunications Commission), 1989 CanLII 67 (SCC), [1989] 1 S.C.R. 1722; CAIMAW v. Paccar of Canada Ltd., 1989 CanLII 49 (SCC), [1989] 2 S.C.R. 983; Gendron v. Supply and Services Union of the Public Service Alliance of Canada, Local 50057, 1990 CanLII 110 (SCC), [1990] 1 S.C.R. 1298; Canada (Attorney General) v. Public Service Alliance of Canada, 1991 CanLII 88 (SCC), [1991] 1 S.C.R. 614; Ontario Human Rights Commission v. Borough of Etobicoke, 1982 CanLII 15 (SCC), [1982] 1 S.C.R. 202; Re Ontario Human Rights Commission and City of North Bay (1977), 1977 CanLII 1253 (ON CA), 17 O.R. (2d) 712; R. v. M. (S.H.), 1989 CanLII 31 (SCC), [1989] 2 S.C.R.. 446; Saskatchewan (Human Rights Commission) v. Saskatoon (City), 1989 CanLII 18 (SCC), [1989] 2 S.C.R. 1297; Central Alberta Dairy Pool v. Alberta (Human Rights Commission), 1990 CanLII 76 (SCC), [1990] 2 S.C.R. 489; Canada (Attorney General) v. Rosin, 1990 CanLII 12957 (FCA), [1991] 1 F.C. 391; Hartford Accident and Indemnity Co. v. Insurance Commissioner of the Commonwealth of Pennsylvania, 482 A.2d 542 (1984); Pennsylvania National Organization for Women v. Commonwealth of Pennsylvania Insurance Department, 551 A.2d 1162 (1988).

Statutes and Regulations Cited

Canadian Human Rights Act, R.S.C. 1985, c. H‑6, ss. 15(a), (g).

Compulsory Automobile Insurance Act, R.S.O. 1980, c. 83, s. 10.

Employment Standards Act, R.S.O. 1980, c. 137, s. 34(2).

Human Rights Code, 1981, S.O. 1981, c. 53, ss. 1, 3, 8, 21, 24(1)(b), (2), 41(1), (3), 46(3).

Insurance Act, R.S.O. 1980, c. 218, ss. 80(1), (2), 393(b)(iii), 396.

Insurance Statute Law Amendment Act, 1990, S.O. 1990, c. 2, ss. 3(1), 74.

Ontario Automobile Insurance Board Act, 1988, S.O. 1988, c. 18, s. 33(1).

Regulation 282 under the Employment Standards Act, R.R.O. 1980.

Authors Cited

Ontario. Inquiry into Motor Vehicle Accident Compensation in Ontario. Report of Inquiry into Motor Vehicle Accident Compensation in Ontario. (The "Osborne Report".) Toronto: Queen's Printer for Ontario, 1988.

National Association of Insurance Commissioners. Report of the rates and Rating Procedures Task Force of the Automobile Insurance (D3) Subcommittee (November, 1978).

Ryan, Stephen R. "The Elimination of Gender Discrimination in Insurance Pricing: Does Automobile Insurance Rate Without Sex?" (1986), 61 Notre Dame L. Rev. 748

APPEAL from a judgment of the Ontario Court of Appeal (1989), 1989 CanLII 4051 (ON CA), 70 O.R. (2d) 639, 49 O.A.C. 361, 45 C.C.L.I. 303, dismissing an appeal from a judgment of the Divisional Court (1987), 1987 CanLII 4058 (ON SC), 58 O.R. (2d) 325, [1987] I.L.R. {PP} 1‑2148, 1987 CanLII 9962 (ON SC), 23 C.C.L.I. 130, allowing an appeal from a judgment of board of inquiry of the Ontario Human Rights Commission (1985), 1985 CanLII 5245 (ON HRT), 6 C.H.R.R. D/2948. Appeal dismissed, L'Heureux‑Dubé and McLachlin JJ. dissenting.

J. E. Minor, T. H. Wickett and Tanya Lee, for the appellant.

J. F. Howard, Q.C., Neil Finkelstein and Jeff Galway, for the respondent.

Beatrice Vizkelety and Madeleine Caron, for the intervener Commission des droits de la personne du Québec.

J. Leslie Wallace and S. E. Fitzgerald, for the intervener Alberta Human Rights Commission.

The judgment of Lamer C.J. and La Forest, Sopinka, Gonthier and Iacobucci JJ. was delivered by

//Sopinka J.//

Judgment of the Court

Sopinka J. -- The issue raised in this appeal is whether differentiation in automobile insurance rates based upon age, sex and marital status is reasonable and bona fide within the meaning of s. 21 of the Ontario Human Rights Code, 1981, S.O. 1981, c. 53 (now R.S.O. 1990, c. H.19, s. 22). The respondent, Zurich Insurance Company ("Zurich"), in setting its rates in 1983, discriminated against single, male drivers under the age of 25 but it claims exemption from liability by virtue of s. 21 of the Code.

The Facts

On April 30, 1984, a Board of Inquiry was appointed pursuant to the Ontario Human Rights Code, 1981 (the "Code") to hear and decide a complaint made in May of 1983 by Mr. Michael G. Bates against the respondent insurance company. This complaint alleged that young, single, male drivers must pay automobile insurance rates which exceed the rates paid by young, single, female drivers, the rates paid by young, married, male drivers and the rates paid by any drivers 25 years of age or over. Mr. Bates alleged that this discrepancy contravened provisions in the Code. The complainant alleged that he had been denied the right to contract on equal terms without discrimination contrary to ss. 3 and 8 of the Code. The complainant also alleged that his right to equal treatment in services, goods and facilities had been infringed contrary to ss. 1 and 8 of the Code.

Zurich concedes that the classification system upon which its premiums were based constitutes a prima facie infringement of the Code but it argues that this infringement is a "distinction, exclusion or preference on reasonable and bona fide grounds" and thus falls within the exception provided by s. 21 of the Code.

The Board of Inquiry concluded that the automobile driver classification for unmarried male drivers under 25 years of age contravened the Code. Zurich appealed this decision and the Divisional Court allowed the appeal. The appellant appealed to the Court of Appeal for Ontario which dismissed the appeal. This appeal comes before the Court with leave.

Judgments

A. Board of Inquiry (1985), 1985 CanLII 5245 (ON HRT), 6 C.H.R.R. D/2948

Board of Inquiry

The Board of Inquiry based its decision on an extension of the reasoning used by this Court in Ontario Human Rights Commission v. Borough of Etobicoke, 1982 CanLII 15 (SCC), [1982] 1 S.C.R. 202. While indicating that no perfect analogy existed for the application of the Etobicoke decision to the insurance context, the Board concluded, at p. D/2962, that it was reasonable to utilize the basic principles of the Etobicoke case:

Just as employers are required to demonstrate that a discriminatory limitation is reasonably necessary to assure the safe, efficient and economic performance of the job in order to establish a "bona fide occupational qualification", so too insurers should be required by section 21 of the Code to demonstrate that a rate classification system, which discriminates on the basis of prohibited group characteristics, is reasonably necessary to ensure the efficient operation of the insurance system.

The Board held that ease of determining rates was not sufficient to satisfy the bona fide requirements of s. 21. It concluded, at p. D/2964, that to be reasonably necessary it was up to Zurich to demonstrate "that it would be impractical or impossible to operate the automobile insurance system with the use of non-discriminatory criteria for rate classification".

The Board of Inquiry then examined whether the method of setting premiums used by Zurich was reasonably necessary. The Board noted that there had been no evidence offered to support the assertion that it has been scientifically proved that there is a direct, causal relationship between the discriminatory group factors used and high risk. It concluded that these factors were mere proxies for other factors. The Board also suggested that there were non-discriminatory alternatives which were possible and desirable in determining the premiums of insured drivers. One alternative mentioned by the Board was to incorporate the non-discriminatory criteria used for setting rates for those 25 years of age or over and apply them to the under 25 category. The Board of Inquiry concluded that Zurich had failed to establish that the very essence of its business would be undermined if it could no longer rely on discriminatory group characteristics for its rate classification system and thus the method of setting rates was not reasonably necessary.

B. Divisional Court (1987), 1987 CanLII 4058 (ON SC), 58 O.R. (2d) 325

Divisional Court

The Divisional Court concluded, at pp. 329-30, that the Board of Inquiry had fallen into error in interpreting s. 21 of the Code:

Thus, in the course of its judgment the board progressed from the statutory requirement that the insurer show that it proceeded on "reasonable and bona fide grounds" to a requirement that the insurer establish that its grounds were "reasonably necessary". After a consideration of American cases, it made a further leap to a requirement that the insurer establish "that the very existence of the business would be undermined if the discriminatory requirements could not be relied upon".

In our view, neither of these leaps can be justified. Something may be reasonable without being necessary. Something may be reasonable without being so essential that the existence of an industry will be undermined if it cannot be relied upon. [Emphasis in original.]

The court held that there was no reason to substitute the plain words of "reasonable and bona fide grounds" for the concept of undermining the existence of a business. It was also unnecessary that the words "reasonably necessary" be imported into s. 21 as a result of the Etobicoke decision. Etobicoke was distinguishable as it was an employment case and the words being construed were "bona fide occupational classification and requirement", the words of which import occupational qualifications. The wording of s. 21 is different and there was no justification in substituting for the words used in s. 21, "reasonable and bona fide grounds", the words "reasonably necessary grounds". In the occupational context, individual assessment is vital; in the insurance context it is impossible. Section 21 recognizes that "the many who are skilled, experienced, careful and lucky will subsidize those who are not" (at p. 330).

The court held that the words "reasonable and bona fide grounds", as used in s. 21, should be construed according to their plain meaning. The court then turned to a consideration of whether the grounds used to set the premiums were reasonable. It concluded, at p. 331, that the distinctions created by the classification system were reasonable distinctions based on age, sex and marital status:

The available statistical information demonstrates clearly that the single male driver under 25 years of age is the highest risk classification. Average costs resulting from claims within that classification are triple those of the average class. The statistics show that premium income from this class is not sufficient to satisfy the loss costs of the class. The premium charged to the single male driver under 25 years of age is, accordingly, subsidized by other classifications. It appears reasonable that the group causing the highest loss costs should pay the highest premium. In short, the current classification system, although subject to future revision and current doubts, does embody distinctions supported by reasonable, actuarially verified statistics.

With respect to the complainant's argument that he may be a much better risk than the other members of his group, the court noted that it was inevitable that in every class there would be drivers who represent worse or better risks than average. Nonetheless, premiums must be based on averages.

The court added that there were, at the relevant time, no statistics available to classify drivers on any other basis. While the Superintendent of Insurance (now the Commissioner) had the power to require statistics to be compiled on another basis, this power had not been utilized. Absent such data, the existing verified statistics provided a suitable framework for reasonable and bona fide distinctions to be drawn through insurance classifications. The absence of any other statistics also supported the "common perception, shared by the industry and its regulators, of the reasonableness of the classifications" (at p. 332). In the alternative, the court was of the view that Zurich had met the test of "reasonable necessity" as it had relied upon the only statistics available to it. These statistics showed that the class into which Mr. Bates fell was the class generating the highest loss cost. The court was of the view that to ignore those statistics would have been irresponsible.

C. Court of Appeal (1989), 1989 CanLII 4051 (ON CA), 70 O.R. (2d) 639

Court of Appeal

The Court of Appeal agreed with the decision and reasons of the Divisional Court and dismissed the appeal.

The Issue

The issue to be determined in this appeal is whether the method by which Zurich set its automobile insurance premiums, which admittedly discriminates on the basis of age, sex and marital status, nonetheless satisfies the "reasonable and bona fide grounds" exemption provided by s. 21 of the Code.

For the purposes of convenience I reproduce the relevant portions of the Code below:

Every person has a right to equal treatment with respect to services, goods and facilities, without discrimination because of race, ancestry, place of origin, colour, ethnic origin, citizenship, creed, sex, age, marital status, family status or handicap.
Every person having legal capacity has a right to contract on equal terms without discrimination because of race, ancestry, place of origin, colour, ethnic origin, citizenship, creed, sex, age, marital status, family status or handicap.
No person shall infringe or do, directly or indirectly, anything that infringes a right under this Part.
The right under sections 1 and 3 to equal treatment with respect to services and to contract on equal terms, without discrimination because of age, sex, marital status, family status or handicap, is not infringed where a contract of automobile, life, accident or sickness or disability insurance or a contract of group insurance between an insurer and an association or person other than an employer, or a life annuity, differentiates or makes a distinction, exclusion or preference on reasonable and bona fide grounds because of age, sex, marital status, family status or handicap.

41.--(1) Any party to a proceeding before a board of inquiry may appeal from a decision or order of the board to the Divisional Court in accordance with the rules of court.

(3) An appeal under this section may be made on questions of law or fact or both and the court may affirm or reverse the decision or order of the board of inquiry or direct the board to make any decision or order that the board is authorized to make under this Act and the court may substitute its opinion for that of the board. [Emphasis added.]

Standard of Review

In considering the manner in which the lower courts have dealt with this proceeding, it is important to keep in mind that the standard of review by the courts in matters of this type is different from normal appellate review. This Court will normally overturn findings of fact only when there has been a palpable and overriding error made by the trier of fact (Fletcher v. Manitoba Public Insurance Co., 1990 CanLII 59 (SCC), [1990] 3 S.C.R. 191, at pp. 202-6). The Code, however, states in s. 41(3) (now s. 42(3)) that an appeal lies to a court on any question of law or fact and that the court may substitute its opinion for that of the Board of Inquiry (see also Etobicoke, supra, at p. 211). Evidently, the legislature was not of the opinion that the conclusions of the Board of Inquiry should be given great deference as a result of accumulated expertise or specialized understanding.

Unlike the situation present in many of the cases considered by this Court in regard to curial deference of administrative action (including the cases of National Corn Growers Assn. v. Canada (Import Tribunal), 1990 CanLII 49 (SCC), [1990] 2 S.C.R. 1324, Lester (W.W.) (1978) Ltd. v. United Association of Journeymen and Apprentices of the Plumbing and Pipefitting Industry, Local 740, 1990 CanLII 22 (SCC), [1990] 3 S.C.R. 644, Gendron v. Supply and Services Union of the Public Service Alliance of Canada, Local 50057, 1990 CanLII 110 (SCC), [1990] 1 S.C.R. 1298, and CAIMAW v. Paccar of Canada Ltd., 1989 CanLII 49 (SCC), [1989] 2 S.C.R. 983, discussed in the reasons of L'Heureux-Dubé J. at p. 000), there is no privative clause, or any other clause signifying an attempt to protect the decision of the Board of Inquiry from attack in the courts, involved in this appeal. In fact, as I have noted above, there is an appeal clause which confers powers on a reviewing court that are considerably wider than the normal powers of an appellate court. This was acknowledged in the Etobicoke decision where McIntyre J. stated, at p. 211, that "[t]he appellate court is specifically empowered to review the evidence and substitute its own findings for those of the board of inquiry...." As a result of this clear statutory and jurisprudential authority, I am unable to agree with my colleague L'Heureux-Dubé J., at p. 000, that "[i]t is not open to the Court to conduct its own re-examination of the evidence, and then to substitute its opinion for that of the tribunal".

In spite of the ability to overturn decisions of the Board on findings of fact, this Court has indicated that some curial deference will apply even to cases without privative clauses to reflect the principle of the specialization of duties (see Bell Canada v. Canada (Canadian Radio‑television and Telecommunications Commission), 1989 CanLII 67 (SCC), [1989] 1 S.C.R. 1722, at p. 1746, Etobicoke, supra, at p. 211). While curial deference will apply to findings of fact, which the Board of Inquiry may have been in a better position to determine, such deference will not apply to findings of law in which the Board has no particular expertise. The record provided to the Court and the judgments below are not, in my view, in conflict on issues of fact but merely in interpreting the legal implications flowing from those facts. Moreover, it is common ground between my colleague and me that the Board applied the wrong test. Its findings, therefore, cannot be accepted without subjecting the facts to an analysis on the basis of the appropriate test. Accordingly, both my colleague and I have reviewed the evidence unconstrained by curial deference. The fact that my colleague's review of the evidence in light of a different test yields a result that accords with the conclusion reached by the Board does not defer to the Board but, rather, urges that the Board came to the right result for the wrong reasons.

Reasonable and Bona Fide Grounds

The determination of insurance rates and benefits does not fit easily within traditional human rights concepts. The underlying philosophy of human rights legislation is that an individual has a right to be dealt with on his or her own merits and not on the basis of group characteristics. Conversely, insurance rates are set based on statistics relating to the degree of risk associated with a class or group of persons. Although not all persons in the class share the same risk characteristics, no one would suggest that each insured be assessed individually. That would be wholly impractical. Sometimes the class or group classification chosen will coincide with a prohibited ground of discrimination, bringing the rating scheme into conflict with human rights legislation. The Code, in s. 21 and other sections, has recognized the special problem of insurance. It exempts an insurer from liability for discrimination if based on reasonable and bona fide grounds. The Board of Inquiry in this appeal determined that these words had the same meaning as the bona fide occupational qualification or requirement provision which applies in employment cases. It is necessary to determine whether the test developed in employment cases can be transplanted to the special field of insurance and, in particular, to the setting of insurance rates.

In approaching the interpretation of a human rights statute, certain special principles must be respected. Human rights legislation is amongst the most pre-eminent category of legislation. It has been described as having a "special nature, not quite constitutional but certainly more than the ordinary..." (Ontario Human Rights Commission v. Simpsons-Sears Ltd., 1985 CanLII 18 (SCC), [1985] 2 S.C.R. 536, at p. 547). One of the reasons such legislation has been so described is that it is often the final refuge of the disadvantaged and the disenfranchised. As the last protection of the most vulnerable members of society, exceptions to such legislation should be narrowly construed (Brossard (Town) v. Quebec (Commission des droits de la personne), 1988 CanLII 7 (SCC), [1988] 2 S.C.R. 279, at p. 307; see also Bhinder v. Canadian National Railway Co., 1985 CanLII 19 (SCC), [1985] 2 S.C.R. 561, at pp. 567 and 589).

In a series of human rights cases, these principles have been applied to determine the meaning of a bona fide occupational qualification or requirement with respect to discrimination in employment. This Court developed a two-pronged test encompassing both a subjective and an objective element. To satisfy the subjective element, a qualification or requirement having a discriminatory effect must have been imposed honestly, in good faith and in the sincerely held belief that such a limitation "is imposed in the interests of the adequate performance of the work involved ... and not for ulterior or extraneous reasons..." (Etobicoke, supra, at p. 208). To satisfy the objective component of the test, it was necessary to show that the qualification or requirement was "reasonably necessary to assure the efficient and economical performance of the job..." (at p. 208; see also Brossard, supra, at pp. 311-12). The language of the Ontario statute which gave birth to this dual test clearly indicated that the qualification or requirement must be "job related" but gave no guidance as to the standard against which that relationship was to be judged. The section of the Ontario Human Rights Code, R.S.O. 1970, c. 318, in question in Etobicoke did not use the word "reasonable". The Court adopted the "reasonable necessity" test which was in current use in the United States. See Griggs v. Duke Power Co., 401 U.S. 424 (1971). The current wording of the employment exemption in the Code is "reasonable and bona fide qualification because of the nature of the employment" (see s. 24(1)(b)).

At bottom, this type of exemption seeks to achieve a balance between the right of the individual to be dealt with on his or her own merits and the needs of the employer to operate his or her business safely, efficiently and economically. The object of the legislation is to eliminate discrimination entirely. To do so, however, would require that each employee be considered on an individual basis. The legislature has recognized that this is not always possible consistent with the reasonable needs of the employer. The exemption therefore permits the employer to apply practices and work rules that may affect some employees in a discriminatory fashion provided they are reasonable. This requirement of reasonableness has two aspects. First, any discriminatory practice or rule must have a substantial connection to the operation of the employer's business and second, it must not discriminate more than is necessary. The second aspect of the requirement is addressed by considering whether there is an alternative to the impugned practice which would be less discriminatory while still achieving the sought after business-related objective. Accordingly, it is insufficient that an employer has adopted a practice or rule that is based on sound business practice and is rationally connected to the efficient operation of the business. Since the least discriminatory method would always be achieved if the employees' rights were considered and respected on an individual basis, the employer who seeks to bring himself or herself within the exemption may fail if he or she does not satisfy a board of inquiry that this would not be practical or feasible. In employment cases, this issue is often cast in terms of the requirement for individual testing.

In Saskatchewan (Human Rights Commission) v. Saskatoon (City), 1989 CanLII 18 (SCC), [1989] 2 S.C.R. 1297, this Court considered a rule in a collective agreement which required firefighters to retire at the age of 60. The requirement for individualized testing was raised as an absolute ingredient in the absence of evidence that it could not be done. We decided that individualized testing was to be treated as a possible alternative to a discriminatory rule. Failure on the part of the employer to show why it was impractical to test employees individually could lead to a determination that the impugned practice was not reasonable. The Court stated, at pp. 1313-14, that "[i]f there is a practical alternative to the adoption of a discriminatory rule, this may lead to a determination that the employer did not act reasonably in not adopting it". See also Central Alberta Dairy Pool v. Alberta (Human Rights Commission), 1990 CanLII 76 (SCC), [1990] 2 S.C.R. 489, at p. 518.

While the words used in s. 21 are similar to those in the employment cases, they are used and applied in a different context. The insurance context is different from the employment context. The legislature has recognized that this is so by adopting special provisions in the Code relating to insurance. There are special exemptions for certain insurance practices. For example, employee group insurance plans are allowed to differentiate on the basis of age and sex, provided that an actuarial basis exists for the differentiation. See s. 24(2) (now s. 25(2)) of the Code, and s. 34(2) of the Employment Standards Act, R.S.O. 1980, c. 137 (now R.S.O. 1990, c. E.14, s. 33(2)) and Regulation 282 of the Act. It is an important principle of insurance practice that premiums charged to individual policy holders vary as much as possible in accordance with the degree of risk posed by the policy holder. In view of the fact that individualized assessment cannot be done, it is necessary to classify the degree of risk on the basis of groups who share characteristics which are material to the risk. It is inevitable that some will be placed in a group who do not share the average characteristics of that group. Rates developed on the basis of the average characteristics of this group will thus discriminate against them. The basic human rights principles referred to above, and applied in the employment cases, must take into account these differences when applied in the context of insurance.

In my opinion, a discriminatory practice is "reasonable" within the meaning of s. 21 of the Code if (a) it is based on a sound and accepted insurance practice; and (b) there is no practical alternative. Under (a), a practice is sound if it is one which it is desirable to adopt for the purpose of achieving the legitimate business objective of charging premiums that are commensurate with risk. Under (b), the availability of a practical alternative is a question of fact to be determined having regard to all of the facts of the case.

In order to meet the test of "bona fides", the practice must be one that was adopted honestly, in the interests of sound and accepted business practice and not for the purpose of defeating the rights protected under the Code.

Before seeking to apply these principles to the facts of this case, it is important to fully understand the context in which the insurance industry sets its automobile premiums, the manner in which information regarding risk is compiled and the highly regulated environment in which the industry operates. This requires a detailed examination of the Statistical Plan and the classes of insurance which were developed under the plan.

The Statistical Plan

All automobile insurers in Ontario are required by the Insurance Act, R.S.O. 1980, c. 218, s. 80(1), (2), (now R.S.O. 1990, c. I.8, s. 101(1), (2)) to compile and report statistics and other insurance data in the form and with the content dictated by the "Statistical Plan" prescribed by the Superintendent of Insurance. It is the Superintendent of Insurance who is ultimately responsible for the supervision of the business of insurance in the Province of Ontario. This information, composed mainly of statistics regarding accident losses, is published by the industry in a document commonly referred to as the "Green Book". The Statistical Plan requires that data be collected in relation to a number of factors including age, sex and marital status. This Statistical Plan was originally developed in the 1920s and early 1930s in order to provide the insurance industry with a sufficient statistical basis to set automobile insurance rates. A similar plan was developed in most Canadian jurisdictions. The entire structure of the automobile rate setting system in Ontario flows from this collective loss experience, compiled within a legislatively determined statistical framework. As well as requiring the collection of certain data, the Insurance Act, through ss. 393(b)(iii) and 396, gives the Superintendent the power to prevent any insurance practices he or she finds to be unfair or deceptive.

While much has been made, in this appeal, of the wish of the Superintendent to move from the current rating practices, no order has ever been made by the Superintendent in regard to the industry practice of setting automobile insurance premiums based on age, sex and marital status. Mr. Newton, the senior actuary in the Office of the Superintendent of Insurance for Ontario, admitted in evidence given before the Board of Inquiry that if the Office had concluded that there was bad faith relating to the structuring of the Statistical Plan, the Superintendent would have directed that the Plan be amended. Equally, the successor to the Superintendent (the Commissioner of Insurance) (see Insurance Statute Law Amendment Act, 1990, S.O. 1990, c. 2, s. 3(1)) has done nothing to prohibit differentiation on the basis of age, sex or marital status despite having wide-ranging powers to do so.

As a result of joint industry and government discussions, the Superintendent of Insurance amended the Statistical Plan described above effective January 1, 1985 to gain information regarding further categories of risk. In spite of these amendments, however, the Plan continued to require data to be collected in relation to loss experience based on age, sex and marital status. In testimony given to the Board of Inquiry, the Superintendent of Insurance indicated that the collection of these additional statistics would allow the industry to investigate the suitability of setting premiums on the basis of criteria other than age, sex and marital status. While interested in setting premiums on the basis of other criteria, the Superintendent indicated that he was satisfied that the changes had gone as far as they could at that time. He indicated ((1985), 1985 CanLII 5245 (ON HRT), 6 C.H.R.R. D/2948), at p. D/2961, that he was:

. . . still concerned about the fact that a decent period of time is necessary for the development of credible statistics to show that the alternate factors that are selected will, in fact, stand up and will, in fact, produce the type of statistical base that the industry requires to change its system.

In fact, the industry and the Superintendent of Insurance have been trying, since at least 1977, to develop acceptable alternatives to the criteria of age, sex and marital status. In this regard, the Office of the Superintendent embarked on a search for effective alternative criteria. In parallel, the industry also undertook an exploration for alternatives. While he sought changes to the manner in which premiums were set, the Superintendent stressed that "a change in the statistical base should be allowed a sufficient length of time to gain credible experience from an actuarial standpoint".

It should be noted that changes in a system as complex as insurance cannot be made instantaneously. Once a change has been made to the Statistical Plan, it takes three years to see the results of that change. It takes this long to publish the Plan changes, implement the changes in company data processing systems and compile and publish a full year of statistical experience. The criteria of age, sex and marital status were retained in the Statistical Plan as, in the absence of a proven alternative statistical base, the effect of dropping those criteria could well have seriously disrupted the automobile insurance marketplace. As stated by the Superintendent,

We should not take steps . . . that would produce a chaotic condition in the marketplace, where, particularly the present under-25 drivers, there is no real basis for judging or calculating what their rates are.

And the effect of not having an alternate system that was developed could well have produced a chaotic effect in the marketplace that may have affected all premium-paying members of the public who buy that insurance.

Even the Superintendent himself utilized the basic structure of the Statistical Plan to "set" rates of insurance. This was done in relation to the Facility Association. The Facility Association is a mechanism, set up under the provisions of the Compulsory Automobile Insurance Act, R.S.O. 1980, c. 83 (now R.S.O. 1990, c. C.25), which pools insurance risks so that high risk drivers can obtain motor vehicle insurance in circumstances where insurance might not otherwise be available. Under s. 10 of that Act, the Superintendent has the power to approve or disallow the rates prepared by the Facility Association. In November of 1982, the Superintendent approved Facility Association automobile insurance rates which included differentiation on the basis of age, sex and marital status. These criteria were approved by the Superintendent because there were no other criteria by which rates could be set, outside of a subjective guess which the Superintendent concluded would have been wholly inadequate. The Superintendent of Insurance was of the view that there was simply insufficient data to set premiums in a manner that did not include differentiation on the basis of age, sex and marital status despite his desire to do so.

A legislative initiative attempted to remove the classification system contained in the current Statistical Plan through s. 33 of the Ontario Automobile Insurance Board Act, 1988, S.O. 1988, c. 18. Section 33(1) stated:

33.‑‑(1) No insurer shall differentiate or make a distinction, exclusion or preference in a contract of automobile insurance on the basis or age, sex, marital status, family status or handicap.

The implementation of this provision was first delayed and then the entire Act was repealed without s. 33(1) ever having come into force. Currently, insurers must apply to the Commissioner of Insurance (the successor to the Superintendent of Insurance) for approval of the classes of risk exposure they intend to use in setting rates for automobile insurance. The Commissioner must refuse to approve any class of risk exposure that is not just or reasonable or does not reasonably predict risk or distinguish fairly between classes of risk exposure (Insurance Statute Law Amendment Act, 1990, s. 74). Despite such wide-ranging powers, the Commissioner has not prohibited auto insurance rates based on age, sex or marital status. There are also other mechanisms in the amended Insurance Act to combat inappropriate insurance criteria and yet there are no prohibitions in either the Insurance Act or in its regulations, nor have any policy statements been issued by the Ministry of Financial Institutions which prohibit age, sex or marital status as rating criteria.

The Insurance Classes

For the time period covered by the application of Mr. Bates for insurance, the insurance classes comprising single males under the age of 25 had the highest claim frequency, the highest loss per car insured and the highest average claim cost of any of the categories for which statistics were kept. Internal statistics kept by Zurich confirmed this profile. In fact, for the relevant time period, the premiums charged by Zurich to drivers in Mr. Bates' class (single males aged 19 or 20) were appreciably less than the losses generated by that class. In the 1981 policy year, Zurich paid 26 percent more in claims to this class than it collected from this class. In the 1982 policy year the class had a deficit of approximately 19 percent. These company shortfalls do not even take into consideration the overhead and other losses suffered by Zurich in insuring these drivers. In fact, the break even point for Zurich is reached when claims total only 75 percent of premiums. The loss cost per car of this class was three times as high as that of the average vehicle. In comparison, during the 1982 policy year, the class composed of married males under 20 paid premiums almost three times the amount of the losses suffered by this class. Other classes showed similar, if not so dramatic, profiles. A review of the material submitted on this appeal confirms that the class in which Mr. Bates was placed had a far worse risk profile than any other comparable class of driver. While this class also paid the highest premiums, these premiums were still insufficient to cover the losses generated. It is therefore not possible to argue that there is no statistical basis for the premium allocation to this class of driver. It is the method by which these premiums were set, however, and not the amount of the premiums that is truly at issue in this appeal.

Application of Principles to the Facts

It is conceded in this appeal that the manner in which the insurance industry as a whole and Zurich in particular set their automobile insurance premiums in 1983 was discriminatory. It is also conceded that this discrimination was on the basis of grounds prohibited by the Code. What remains to be determined is whether such discrimination fits within the exception provided by s. 21 of the Code.

Moreover, there does not appear to be any dispute that Zurich acted bona fide in setting its insurance premiums. The Board of Inquiry specifically held at p. D/2966 that Zurich acted in good faith and that the classification system was developed for legitimate economic and business reasons. This finding is fully supported by the evidence and I agree with it. The issue then becomes whether the setting of premiums on the basis of age, sex and marital status was reasonable in the circumstances.

Not only did the respondent honestly believe that in setting the premium for Mr. Bates it was acting in the best interests of its insurance business, its decision was also based on credible actuarial evidence. This was undoubtedly sound and accepted business practice and satisfies the first requirement of the reasonableness test to which I have referred. The relationship between risk and the age and sex of drivers was discussed in the Report of Inquiry into Motor Vehicle Accident Compensation in Ontario, 1988, Chapter 6 (the "Osborne Report"). This inquiry investigated, inter alia, the correlation between insurance costs and criteria such as age, sex and marital status. The report, at p. 202, indicates that "[y]oung drivers are involved in proportionately more accidents, and more severe accidents than other drivers. The data are overwhelming on this issue". The report goes on to conclude that being male or female, or young or old does not cause accidents but a statistical correlation exists between these criteria and insurance losses.

This statistical analysis does not, however, fully satisfy the reasonableness test required by s. 21. Human rights values cannot be over-ridden by business expediency alone. To allow "statistically supportable" discrimination would undermine the intent of human rights legislation which attempts to protect individuals from collective fault. To allow discrimination simply on the basis of statistical averages would only serve to perpetuate traditional stereotypes with all of their invidious prejudices. Society has decided not to hold the individual responsible for the sins of his or her "group" and the courts must seek to further rather than restrict this decision. It is therefore necessary to consider whether there was an alternative which in all the circumstances was practicable. Did the industry as a whole have a reasonable alternative which would have avoided discrimination and, more specifically, did the respondent have such an alternative?

The appellant submits that there was an alternative: the industry and the respondent could have used neutral factors such as driving record, experience and vehicle use, all of which are currently employed in relation to policy-holders over 25 years of age. The Board accepted this submission. It dealt with this issue, at p. D/2965, by asking itself the following question:

In narrower terms, has Zurich established that the very essence of its business would be undermined if it could no longer rely on discriminatory group characteristics for its rate classification system?

In my view, this set too high a standard and not one that s. 21 requires. It is a higher standard than the employment cases in Canada have demanded and appears to have been adopted from employment cases in the United States which are based on different legislation, utilizing different wording, employed in a different setting. This standard ascribes too narrow a meaning to what constitutes a practical alternative. An alternative may be impractical even though its adoption would not undermine the very essence of a business. By applying this standard, the Board gave insufficient weight to the difficulties inherent in attempting to adopt new criteria in the absence of an adequate statistical base. Furthermore, this standard tended to minimize the importance of the statutory framework within which the industry and the respondent were required to operate. In February of 1983, when this complaint arose, statistics were compiled as required by law on the basis of criteria that employed the impugned characteristics. No action was taken by the Superintendent to require collection of statistics on a different basis until the amended Statistical Plan was put in place in January of 1985. When the Board decided this case in August of 1985, the plan had only been in effect some seven months, well short of the three years which, on the evidence, are required to obtain meaningful statistics. In these circumstances, I am satisfied that setting premiums on a basis that did not rely on the classifications then in use was not a practical alternative for the industry.

It would have been even more impractical for the respondent to adopt unilaterally its own classifications based on non-discriminatory criteria. It was submitted by the appellant that Zurich could have collected additional statistics and based its premiums on them. This argument ignores the fact that Zurich is a single company in a highly competitive and complex industry. In Ontario alone there are over 180 insurers licensed to sell automobile insurance, with about 50 companies being major players. The evidence indicates that Zurich insures only 116,000 cars in a jurisdiction where over 3.5 million cars are insured in total. This represents only 3.6 percent of the market. The Statistical Plan not only combines the data of all insurers but the Plan also represents the cumulative loss experience of over 50 years of insurance. It would have been impractical, if not impossible, for Zurich to generate actuarially reliable figures on only its own data. Zurich required a wider pool of information in order to set its premiums in any meaningful way. The fact that it was theoretically possible to collect data based on other criteria does not establish that reasonable alternatives to the current system exist. Neither does the presence of data from U.S. jurisdictions which may or may not reflect the risks in Ontario. In any event, I am unwilling to accept the possibility of an alternative based on U.S. data that was not before the Board and is not before this Court.

My colleague discusses the Pennsylvania case of Hartford Accident and Indemnity Co. v. Insurance Commissioner of the Commonwealth of Pennsylvania, 482 A.2d 542 (Pa. 1984), and concludes at p. 000 that "American data also demonstrates that distance driven is a more accurate predictor of accident rates than gender". I simply wish to point out that that same jurisdiction, in the case of Pennsylvania National Organization for Women v. Commonwealth of Pennsylvania Insurance Department, 551 A.2d 1162 (Pa. 1988), at p. 1167, upheld the conclusion that "there is no direct correlation between mileage and insurance costs". The complaint in that case was that flat rates for both men and women discriminated against women because the premiums did "not respond to the disparity in mileage between men and women" (at pp. 1166-67).

The auto insurance industry is a highly competitive industry and to expect a single company to strike out on its own and take such a risk on behalf of itself and its clients is unreasonable. Not only would Zurich have lacked the volume of statistics to analyze but it would have taken a long period of time to determine a suitable alternative rating scheme, if in fact one exists. In the absence of alternative criteria to set premiums, it is unreasonable to expect Zurich to simply guess at an equitable distribution of premiums among the consumers it insures.

As previously noted, statistics are now being collected on a wide range of possible risk factors. This process was initiated in 1985 and would only have generated meaningful statistics in 1988. It may well be that there now exists a statistical basis for determining insurance premiums in a non-discriminatory manner. However, this appeal is limited to the situation as it existed in 1983 and not as it existed in 1988 or as it currently exists today. There is no guarantee, in fact, that this additional data will provide the industry with a mechanism in which to set premiums on non-discriminatory criteria. Mr. Newton, the senior actuary in the Office of the Superintendent indicated in his evidence to the Board of Inquiry that there were still doubts as to whether the 1985 changes would lead to effective non-discriminatory rating criteria.

One should also not lose sight of the effect of simply submerging high risk drivers into pools of lower risk drivers. If the premiums of Mr. Bates' class are arbitrarily spread out amongst other classes, the net effect of this change is that drivers in other classes will be forced to subsidize further a group of drivers who already fail to pay premiums sufficient to cover their insurance losses. It is not reasonable simply to submerge the risks associated with this group into other classes with no statistics to support the classes into which these high-risk drivers are placed.

The insurance industry must be allowed time to determine whether it can restructure its classification system in a manner that will eliminate discrimination based on enumerated group characteristics and still reflect the disparate risks of different classes of drivers. It would therefore be inappropriate for this Court to find a particular practice to be unreasonable when no reasonable alternative exists. While the situation as it existed in 1983 did not provide a reasonable alternative to setting premiums based on age, sex and marital status, the situation today and in the future may be quite different. The insurance industry must strive to avoid setting premiums based on enumerated grounds.

Disposition

Zurich set its premiums on the basis of the only statistics available to the insurance industry at the time in question. These statistics supported the imposition of higher premiums on certain classes of drivers whose cumulative accident history suggested an increased insurance cost. Faced with an absence of any other criteria in which to set insurance rates, I am satisfied that Zurich set its rates on reasonable and bona fide grounds as those terms are used in s. 21 of the Code. Zurich was, in short, forced to use the industry wide statistics contained in the Superintendent's Statistical Plan to set its premiums. While it is not enough that insurance premiums be set in good faith and be supported by credible actuarial evidence, I find that Zurich has acted within the limits of s. 21 of the Ontario Human Rights Code, 1981 in all the circumstances of this case. In view of this conclusion, it will not be necessary to consider the effect of s. 46(3) of the Code. I would therefore dismiss the appeal with costs.

The following are the reasons delivered by

//L'Heureux-Dubé J.//

Dissenting Reasons of L'Heureux-Dubé J.

L'Heureux-Dubé J. (dissenting) -- This case concerns the interpretation of s. 21 of the Ontario Human Rights Code, 1981, S.O. 1981, c. 53 (the "Code") (now R.S.O. 1990, c. H.19, s. 22), which permits insurers to make distinctions in insurance policies on reasonable and bona fide grounds on account of age, sex, marital status, family status or handicap.

I have read my colleague Justice Sopinka's reasons, and I agree with his comprehensive examination of both the history of the classification system in Ontario and the background of the complaint in this case. While I also agree with his view that there exists a distinction between discrimination in employment and in insurance contracts, and with his analysis of the insurance context, I respectfully disagree with his interpretation of s. 21 of the Code, and with his treatment of the findings of fact by the Board of Inquiry.

I shall briefly allude to the facts before analyzing the decisions rendered below in light of my own view of the importance of the findings of the Board of Inquiry ((1985), 1985 CanLII 5245 (ON HRT), 6 C.H.R.R. D/2948).

On April 30, 1984, a Board of Inquiry was appointed pursuant to the Code to hear and decide a complaint made by Michael G. Bates against the respondent insurance company. The complainant alleged that young, single male drivers must pay automobile insurance rates which exceed those of young, single female drivers and young, married male drivers, as well as rates set for all drivers over the age of 25, and that this discrepancy contravenes the provisions of the Code. Specifically, he submitted that he had been denied the right to contract on equal terms without discrimination and the right to equal treatment in services, goods and facilities, contrary to ss. 1, 3 and 8 of the Code, as my colleague Sopinka J. points out. The respondent has throughout these proceedings conceded that the classification system upon which the premiums were based constitutes a prima facie infringement of the Code, but contends that this infringement falls within the exception set out in s. 21 of the Code.

Judgments

The Board of Inquiry relied on the line of cases from this Court interpreting the "bona fide occupational qualification" exception, and concluded that the distinctions in the classification scheme would be based on reasonable and bona fide grounds only if they were "reasonably necessary". As formulated by the Board, at p. D/2964, the test under s. 21 of the Code emphasized the need for insurers to show the absence of practical alternatives to discriminatory practices, all the while recognizing the impossibility of individual testing for establishing insurance rates:

While, as stated above, it is unreasonable to expect automobile insurance companies to do individual testing, they should be scrutinized in terms of reasonable alternatives which may indicate whether the present scheme of rate classification is "reasonably necessary". Without requiring that insurance companies assess individual characteristics for everyone, it is reasonable to inquire if it is possible or practical to deal with applicants for insurance under‑25 in the context of a classification system that relies on characteristics other than age, sex or marital status. In other words, are there any characteristics which can be used which are non‑discriminatory in the sense that they are not prohibited by sections 1 and 3 of the Code? We can ask whether Zurich has demonstrated that it would be impractical or impossible to operate the automobile insurance system with the use of non‑discriminatory criteria for rate classification. [Emphasis added.]

The Board decided that the evidence proffered by the respondent did not satisfy that burden. Several findings of fact were crucial to this decision. First, the Board said at p. D/2965:

There has simply been no evidence, in my opinion, offered to support the assertion that it has been scientifically proven that there is a direct, causal relationship between the discriminatory group factors used ‑ age, sex and marital status and high risk. [Emphasis added.]

In fact, the Board noted that the evidence demonstrated that age, sex and marital status were likely only proxy factors, and had never been controlled or isolated in the statistics to determine whether a causal connection exists.

Second, the Board determined that there was no statistical basis for choosing the age of 25 as the dividing line, nor did other elements of the system, including the surcharge for convictions, multi‑vehicle discounts and driver‑training discounts, rest on any statistical basis. In the face of these anomalies, the Board concluded at p. D/2965 that:

. . . the automobile premium rate-structure in Ontario is as much if not more based upon economic and business decisions than upon the Statistical Plan.

Finally, the Board held that the respondent had clearly not proven that there were no other group factors which are non‑discriminatory. The Superintendent of Insurance had been urging change for years, demonstrating that, in his view, feasible and desirable alternatives existed. The Board emphatically rejected the respondent's contention that the system was reasonable because it had been the only system in place for 50 years. "The insurance industry cannot", the Board held, "rely on its inaction and tradition to support a discriminatory rate classification system". Most importantly, the Board held at p. D/2966 that the evidence demonstrated that an alternative to the discriminatory rate classification system existed and was being used at the time of the hearing before the Board:

For those insured persons who are 25 or older, the industry relies solely on non‑discriminatory, causally connected factors such as driving record, driving experience and vehicle use to establish rate differentials. The evidence indicates that this system has worked well. There has been no evidence offered to suggest that these factors cannot be used for insured persons who are under-25.

The Divisional Court of Ontario allowed the appeal from the decision of the Board of Inquiry ((1987), 1987 CanLII 4058 (ON SC), 58 O.R. (2d) 325). It rejected the Board's interpretation of s. 21 and held that the words "reasonable and bona fide grounds" should be construed according to their plain meaning. The Court concluded that the distinctions created by the classification system were reasonable distinctions since they were supported by reasonable, actuarially verified statistics, showing a higher rate of risk for single male drivers under the age of 25. In the alternative, the Divisional Court held that Zurich had met the test of reasonable necessity because it relied upon the only statistics available to it. The Court of Appeal affirmed the decision of the Divisional Court in a brief endorsement ((1989), 1989 CanLII 4051 (ON CA), 70 O.R. (2d) 639).

Relevant Statutory Provisions

At the time of the complaint, the relevant provisions of the Human Rights Code, 1981 read as follows:

Every person has a right to equal treatment with respect to services, goods and facilities, without discrimination because of race, ancestry, place of origin, colour, ethnic origin, citizenship, creed, sex, age, marital status, family status or handicap.
Every person having legal capacity has a right to contract on equal terms without discrimination because of race, ancestry, place of origin, colour, ethnic origin, citizenship, creed, sex, age, marital status, family status or handicap.
No person shall infringe or do, directly or indirectly, anything that infringes a right under this Part.
The right under sections 1 and 3 to equal treatment with respect to services and to contract on equal terms, without discrimination because of age, sex, marital status, family status or handicap, is not infringed where a contract of automobile, life, accident or sickness or disability insurance or a contract of group insurance between an insurer and an association or person other than an employer, or a life annuity, differentiates or makes a distinction, exclusion or preference on reasonable and bona fide grounds because of age, sex, marital status, family status or handicap. [Emphasis added.]

Sections 1 and 3 have subsequently been amended to add sexual orientation as a prohibited ground of discrimination (Equality Rights Statute Law Amendment Act, 1986, S.O. 1986, c. 64, s. 18).

Analysis

This Court is required to interpret, for the first time, the term "reasonable and bona fide grounds" in s. 21 of the Ontario Human Rights Code, 1981 as it applies to the facts of this case.

General Principles

The starting point for any analysis of human rights legislation is the recognition that the purpose of such legislation is the protection of fundamental individual rights. These rights are violated if stereotypical group characteristics are ascribed to individuals. Since the protection of human rights is vitally important in our society, legislation prohibiting discrimination must be construed broadly using a purposive approach. The approach of this Court to the interpretation of human rights legislation is typified by the statement of McIntyre J. in Ontario Human Rights Commission v. Simpsons‑Sears Ltd., 1985 CanLII 18 (SCC), [1985] 2 S.C.R. 536, at p. 547:

Legislation of this type is of a special nature, not quite constitutional but certainly more than the ordinary -‑ and it is for the courts to seek out its purpose and give it effect.

A logical corollary to this purposive approach is the narrow interpretation of exceptions within human rights legislation. Defences to discrimination under the Code must be narrowly construed so that the larger objects of the Code are not frustrated. In this regard, the comments of Justice Lamer, now Chief Justice, in Insurance Corporation of British Columbia v. Heerspink, 1982 CanLII 27 (SCC), [1982] 2 S.C.R. 145, at pp. 157‑58, are apposite:

When the subject matter of a law is said to be the comprehensive statement of the "human rights" of the people living in that jurisdiction, then there is no doubt in my mind that the people of that jurisdiction have through their legislature clearly indicated that they consider that law, and the values it endeavours to buttress and protect, are, save their constitutional laws, more important than all others. Therefore, short of that legislature speaking to the contrary in express and unequivocal language in the Code or in some other enactment, it is intended that the Code supersede all other laws when conflict arises. [Emphasis added.]

Similarly, Beetz J. wrote in Brossard (Town) v. Quebec (Commission des droits de la personne), 1988 CanLII 7 (SCC), [1988] 2 S.C.R. 279, at p. 307:

It has been decided by this Court in Ontario Human Rights Commission v. Borough of Etobicoke, 1982 CanLII 15 (SCC), [1982] 1 S.C.R. 202, and reaffirmed in Bhinder v. Canadian National Railway Co., 1985 CanLII 19 (SCC), [1985] 2 S.C.R. 561, that bona fide occupational qualification exceptions in human rights legislation should, in principle, be interpreted restrictively since they take away rights which otherwise benefit from a liberal interpretation.

See also Bhinder v. Canadian National Railway Co., 1985 CanLII 19 (SCC), [1985] 2 S.C.R. 561, at pp. 567‑69 and 589.

It is noteworthy that the exception in s. 21 of the Code is unlike the "exceptions" at issue in Caldwell v. Stuart, 1984 CanLII 128 (SCC), [1984] 2 S.C.R. 603, and the second branch of s. 20 in Brossard. In those cases, the provisions in question limited rights and also conferred rights. As stated by McIntyre J. in Caldwell, at p. 626:

. . . the courts should not in construing s. 22 consider it merely as a limiting section deserving of a narrow construction. This section, while indeed imposing a limitation on rights in cases where it applies, also confers and protects rights.

Section 21, in contrast, confers no rights. It simply provides a defence to what would otherwise be discrimination contrary to the Code, and must therefore be narrowly construed.

Before turning to the interpretation of s. 21, it is also important to emphasize the general principle of deference to administrative tribunals. Curial deference to the findings of the Board of Inquiry is appropriate both because of their expertise in matters of human rights and because the board was in the best position to assess the evidence before it.

These principles have been well established by this Court. Most recently, this Court considered the principle of curial deference in National Corn Growers Assn. v. Canada (Import Tribunal), 1990 CanLII 49 (SCC), [1990] 2 S.C.R. 1324. In a concurring opinion, Wilson J. at p. 1336 reviewed the evolution of this principle in Canadian law and the growing reluctance of courts to interfere with decisions made by administrative tribunals within their areas of expertise:

Part of this process has involved a growing recognition on the part of courts that they may simply not be as well equipped as administrative tribunals or agencies to deal with issues which Parliament has chosen to regulate through bodies exercising delegated power, e.g., labour relations, telecommunications, financial markets and international economic relations. Careful management of these sectors often requires the use of experts who have accumulated years of experience and a specialized understanding of the activities they supervise.

Courts have also come to accept that they may not be as well qualified as a given agency to provide interpretations of that agency's constitutive statute that make sense given the broad policy context within which that agency must work. [Emphasis added.]

Similarly, in Lester (W.W.) (1978) Ltd. v. United Association of Journeymen and Apprentices of the Plumbing and Pipefitting Industry, Local 740, 1990 CanLII 22 (SCC), [1990] 3 S.C.R. 644, Justice McLachlin for the majority held at p. 669:

Courts should exercise caution and deference in reviewing the decisions of specialized administrative tribunals, such as the Labour Board in this case. This deference extends both to the determination of the facts and the interpretation of the law.

Wilson J. agreed with this principle, although she dissented on its application to the facts of the case at p. 650:

The principle of curial deference to decisions of administrative tribunals is grounded not only on sound judicial policy but also on good common sense. It is quite unrealistic in this age of increasingly complex and highly specialized regulatory regimes to expect the courts to have the requisite knowledge and skill to adjudicate properly on some of those regimes.

My colleague Gonthier J. made the appropriate distinctions in Bell Canada v. Canada (Canadian Radio-television and Telecommunications Commission), 1989 CanLII 67 (SCC), [1989] 1 S.C.R. 1722, at pp. 1745‑46:

It is trite to say that the jurisdiction of a court on appeal is much broader than the jurisdiction of a court on judicial review. In principle, a court is entitled, on appeal, to disagree with the reasoning of the lower tribunal.

However, within the context of a statutory appeal from an administrative tribunal, additional consideration must be given to the principle of specialization of duties. Although an appeal tribunal has the right to disagree with the lower tribunal on issues which fall within the scope of the statutory appeal, curial deference should be given to the opinion of the lower tribunal on issues which fall squarely within its area of expertise.

See also CAIMAW v. Paccar of Canada Ltd., 1989 CanLII 49 (SCC), [1989] 2 S.C.R. 983, at pp. 1002 (La Forest J.), 1020 (Wilson J.) and 1033 (L'Heureux‑Dubé J.); Gendron v. Supply and Services Union of the Public Service Alliance of Canada, Local 50057, 1990 CanLII 110 (SCC), [1990] 1 S.C.R. 1298, at pp. 1322‑26; and Canada (Attorney General) v. Public Service Alliance of Canada, 1991 CanLII 88 (SCC), [1991] 1 S.C.R. 614, at p. 629.

The principle of curial deference is thus applied most stringently in the judicial review of administrative action. The findings of a specialized tribunal can only be overturned by a court if they are patently unreasonable.

The principle of curial deference is also applicable in cases where appellate courts are expressly given a broader power of review, here the Board of Inquiry under s. 41(3) of the Code. The importance of curial deference to the Board of Inquiry was recognized by this Court in Ontario Human Rights Commission v. Borough of Etobicoke, 1982 CanLII 15 (SCC), [1982] 1 S.C.R. 202, at p. 211. McIntyre J. there noted that because of the broader power of review, it would not always be an error for the appellate court to substitute its findings for those of the board. However, he acknowledged the importance of the principle set out in Re Ontario Human Rights Commission and City of North Bay (1977), 1977 CanLII 1253 (ON CA), 17 O.R. (2d) 712, at p. 716, where Hughes J. held that:

. . . the learned Inquiry Officer's finding that the mandatory retirement provision was a bona fide occupational qualification and requirement for Cosgrove's employment was a finding of fact. ... It was he who heard the evidence and under the circumstances was the only authority capable of judging the good faith of the assertions made before him. No appellate Court, in my respectful view, should overlook this signal advantage.

Thus, in my view, particular attention should be paid to the findings of the Board of Inquiry and both a court of appeal and this Court should, absent error, be loathe simply to substitute their own opinions for well‑considered views and findings of the Board of Inquiry, even in cases like this one where, as my colleague points out, there is no privative clause and the appeal provision is wider than the normal powers of an appellate court. This, en passant, was also the case in Etobicoke, supra. I still hold to the opinion expressed in R. v. M. (S.H.), 1989 CanLII 31 (SCC), [1989] 2 S.C.R. 446, as regards powers of appeal courts in such circumstances.

It is against this background that this case must be examined.

As stated earlier, the respondent has conceded that the classification system upon which its premiums were based constitutes a prima facie infringement of the Code. The sole issue, therefore, is whether the discriminatory classification scheme is saved under the exception set out in s. 21.

"Reasonable and bona fide grounds" under s. 21

“Reasonable and bona fide grounds” under s. 21

There has been no judicial interpretation of s. 21 of the Code. Similar exceptions have, however, been considered by this Court. The interpretation of an exception for "bona fide occupational qualification and requirement" was at issue in Etobicoke. In that case, McIntyre J. held at p. 208 that:

To be a bona fide occupational qualification and requirement a limitation, such as a mandatory retirement at a fixed age, must be imposed honestly, in good faith, and in the sincerely held belief that such limitation is imposed in the interests of the adequate performance of the work involved with all reasonable dispatch, safety and economy, and not for ulterior or extraneous reasons aimed at objectives which could defeat the purpose of the Code. In addition it must be related in an objective sense to the performance of the employment concerned, in that it is reasonably necessary to assure the efficient and economical performance of the job without endangering the employee, his fellow employees and the general public. [Emphasis added.]

In Brossard, the legislation provided an exception for "aptitudes or qualifications required in good faith". Beetz J. articulated two criteria for establishing "reasonable necessity", the second branch of the Etobicoke test at pp. 311‑12:

(1) Is the aptitude or qualification rationally connected to the employment concerned? This allows us to determine whether the employer's purpose in establishing the requirement is appropriate in an objective sense to the job in question. In Etobicoke, for example, physical strength evaluated as a function of age was rationally connected to the work of being a fireman.

(2) Is the rule properly designed to ensure that the aptitude or qualification is met without placing an undue burden on those to whom the rule applies? This allows us to inquire as to the reasonableness of the means the employer chooses to test for the presence of the requirement for the employment in question. The sixty‑year mandatory requirement age in Etobicoke was disproportionately stringent, for example, in respect of its objective which was to ensure that all firemen have the necessary physical strength for the job.

Since Etobicoke and Brossard, the Court has emphasized the requirement, implicit in the second criterion, that the party who seeks to justify discrimination against an individual must show that there is no reasonable or practical alternative. In Saskatchewan (Human Rights Commission) v. Saskatoon (City), 1989 CanLII 18 (SCC), [1989] 2 S.C.R. 1297, Sopinka J. wrote for the Court at pp. 1313‑14:

While it is not an absolute requirement that employees be individually tested, the employer may not satisfy the burden of proof of establishing the reasonableness of the requirement if he fails to deal satisfactorily with the question as to why it was not possible to deal with employees on an individual basis by, inter alia, individual testing. If there is a practical alternative to the adoption of a discriminatory rule, this may lead to a determination that the employer did not act reasonably in not adopting it. [Emphasis added.]

Or, as Wilson J. said in the recent judgment in Central Alberta Dairy Pool v. Alberta (Human Rights Commission), 1990 CanLII 76 (SCC), [1990] 2 S.C.R. 489, at p. 518:

If a reasonable alternative exists to burdening members of a group with a given rule, that rule will not be bona fide.

(See also the concurring reasons of Sopinka J. in that case at pp. 526‑27.)

While these tests are helpful, they have clearly been developed by this Court in the employment context, as my colleague Sopinka J. rightly points out. Should this test form the basis for interpretation of "reasonable and bona fide grounds" in s. 21 of the Code, and if so, should the test be modified? If not, what test should apply?

My colleague Sopinka J. proposes, at p. 000, a modification of the Brossard test applicable to the insurance context so that, under s. 21, a discriminatory practice would be reasonable if the insurer could show that it was based on a "sound and accepted insurance practice" and there is no practical alternative. He defines, at p. 000, a practice as sound if it is "desirable to adopt for the purpose of achieving the legitimate business objective of charging premiums that are commensurate with risk". As well, the insurer must prove bona fides by showing that the practice was adopted honestly, in the interests of sound and accepted business practice and not for the purpose of defeating the rights protected under the Code. My own view is that the proper test should be less stringent than the one used by the Board of Inquiry, but more stringent than that devised by the Divisional Court and my colleague's proposed test.

The reasons of the Board of Inquiry and of the Divisional Court reflect two radically different approaches to s. 21 of the Code. The Divisional Court rejected the applicability of the "bona fide occupational qualification" jurisprudence to the case at bar. In its view, a stringent test was required in the employment context because individual assessment was generally desirable and feasible. Individual assessment is not possible in insurance, and in fact, defeats the purpose of insurance. Therefore, in the view of the Divisional Court, the "bona fide occupational qualification" test developed in the employment context was of no assistance in the interpretation of "reasonable and bona fide grounds" in the insurance context.

The Board of Inquiry, by contrast, adopted at p. D/2965 the bona fide occupational qualification test and then expanded it, based on American authorities, so that the respondent was required to demonstrate that "the very essence of its business would be undermined if it could no longer rely on discriminatory group characteristics". It went as far as to hold, at p. D/2964, that the burden for the respondent was to demonstrate that "it would be impractical or impossible to operate the automobile insurance system with the use of non‑discriminatory criteria for rate classification".

In my opinion, both the Divisional Court and the Board of Inquiry erred in their formulations of the requisite tests. By relying on stereotype and entrenched practice, the approach of the Divisional Court subverts the very purpose of human rights legislation. The test adopted by the Board of Inquiry, at the other extreme, would make it practically impossible for an insurance company to justify a distinction based on a prohibited ground. This would render s. 21 meaningless and surely could not have been the intention of the Legislature.

In my view, the test for "reasonable and bona fide grounds" should be based on that established by this Court for "bona fide occupational qualification", "bona fide occupational requirement" and "aptitudes and qualifications required in good faith". The terms "reasonable and bona fide" are stronger than the single term "bona fide"; therefore, there is no basis for concluding, as the Divisional Court did, that the former is a less stringent exception than the latter.

Moreover, the protection of individual rights is equally important in employment and insurance contexts. As the Alberta Human Rights Commission submitted:

The Code, on its face prohibits discrimination in public services, contract, and employment with equal vigour. Nothing in the statute or its history, nor that of other Canadian human rights statutes, suggests a diminished standard of interpretation for defence sections outside of the employment area.

Support for the broader applicability of the "bona fide occupational qualification" test can also be found in Canada (Attorney General) v. Rosin, 1990 CanLII 12957 (FCA), [1991] 1 F.C. 391 (C.A.), at pp. 408‑9. In that case, the court compared the defences in paras. 15(a) and (g) of the Canadian Human Rights Act, R.S.C., 1985, c. H‑6, and determined that despite the differences in wording, the defences should be given the same interpretation:

Similarly, it might be concluded that the two phrases -- "bona fide occupational requirement" (as in paragraph 15(a)) and "bona fide justification" (as in paragraph 15(g)) convey the same meaning, except that the former is applicable to employment situations, whereas the latter is used in other contexts. The choice of these different words used to justify prima facie discrimination, therefore, are matters of style rather than of substance.

In my opinion the test set out by Beetz J. in Brossard, supra, should form the basis for determining whether the discriminatory classifications in this case are based on "reasonable and bona fide grounds". Given, however, the context of the Brossard case, some modifications are required in the present context, which the legislature itself recognized. I would therefore adopt the basic framework from Brossard, and restate the test to be applied under s. 21 of the Code. Like my colleague Sopinka J., I would adopt a two‑pronged test with subjective and objective components.

As to the subjective component of the test, to be based on reasonable and bona fide grounds, a distinction, exclusion or preference because of age, sex, marital status, family status or handicap in an insurance contract must be imposed honestly, in good faith, and in the sincerely held belief that such distinction, exclusion or preference accurately reflects the cost of the risk insured, and not for ulterior or extraneous reasons aimed at objectives which could defeat the purpose of the Code.

The objective component of the test requires that the distinction, exclusion or preference be related in an objective sense to the insured risk, in that it is reasonably necessary to assure the proper allocation of risk among insured groups. Whether the distinction, exclusion or preference is reasonably necessary can be determined by considering the following two guidelines:

(1) Is the distinction, exclusion or preference rationally connected to the insured risk? This rational connection must be something more than a statistical correlation; a causal connection between the distinction and the insured risk must be established. Proof of a rational connection must also consist of more than a simple reliance on traditional or accepted insurance practices which may be nonetheless needlessly discriminatory.

(2) Is the distinction, exclusion of preference a reasonable means of identifying and classifying similar risks? Under this section of the test, the availability of alternative means of ascertaining the risk can be measured.

I recognize, as did the Divisional Court, that while individual testing may be both feasible and desirable in the employment context, it is not possible in the insurance context where, as my colleague points out, rates are based on characteristics shared by members of an identifiable group. I agree with Sopinka J. that it is inevitable that some individuals placed in a particular group will not share the average characteristics of that group, will pose less of a risk than the average member and will hence suffer from inclusion in the group. However, in keeping with the spirit of the Court's comments in Saskatchewan (Human Rights Commission) v. Saskatoon (City) and Central Alberta Dairy Pool v. Alberta (Human Rights Commission), supra, the desirability of placing the fewest individuals in this position by identifying groups that are as little over‑inclusive as possible is important in applying the test, particularly in the evaluation of available alternatives and the proportionality of the discriminatory classification.

Although the Board applied the "reasonably necessary" component of the test, it may have required too onerous a burden of proof from the respondent in so far as it required Zurich to show that it would be "impractical or impossible" to operate its business with the use of non‑discriminatory criteria for rate classification. Accordingly, it is necessary to review the Board's findings of fact, bearing in mind the less stringent test that I have devised, in order to ascertain whether the respondent has met the appropriate burden of proof.

Application

Since the appellant has established a prima facie case of discrimination, the burden rests on the respondent to demonstrate the "reasonable and bona fide grounds" upon which the discriminatory classification system rests. As McIntyre J. held in Etobicoke, at p. 208:

Once a complainant has established before a board of inquiry a prima facie case of discrimination. . . he is entitled to relief in the absence of justification by the employer. The only justification which can avail the employer in the case at bar, is the proof, the burden of which lies upon him, that such compulsory retirement is a bona fide occupational qualification and requirement for the employment concerned. The proof, in my view, must be made according to the ordinary civil standard of proof, that is upon a balance of probabilities.

It has been accepted by all the parties that the discriminatory classification scheme used by the respondent was chosen in good faith. There is no question that the first, subjective component of the applicable test has been satisfied.

The central question in this appeal is whether the second, objective component of the test is met. The first branch of the Brossard test requires the respondent to demonstrate that there is a rational connection between the discriminatory rate classification system and high risk. The Divisional Court concluded that such a rational connection existed since the evidence demonstrated a statistical correlation between young, single males and high risk. I am of the opinion that this conclusion is dangerously and fundamentally flawed.

The mere statistical correlation between a group and higher risk cannot suffice to justify discrimination on prohibited grounds. Such correlation accepts the very stereotyping that is deemed unacceptable by human rights legislation: prohibited grounds of discrimination are used to ascribe the characteristics of the group to all individuals in the class. I agree with the intervener the Alberta Human Rights Commission that:

All generalizations on prohibited grounds are presumptively objectionable, whether they are capable of being reduced to statistics or not. It is the blind application of the stereotype to the individual, not the untruth of the stereotype, that makes such a generalization objectionable.

Discrimination based on statistical correlation is simply discrimination in a more invidious form.

This is not to say that statistics could not be used to justify the discriminatory rate classification system. On the contrary, strong statistical proof is required to demonstrate a rational connection between the discriminatory classification and high risk. This proof, however, must not simply be one of correlation, but one of causal connection. It is this causal connection that the respondent has clearly failed to establish here.

The respondent, which has the burden of proof in this matter, has failed to show that there is any causal link between age, sex or marital status and an insured's likelihood of having a car accident. Indeed, the Board of Inquiry found, at p. D/2965:

Since, as the Board noted, age, sex and marital status had never been controlled or isolated in the statistics to determine whether a causal connection exists, it is not surprising that the respondent has been unable to demonstrate that such a causal connection exists. The insurance industry simply does not know whether age, sex and marital status are causally connected to higher risk and is attempting to bridge this gap in knowledge by reliance on myth and stereotype. This impressionistic approach does not satisfy the burden of proof and constitutes impermissible discrimination, in my view.

The Supreme Court of Pennsylvania came to the same conclusion in Hartford Accident and Indemnity Co. v. Insurance Commissioner of the Commonwealth of Pennsylvania, 482 A.2d 542 (Pa. 1984). In that case, a young, male automobile insurance policy holder complained to the Insurance Commissioner that the premiums he was charged were "unfairly discriminatory", contrary to the state Casualty and Surety Rate Regulation Act. The Insurance Commissioner held that "discriminatory" should be interpreted in the context of the state Equal Rights Amendment and concluded that the rates were unfairly discriminatory. The Supreme Court upheld the decision, concluding that in order to justify the discriminatory classification scheme, the insurance company was required to demonstrate a causal connection between the classifications and the risk of accident. Nix C.J. reproduced the recommendations of the National Association of Insurance Commissioners' Report of the Rates and Rating Procedures Task Force of the Automobile Insurance (D3) Subcommittee (November, 1978) at pp. 548‑49:

In terms of simplicity and consistency (i.e., stability and ease of verification), age, sex and marital status receive high marks as rating factors. This is not the case from the viewpoint of causality. Causality refers to the actual or implied behavioral relationship between a particular rating factor and loss potential. The longer a vehicle is on the road, for example, the more likely it is that the vehicle may be involved in a random traffic accident; thus, daily or annual total mileage may be viewed as a causal rating factor. To the extent that sex and marital status classifications may be defended on causal grounds, the implied behavioral relationships rely largely on questionable social stereotypes... . Given the significant changes in traditional sex roles and social attitudes which have occurred in recent years, justifications for rating plans on the grounds of such implied assumptions are unacceptable.

. . . (P)ublic policy considerations require more adequate justification for rating factors than simple statistical correlation with loss; in this regard, the task force recommends consideration of criteria such as causality, reliability, social acceptability, and incentive value in judging the reasonableness of a classification system. Based on these criteria, the task force concludes that as rating characteristics, sex and marital status are seriously lacking in justification and are subject to strong public opposition, and should therefore be prohibited as classification factors. [Emphasis added.]

Based on these findings and the evidence before the court, Hutchinson J. concluded at p. 550:

Absent at least a causal relation between sex and accident incidence a difference in auto insurance rates between men and women is plainly an unfair discrimination based on sex. No causal connection is shown on this record. What does appear is only a statistical correlation between sex and the incidence of auto accidents. This correlation simply provides a convenient measuring rod for setting rate differentials occasioned by other factors not so easily identified or quantified. Such considerations of convenience are not enough to stand in the face of our ERA. [Emphasis added.]

Applying the "reasonably necessary" component of the objective test in the case at bar, the Board of Inquiry wrote at p. D/2965:

There has also been no evidence offered by Zurich to show that unmarried males under the age of 25 constitute a high risk. There was evidence which was offered in support of the contrary view that the factors of age, sex and marital status are, very likely, mere proxy factors which are not, in fact, causally related to the differentials in the claims experience demonstrated for the various groupings. . . . Thus, as age, sex and marital status have never been controlled or isolated in the statistics to ascertain if they are high risk factors, it has never been shown that there is reasonable cause to assume that all or substantially all unmarried males under-25 present a high risk. Many anomalies and non‑statistical factors have raised doubts as to whether the present system is, in fact, scientifically or statistically based and, thus, whether statistics should be used to justify discrimination. There is no statistical basis to support a surcharge for convictions, multi-vehicle discounts, driver‑training discounts or good‑student discounts. Similarly, there is no statistical basis for the view that administrative costs are higher for the under‑25 group. Furthermore, there are certain anomalies, such as the inconsistency between companies vis‑a‑vis their forgiveness policies, and the fact that a principal driver over‑25 in combination with an occasional driver under‑25 will pay more than a principal driver under‑25 with an occasional driver over‑25. These anomalies reinforce my opinion that the automobile premium rate‑structure in Ontario is as much if not more based upon economic and business decisions than upon the Statistical Plan.

Is the present anomalous system nevertheless reasonably necessary to assure the accomplishment of the objectives of the industry? Certainly, if the central objective is the ease of establishing insurance rates ... . But with the statutory duty of providing insurance rates that differentiate only on certain limited, reasonable and bona fide grounds, the ease of establishing rates is not an acceptable objective. [Emphasis added.]

My own analysis of the evidence dictates the same conclusion in regard to the lack of causal connection, bearing in mind the appropriate test.

In Hartford, the Pennsylvania court held, at p. 547, that the phrase "unfairly discriminatory" must be interpreted in view of the state's adoption of the Equal Rights Amendment, which provides that "[e]quality of rights under the law shall not be denied or abridged ... because of the sex of the individual". While, no doubt, differences in the statistical evidence as well as in the wording of the Act can be found between our two countries, it is the reasoning and result in Hartford which are particularly compelling for our purposes because in the present case, we are also faced with a legislative provision permitting discrimination on "reasonable and bona fide grounds" (and therefore forbidding "unfair discrimination") in the broader context of a human rights code which prohibits any discrimination on the grounds of gender, age or marital status.

It may well be that, based on proper statistical evidence rather than stereotypes, there is a causal link between the criteria presently used by Zurich to establish rates and the high risk such that the higher rates paid by young, single, male drivers may be justified. However, the respondent has not produced evidence showing this link and, since it bears the evidentiary burden under s. 21 of the Code, it was not able to justify its use of the discriminatory scheme, as the Board of Inquiry properly found.

I therefore conclude that the respondent has not demonstrated on a balance of probabilities that there is a rational connection between the discriminatory classification scheme and higher risk. Although the Board of Inquiry applied a test under s. 21 that was too stringent, in my opinion, on the evidence before the Board the respondent would not have been able to meet the burden of proof on the less stringent test which I have devised. Accordingly, there is no need to remit this matter to the Board, except for the determination of the appropriate remedy.

This is sufficient to dispose of this appeal. However, in light of the reasons of the Divisional Court, and the reasons of my colleague, I will address the second branch of the objective component of the Brossard test.

The second branch, as I have reformulated it in the context of s. 21 of the Code, requires the respondent to demonstrate that the discriminatory classification scheme is a reasonable means to achieve its goal of matching group classification to higher risk. Central to a consideration of this question is the availability of alternative means of classifying individuals into groups. The Divisional Court concluded, and my colleague agrees, that at the time of the complaint no such alternatives existed and therefore the means chosen by the respondent were reasonable. I disagree.

It is true that, at the time of this complaint, collection of statistics which would have provided the basis for an alternative classification system was not required by the Superintendent of Insurance. This, in my view, is not an adequate answer to the complaint of discrimination. While my colleague places great reliance on the testimony of the Superintendent and his failure to force insurers to adopt a non‑discriminatory rating scheme, it is clear that it is not up to the Superintendent to decide whether or not there was an infringement of the Code.

Moreover, after hearing testimony from the Superintendent as well as witnesses appearing on behalf of Zurich, the Board determined that the Superintendent was in favour of a non‑discriminatory position, but thought it preferable to adopt a non‑confrontational approach to insurance companies, in the belief that it would be better if the insurers themselves took the initiative to change the discriminatory rating system without being forced under law to do so. The Board of Inquiry, of course, is there to apply the law, not to look for a consensus, once it has concluded that a consensus cannot be achieved, which is clearly the case here.

The Board thus held, at pp. D/2965-66, that the Superintendent's decision not to compel insurance companies to collect statistics which would allow them to set rates on a non‑discriminatory basis did not excuse Zurich's failure to take any steps to investigate alternative ratings systems:

First since 1977, there have been many meetings between the Superintendent's Office and the industry with the express purpose of amending the industry's classification system. The thrust of the Superintendent's position, in accordance with the comments of the Select Committee of the Legislature on Company Law, has been to de‑emphasize or to eliminate entirely the factors of sex, marital status and age from the classification system for the under‑25 group. Thus, the Superintendent feels that there are various alternatives which are possible and desirable.

Counsel for Zurich maintained that the Superintendent's Office supported the discriminatory rate classification system as it did not legislate against it. While it is true that, aside from the Code, there has been no specific legislative reform, this absence is not absolute proof that the Superintendent's Office supports the present classification system. Mr. Newton testified that, despite the Superintendent's views agains [sic] the classification system, legislation has not been imposed due to the working philosophy of mutual cooperation rather than confrontation with the industry. Mr. Newton also testified that it is the Superintendent's hope that the industry will be able to change the system on its own, rather than being forced by the Government. [Emphasis added.]

In keeping with curial deference to the findings of fact of specialized tribunals, in my opinion the reliance by the Board in this respect on the testimony it recounts should not be second‑guessed by this Court in the absence of a clear error or misapprehension on the part of the Board. It is not open to the Court to conduct its own re‑examination of the evidence, and then to substitute its opinion for that of the tribunal which had the privilege of seeing and hearing witnesses.

Moreover, I agree with and adopt the conclusion of the Board of Inquiry on that particular issue at p. D/2966:

Simply asserting that a system has been in place for fifty years is an inadequate response to the argument that there are alternatives to the system. The insurance industry cannot rely on its inaction and tradition to support a discriminatory rate classification system.

The fact is, Zurich, as well as other Canadian insurance companies, have too long relied upon the tradition of discriminatory practices in their industry in order to justify their refusal to take the first step in implementing a less discriminatory system. As I mentioned earlier, the Superintendent has been urging a change since 1977. Having done nothing to comply with the spirit of the Code, Zurich and its competitors are ill‑placed to defend their practices on the grounds of a lack of statistical data. No human rights legislation could ever attain its objectives if discrimination could be justified by the self‑serving claim that a practice "has always been done this way", which is at the heart of the respondent's argument relying on the absence of other classification schemes. If this were so, complacency and a history of discrimination would be rewarded at the cost of progress and the recognition of higher societal norms of behaviour.

Besides, I do not share my colleague's view, at p. 000, that Zurich was not in a position to adopt a non‑discriminatory classification system unilaterally, since it is "a single company in a highly competitive and complex industry". This argument could be made in any case where there are a number of industries in the same predicament and would totally defeat the purpose of the Code. It can very well be assumed that the whole industry would have to conform to the decision under the threat of many similar complaints of discrimination by other young, single male drivers if Zurich were found to be in breach of the Code.

In any event, the evidence is clear that an alternative to the discriminatory classification system existed at the time of the complaint. Drivers over the age of 25 were not, and are not now, classified according to age, sex or marital status. Rather, the premiums for these drivers are set according to the distances they drive and their accident records, completely non‑discriminatory classifications. As the Board of Inquiry pointed out at p. D/2965:

Age, sex and marital status are used for rate classification purposes only for those drivers under‑25. Where did this distinction between those over and under the age of 25 originate? According to the evidence, there is no statistical or actuarial basis for choosing the age of 25 as the "magic age". This distinction appears to be based on little more than historical accident. Apparently, the age of maturity for insurance purposes could just as easily have been established at 23, 27, or 30. . . . Any of these ages would likely reveal a similar differential in terms of claims experience, to that revealed currently between the under‑25 and over‑25 group.

The Board concluded at p. D/2966:

There is, in fact, an alternative to discriminatory rate classification factors which is being used today by the automobile insurance industry. For those insured persons who are 25 or older, the industry relies solely on non‑discriminatory, causally connected factors such as driving record, driving experience and vehicle use to establish rate differentials. The evidence indicates that this system has worked well. There has been no evidence offered to suggest that these factors cannot be used for insured persons who are under‑25. [Emphasis added.]

While in the case of Pennsylvania National Organization for Women v. Commonwealth of Pennsylvania Insurance Department, 551 A.2d 1162 (Pa. 1988), a case cited by my colleague Sopinka, it may well be said that "there is no correlation between mileage and insurance costs", other American data demonstrate that distance is a more accurate predictor of accident rates than gender. Ryan, in a note entitled "The Elimination of Gender Discrimination in Insurance Pricing: Does Automobile Insurance Rate Without Sex?" (1986), 61 Notre Dame L. Rev. 748, confirms at p. 763 that mileage as a rating factor would be as accurate as discriminatory factors:

State Department of Insurance Studies in Massachusetts, New Jersey, and Florida have come to similar conclusions. In addition, driving statistics reveal that the differences in actual miles driven between males and females explains the difference in accident rates. Male accident rates remain higher than those of female drivers because men, as a group, drive twice as many miles as women. When accident figures are calculated on a per‑mile basis, the rates are comparable.

As Ryan points out at p. 765, classification by mileage is a particularly desirable scheme because it emphasizes individual, rather than group, characteristics:

One of the strongest arguments against sex as a valid rating criteria is its failure to account for the wide individual variances within a class. Indeed, the backbone of the unisex insurance movement is the quest for individual treatment rather than broad brushed group classifications. While the very concept of risk distribution prevents any factor from achieving perfect homogeneity with a group, the use of mileage as a rating variable moves closer to this goal through its emphasis of individual driving characteristics instead of group stereotypes. [Emphasis in original.]

Hence, in addition to mileage's being a more accurate predictor, and one causally connected to enhanced risk, mileage as a rating variable allows insurers to identify less over‑inclusive groups, and thus penalize fewer individuals who would suffer from being included in a group composed of drivers with a higher average risk than that individual. Given that individual testing is not possible in the context of insurance, the existence of an alternative rating variable which emphasizes individual characteristics rather than stereotypes associated with large groups presents a strong argument against the justifiability of the discriminatory classification used by Zurich under s. 21 of the Code.

In conclusion, since an alternative to the existing discriminatory classification scheme did exist, and the respondent has not, as the Board of Inquiry pointed out, presented any evidence that the classification system currently in place for drivers over the age of 25 cannot be used for insured persons who are under the age of 25, I must conclude that the respondent has also failed to establish that the discriminatory system was a reasonable basis for classification of drivers.

The respondent made extensive submissions on the necessity for what it calls an "orderly transition of values" and warns of impending chaos if this appeal is allowed. In my view, the Ontario Human Rights Commission is correct that these matters address the issue of remedy, which is not before this Court.

Disposition

I would allow the appeal of the Ontario Human Rights Commission with costs, reinstate the order of the Board of Inquiry and refer the matter back to the Board of Inquiry to determine the appropriate remedy.

The following are the reasons delivered by

//McLachlin J.//

Dissenting Reasons of McLachlin J.

McLachlin J. (dissenting) -- I have had the advantage of reading the reasons of my colleagues Justice L'Heureux-Dubé and Justice Sopinka. I agree with Sopinka J.'s analysis of legal principles applicable in this case. I part company from him on the application of that test to the facts of this case, and in particular whether the insurer has demonstrated that there was no practical alternative in 1983 to basing its premiums for young men on age, sex and marital status. I agree with the conclusion of L'Heureux-Dubé J. that the insurer has not demonstrated this.

The rule is clear. The law, as Sopinka J. accepts, forbids discrimination on the grounds of age, sex and marital status, unless the offender can establish -- and the onus lies on the offender -- that its intentions were bona fide and that, on an objective test, there was no practical or reasonable alternative to distinguishing, i.e., discriminating, on those grounds. The critical question raised by this case is what constitutes a "practical alternative". The law does not forbid an insurance company from using age, sex or marital status as a basis for its premiums. But it says that if the insurer uses them, it must be able to show that there was no other reasonable or practical basis for determining the rates.

Zurich says there is no practical alternative to using age, sex and marital status as the basis of its premiums in 1983 because it did not have in its possession in 1983 statistics based on non-discriminatory grounds. The reason it did not have those statistics is that it (and other insurers in the business) had never collected data on non-discriminatory bases. This failure, in turn, is blamed on the failure of the Ontario Superintendent for Insurance (now Commissioner) to require insurers to collect non-discriminatory data. Sopinka J., in the end, accepts that the insurer's own default in these circumstances establishes that there was no "practical alternative" to discriminating on the basis of age, sex and marital status. More precisely, there was no alternative classification which would have achieved effective separation of risk between categories.

This brings us to the heart of the case. Can a person who uses age, sex and marital status as a basis of discriminating against a group meet the requirement of showing there was no other reasonable or practical alternative by saying it possesses no statistics on other legitimate bases? Initially I was inclined to say yes. After all, insurance is a statistic-driven industry, and without statistics, rates cannot be set. If there are no statistics other than those based on age, sex and marital status, then there is no practical alternative to using those statistics.

Upon reflection, however, I have concluded that the foregoing argument misses the mark. An absence of statistics based on grounds other than age, sex and marital status does not establish that there are no risk factors other than age, sex and marital status which would achieve the objective of charging premiums that are commensurate with risk. All it establishes is that we do not know whether viable alternative bases of risk evaluation exist. Under the Human Rights Code, 1981, S.O. 1981, c. 53, the insurer bears the burden of showing no reasonable alternative exists to calculating rates on the basis of age, sex and marital status. An absence of statistics does not establish that there is no other alternative. It establishes merely that the insurer does not have the means to establish that no other alternatives exist. As a matter of logic, the absence of alternative statistics is equally consistent with the inference that alternatives exist as with the inference that alternatives do not exist. In short, the absence of alternative statistics establishes that Zurich cannot demonstrate that no practical alternative existed in 1983; it does not establish that no practical alternative in fact existed. We must not confuse the absence of evidence about alternatives with the absence of alternatives. The bottom line in this case is that the insurer bears the burden of showing no reasonable alternative exists, and, through its own failure to collect the required data, it has failed to meet this burden. The ineluctability of this reasoning is demonstrated by the practical fact that Zurich could not say, even when the appeal was argued before us, whether it would be reasonably practical to calculate the rates on criteria other than age, sex and marital status; the best evidence it could adduce, as Sopinka J. notes, was that it was a matter of some difficulty. Its defence was simply that it did not know and still does not know if practical alternatives exist to rating on the basis of age, sex and marital status. Zurich cannot discharge the onus on it now; it certainly could not in 1983.

The effect of confusing absence of a reasonable alternative with absence of proof of a reasonable alternative is to remove the burden of proof from the shoulders of the person who prima facie violates the Code and to place it on the shoulders of the person who complains of discrimination. All the offender need do is show that it cannot produce evidence as to the existence or non-existence of a reasonable alternative. Unless the complainant can then prove that a reasonable alternative exists, an almost impossible task for the lay person pitted in a rate dispute against the insurer, the complainant will fail. This is contrary to human rights tradition and contrary to the rule of evidence that the burden of proof should lie on the person most likely to be in possession of the relevant facts.

On the plane of policy as well as logic, I am persuaded that there is danger in permitting a person who is prima facie in breach of the Code to say there is no practical alternative on the basis that there are no statistics to demonstrate a practical alternative. Such an approach would encourage prima facie offenders to continue in their traditional discriminatory ways rather than reforming their practices in accordance with the object of the Code. For example, if no violation is found in this case, there is little inducement to explore alternative, non-pejorative bases of rate-fixing. The industry is in effect encouraged to continue using the proscribed criteria for making its decisions. In response, it might be suggested that at some point the continued use of age, sex and marital status in setting premiums would become a violation. But this places the power to "create" a violation in the hands of the tribunals and courts. It says, in effect, we will give you a little more time not to be guilty. After a while, we will find you guilty. Such uncertainty is undesirable, both from the point of view of the potential offender, who needs to know in simple and clear terms what conduct is tolerated and what conduct is wrong, and from the point of view of the public interest in deterring practices which discriminate on the basis of inappropriate stereotypes. In the end, it seems to me that the law should not allow difficulties of proving the non-existence of reasonable alternatives to stand as a defence to a charge under the Code. To the extent that it may take time for industry to convert from ingrained discriminatory practices to non-discriminatory alternatives (or alternatively to show that none is possible), those difficulties are better dealt with under the rubric of remedies than by saying there is no violation of the Code.

The undesirable effects of deciding the contrary may extend to other areas of human rights law. Let us consider the effect of taking a similar course with respect to discrimination in the workplace. The law in force in all jurisdictions of Canada says one cannot discriminate against a person in the workplace on the ground of age, sex, religion etc., unless it is done bona fide and the discriminator shows there is no practical alternative, the same test as is applicable in the type of discrimination here at issue. This, in practical terms, requires the employer to make "reasonable accommodation". If the court below is correct, the employer by analogy could arguably escape a finding of contravention of the Code by showing that there is no basis for concluding that it could carry on its business as it wishes if it accommodated the employee. For example, an employer might argue that in the absence of data showing that it could carry on its business effectively if it employed handicapped persons, refusal to employ such persons does not violate the Code. In response, it might be argued that this case is distinct, given that insurance is a statistic-driven industry. But businesses of all types rely on experience and statistics to make their decisions. If one industry is to be exempted from compliance with the Code on the basis of the absence of statistics or experience supporting the viability of an alternative non-pejorative way of doing business, why not others? Put another way, if we relieve the prima facie offender of the burden of establishing the absence of practical alternatives on the basis of lack of statistics or experience in this case, we may open the door to the burden shifting in other areas of human rights law.

If the concerns I have identified are significant, as I believe them to be, then the only question is whether a person who makes a distinction on grounds of age, sex and marital status, can be excused from the usual requirement of showing that there was no reasonable alternative open to it. Zurich argues it should be so excused. Its argument may be summarized as follows. There may, Zurich concedes, be grounds upon which premiums for these groups can be practically calculated other than age, sex and marital status. We cannot tell you, Zurich says, whether there are alternative grounds because we have confined our statistic gathering to the grounds of age, sex and marital status. We have not bothered to get statistics on other grounds for a number of reasons. The first and most important is that the Superintendent of Insurance has not, until recently, required us to do so. Additional difficulties arise from the fact that we are too small to collect meaningful statistics on our own and so are compelled to use industry-wide statistics which are based on age, sex and marital status and from the difficulty of devising alternative bases of calculating premiums.

In my view, Zurich cannot be excused for its failure to demonstrate reasonable alternatives on any of these grounds. First, I note that there is no precedent in human rights cases for excusing a person who is guilty of prima facie discrimination on any ground. The prima facie offender is under a duty to demonstrate the absence of reasonable alternatives. If it fails to do so, then it violates the Code. If excuses are relevant anywhere, it is under the rubric of penalty.

Second, assuming excuses for failure to demonstrate the absence of a "reasonable alternative" could function as a defence to a charge of violating the Code, the excuses adduced in this case would not suffice. The excuse which bulks the largest is the failure of the Superintendent of Insurance until recently to require insurance companies to collect statistics on bases other than age, sex and marital status. This argument amounts to saying, "We did what the state required; therefore we cannot be found in violation of the Code". The problem with the argument is that it assumes that the only responsibility of an insurance company doing business in Ontario is to comply with the requirements of the Superintendent of Insurance. This is not so. The corporation is obliged, as a condition of doing business in the province, to conform to all the laws of the province, including the Code. It must accept the responsibility of complying with the human rights law in force in the province, or face the alternative -- censure under the Code.

Only if the Superintendent could be regarded as having the power to decide whether a rate structure was reasonable under the Code, could it be argued that compliance with his requirements satisfies the Code. Sopinka J. seems to suggest at p. 000 that this is the case when he says: "The Commissioner must refuse to approve any class of risk exposure that is not just or reasonable or does not reasonably predict risk or distinguish fairly between classes of risk exposure.... Despite such wide-ranging powers, the Commissioner has not prohibited auto insurance rates based on age, sex or marital status."

With respect, I think it is fallacious to accept that satisfying the Superintendent amounts to satisfying the requirements of the Code. The Superintendent does not have the power to make determinations under the Code, and for good reason. The primary responsibility of the Superintendent is to insure the orderly administration of the insurance industry. This goal may sometimes conflict with insuring that human rights are respected, which is why the latter task is entrusted to a separate agency. The representative of the Superintendent testified that the Superintendent's concern was to work with the industry in developing insurance policy. This tendency of regulators of an industry to see the world through the eyes of that industry, sometimes referred to as "industry capture", is understandable and in some respects may be beneficial. But it provides good reason for the legislature's placing the determination of human rights violations not in the hands of industry regulators, but in the hands of a separate body. In effect, the reasons of Sopinka J. delegate to the Superintendent of Insurance the task of determining the existence of "a reasonable alternative" which the law assigns to the Human Rights tribunal. This comes perilously close to saying that if a government regulator permits you to discriminate, you are entitled to discriminate. We know this is not the case, or there would be no need for the prohibitions on state discrimination found in s. 15 of the Canadian Charter of Rights and Freedoms.

A second excuse is that, in the words of Sopinka J. at p. 000, it was "impractical, if not impossible", for Zurich to generate actuarially reliable figures on its own data, and that "Zurich required a wider pool of information in order to set its premiums in any meaningful way." I note initially that the conclusion that it would have been "impractical, if not impossible" for Zurich to collect the necessary information on a non-discriminatory basis seems far from certain. The Board below (1985), 1985 CanLII 5245 (ON HRT), 6 C.H.R.R. D/2948, made no such finding and, with respect to the industry as a whole, it commented at p. D/2964 that Zurich failed to demonstrate "that it would be impractical or impossible to operate the automobile insurance system with the use of non-discriminatory criteria for rate classification".

But even if it were established that Zurich was too small to collect useful statistics on its own, the argument that Zurich was too small to fix its rates on non-discriminatory statistics should not, in my opinion, be accepted as a defence to a change under the Code. The argument would create two standards of human rights -- one for small businesses, another for large businesses. I cannot accept such a conclusion. If Zurich chooses to do any amount of business in Ontario, large or small, it must be prepared to adhere to the laws of the province, including the Code.

A variant on this argument is the "industry standard" argument. It suggests that in the absence of non-discriminatory practices in the industry as a whole, small players are entitled to discriminate. But the failure of others to adhere to human rights standards has never provided a defence to charges under the Code. Each person is responsible for his or her own conduct, and cannot rely on what the competition does or does not do. An employer is not excused from failing to accommodate handicapped people or women reasonably because its competitors do not do so. No more should an insurance company be excused for setting rates based on discriminatory criteria merely because its competitors do the same. It is a fact of life that discriminatory practices have often in the past been so widespread as to be perceived as reasonable and justifiable. The purpose of human rights legislation is to require the prima facie offender to provide actual justification for what may in the past have been assumed to be justified. To accept less in this case would constitute an about-face' in human rights law, an about-face' moreover which possesses the potential to undermine seriously the effectiveness of the protection which human rights codes afford to individuals in a broad range of cases.

A third excuse offered for the failure to produce evidence as to the non-existence of reasonable alternatives is that the development of alternatives is a difficult matter. Thus Sopinka J. notes the difficulties have been encountered in subsequent years in developing non-discriminatory criteria for fixing premiums for young drivers (at p. 000). I am prepared to concede that the development of alternative bases of setting premiums may be difficult and may take time. I am even prepared to concede that in the end it may prove impossible, with the result that insurers will be found entitled to continue to calculate their premiums for this age group on the basis of age, sex and marital status. But such contentions miss the mark. The law requires the insurer to prove there was no other practical alternative to calculating premiums on the basis of age, sex or marital status. The insurer cannot discharge that burden merely by showing that it is difficult to ascertain whether there are reasonable alternatives and if so, what they are. We come once again to the confusion I alluded to at the outset between the absence of reasonable alternatives and the absence of proof of reasonable alternatives. Difficulty alone has never been accepted as an excuse for discriminatory conduct contrary to human rights legislation. As the Board noted below, insurers who find it difficult to justify their discrimination have the option of ceasing to discriminate, that is, applying the non-discriminatory criteria generally applicable to drivers to male drivers under 25.

Finally, it is suggested, at p. 000, by way of exoneration that the consequence of finding that Zurich has failed to establish a practical alternative is "the effect of simply submerging high risk drivers into pools of lower risk drivers". This argument, if it is to assist, must be taken to suggest that there will be no violation of the Code, if the result would be to penalize persons who are not complaining of the discrimination. I note firstly that there is no precedent in human rights law for accepting such an argument. I note secondly that the Code, in requiring the premiums be set in accordance with its criteria, necessarily contemplates that the striking down of an offending classification might adversely affect members of another classification; this follows from the basis of insurance in the concept of loss-pooling. In other words, the Code requires the classifications on which premiums are set to be rationally defensible. If they are not, they must be found to violate, even if other classes are negatively affected. Any adverse effect on drivers in other categories arises from Zurich's failure to shoulder the burden which the law imposes of conforming to the Code or demonstrating the absence of reasonable alternatives. While it is unfortunate that third parties may be adversely affected, the legislature has deemed it more unfortunate that those carrying on business in Ontario should be allowed to evade their responsibilities under the Code. It is not open to the courts, even if they wished, to decline to enforce those responsibilities on the ground that third parties may be adversely affected.

I conclude that Zurich failed to discharge the burden upon it of showing that in 1983 it had no reasonable alternative but to calculate its premiums for under-25 drivers on the basis of age, sex and marital status. As the Board found, it remains open on the record to conclude that such alternatives might have existed, although Zurich, for want of statistics, was unable to demonstrate whether they did or did not. In these circumstances, the principles of human rights law require that Zurich be found to be in violation of the Code.

That said, I agree with Sopinka J. at p. 000 that "[t]he insurance industry must be allowed time to determine whether it can restructure its classification system in a manner that will eliminate discrimination based on enumerated group characteristics and still reflect the disparate risks of different classes of drivers". Such concerns should be reflected in the remedy granted by the Board. Its remedial powers are broad and contemplate situations such as this where time may be required to effect compliance with the Act. It may:

-- (1) ...

(a) direct the [offending] party to do anything that, in the opinion of the board, the party ought to do to achieve compliance with this Act, both in respect of the complaint and in respect of future practices....

(Now Human Rights Code, R.S.O. 1990, c. H.19, s. 41(1)(a).)

I would remit the matter to the Board for further consideration of what remedies are appropriate having regard to the situation in 1983 and developments since then. Without tying the Board's hands, I would suggest that it might wish in cases such as this to consider setting a time during which the offender may collect data relevant to the existence of reasonable alternatives. For this period, no penalty would be imposed on account of the violation; after its expiry, the Board could determine the appropriate penalty or other remedy. It must be remembered that the object of the Code is remedial; it is not directed to imposing penalties so much as bringing business practices into conformity with its precepts.

Like L'Heureux-Dubé J. I would allow the appeal with costs, reinstate the order of the Board of Inquiry and refer the matter back to the Board to determine the appropriate remedy.

Appeal dismissed, L'Heureux‑Dubé and McLachlin JJ. dissenting.

Solicitor for the appellant: The Ministry of the Attorney General, Toronto.

Solicitors for the respondent: Blake, Cassels & Graydon, Toronto.

Solicitor for the intervener Commission des droits de la personne du Québec: Commission des droits de la personne du Québec, Montreal.

Solicitor for the intervener Alberta Human Rights Commission: Alberta Human Rights Commission, Edmonton.

Judgment of the Court

The Facts

Judgments

Board of Inquiry

Divisional Court

Court of Appeal

The Issue

Standard of Review

Reasonable and Bona Fide Grounds

The Statistical Plan

The Insurance Classes

Application of Principles to the Facts

Disposition

Dissenting Reasons of L'Heureux-Dubé J.

Judgments

Relevant Statutory Provisions

Analysis

General Principles

“Reasonable and bona fide grounds” under s. 21

Application

Disposition

Dissenting Reasons of McLachlin J.

Judgment of the Court

The Facts

Judgments

Board of Inquiry

Divisional Court

Court of Appeal

The Issue

Standard of Review

Reasonable and Bona Fide Grounds

The Statistical Plan

The Insurance Classes

Application of Principles to the Facts

Disposition

Dissenting Reasons of L'Heureux-Dubé J.

Judgments

Relevant Statutory Provisions

Analysis

General Principles

“Reasonable and bona fide grounds” under s. 21

Application

Disposition

Dissenting Reasons of McLachlin J.

Cited By (14 cases)

Security National Insurance Co. v. The Wawanesa Mutual Insurance Company, 2013 ONSC 7589

Saskatchewan (Human Rights Commission) v. Whatcott, 2013 SCC 11, [2013] 1 S.C.R. 467

Power Tax Corporation v. Millar et al., 2013 ONSC 135

New Brunswick (Human Rights Commission) v. Potash Corporation of Saskatchewan Inc., [2008] 2 S.C.R. 604, 2008 SCC 45

Tranchemontagne v. Ontario (Director, Disability Support Program), [2006] 1 S.C.R. 513, 2006 SCC 14

Canada (House of Commons) v. Vaid, [2005] 1 S.C.R. 667, 2005 SCC 30

Barrie Public Utilities v. Canadian Cable Television Assn., 2003 SCC 28

Quebec (Commission des droits de la personne et des droits de la jeunesse) v. Montréal (City); Quebec (Commission des droits de la personne et des droits de la jeunesse) v. Boisbriand (City), 2000 SCC 27

British Columbia (Public Service Employee Relations Commission) v. BCGSEU, 1999 CanLII 652 (SCC), [1999] 3 SCR 3

Battlefords and District Co-operative Ltd. v. Gibbs, 1996 CanLII 187 (SCC), [1996] 3 SCR 566

Large v. Stratford (City), 1995 CanLII 73 (SCC), [1995] 3 SCR 733

Pezim v. British Columbia (Superintendent of Brokers), 1994 CanLII 103 (SCC), [1994] 2 SCR 557

United Brotherhood of Carpenters and Joiners of America, Local 579 v. Bradco Construction Ltd., 1993 CanLII 88 (SCC), [1993] 2 SCR 316

Dickason v. University of Alberta, 1992 CanLII 30 (SCC), [1992] 2 SCR 1103

Cited By (14 cases)

Security National Insurance Co. v. The Wawanesa Mutual Insurance Company, 2013 ONSC 7589

Saskatchewan (Human Rights Commission) v. Whatcott, 2013 SCC 11, [2013] 1 S.C.R. 467

Power Tax Corporation v. Millar et al., 2013 ONSC 135

New Brunswick (Human Rights Commission) v. Potash Corporation of Saskatchewan Inc., [2008] 2 S.C.R. 604, 2008 SCC 45

Tranchemontagne v. Ontario (Director, Disability Support Program), [2006] 1 S.C.R. 513, 2006 SCC 14

Canada (House of Commons) v. Vaid, [2005] 1 S.C.R. 667, 2005 SCC 30

Barrie Public Utilities v. Canadian Cable Television Assn., 2003 SCC 28

Quebec (Commission des droits de la personne et des droits de la jeunesse) v. Montréal (City); Quebec (Commission des droits de la personne et des droits de la jeunesse) v. Boisbriand (City), 2000 SCC 27

British Columbia (Public Service Employee Relations Commission) v. BCGSEU, 1999 CanLII 652 (SCC), [1999] 3 SCR 3

Battlefords and District Co-operative Ltd. v. Gibbs, 1996 CanLII 187 (SCC), [1996] 3 SCR 566

Large v. Stratford (City), 1995 CanLII 73 (SCC), [1995] 3 SCR 733

Pezim v. British Columbia (Superintendent of Brokers), 1994 CanLII 103 (SCC), [1994] 2 SCR 557

United Brotherhood of Carpenters and Joiners of America, Local 579 v. Bradco Construction Ltd., 1993 CanLII 88 (SCC), [1993] 2 SCR 316

Dickason v. University of Alberta, 1992 CanLII 30 (SCC), [1992] 2 SCR 1103