Auto insurance

The Insurance Bureau of Canada (IBC) affirms that it represents 90 per cent of the auto insurance industry in Ontario. IBC provides a Consumer Information Centre service and has a mandate to coordinate industry views and present them at consultations such as the one conducted by the Commission. Data studies are also carried out to analyze what variables most effectively predict driver risk.

The Commission received a joint submission from IBC and the Association of Canadian Insurers and met with IBC representatives and several of their member organizations in February 2000 to discuss issues raised in IBC’s response submission to the Commission’s Discussion Paper. Other submissions and comments dealing with auto insurance were also provided to the Commission, including valuable input from FSCO.

"Bona Fide" and Reasonable Risk Assessment

Auto insurance premiums are typically set prospectively. This means that certain drivers are charged higher rates because, as a group - classified by many variables but predominately by age, sex and/or marital status - they are statistically shown to be at higher risk for accidents and claims.

At the same time, industry representatives explain that an individual’s driving accident record is used to fine-tune and individualize premium rates as much as possible within the general group. However, the problem is that the frequency rate of at-fault-accidents is not significant enough in order to use individual driving record as a decisive measure for risk classification. What is left are some variables that classify individuals, such as young drivers, not by criteria that have a causal relationship with the purpose of the insurance product, but rather, by personal characteristics identified by enumerated grounds of discrimination under human rights law that correlate strongly with accident risk and higher expected costs across an entire group.

Industry representatives argue that the use of such criteria is bona fide and reasonable and that the removal of any of the variables of sex, age or marital status without suitable alternatives would impede the viable delivery of auto insurance.

The industry believes that it has been diligent in its use of sound and reasonable business practices to gather data on age, sex and marital status and alternative variables. At the same time, it argues it has attempted to balance changes in government policy with ensuring a wide availability of affordable auto insurance products to consumers.

It is the position of IBC that the 1992 Supreme Court of Canada decision in Zurich clearly upholds the defence in Ontario’s Human Rights Code for the industry’s use of age, sex and marital status as bona fide and reasonable business practices for the purpose of setting premiums commensurate with risk.[9] IBC comments that, at the same time, the industry is open to using new assessment methods and data. And though the 1992 Zurich decision is instructive as it encouraged the industry to look for alternatives, the research to date, according to IBC, seems to indicate that there may be no better alternatives.

Looking for Alternatives

There is some history in the Canadian insurance industry of attempting to use alternative risk classification variables.

In the late1980s, the former Ontario Auto Insurance Board was given the responsibility by government to design an auto insurance rate classification plan that would not include use of age, sex and marital status. FSCO reports that the government of the day declined to implement the public policy scheme for several reasons: rate “dislocation” would occur without these variables; higher costs would be passed on to consumers; there would be costs to the industry associated with the research and implementation of a revised system; and, significant information technology resources would be required.

The government of the Yukon Territories tried to eliminate age, sex and marital status from rate setting as well, and abandoned the attempt because it was not successful.

One representative from a large insurer reported that their company also looked at eliminating age, sex and marital status from rate setting. The results showed that though young single males for example would achieve a modest reduction in their premiums, older married males would experience an increase of approximately 60%. This “dislocation”[10] effect would result in higher costs that would not be commensurate with risk for certain groups.

Two jurisdictions that have chosen not to use age, sex or marital status for auto insurance rate setting, are British Columbia and Massachusetts. Both are held by the insurance industry as poor and unsuccessful examples of public policy that would have good drivers substantially and disproportionately subsidize bad drivers.

Since 1974, all vehicle owners in British Columbia have been required to buy a package of basic auto insurance from the Insurance Corporation of British Columbia (ICBC). ICBC's Basic Autoplan coverage ensures that every B.C. motorist carries a minimum amount of liability insurance, as well as insurance to help if they are involved in a motor vehicle accident resulting in their injury or death.

ICBC does not use age, sex, or marital status to measure risk or set premiums. Instead, average claim, repair history, engine size and safety features of each car model are assessed. How the vehicle is used and the territory driven in is also considered. Most significant, ICBC uses a Claim-Rated Scale where all drivers start at a base rate and then cumulative discounts over time are provided to drivers who aren't responsible for motor vehicle crashes. It also raises the cost of insurance for drivers who cause crashes and have at-fault claims. As a result, “years of driving” accident-free becomes a significant risk factor for premium rate setting. Such a provision applies to all drivers equally, regardless of age, but will affect young drivers disproportionately since they account for most of the novice driver population. However, “years of driving” is a variable that appears to have a causal relationship with the purpose of auto insurance. Using bona fide and reasonable risk factors that are rationally connected to their intended purpose, even though an adverse impact might result, is preferable to directly classifying individuals on the basis of enumerated grounds of discrimination.

As a public corporation, ICBC does not turn people away and therefore cannot select risk. ICBC explains that private insurers can compete to offer consumers higher optional coverage for liability and collision but they can also “anti-select” bad risk drivers back over to ICBC for coverage.

IBC points out that the system in British Columbia is a government monopoly where good drivers end up cross subsidizing bad drivers, resulting in young males paying less. It remarks that private insurers in B.C. compete at great disadvantage to the government monopoly, which uses its dominant role to undercut the competitive marketplace. IBC also states that ICBC is not forthcoming in sharing data with private sector competitors. IBC reports it recently tried to measure B.C. variables and concluded that some could not be used in the competitive market place.

In addition, IBC contends that Massachusetts’ non-discriminatory classification scheme has resulted in substantial dislocation and insurance availability problems for a large residual market of certain groups of drivers paying higher premiums. In any case, safe driver plans as used in B.C. and in Massachusetts are also used in other jurisdictions in Canada as well and are part of the mix in premium rate setting.

Another variable in the risk classification mix is the Canadian Loss Experience Rating System (CLEAR), a methodology approved by the former Ontario Insurance Commission in 1994. Most insurance companies introduce CLEAR into their classification systems to assess the loss history of vehicles by make and model for risk and premium rate setting.

Other variables such as years of driving or annual driving distance have also been put forward as a suitable alternative to age. However, actuarial data to date has not shown annual driving distance to be a useful indicator of risk. With this indicator, drivers more or less fall into two categories - those that drive a lot and those that don’t. Two categories do not sufficiently split the population for risk classification. And, although data on annual driving distance is still tracked by some, most jurisdictions in Canada do not differentiate on this basis.

Data Requirements

IBC re-iterated that the industry continues to be interested in looking at different ways of alternative rating. However, this has been difficult because the insurance industry could be characterized as volatile over the last 15 years with four system changes legislated by government. IBC cautioned that it takes several years for the market to adapt to major changes before reliable statistics are available. At least three to five years are needed to capture measurable statistics for a new rating variable.

All insurance companies have to collect and submit data, such as driver accident record data, to the Insurance Information Division (IID) of IBC. [11] Although insurers’ rating plans can vary, data submitted to IID must be uniform. FSCO sets the data plan for Ontario and approves new variables and rating systems. New variables were last added in 1985.

Some data going to IID is “scrubbed”, such as driver vehicle class, because of inaccurate or ineffective company reporting. Other variables, like annual driving distance, even though a potential surrogate for age/sex/marital status, are not part of the data plan and therefore not coded.

FSCO does not permit the industry to use alternative variables such as income, credit history and housing accommodation status (e.g. asking whether you rent or own your own home) because of potential unfairness. Such variables might also correlate highly with the socio-economic status of individuals protected by certain grounds under the Code such as place of origin, citizenship, marital status, same-sex partnership status, gender, etc. IBC contends that FSCO’s stance makes it difficult for the industry to develop alternative risk variables.

IBC Auto Insurance Rate Criteria Study 1995/2000

IBC stated that prior to and since the 1992 Zurich decision, the insurance industry has continuously collected data on alternate variables related to drivers and risk. In 1995, IBC conducted a study in several jurisdictions in Canada, based on 1991-93 data that looked at nine alternate variables for their effectiveness in predicting driver risk:

  • Years licensed, principle operator
  • Gender, principle operator
  • Vehicle use (e.g. pleasure, business)
  • Number of years claims-free
  • Years licensed, other operators
  • Age range, principle operator
  • Driver training, principle operator
  • Number of claims in last 6 years
  • Number of other operators

IBC concluded that while these criteria improved the measurement of risk, they were at least imperfect substitutes for age and sex, which correlate highly and significantly in the accurate determination of accident risk (though no “causal” relationship is demonstrated). Furthermore, elimination of age and sex would disrupt current pricing and result in significant “dislocation”, subsidizing younger male drivers at the expense of higher costs to older female drivers, for example, below the actual risk assumed. In IBC’s view, this would be fundamentally unfair and would contravene the Court’s direction in Zurich to accurately “reflect the disparate risks of different classes of drivers.”

In June 2000, IBC updated its 1995 study of alternative risk variables following the Commission’s Discussion Paper, Human Rights in Insurance, which urged the insurance industry to find alternatives. The updated study reaffirms IBC’s view that the use of age, sex and marital status as auto insurance rating criteria continue to be justified under the Code.

IBC points out that there are other variables available to the insurance industry. These include:

  • Statistical Territory (except for the split to Urban/Rural)
  • Annual Distance
  • Commute Frequency and/or Distance
  • “Driving Record” as traditionally used
  • Conviction History
  • Third Party Liability Limit
  • Multiple Vehicles
  • “Rate Group” through Vehicle Code

IBC notes that these variables are not used consistently across the insurance industry and therefore actuarial analysis could not be undertaken to prove or disprove their appropriateness as rating variables. As a result, the industry has continued to rely on age, sex and marital status as the most reliable variables to assess risk. At the same time, IBC states that it is committed to improving rating practices as new data and techniques become available.

Innovation and Moving Forward

Industry representatives stated that innovation in risk classification comes about through a competitive market. Competition means companies have an incentive to find data that accurately measures risk. And, competition keeps costs lower for consumers. At the same time, it is acknowledged that risk classification and innovation are expenses for companies.

The industry explains that there are no patents on rating variables and most companies use similar ones including age, sex and marital status. Companies have to do the research and produce the data themselves. Any other company can mimic the rating variables used. Yet at the same time, the industry acknowledges that there is a lot of variance between companies in how variables are used. However, one uniform practice is the industry-wide use of a rate differential at age 25.

Short of a mandated change in government social policy, to the extent of British Columbia or Massachusetts for example, the industry concludes that in order to ensure the availability of viable and affordable auto insurance products in a competitive private sector market, the use of age, sex and marital status is still necessary.

The last major change in government social policy was as a result of Bill 59, the Automobile Insurance Rate Stability Act [12] , which went into effect on November 1, 1996. While data for 1999 is now available, industry experts note that a few years must pass after a change to the system before reliable and meaningful data can be assessed.

In 1998, the Ministry of Transportation released an interim report[13] evaluating driver accident data under the Graduated Licensing System, which was implemented in 1995. The study reveals that accidents did decline as a result of the new graduated program. The data also confirms that males continue to be involved in proportionally more collisions than females, and that collision rates continue to decrease with age. However, graduated licensing resulted in greater age-based improvements for younger age groups for both males and females. In fact, the collision rate for novice women drivers mostly levels off at age 20 and beyond. In addition, collision and fatal collision rates fell dramatically for 16 year olds after the implementation of graduated licensing to levels comparable to that of the general driving population. The Ministry will be completing a full evaluation, with a report expected later in 2002. Data on the effectiveness of graduated licensing preventing driving accidents should be of assistance to IBC in any future study of the use of risk classification variables.

IBC concludes that the variables currently in use are those that most accurately measure risk. At the same time, the industry believes that it has been a leader in the past in looking for alternative risk classification variables and will continue to repeat its studies in this area.

[9] A 1993 decision by the Alberta Court of Appeal involving a similar case ruled in accordance with the Supreme Court in Zurich [Co-Operators General Insurance Co. v. Alberta (Human Rights Commission),[1993] A.J. No. 828, DRS 95-02920, Appeal No.9103-0466-AC (Alta. C.A.)]
[10] Rate dislocation means that higher costs are not proportionate with identified risk for certain groups
[11]The Insurance Information Centre of Canada merged with the Insurance Bureau of Canada as of July 1, 2001 and will henceforth be known as the Insurance Information Division (IID) of IBC.
[12] Automobile Insurance Rate Stability Act 1996 – requires insurers to offer “retiree” discounts; restores the right to sue for economic loss while preserving basic no-fault accident benefits; provides tools to control fraud and overcompensation; anticipated that drivers with good records and no claims will experience rate reductions.
[13]Graduated Licensing System Evaluation: Interim Report 1998, Ministry of Transportation of Ontario.