Settle the Score

Be it praised or maligned, credit scoring, used as a predictor tool for future loss, is again back in the news.

A recent ruling by the Office of the Privacy Commissioner of Canada (OPC) could return the issue to the front burner, following the cooling effect applied last fall when a Canadian Council of Insurance Regulators (CCIR) working group was unable to come up with concrete guidance.

OPC issued what amounted to a “tsk-tsk” for an unnamed insurer’s failure to emphasize transparency in providing information to policyholders — something it contends should be regarded as the top priority.

Ruling on a 2009 complaint involving an Ontario couple that saw their home insurance premium rise sharply, OPC noted its concern turns on transparency related to how the information was used and how the insurer informed the policyholders, as well as the belief that most consumers are unaware of how their information is being used.

The CCIR working group said it did not have sufficient information to determine the actual and current, rather than potential, “harms that may be accruing to consumers from the use of credit scores by insurers.” Policymakers should have the final say.

Prince Edward Island wasted little time in speaking its mind, releasing a consultation paper and penning draft regulations that would ban insurers from using personal financial information to determine eligibility for insurance coverage. (Credit scoring is off limits in some provinces; for auto premiums, it is banned in Ontario and Quebec.)

PEI moved after receiving complaints that use of personal financial information by insurance companies resulted in some individuals not being offered property or auto insurance, or being offered insurance at unaffordable rates.

With companies increasingly making use of personal information, “that, in turn, may lead to significant availability and affordability issues within the province, with respect to these insurance products.”

While OPC had no quarrel with using credit scoring, others do. “Consumers are not being protected under the current regulations, and we need the Ontario government to take action to ban the use of credit scoring in insurance,” argues Randy Carroll, CEO of the Insurance Brokers Association of Ontario.

But some insurers see gold in the data. They contend that how customers manage their finances is among the most accurate indicators of risk, and there is a strong correlation between credit score and insurance losses.

The Co-operators, for example, uses credit score in addition to more traditional rating factors like claims history, age of home and geographical location to determine risk level and, in turn, appropriate home insurance premium, notes information from the company. “Analysis shows a direct link between a person’s credit score and the frequency and severity of claims. By adopting this factor, we are able to more accurately rate individual clients by customizing rates and charging an appropriate premium for the risk.”

Insurers have (and perhaps should) argue it makes little sense not to use every available tool to come to the most accurate determination of risk. Failing that, overall costs can be higher and consumers who manage their finances responsibly can be penalized.

But are there better, more accurate tools available?

Brian Stoll, a director at Towers Watson, suggested last fall that telematics (providing details such as vehicle speed, distance travelled and braking pattern information that focus on actual driving behaviour) may reduce the need for proxy underwriting factors, such as credit scoring, in the future.

Use of telematics is potentially “a gamer changer from the automobile point of view,” Brigid Murphy, president and CEO of The Dominion, commented in a recent speech. It “eliminates the whole credit score argument on automobile insurance because if you can see the actual risk, then you don’t need to use a reasonable facsimile,” Murphy said.

Although there is no equivalent to telematics in property lines, that may not always be so. And that may open the door to exploring how closely aligned to behaviour these predictor tools need to be.