NAII Says Industry Critic’s Report on Credits Scores Misleading

April 1, 2003

A report released this month regarding Ohio insurers’ use of credit-based insurance scores incorrectly asserts that such scores intentionally and unfairly create a disproportionate impact on minority and low-income groups.

“Insurance scores cannot be used as a proxy to determine a person’s ethnicity or income,” said Diana Lee, senior research consultant for the National Association of Independent Insurers (NAII). “The report’s theory of disparate impact is flawed, and has no basis for evaluating the effectiveness of objective underwriting factors such as insurance scores.”

Industry critic Birny Birnbaum, who authored the report to the Ohio Civil Rights Commission, alleges that Ohio insurers’ use of insurance scoring in underwriting and rating homeowners policies “very likely” has a disparate impact on poor and minority populations in Ohio. Birnbaum states that the use of insurance scores makes insurance less available and more expensive for ethnic and low-income consumers in Ohio.

“Insurance companies do not intentionally and unfairly discriminate against any particular group based on race or income,” Lee said. “The premise behind insurance is to make distinctions among different groups based on their loss likelihood. These distinctions are based on statistical loss costs, and are not made randomly or unfairly.

“Insurance companies do not know whether the use of credit history disadvantages any specific group of consumers since insurers don’t collect information based on these demographic traits. What the insurance industry does know is that the use of credit has benefited most policyholders.”

An insurance score measures risk based on an individual’s credit history and does not include any data on an individual’s race, income, gender, religion or other demographic information. A Company only sees the computer-generated score—specific information contained in an individual’s credit report remains private.

Contrary to allegations in the report, at least four independent studies have demonstrated a connection between consumer credit information and risk of loss, according to the NAII. Most recently, a University of Texas study shows that “using logistic and multiple regression analyses…the credit score for the named insured on a policy was significantly related to incurred losses for that policy.” Also, the study revealed analysis that insurance scores yield new information about incurred losses not contained in existing underwriting variables.

Birnbaum also asserts that a credit score is the last factor applied to the premium in the rate development, and that the scores are correlated to the profitability of the consumer.

“This assertion that insurance scoring is the only factor that determines the rating tier is incorrect, as other factors—such as motor vehicle records—can also determine under which rating tier an individual will be placed,” Lee said. “Also, a consumer’s income level is not revealed to an insurer. Credit problems are not unique to only the lower income groups; rather, they persist among all groups. Individuals in the higher income groups are also treated the same if their credit records are equivalent to those in the lower income groups.”

Even though different credit scoring models are used by different insurance companies, the use of insurance scores is still valid. Policyholders insured by the same company or one group are treated the same in regards to tier placement and rates. Individual companies or a group of companies use the same procedure and formula to determine a score and rate.

Also, Birnbaum states that an increase in bankruptcies should result in an increase in the private passenger automobile insurance loss ratios. He demonstrates that this is not the case and attempts to cast doubt that a correlation between credit and loss likelihood exists. However, decline in loss ratio is a positive consequence as it shows that credit information has helped companies to more accurately determine the appropriate amount of premium in light of existing losses, Lee said. This reduction demonstrates that loss ratios are now closer to their anticipated level, Lee added.

“Birnbaum’s use of loss ratio in this report is flawed,” Lee said. “Rather than comparing bankruptcies to loss ratios, they should be compared to loss costs because loss costs actually represent the average paid losses per insured vehicle.

“When loss cost data are evaluated it will be seen that personal auto loss costs have been steadily increasing—up 73 percent from 1985 to 1999. Hence, a positive correlation between losses and bankruptcies most certainly does exist,” Lee added.

Birnbaum also attacks the industry for asserting that it does not know why poor insurance scores cause higher claims.

“Like other rating factors, no one fully understands ‘causality,'” Lee said. “The argument provided by insurers that people who can handle their financial assets responsibly are more likely to handle other aspects of their lives responsibly is just as sound as claiming that young males are inherently more reckless than young females, or people who live in urban areas are more apt to have their cars stolen. A correlation exists, not necessarily a cause-and-effect relationship, as statistics do show that a group of policyholders with certain traits are more likely to make an insurance claim.”

Lee also notes that the main point to consider with insurers’ use of credit scores is that people with bad, indeterminate and good credit exist within each ethnic group and income level.

“Individuals with bad or indeterminate credit, regardless of the ethnic and income group in which they are classified, experience problems obtaining insurance or pay more for insurance,” she said. “The distinctions insurers make between good and poor credit are fair. All individuals, no matter their race or income level, are treated the same.”

Was this article valuable?

Here are more articles you may enjoy.