"No harm in credit scoring if insurers apply it to all
Disparities among groups don't equal discrimination"

The following is an excerpt from a January 19, 2005 Houston Chronicle as an Op-Ed piece

A new study by the Texas Department of Insurance confirms what insurance companies have known for years: Individuals with poor credit scores tend to file significantly more auto and homeowners insurance claims than those with good credit scores. A report last year by the University of Texas' Bureau of Business Research reached the same conclusion.

Unlike the UT study, the TDI study also analyzed demographic data to determine whether a relationship exists between credit scores and race, ethnicity or income. The TDI found that whites, Asians and high-income policyholders tend to have better credit scores than black, Hispanic and low- to moderate-income policyholders. Members of the latter groups were found to be statistically overrepresented in the worse credit score categories and underrepresented in the better credit score categories.

Insurers use credit information to assess risk more accurately and to price insurance coverage accordingly. Because the studies leave no doubt that a person's credit score is a valid predictor of his propensity for filing claims, the demographic characteristics of various risk classes is irrelevant from an insurance standpoint. Yet some insurance industry critics have chosen to ignore the study's primary conclusion, seizing instead on the credit score disparities among demographic groups.

Last week, for example, Sen. Rodney Ellis of Houston and three other state lawmakers announced plans to introduce a bill to ban credit-based insurance scoring, claiming the practice discriminates against minorities and the poor. "Our skin color should not make us a higher risk," declared LULAC spokeswoman Ana Correra, who supports the proposed ban.

The critics would have a point if insurers subjected only minority or low-income consumers to credit scoring, or if they applied more stringent credit standards to minorities and the poor. In that case, insurers would indeed be guilty of unfair discrimination, because they would be treating people differently based on race, ethnicity and income. But credit-based insurance scoring is utterly blind to these factors, and the insurers who use credit information as an underwriting tool apply the same standards to all their customers.

The notion that insurance scoring is unfair to certain minority populations is based entirely on the disparate-impact theory of discrimination, which holds that a business practice is unfairly discriminatory if the percentage of those who are adversely affected by it is larger for some groups than for others - despite the fact that the practice treats each individual equally and was never intended to discriminate on the basis of race or ethnicity.

Applied to insurance underwriting and pricing, disparate-impact analysis simply doesn't make sense. To understand why, consider the hypothetical case of two applicants for insurance:

Dick and Jane have the same, relatively low insurance score, because of their poor credit histories. Their insurer therefore charges each of them the same higher-than-average rate for insurance coverage. The insurer treated Dick and Jane equally - that is, it subjected them to the same risk-assessment standard, with the same result. This is how insurance scoring operates in practice.

As long as Dick and Jane share the same demographic characteristics, there's no problem, according to the disparate-impact theory of discrimination. But suppose Dick and Jane belong to different racial groups. Suppose further that 30 percent of Dick's group have low insurance scores, while only 10 percent of Jane's group have low scores. Disparate-impact analysis would proclaim that under these circumstances, insurance scoring is unfairly discriminatory.

So who, exactly, is being discriminated against? After all, Dick and Jane were treated equally. According to the theory, it is Dick's group that is the object of discrimination - even though Dick himself, along with every other member of his group, was treated no differently than anyone else. Proponents of the disparate-impact theory of discrimination pay no attention to whether individuals are treated in a fair and nondiscriminatory fashion. Instead, they measure fairness by the presence or absence of statistical parity among demographic groups.

Applying this odd definition of discrimination to insurance scoring would actually harm many of the very people - minorities and the poor - whom disparate-impact proponents seek to protect. After all, just as many whites, Asians and upper-income individuals have poor credit scores, it is also true that many blacks, Hispanics and low-income individuals have good credit scores. All of these consumers stand to benefit from insurance scoring in the form of lower rates and more coverage options. In the end, that is the most meaningful conclusion to be drawn from the TDI study.

Detlefsen is the director of public policy at the National Association of Mutual Insurance Companies. Readers may e-mail him at rdetlefsen@namic.org.