Many insurers use credit scores to help underwrite personal lines policies such as automobile and homeowner’s insurance. Although three states prohibit use of credit scores, most do not. A bill pending in Congress will end the practice, however, if the Federal Trade Commission finds that use of this sort of “consumer information” is a substitute for race or ethnicity even if the effect is small. In this article, Professor Seth J. Chandler analyzes the bill and examines the evidence to date on the relationship between race./ethnicity and credit score.

Professor Chandler notes that although the serious work on the subject is thinner than one would imagine, the evidence to date coupled with the definition of “proxy effect” contained in the bill, makes it likely that the bill would prohibit use of credit scores. He writes: “Although the published and peer reviewed evidence on this point is a bit thinner and bit more equivocal than one might believe precisely because of the understandable sensitivity about collection of race-correlated credit scoring data, most of the better work that has been done on this point suggests that use of credit scores will, on balance, hurt African Americans and Hispanics.”

The professor also notes a tricky legal issue embedded in the bill. “It appears that the intent of the bill is that, if there is some statistically significant effect that results from the use of credit scores on the prices paid for personal lines insurance by the “average” member of some racial or ethnic group, the use of this sort of consumer information will be prohibited. Although it is possible that an agency or court unimpressed with the ban would require something more than correlation, there is little in the statutory language that would justify a tougher standard. The difficult case will come if the evidence shows that, yes, credit score and race are correlated but not after income and other correlated variables are taken into account. Will it then be the case that credit score is a “substitute” or “stand-in” for race? Statistical concepts are slippery beasts.”

The article further notes the tension between H.R. 5633 and the traditions of American insurance regulation embodied in the McCarran-Ferguson Act, the ability of the Federal Trade Commission to trigger the band without traditional administrative procedures and the difficulties of starting down a slope under which a statistical correlation with race/ethnicity results in a prohibition on the use of an underwriting criterion. 

Access the complete commentary on lexis.com