Do Credit Scoring Models Treat Minorities Unfairly?


Credit scoring models are often criticized for almost everything imaginable under the sun. They treat medical collections too harshly.  They aren’t a valid measurement of consumer credit risk.  They’re used by too many industries to make too many decisions.  Of course none of this is actually true, but it does make for good fodder.

Credit scoring models have also been targeted by advocacy groups suggesting they treat protected classes unfairly relative to the general population. The operative term here is “disparate impact”, and it’s defined as a practice that has a disproportionate adverse effect on those in a protected class, like the elderly or minorities. In short, there are some who suggest that credit scores treat minorities unfairly.

Credit scoring systems are colorblind. They do not take into consideration anything in the consumer identification section of a credit report. And even if they did, there’s nothing on a credit report that identifies the race or national origin of consumers. From a systemic perspective the only reason a credit-scoring model would treat anyone adversely is if they’ve got a risky credit profile. In that case you can make the argument that the consumer deserved the lower credit score because of their poor credit management practices.

The problem is that anyone can allege that credit scores produce a disparate impact and don’t really have to provide any basis to their allegation because the allegation is sexy.  Those who defend the use of scoring systems, however, do have to produce some evidence that their use isn’t unfair to protected classes. In the world of credit scoring this very thing has occurred.  The critics of credit scoring have produced no evidence that disparate impact occurs while many other reputable parties have produced evidence that it does not.

In 2007 the Board of Governors of the Federal Reserve System submitted a 304 page report to Congress concluding that disparate impact does not occur in credit scoring. And if that study wasn’t good enough, the Board of Governors of the Federal Reserve System generated a second report in 2010 saying the same thing, which is, “Our examination yields no evidence of disparate impact by race or ethnicity or gender.”