On May 20, 2014, the Consumer Financial Protection Bureau (CFPB) issued a data point report that focused on whether consumers with medical debts were overly penalized for medical debts that go into collection and show up on their credit reports. The CFPB confronted two questions in the report: (1) “whether medical and non-medical collection are equally predictive about the subsequent respective credit performance of consumers with these different types of accounts” and (2) “whether paid and unpaid medical collections are equally predictive of consumer delinquency rates.” In assessing these questions, the CFPB reviewed about five million anonymized credit records. The answer to both questions—according to the CFPB—is no.
With regard to the first question, the CFPB accumulated data on consumers with mostly-medical debt (MM consumers) and consumers with mostly non-medical debt (MNM consumers); the result: MNM consumers had (on average) a higher estimated delinquency rate than MM consumers. The median MM consumer had a delinquency rate that corresponded more closely with a credit score of 8-10 points above the consumer’s actual score. On the other hand, the median MNM consumer had a delinquency rate that corresponded with a credit score between 9 points below and 1 point above the consumer’s actual score.
Turning to the second question, the CFPB reached a similar result. The median mostly paid medical consumer (MPM consumer) had an estimated delinquency rate corresponding to a credit score of between 16 and 22 above the consumer’s actual score. The median mostly unpaid consumer (MUM consumer) was close to properly scored.
It is not clear from the study what, if anything, the CFPB intends to do in response. Based on statements from CFPB Director Richard Cordray, the CFPB will likely seek to have credit scoring models differentiate between medical and non-medical debt in the future, as well as differentiate between paid medical debts versus unpaid medical debts. Unfortunately, the study fails to define “medical debt,” and thus, we are left—at least to some extent—to guess what such a differentiation would entail.