The three major credit bureaus announced earlier this year that they will be dropping most medical debt from credit reports. Both paid and unpaid collections under $500 will be dropped in the coming year, and the reporting requirements around such debt will be made more favorable to consumers (i.e., adding a one-year hold period before furnishing such data). At first glance this looks like a clear win-win; we know there is a high degree of inaccuracy in medical debt reporting that harms consumers (through decreased access and increased cost of credit), lenders (who make less money by being unnecessarily restrictive in underwriting), and bureaus (whose product is accurate data). And we know that most collections (by count) are for small-dollar medical debt, making the reach of the problem significant (43 million individuals). What’s not to like?

Unfortunately the campaign driven by the CFPB (Consumer Financial Protection Bureau) to change medical debt reporting will do the very opposite of what was intended. It’s likely to decrease accuracy and increase the cost of credit for non-prime U.S. consumers (generally the middle class, the young, and the working poor). By reading the primary sources published by the CFPB and their claims, we can better understand how and why.

The most recent study of medical debt by the CFPB states the challenges it sees with current reporting practices and identifies disparate impact and diminished predictive value as the primary consumer harms they wish to see rectified by industry. Solving for the former can be a philosophical question of equality of process vs. equality of outcome, while solving for the latter is very much in the interest of all parties (consumers, lenders, and bureaus).

We focus our attention on the latter, the source of which is this CFPB study from 2014, which is cited to back up the claim that “Medical debt collections are less predictive of future payment problems than other debt collections are [1].” The cited study comes to two conclusions: that both paid medical collections are less predictive of future defaults and that all medical collections taken together (paid and unpaid) are less predictive of future defaults than non-medical ones. However, the very same study also states that “[For consumers with mostly unpaid medical debt] The new account performance measure yields predominantly negative score differentials, suggesting that the scores of these consumers were higher than they should have been given their subsequent performance. That is, they appear to have been under-penalized” [2] (emphasis added).

So we know from the CFPB’s own data that unpaid medical collections appear to “under-penalize” consumers while paid medical collections appear to “over-penalize” consumers. Thus the logical conclusion is that accuracy could be improved by removing only paid medical collections from reports or changing the way that FICO and Vantage scoring models count these collections. Credit scores that include unpaid medical collections predict the probability of future delinquency better than those that don’t, and should therefore remain a part of credit reports and FICO/Vantage scores. Bureau changes to remove both paid and unpaid medical collections are likely to reduce accuracy, thereby increasing the cost of debt for non-prime consumers (through higher interest rates to compensate for higher loss rates that are harder to predict). The changes could also potentially increase the costs of care (e.g., a lot of medical bills may suddenly add up to $501, as providers become aware of the new threshold and look to bump prices to ensure patients have an incentive to pay on time). 

The three major credit bureaus announced earlier this year that they will be dropping most medical debt from credit reports. Both paid and unpaid collections under $500 will be dropped in the coming year, and the reporting requirements around such debt will be made more favorable to consumers (i.e., adding a one-year hold period before furnishing such data). At first glance this looks like a clear win-win; we know there is a high degree of inaccuracy in medical debt reporting that harms consumers (through decreased access and increased cost of credit), lenders (who make less money by being unnecessarily restrictive in underwriting), and bureaus (whose product is accurate data). And we know that most collections (by count) are for small-dollar medical debt, making the reach of the problem significant (43 million individuals). What’s not to like?

Unfortunately the campaign driven by the CFPB (Consumer Financial Protection Bureau) to change medical debt reporting will do the very opposite of what was intended. It’s likely to decrease accuracy and increase the cost of credit for non-prime U.S. consumers (generally the middle class, the young, and the working poor). By reading the primary sources published by the CFPB and their claims, we can better understand how and why.

The most recent study of medical debt by the CFPB states the challenges it sees with current reporting practices and identifies disparate impact and diminished predictive value as the primary consumer harms they wish to see rectified by industry. Solving for the former can be a philosophical question of equality of process vs. equality of outcome, while solving for the latter is very much in the interest of all parties (consumers, lenders, and bureaus).

We focus our attention on the latter, the source of which is this CFPB study from 2014, which is cited to back up the claim that “Medical debt collections are less predictive of future payment problems than other debt collections are [1].” The cited study comes to two conclusions: that both paid medical collections are less predictive of future defaults and that all medical collections taken together (paid and unpaid) are less predictive of future defaults than non-medical ones. However, the very same study also states that “[For consumers with mostly unpaid medical debt] The new account performance measure yields predominantly negative score differentials, suggesting that the scores of these consumers were higher than they should have been given their subsequent performance. That is, they appear to have been under-penalized” [2] (emphasis added).

So we know from the CFPB’s own data that unpaid medical collections appear to “under-penalize” consumers while paid medical collections appear to “over-penalize” consumers. Thus the logical conclusion is that accuracy could be improved by removing only paid medical collections from reports or changing the way that FICO and Vantage scoring models count these collections. Credit scores that include unpaid medical collections predict the probability of future delinquency better than those that don’t, and should therefore remain a part of credit reports and FICO/Vantage scores. Bureau changes to remove both paid and unpaid medical collections are likely to reduce accuracy, thereby increasing the cost of debt for non-prime consumers (through higher interest rates to compensate for higher loss rates that are harder to predict). The changes could also potentially increase the costs of care (e.g., a lot of medical bills may suddenly add up to $501, as providers become aware of the new threshold and look to bump prices to ensure patients have an incentive to pay on time).