eLaws | eCases | United States Code | Federal Courts |
Home » 2025 Issues » 90 FR (01/14/2025) » 2024-30824. Prohibition on Creditors and Consumer Reporting Agencies Concerning Medical Information (Regulation V)

  2024-30824. Prohibition on Creditors and Consumer Reporting Agencies Concerning Medical Information (Regulation V)  

  • Table 1—Estimated Changes in the Number of Originated Loans Under the Rule by Credit Account Type 318
    (1) Account type (2) Estimated coefficient (3) Baseline inquiry success rate (%) (4) Expected percent change in originated accounts (5) Annual number of originated accounts (6) Expected change in annual originated accounts
    Credit card ***−0.047 26.0 18.1 2,014,427 364,611
    Mortgage *−0.026 17.2 15.1 144,915 21,882
    Other loans *−0.014 23.9 5.9 1,083,879 63,949
    Estimates marked with *** are statistically significantly different from zero at the one percent confidence level. Estimates marked with * are statistically different from zero at the 10 percent confidence level.

    For all credit account categories, the CFPB expects that more loans would be originated if all medical collections were removed from consumer reports provided to creditors under the rule. The estimates in Columns 5 and 6 are underestimates because not all originated loans can be connected to an inquiry in the CFPB's CCIP, as the data only include inquiries made to one NCRA, and many non-mortgage creditors pull consumer reports from only one or two NCRAs. Additionally, these estimates assume that credit demand would not change under the rule. The CFPB's research in the Technical Appendix finds that consumers are more likely to apply for ( print page 3337) credit in the weeks before a medical collection is added to their consumer report than in the weeks after. However, the characteristics of credit applications made before and after a medical collection is added (and their associated consumers) do not appear to have any statistically distinguishable differences between them. This finding suggests that any increase in credit demand under the rule will not lead to declines in credit application quality.

    The results in Table 1 provide evidence that creditors will provide more credit to consumers with medical collections under the rule. At baseline, the CFPB assumes that creditors only make loans to people with reported medical collections if those loans are profitable on average, holding consumer report characteristics constant. If the marginal loans that would be made under the rule have similar revenue potential to those made to consumers with reported medical collections at baseline, the increase in the number of loans made to people with medical collections would increase creditor profits. To estimate the revenue potential of originated accounts, the CFPB estimates the likelihood of serious delinquency within two years of a credit account's origination date for accounts that are opened in connection with an inquiry made in the 180 days before or after a medical collection is included on a consumer report. If creditors' use of medical collections information in their underwriting decisions reduces the delinquency risk of newly opened accounts, one would expect that credit provided to consumers with outstanding, but unreported, medical collections will have higher delinquency propensity than credit provided to consumers with outstanding and reported medical collections.

    The CFPB tests this hypothesis, and estimates ranges for the number of delinquent loans that would be issued if medical collections were not included on consumer reports, as when the rule is finalized, in Table 2. These ranges also incorporate the evidence from the Technical Appendix on how the number of newly originated loans would change, shown above in Table 1. The estimated coefficients from Column 1 of Table 8 in the Technical Appendix are listed in Table 2 in Column 2.

    Table 2—Estimated Changes in the Number of Seriously Delinquent Loans Under the Rule by Credit Account Type 319
    (1) Account type (2) Estimated coefficient (3) Baseline D90+ rate (%) (4) Expected change in annual originated accounts (5) Expected number of D90+ accounts within two years of origination at baseline D90+ rate (6) Expected number of annual D90+ accounts within two years of origination at estimated delinquency rate for unreported medical collections
    Credit card 0.000 20.7 364,611 75,474 75,474
    Mortgage 0.011 3.1 21,882 678 438
    Other loans 0.012 17.1 63,949 10,935 10,168
    None of the estimated coefficients are statistically significantly different from zero.

    Table 2 shows that, for mortgages and other (not credit card and not mortgage) account types, accounts originated by consumers with reported medical collections have slightly higher delinquency propensity than accounts originated by consumers with unreported medical collections. This is not consistent with creditors using medical collections information to reduce the delinquency risk of originated accounts. The coefficients are not statistically distinguishable from zero, so the evidence is only suggestive, rather than conclusive, that the expansion of credit under the rule would yield a rate of serious delinquency that is lower than the rate of serious delinquency currently faced by creditors for accounts they provide to consumers with reported medical collections. The CFPB interprets its findings as evidence against any significant increase in the rate of serious delinquency as compared to the rate of serious delinquency for accounts provided to consumers with reported medical collections at baseline. The CFPB notes that this claim holds if all else is equal under the rule.

    If consumer demand for credit is affected by the rule, the credit applications that creditors receive may have different underlying delinquency risk. Some consumers may avoid applying for credit when a medical collection appears on their consumer report if they understand that this information lowers the likelihood that their credit application will be approved or provided with favorable terms. Removing medical collections from consumer reports used in credit eligibility determinations may lead these consumers to submit credit applications, which could lead to an increase or decrease in the delinquency risk of applicant pools. As discussed in the Technical Appendix, the CFPB finds that consumers are less likely to apply for credit after a medical collection is added to their consumer report; however, the underlying delinquency risk of the remaining credit applications is not statistically distinguishable from the delinquency risk of credit applications made before the medical collection is reported.

    To provide further evidence for how credit demand may respond to the rule, the CFPB used data from the CCIP to estimate if the NCRAs' voluntary removal of medical collections under $500 in April 2023 was associated with ( print page 3338) increased credit demand.[320] The CFPB found that consumers who had medical collections under $500 included on their consumer reports in the first quarter of 2023 were just 0.07 percent less likely to have an inquiry in the six months after medical collections under $500 were removed from their consumer reports. This suggests that credit demand is not responsive to the removal of medical collections from consumer reports, at least in the short run. In the long-run equilibrium, the CFPB expects that consumer demand for credit may increase without the use of medical collections information in underwriting, but the CFPB does not expect that either the underlying delinquency risk of consumers with medical collections that apply for credit, or creditors' ability to predict those consumers' delinquency risk, will change under the rule.

    Creditors may change their underwriting processes in response to the rule, which may impact the allocation of credit. The CFPB's research in the Technical Appendix analyzed inquiries that were made when a subset of medical debt information was available to creditors on consumer reports. If creditors instead knew that they could not generally use any medical debt information in their underwriting processes, they may change their underwriting models to put more weight on other variables. The CFPB expects that these changes would improve model performance relative to the baseline, and as a result, delinquency rates may be lower under the rule.

    Although the CFPB does not estimate that there will be a significant number of additional seriously delinquent accounts as a result of the rule, the CFPB does not have data available that would enable it to calculate the monetary cost to creditors of potential additional delinquencies. The CFPB requested information on the dollar cost to creditors of an account that becomes seriously delinquent within two years of its origination. A researcher commenter stated that even if the probability of serious delinquency or default did not change under the rule, the Exposure at Default, or the dollar amount that the consumer is delinquent for, may be higher for credit card lenders under the rule. This would lead to a higher expected loss and reduced revenues.

    The CFPB agrees that the profitability of a loan is not solely defined by whether it becomes delinquent or not. However, the other factors that determine profitability do not unambiguously point toward lower revenue for creditors due to the rule, all else equal. For example, credit card borrowers who carry a balance month-to-month (often termed revolvers), are more profitable for credit card companies than other types of consumers.[321] Therefore, higher credit limits for these consumers under the rule may lead to higher revenue. If this source of higher revenue does not balance higher costs of default, the CFPB expects that credit card lenders may reduce the credit limits they provide under the rule. The CFPB does not expect this outcome to be likely because credit card lenders already do not observe most medical debts, as most medical debts are not included on consumer reports and credit card lenders do not generally explicitly ask for medical debt information on credit applications.

    Increases in access to credit may revert over the long run if credit scoring companies change their models or creditors change their underwriting practices in response to the rule. Other information on consumer reports could receive different weights to compensate for the loss of medical collection information, which could attenuate these increases or even reduce access to credit for some consumers. However, the CFPB understands that credit scoring companies and creditors would only implement these changes if the benefit from doing so outweighed the costs of changing these models and procedures. The results in the Technical Appendix suggest that medical collections reporting does not enable creditors to make fewer delinquent loans, implying that creditors on average would not experience any decline in revenue from the absence of this information. Accordingly, the expected small (or zero) benefit of recalibrating credit scoring models and underwriting practices may lead to longer-term increases in access to credit for consumers with medical debt.

    Because commonly used commercial credit scoring models require a minimal number of credit tradelines to generate a score, some consumers may lose their credit scores if medical collections are removed from their consumer reports. For instance, FICO will only provide a credit score if the consumer has at least one credit account that is at least six months old and there has been activity on the credit account in the previous six months.[322] Similarly, VantageScore requires at least one tradeline with any activity before providing a score.[323] For consumers with few tradelines, the removal of medical collections could lead them to lose their credit score.

    Multiple commenters, including at least one debt collector, health care provider, and an individual, agreed with this point, stating that the proposed rule would cause more people to have thin credit files.[324] One NCRA commenter estimated that over three million consumers would lose their credit score under the proposed rule, though the commenter did not list the scoring model or models that provided the basis for their estimation.

    To provide evidence for the scale of this effect, the CFPB analyzed CCIP data from the months immediately before and after the NCRAs' voluntary removal of medical collections under $500 in April 2023. This internal analysis estimated that these reporting changes caused approximately 5,500 consumers to lose their credit score, representing 0.03 percent of consumers who had all their medical collections removed because of the April 2023 reporting changes. The median credit score for these consumers before their medical collections were removed was 581. The CFPB estimated using consumer reports from January 2024 in CFPB's CCIP as the baseline that fewer than 1,000 consumers may lose their credit scores if all medical collections were to be removed from consumer reports. The median credit score for these consumers in January 2024 was 573. Though not having a credit score can reduce access to credit, so too does having a subprime ( print page 3339) credit score, and the generally low baseline credit scores of affected consumers indicate that any increase in the number of consumers without credit scores under the rule may not lead to an overall reduction in consumers' access to credit. Indeed, as stated by one NCRA, generally “no credit is better than bad credit” for the purposes of accessing credit.[325] Based on CFPB's analysis of the CCIP data, the CFPB expects that any reduction in access to credit because of an increase in the population of consumers without credit scores would be very small.

    Despite these potential negative effects, the CFPB expects that consumers who at baseline have medical collections on their consumer reports would generally experience increased access to credit under the rule, in part caused by increases in their credit scores. Multiple commenters, including consumer advocates, at least one research institute, and at least one researcher, stated that the proposed rule would lead to higher credit scores for consumers, though a bank trade association stated that any expected gains are speculative without access to credit scoring models. The CFPB agrees that existing research cannot conclusively describe how credit scores will be impacted by the rule, but the research sheds light on potential changes.

    Consumers with medical collections on their consumer reports in August 2022 had credit scores that were 30 points higher in August 2023 than in August 2022, after the implementation of the voluntary removal of medical collections under $500 in April 2023; consumers without medical collections on their consumer reports in August 2022 experienced a one-point decline in their average credit scores by August 2023.[326] This suggests that the removal of medical collections under $500 may have increased credit scores by at least 30 points, on average, though there may be other differences between consumers with and without medical collections on their consumer reports in August 2022 that explain part of the difference in their credit scores.

    Evidence from CFPB research suggests that consumers experience a 25-point increase in their credit score, on average, after their last medical collection is removed from their consumer report.[327] However, the causes of the studied medical collection removals were unknown, and there may be unobservable factors that caused both the medical collection removal and increases in consumer credit scores, so these results cannot be interpreted causally.

    One researcher commenter identified several methodological limitations of the study that the commenter stated meant that the CFPB study did not capture the causal effect of removing medical collections on consumers' credit scores. While the CFPB does not agree with many of the specific methodological critiques made by the commenter, the CFPB stated in the proposal and again above that it does not see this evidence as causal and agrees with the commenter on this broader point.

    Other CFPB research has leveraged the recent voluntary removal of medical collections tradelines below $500, finding that consumers for whom all medical collections were below $500 prior to the changes saw their credit scores increase 20 points more than consumers who had some medical collections tradelines above $500.[328]

    One researcher commenter stated that this conclusion could not be validated because there is no control group.[329] The researcher commenter further stated that any increase in credit scores would lead to higher delinquency rates. With respect to the consequences of increases in credit scores for creditors, the CFPB does not agree that creditors will face higher delinquency rates as a result. When a credit scoring company removed medical collections from its model in 2023, the company reported that there was minimal change in predictive performance.[330]

    For a sample of fewer than 3,000 consumers who had their medical debts removed from their consumer reports after their debt was relieved by a nonprofit organization, Kluender et al. (2024) found that credit scores increased by an average of just three points; however, this sample may not be representative of all consumers with medical debts, as the reported collections were much older on average than most medical collections on consumer reports.[331]

    VantageScore removed all medical collections from its credit scoring model in 2022 and reported that “millions of consumers may see an increase of up to 20 points in their VantageScore credit scores.” [332] The CFPB expects that consumers may experience similar increases in their credit scores from other credit scoring companies if medical debt information is removed from consumer reports under the rule. Higher credit scores can lead to higher loan approval rates and more favorable terms.[333] The CFPB requested information on the dollar value to consumers of higher credit scores but did not receive relevant comments.

    Several commenters, including debt collectors, individuals, and at least one health care provider, one researcher, and one debt collector trade association, commented that the proposed rule will make it more challenging for consumers to repair their credit scores. The commenters stated that “clearing” or “resolving” medical debts from a consumer report signals that a consumer is a good credit risk. The CFPB understands that this concern is not accurate based on the way credit scores currently operate. In the past, it was possible for consumers to increase their credit scores by paying a collections item, as paid collections tradelines are typically treated as less negative than unpaid collections tradelines. However, currently there is no mechanism for ( print page 3340) paid medical collections to appear as a positive indicator because the NCRAs voluntarily removed paid medical collections from consumer reports in June 2022. Even before this change, the removal of a medical collection tradeline would typically be better for consumers' credit scores than a paid medical collections tradeline. Furthermore, it would typically be even better for the consumer to never have the medical collection appear on their consumer report in the first place, as under the rule.

    At baseline, debt collectors may use information from consumer reports to determine a consumer's ability to pay the collection amount and to guide what collection practices will be most cost-effective. Debt collector small entity representatives, in their submitted comments during the SBREFA process, stated that they found medical debt information on consumer reports to be relevant to estimating whether a consumer will repay a debt that is in collections.[334] Under the rule, debt collectors will continue to be permitted to use medical debt information on consumer reports because debt collection is not considered a credit eligibility determination. The CFPB expects that, if medical debt information is a critical indicator of a consumer's ability to repay a debt in collections, debt collectors will continue to furnish medical debt information to consumer reporting agencies. Furthermore, debt collectors and health care providers may continue to furnish medical collections to consumer reporting agencies because consumer reporting agencies would still be able to include medical collections information on the reports that they provide for other non-credit eligibility determination purposes such as employment or insurance, or to consumers seeking a copy of their own consumer reports.

    6. Compliance Costs

    Under the rule, consumer reporting agencies will need to ensure that medical information is removed from consumer reports that are provided to creditors for credit eligibility determinations, which will require significant changes to existing consumer reporting systems and databases. Compliance costs will be low because furnishers have an obligation to notify consumer reporting agencies of their status,[335] thus making their removal from certain consumer reports a simple coding adjustment, which the consumer reporting agencies will already have to do to comply with several states' laws. The rule also has the potential to reduce consumer reporting agencies' costs by reducing the number of consumer disputes. Creditors may incur compliance costs to update their underwriting systems. Creditors may also need to train and staff attorneys to ensure they continue to meet their obligations to assess a consumer's ability to repay under the Truth in Lending Act and Regulation Z.

    Commenters noted that consumer reporting agencies may incur one-time costs to develop practices to comply with the rule. At least one credit union commenter and consumer reporting agency trade group commenter stated that removing medical debt from consumer reports would require significant changes to existing consumer reporting systems and practices. One consumer reporting agency commenter wrote that implementing the NCAP required working with furnishers and took two years to complete. The commenter stated they would require at least the same amount of time to implement the changes needed to comply with the proposed rule.

    The CFPB agrees that some consumer reporting agencies may need to add new code to computer systems and databases such that no medical debt information is contained in consumer reports provided to creditors for credit eligibility determinations. However, some operational and compliance costs that may have otherwise been caused by the rule may have already been incurred to some degree to comply with certain State laws as well as the NCAP changes, and the consumer reporting agencies should be able to scale those coding changes nationwide. The CFPB requested data that could be used to quantify costs that may be incurred or have already been incurred by consumer reporting agencies but did not receive relevant quantitative information.

    A SBREFA commenter, not representing the NCRAs, posited that making the necessary changes would be a significant undertaking in terms of time and cost and that the NCRAs would have to reconfigure, test, and validate their current compliance programs. The CFPB agrees that compliance costs may be different for the three NCRAs (Equifax, Experian, and TransUnion) and Innovis compared to other consumer reporting agencies. The NCRAs and Innovis are known to provide a standardized reporting format to be used by furnishers, called Metro 2, and have organized their databases to process and screen data furnished in this format.[336] The Metro 2 format that the NCRAs and Innovis currently provide to furnishers may help facilitate compliance because tradeline information submitted by furnishers is already required to include codes that specify when a debt is a medical debt.[337]

    Additionally, the three NCRAs currently do not include medical collections under $500, medical collections that are less than one year past due, or paid medical collections on any consumer report provided to third parties. The NCRAs also remove medical collections as required by State laws.[338] It is likely that the NCRAs already have systems in place to screen out any furnished medical collections that may violate these conditions. These systems may be specific to removing furnished medical collections from all consumer reports, rather than only from consumer reports provided to creditors for the purposes of a credit eligibility determination as required under the rule. It is possible that the NCRAs' and Innovis's screening process may have to be expanded such that they only selectively remove medical collections information from consumer reports as required by the rule, or they may choose to remove medical collections from all consumer reports. The CFPB does not have information that would allow it to predict how the NCRAs and Innovis would choose to comply with the rule.

    Consumer reporting agencies that have different screening processes and databases that do not rely on the Metro 2 format may incur different compliance costs associated with their own systems, though, as noted above, some compliance costs may already have been incurred to comply with State laws, and efforts to comply with those state laws are likely replicable or scalable for a nationwide change. Consumer reporting agencies may incur costs to screen medical information provided by such furnishers, or for which there is no medical information furnisher within the meaning of FCRA section 623(a)(9), from consumer reports provided to creditors for credit eligibility ( print page 3341) determinations. The CFPB requested comment and information on this potential compliance cost, general operational and compliance costs, and other possible one-time costs for consumer reporting agencies, but it did not receive relevant information.

    The U.S. Chamber of Commerce and an individual commenter stated that determining which debts are “medical” could be complex and that any broadening to include debts owed to third-party lenders would result in debts not connected to the provision of health care being removed from consumer reports. The CFPB is not broadening the rule to include debts owed to third-party lenders, however, and there is no indication that consumer reporting agencies currently have difficulty determining which debts are “medical.”

    The removal of medical collections information from consumer reports provided to creditors for the purpose of credit eligibility determinations may reduce consumer reporting agencies' costs by reducing the number of accounts that consumer reporting agencies must screen or conduct accuracy checks for, and the number of consumer disputes that they may need to resolve. Consumer reporting agencies regularly process significant amounts of data. For example, the NCRAs receive information on over 1 billion tradelines each month and must accurately compile this information for each consumer.[339] Under the FCRA, consumers have the right to dispute inaccuracies on their consumer reports, and consumer reporting agencies are obligated to investigate and resolve disputes if necessary.[340] This dispute resolution process imposes costs on consumer reporting agencies. A CFPB analysis shows that 5.7 percent of reported medical collections tradelines have had a dispute flag, much higher than the rate of dispute flags for credit cards and student loans.[341] One NCRA commenter reported that their data shows that while consumers dispute medical collections tradelines more often than other collections tradelines, they do so at a similar rate to consumers disputing other delinquent non-collections tradelines.

    To the extent that medical collections tradelines contribute to the number of disputes that consumer reporting agencies must address and, if possible, resolve, removing medical collections information from consumer reports may reduce consumer reporting agencies' costs associated with addressing disputes and dispute resolution. However, the CFPB does not have data to estimate the cost reduction arising from dispute management that consumer reporting agencies may experience if medical debt information is prohibited from appearing on most consumer reports provided to creditors. The CFPB requested data to quantify these potential cost-reducing benefits but did not receive relevant information.

    One credit union commenter wrote that consumer reporting agencies may provide medical debt information in violation of the rule, which puts the recipient creditors at legal risk if they then rely on that information for credit eligibility determinations. The CFPB, however, has no reason to believe that consumer reporting agencies will fail to comply with their obligations under § 1022.38 to exclude medical debt information from consumer reports furnished to creditors, and assuming such compliance on the part of consumer reporting agencies, there will generally be no costs to creditors arising from litigation concerning medical information when they rely on consumer reports from consumer reporting agencies.

    One credit union trade association commented that, for creditors that use information beyond consumer reports, there will be costs associated with excluding medical information. At least one credit union commenter and one credit union trade association commenter stated that financial institutions would incur significant costs to update their underwriting systems, consumer reporting systems, and practices, and that there is a risk of errors during the removal process which could lead to further complications and disputes. The SBA Office of Advocacy commented that it may be challenging for creditors to prove whether medical debt information was disclosed by consumers themselves, and making these determinations would require training and staffing attorneys.

    The CFPB agrees that creditors may incur compliance costs from the rule. Creditors will need to ensure that they are not unintentionally using medical information in making lending determinations in circumstances that fall outside the exceptions to the creditor prohibition. The CFPB has determined that costs related to ensuring that no medical information is unintentionally used in lending determinations should be minor to the extent that creditors currently only utilize medical debt information provided through consumer reports. In such cases, so long as the consumer reporting agency providing the consumer report has complied with the rule, no medical debt information would be conveyed to the creditor, unless the consumer reporting agency has reason to believe the creditor intends to use the medical debt information in a manner not prohibited by the creditor prohibition. Creditors who use consumer reports may have additional costs if they utilize consumer reports from which the consumer reporting agency has not excluded medical debt information in compliance with § 1022.38. In such cases creditors would need to employ systems and staff time to identify and exclude that information. But as explained above, the CFPB has no reason to believe that consumer reporting agencies will fail to comply with their obligations under § 1022.38.

    In addition, creditors that rely on information outside of consumer reports will face compliance costs related to identifying medical information from other sources and excluding it from their underwriting (except as permitted by an exception to the creditor prohibition). The CFPB does not have data available to quantify the extent or dollar amount of any of these compliance costs, and requested comment on this issue but did not receive relevant data or estimates.

    Commenters including health care providers, a researcher, debt collectors, credit unions, the U.S. Chamber of Commerce, and the SBA Office of Advocacy commented that the proposed rule would create conflicting obligations for creditors under the Truth in Lending Act (TILA) and Regulation Z, particularly with respect to the ability-to-repay provisions. They stated that it would be more difficult to respond to ability-to-repay laws under the proposed rule.

    The rule, however, allows creditors to obtain and use medical information to comply with applicable requirements of local, State, or Federal laws, including ability-to-repay laws, and provides an example of how a creditor can consider consumers' self-reported medical debt information to comply with such laws. The CFPB thus concludes that creditors can comply with both the rule and the requirements of ability-to-repay laws. ( print page 3342)

    One credit union commented that, by complying with the proposed rule, they would be at risk of losing a member with medical debt who receives credit from the credit union that they would not have received otherwise and who feels that they were not adequately educated or protected. The CFPB does not expect that consumers with medical debt would be provided credit they cannot afford under the rule.

    One bank trade association expressed confusion about including certain types of medical payments in underwriting. They stated that transaction data includes payments to medical providers, and they were unclear if this information could be used by creditors for the purpose of credit eligibility determinations under the proposed rule. The CFPB has permitted the use of medical information that is included in the transaction information of an account by creditors for the purpose of credit eligibility determinations in the final rule.

    7. Inaccurate Billing

    The CFPB understands that many medical collections included on consumer reports reflect incorrect billing, including debts that were already paid by either the consumer or by their insurance company, or debts that are not owed by the consumer. Nearly half of consumers who made formal complaints to the CFPB about medical debt collection in 2021 reported that they did not owe the debt, and many consumers did not know that they had outstanding medical debt until they discovered a collections tradeline on their consumer report.[342]

    Numerous commenters, including individuals, debt collectors, health care providers, and health care trade associations, disputed the prevalence of inaccurate medical billing as described in the proposed rule. These commenters stated that most patient accounts are billed accurately and that the CFPB's complaint database, which was cited as evidence of inaccurate medical billing in the proposed rule, does not reflect health care provider perspectives. Multiple individuals, debt collectors, health care providers, NCRAs, and debt collection trade associations commented that most medical bills are accurate and that there is no evidence that bills are inaccurate. One debt collection trade association commented that disputes are generally the result of conflicts between health insurers and consumers, so the fault for inaccurate medical billing lies with health insurers rather than with debt collectors. An NCRA commented that medical collections are disputed less frequently than other collections, and when disputed, are verified at higher rates.

    One debt collector commented that consumer reporting agencies already have methods for consumers to dispute and pursue legal remedies for inaccurate data. A financial trade association noted that the proposed rule referenced a study in which, of the 43 percent of consumers that reported receiving a medical bill that they believed contained an error, 79 percent took actions to dispute the mistake with their insurer or health care provider.[343] Seventy percent of those disputes led to a successful resolution, which the commenter interpreted as evidence that there are already measures in place within the health care system to address erroneous billing.

    At least one consumer advocate stated that, citing public statements from medical billing advocate groups, 60 to 80 percent of hospital medical bills have errors, with multiple individuals, research centers, consumer advocates, and law firms stating that inaccuracies in medical billing data are pervasive. At least one consumer advocate cited survey results that found most consumers have received a medical bill they believe to have errors.

    The CFPB acknowledges that its complaint database is centered around consumers' negative experiences with medical debt, and the database cannot provide an estimate of the share of medical collections that result from inaccurate billing. However, even though there are existing mechanisms for consumers to dispute inaccurate medical bills with health care providers, debt collectors, and consumer reporting agencies, consumers will benefit from not needing to dispute these debts under the rule in order to avoid inaccurate negative information on their credit reports.

    At baseline, consumers may pay debts they do not owe to remove them from their consumer report. The CFPB does not have information available to estimate how many medical debts are paid despite containing inaccurate information but expects that fewer of these erroneous debts will be paid under the final rule. The CFPB requested comment and submissions of data, or any other relevant information, that may be helpful in estimating this reduction in erroneous debts paid but did not receive data or evidence.

    At least one debt collector commented that consumer privacy would be harmed under the proposal because many entities would need to handle sensitive information. The commenter did not explain why this would be the case. In fact, the CFPB expects that fewer entities would need to handle sensitive information under the rule because medical information would no longer be provided to creditors on consumer reports.

    A researcher commenter stated that consumer privacy would be harmed by increased use of litigation under the rule, because litigation can lead to the formation of public records, unlike consumer reporting. The CFPB agrees that this is a potential cost of the rule but expects that the rule will not greatly increase the number of consumers that are subject to litigation.

    8. Alternatives Considered

    Government officials and consumer advocate commenters recommended extending the rule to include medical credit cards, medical financing plans, and medical information on general-purpose credit cards. Under this alternative construction of the rule, consumer reporting agencies would not be permitted to provide this information to creditors, and creditors could not use this information in their credit eligibility determinations. One consumer advocate commented that, in addition, the CFPB should prohibit common features of medical payment products that can lead to consumer harm, including deferred interest, charging for services before they are rendered, and issuing payment products to consumers whose insurance would otherwise pay or who qualify for financial assistance. One government official suggested that this alternative may be preferable in part because card issuers' merchant category code system includes categories that would be similar to those needed to label medical information as such, simplifying the process by which creditors would be required to identify medical information.

    The CFPB's own research has shown that medical payment products can pose financial risk to consumers.[344] The CFPB is working with the U.S. ( print page 3343) Departments of Health and Human Services and Treasury to monitor the relationships between financial institutions and health care providers and gather relevant information.[345] The CFPB has also considered the use of medical transaction information in credit eligibility determinations but understands that most creditors do not use granular transaction data. The CFPB has determined that there is not yet substantial evidence that the inclusion of medical payment products information on consumer reports, or its use in underwriting, leads to consumer harm and has chosen not to include this information in the rule.

    Government officials and consumer advocates recommended extending the rule to consumer reports used for employment and tenant screening, rather than limiting the prohibition to medical information provided on consumer reports for the purpose of credit eligibility determinations as proposed.

    The CFPB does not have insight into the use of medical information by employers or landlords, but it did study its use by creditors to deny access to credit through its CCIP, as discussed in part VII.E.5. This evidence motivates the rule's focus on consumer reports provided to creditors for the purpose of credit eligibility determinations. The CFPB has determined that while these proposals might have additional benefits for consumers, they are beyond the scope of this rulemaking.

    Commenters including a debt collector trade association and multiple credit union trade associations stated that the CFPB should provide guidance to, or increase its enforcement of, relevant entities instead of issuing the final rule. The SBA Office of Advocacy commented that CFPB should consider using enforcement actions with respect to businesses that furnish inaccurate medical debt information instead of the proposed rule. A debt collector trade association commented that the CFPB should provide guidance to medical debt collectors covering the inclusion of financial assistance policies in debt collection communications under the safe harbor provisions of Regulation F. The commenter also stated that the CFPB should better enforce the FCRA consumer dispute provisions to ensure the accuracy of medical debt reporting and should work with the U.S. Department of Health and Human Services to provide information about financial assistance to consumers who may qualify. One credit union trade association commented that the CFPB should provide a safe harbor provision for credit unions that unintentionally possess medical debt information. A second credit union trade association commented that the CFPB should issue guidance to financial institutions to help them better understand the predictive value of medical debt or permit lenders to use medical debt as long as it is assigned a lower weight in credit eligibility determinations.

    The CFPB has determined that these proposed alternatives may be marginal improvements toward the intended goals of the rulemaking but would not fully realize the full scope of the rule's benefits for consumers. As such, it has decided not to implement these suggestions.

    Multiple commenters suggested alternatives that are beyond the jurisdiction of the CFPB. A debt collector trade association commented that it would be preferable to target health plan cost-sharing and policies that impact consumers' ability to pay large bills from any source, not just from health care. A debt collector commented that the CFPB should provide financial assistance programs, improve health insurance coverage, and simplify billing processes. A different debt collector stated that the CFPB should encourage health insurers to improve their health care coverage, so consumers incur less medical debt in the first place. A credit union trade association stated that the CFPB should require health care providers to require transparency in medical pricing and billing.

    These alternatives may achieve some of the goals of the rulemaking, but the CFPB does not have the regulatory authority to implement them.

    Commenters suggested that the CFPB conduct additional research before finalizing the proposed rule to evaluate whether the rule is necessary. A debt collector trade association commented that the CFPB could evaluate the benefit of the No Surprises Act after it has been fully implemented. A debt collector stated that the CFPB should not finalize the rule before it studies the impacts of the voluntary NCRA reporting changes, while another debt collector stated that the CFPB should first study how the marketplace responds to credit scoring models that reduce the weight that medical collection information receives.

    The CFPB shares the commenters' interest in ensuring the rule is supported by research. The evidence in the Technical Appendix shows that the inclusion of medical information on consumer reports reduces consumer access to credit without lowering creditors' delinquency risk. As such, the CFPB does not believe that additional research is needed.

    Two debt collectors commented that the CFPB should differentiate between consumers who can and cannot pay under the rule. One debt collector recommended making this differentiation by consumers that are insured versus uninsured, while the other recommended finding alternative measures to differentiate between consumers that are unwilling to pay versus those that are unable to pay.

    Information about consumers' insured status or that specifically addresses consumers' ability to pay is not commonly available on consumer reports. Including insurance information on consumer reports would impose substantial costs on consumer reporting agencies and on health insurers that would presumably be responsible for furnishing health insurance information, and it would exacerbate the privacy concerns that this rule aims to address. Including information on consumer ability to pay may pose even more challenges as many consumers' incomes and financial responsibilities are not included on consumer reports, and numerous entities that do not commonly furnish to consumer reporting agencies, such as landlords and employers, would be required to begin doing so. This would also not resolve the privacy concerns that this rule aims to address. As such, the CFPB has decided not to differentiate between groups of consumers in the rule.

    A debt collector commented that the CFPB should allow for positive consumer reporting of medical debts, such that consumers that make payments on medical bills would have those payments reported as positive information demonstrating an ability to pay debts.

    The CFPB understands that, at baseline, most medical debts are furnished to consumer reporting agencies by debt collectors rather than by health care providers. If consumers are more likely to make on-time payments to health care providers before the debt is placed with a debt collector, this would impose costly furnishing requirements on health care providers. It would also impose furnishing costs on debt collectors that, at baseline, often only furnish medical debts a few times, as discussed above. The CFPB does not have any evidence that paid medical collection items are treated positively in any lending models, and reporting positive medical ( print page 3344) payment information also would only add to the privacy concerns that this rule seeks to address.

    A commenter stated that the CFPB should allow medical debt to remain on consumer reports but require that it is given less weight than other debts.

    It would be impracticable for the CFPB to dictate a precise weight that creditors may or may not give to medical debts in their underwriting or how credit scoring companies weight their algorithms. In addition, it is unclear to what debts the medical debt weights should be compared. Furthermore, reducing the weight on medical debts would not resolve the privacy concerns that this rule aims to address.

    A commenter stated that the CFPB should prevent health care providers and debt collectors from reporting medical debts as long as the consumer makes minimum payments.

    The CFPB understands that, at baseline, most medical debt furnishers use consumer reporting as a mechanism to induce payment. Therefore, it is unlikely that consumers making minimum payments on their medical debt would have it furnished to a consumer reporting agency. This suggested alternative to the rule would not achieve the same benefit to consumers as would the rule.

    A debt collector commented that the CFPB should implement a waiting period before medical debt can be reported and remove paid medical debt from consumer reports instead of finalizing the proposed rule.

    The NCRAs have implemented these changes voluntarily, so limiting the rule to these changes would not benefit consumers relative to the current baseline.

    A credit union trade association commented that the CFPB should require medical debt furnishers to ensure the accuracy of the information they provide to consumer reporting agencies.

    The CFPB agrees with this comment and issued similar guidance to medical debt collectors in October 2024.[346]

    F. Specific Impacts on Consumers in Rural Areas

    The costs and benefits to consumers of the rule will likely be the same, on average, for consumers regardless of where they reside. However, consumers who have outstanding medical debt may be more likely to be affected by the rule. Research by the CFPB and others shows that medical collections on consumer reports are more common for consumers who reside in rural areas, compared to those who reside in non-rural areas.[347] Therefore, in the aggregate, the rule may have a disproportionate impact on consumers in rural areas. Additionally, to the extent that the rule will lead to consumers being denied services by a health care provider, that cost could be greater for consumers in rural areas, where there are often fewer options for medical care.

    Several commenters, including numerous debt collectors, multiple health care providers, at least one health care administration trade association, at least one debt collector trade association, and at least one individual, stated that the proposed rule may decrease access to health care in rural settings. At least one health care provider trade association commenter and at least one individual commenter stated that small rural health care providers are at a disadvantage at baseline and would face more challenges under the proposed rule. Commenters including multiple health care providers, at least one health care administration trade association, several debt collectors, at least one debt collector trade association, and at least one researcher, stated that the proposed rule may lead to increased closures among providers in rural areas and may result in patients needing to travel longer distances for treatment, or may force them to use emergency rooms for non-emergency care. At least one individual commenter, at least one debt collector trade association commenter, and at least one health care provider commenter cited that over 700 rural hospitals are already at risk of closure, and more than half of the 700 face an immediate risk of closure.[348] More than one health care provider commenter stated that hospital closures in rural areas will lead to worse health outcomes or more deaths. At least one consumer advocate commenter stated that closures of health care facilities lead to longer travel distances for consumers in rural areas, and that for some consumers, longer travel times can increase unpaid time off from work and paying for childcare, in addition to the cost of health care received.

    The CFPB does not expect that the rule will result in increased closures of rural health care providers. The CFPB expects that rural health care providers would only close if their revenue decreases significantly. As discussed above, the rule is unlikely to substantially impact revenue for health care providers in the aggregate, as most health care revenue does not consist of consumers paying their bills after receiving treatment and the CFPB does not expect that there will be significantly reduced incentive to pay medical debts as a result of the rule. Additionally, the CFPB does not expect rural providers' revenue to be differentially impacted by the rule. Therefore, the CFPB does not expect increased closures of rural health care providers and significant changes to access to health care in rural settings.

    G. Specific Impacts on Depository Institutions With $10 Billion or Less in Assets

    The CFPB does not expect that the rule will have significantly different impacts on depository institutions with $10 billion or less in assets, compared to larger institutions. The CFPB concludes that the costs to creditors, described above, would apply equally to these smaller institutions.

    Several commenters, including at least one credit union trade association and at least one bank trade association, highlighted that small institutions, including some credit unions, lack the same risk mitigation resources as larger institutions. These commenters stated that the proposed rule would have a disparate negative impact on smaller institutions in terms of risk mitigation. At least one credit union trade association commenter stated that the proposal would likely lead to a scenario where small lenders decide the risk is too great and leave the lending market. At least one bank trade association commenter predicted that community banks would need to reduce their lending the most, leading to competitive losses and operational and compliance costs under the proposed rule. The commenter did not provide evidence for why community banks would be disproportionately impacted by the proposed rule.

    The CFPB finds in the Technical Appendix that the use of medical collections information in underwriting does not reduce the delinquency risk of accounts originated to consumers with reported medical collections relative to consumers with unreported medical collections, and therefore expects that ( print page 3345) removing medical collection information from consumer reports will not reduce the ability of institutions to assess delinquency risk. The CFPB does not expect the impact to vary by the size of institution. Thus, the CFPB does not expect significantly different impacts on depository institutions with $10 billion or less in assets.

    H. Specific Impacts on Access to Credit

    The CFPB discusses impacts on access to credit in detail above. In brief, the CFPB expects that some consumers will lose their credit score, although it is unclear whether this will decrease these consumers' access to credit relative to only having medical collections tradelines. Other consumers will likely see increased access to credit due in part to increased credit scores.

    VIII. Regulatory Flexibility Act Analysis

    The Regulatory Flexibility Act (RFA) generally requires the CFPB to conduct an initial regulatory flexibility analysis (IRFA) and a final regulatory flexibility analysis (FRFA) and convene a panel to consult with small entity representatives before proposing a rule subject to notice-and-comment requirements,[349] unless it certifies that the rule will not have a significant economic impact on a substantial number of small entities.[350] The CFPB provided its analysis to “describe the impact of the rule on small entities” in the NPRM and requested public comment.[351]

    In the NPRM, the CFPB Director certified that the proposed rule would not have a significant economic impact on a substantial number of small entities within the meaning of the RFA. Thus, neither an IRFA nor a Small Business Advisory Review Panel (SBREFA Panel) was required. Nonetheless, the CFPB decided in an abundance of caution to include the proposed rule in the SBREFA Panel convened to address a number of topics under the FCRA on October 18 and 19, 2023, and to provide an analysis consistent with the requirements of an IRFA. In response to the NPRM, the CFPB received comments relevant to the IRFA, which are reflected in the FRFA set forth in part VIII.B.

    The Small Business Review Panel for this rule is discussed in part VIII.A. Among other things, the FRFA contains a statement of the significant issues raised by the public comments in response to the IRFA, a statement of the assessment of the agency of such issues, a statement of any changes made in the proposed rule as a result of such comments, the response of the CFPB to comments filed by the Chief Counsel for Advocacy of the Small Business Administration in response to the proposed rule, and estimates of the number of small entities that may be subject to the rule and descriptions of the impact on those entities. The FRFA for this rule is set forth in part VIII.B.

    A. Small Business Review Panel

    Under section 609(b) of the RFA, as amended by SBREFA and the CFPA, the CFPB must seek, prior to publishing the IRFA, information from representatives of small entities that may potentially be affected by its proposed rule to assess the potential impacts of that rule on such small entities. The CFPB complied with this requirement when it included the proposed rule in the Small Business Review Panel convened on October 18 and 19, 2023.

    B. Final Regulatory Flexibility Analysis

    1. Statement of the Need for, and Objectives of, the Rule

    The creditor prohibition in section 604(g)(2) of the FCRA reflects Congress's intention to protect the privacy of sensitive medical information.[352] The creditor prohibition generally prevents creditors from considering medical information pertaining to a consumer in determining the consumer's eligibility, or continued eligibility, for credit. As described in more detail in part IV.B, Congress allowed certain Agencies, and later the CFPB, to make exceptions to this prohibition, consistent with the congressional intent “to restrict the use of medical information for inappropriate purposes.” [353] In 2005, the Federal financial agencies and the National Credit Union Administration promulgated the financial information exception, restated in the CFPB's regulations at § 1022.30(d), which allows a creditor to consider certain medical information, including medical debt information and information relating to expenses, assets, and collateral, pertaining to a consumer in crediting decisions, provided the conditions of a three-part test are met.[354] The CFPB has determined that an exception for creditors to consider this type of medical information for credit eligibility determinations is not “necessary and appropriate” to protect legitimate operational, transactional, risk, consumer, or other needs, nor is an exception consistent with the intent of the creditor prohibition to restrict the use of medical information for inappropriate purposes as required for an exception under FCRA section 604(g)(5). The CFPB has also determined that an exception for creditors to consider medical information relating to a consumer's expenses, assets, and collateral would not meet the requirements for an exception under FCRA section 604(g)(5). As a result, the CFPB is removing the financial information exception and limiting the circumstances under which consumer reporting agencies can include medical collections information in consumer reports provided to creditors. Further details may be found in parts I.B and IV.

    The primary objectives of this rule are to enhance consumer privacy with respect to sensitive medical information and enable creditors to make appropriate credit decisions based on accurate information, in line with the purposes of the FCRA. The CFPB is authorized under section 604(g)(5) of the FCRA to promulgate exceptions to the creditor prohibition “that are determined to be necessary and appropriate to protect legitimate operational, transactional, risk, consumer, and other needs . . . consistent with the intent of [the prohibition] to restrict the use of medical information for inappropriate purposes.” The CFPB also has authority under section 621(e) of the FCRA to issue regulations to carry out the purposes and objectives of, and to prevent evasions of or to facilitate compliance with, the FCRA. A discussion of the background leading to the rule may be found in part I, and a discussion of the legal authority relevant to this rule may be found in part III.

    2. Significant Issues Raised by Public Comments in Response to the IRFA, a Statement of the Assessment of the Agency of Such Issues, and a Statement of Any Changes Made in the Proposed Rule as a Result of Such Comments

    The CFPB received few comments that were explicitly in response to the IRFA of the proposed rule. Commenters, including small entity representatives in the SBREFA process and debt collectors, stated that the SBREFA process was rushed and that they did not have enough information to provide input on the proposed rule. Commenters also stated that some types of entities that would be affected by the proposed rule were not considered in the IRFA, such as nonbank lenders, health care ( print page 3346) providers, and payors. A debt collector trade organization stated that initial compliance costs would be about $100,000 for each of its member debt collectors, most of which, according to the commenter, are small entities. Several commenters, including debt collectors and a debt collector trade association, stated that the CFPB should consider some exemptions or longer implementation timelines.

    In response to these comments, the CFPB respectfully disagrees that the SBREFA process was rushed or that participants needed more information—the proposal relevant to this rulemaking was straightforward, the CFPB gave participants an outline summary of the proposal one month in advance of hosting the panel and gave participants an opportunity to provide written feedback three weeks after the panel. Additionally, with respect to the IRFA, the CFPB has revised its estimate of the number of small entities that may be affected by the rule to include debt collectors and health care providers in addition to the consumer reporting agencies and creditors listed in the IRFA. In its discussion of projected reporting, recordkeeping and compliance costs, the CFPB includes estimates provided by commenters. The CFPB also includes a discussion of the alternatives proposed by commenters in its description of significant alternatives to the rule.

    3. Response of the Agency to Any Comments Filed by the Chief Counsel for Advocacy of the Small Business Administration in Response to the Proposed Rule, and a Detailed Statement of Any Change Made to the Proposed Rule in the Final Rule as a Result of the Comments

    The Chief Counsel for Advocacy of the Small Business Association (Advocacy) provided comments on several aspects of the proposal, which generally echoed comments received from both small and large industry entities. Advocacy stated that the rule will significantly impact small entities involved in debt collection and that the CFPB has underestimated the number of small entities that may be impacted. Advocacy stated that the CFPB did not provide sufficient information to meet the requirements of a certification of no significant economic impact on a substantial number of small entities, and that the IRFA did not contain economic information on the projected reporting, recordkeeping, and other compliance costs. Advocacy also commented that the rule will increase litigation, causing harm to small entities and consumers because litigation is costly. Advocacy stated that the rule will lead to conflicts with other laws, including TILA and Regulation Z, as well as applicable State laws. Furthermore, Advocacy stated that the rule is redundant in light of changes to industry practices and State laws. In their comment, Advocacy stated that the CFPB should issue clarifications on which laws are controlling so as to mitigate litigation risks for small entities, including creditors who have ability-to-repay requirements under TILA and Regulation Z, and debt collectors who operate in states with their own medical debt collection laws. Advocacy also stated, based on feedback from small entity representatives during the SBREFA Panel, that the cost of credit for small entities may be affected by the rule because removing medical collections from consumer reports may increase credit scores and cause creditors to increase their underwriting standards.[355] Finally, Advocacy suggests that the CFPB provide guidance to small entities for complying with the rule, and develop a mechanism to ensure that small entities are not penalized for not including medical debt in their ability-to-repay determinations.

    In this FRFA, the CFPB has considered indirect impacts to small entities that are health care providers and debt collectors, in addition to the direct impacts to consumer reporting agencies and creditors considered in the IRFA.[356] By examining all credit inquiries made between July 2023 and December 2023 contained in the CCIP, the CFPB determined that most small creditors receive few applications from consumers with medical collections that appear on their consumer reports. In order for the rule to create a significant reduction in revenue, consumers with medical collections would have to experience unreasonably high default rates. Thus, the CFPB has determined that the rule will not have a significant economic impact on a substantial number of small entities directly impacted by the rule, specifically, consumer reporting agencies and creditors. The rule will not directly impact the behavior of medical debt holders such as health care providers and debt collectors since the rule will not affect their ability to furnish medical debt information to consumer reporting agencies. However, to the extent that furnishing becomes a less effective means of inducing payment, health care providers and debt collectors may incur costs associated with their use of other collection mechanisms as well as potential reductions in revenue. For these reasons, the CFPB acknowledges the possibility of indirect economic impacts on small entities that are health care providers or debt collectors. In some parts of the FRFA, the CFPB references the impact analysis part of this rule and presents quantitative estimates when available, including estimates provided by commenters in response to the proposed rule. In addition, the CFPB has revised its estimates of the number of small entities that are creditors that will be affected by the rule.

    With regard to the harm to consumers and small entities from litigation, the CFPB has considered the extent to which litigation might increase as a means of inducing payment of medical debt under the rule. As discussed in part VII.E.4, debt collection litigation is already a collection mechanism used at baseline, and the rule might increase debt collection litigation. Increased debt collection litigation may be most likely to occur in States that have not already passed laws prohibiting medical collections from appearing in consumer reports, and also for small entities collecting on medical debts that are over $500.

    In the proposed rule, the CFPB included an example in proposed § 1022.30(e)(6) to direct creditors and card issuers that are creditors regarding how they may use medical information provided by the consumer in compliance with TILA and Regulation Z, as set forth in § 1022.30(e)(1)(ii), for purposes of compliance with the ability-to-repay rule under § 1026.43(c) for closed-end mortgages, the repayment ability rule under § 1026.34(a)(4) for open-end, high-cost mortgages, and the ability-to-pay rule under § 1026.51(a) for open-end (not home-secured) credit card accounts. With respect to Advocacy's comment that the rule is redundant in light of changes to State ( print page 3347) laws and industry practices, the CFPB's expectation is that the rule will provide clarity and uniformity in the treatment of medical collections on consumer reports across the US. The rule will also complement the voluntary NCRA changes that removed medical collections from consumer reports under $500 as well as paid medical collections, which are industry practices that apply only to medical collections furnished to the NCRAs and can be reversed at any time.

    The CFPB acknowledges that it is possible that underwriting standards might tighten if the rule causes credit scores to increase for a substantial fraction of the population. However, the CFPB's recent research shows that only 5 percent of consumers still have medical debt on their consumer reports at baseline.[357] The CFPB expects that any increase in credit scores may represent a more accurate reflection of credit risk, and that it is unlikely that creditors will raise underwriting standards sufficiently to cause a significant impact on the cost of credit for small entities.

    4. Description and, Where Feasible, Provision of an Estimate of the Number of Small Entities to Which the Rule Will Apply

    The rule will directly affect small entities that participate as creditors as that term is defined in section 702 of the ECOA, except for small entities excluded from coverage by section 1029 of the CFPA, because it will prohibit them from considering certain medical information in their underwriting decisions. This information has been available to creditors under the financial information exception. In limiting the circumstances under which medical debt information can be included on consumer reports, the rule will also directly affect some small consumer reporting agencies. Specifically, consumer reporting agencies that currently provide medical debt information to creditors for credit eligibility determinations will generally no longer be able to do so.

    For the purposes of assessing the impacts of the rule on small entities, “small entities” are defined in the RFA to include small businesses, small nonprofit organizations, and small government jurisdictions.[358] A “small business” is determined by application of Small Business Administration (SBA) regulations in reference to the North American Industry Classification System (NAICS) classification and size standards.[359]

    There are several NAICS industries with small entities that may be subject to this rule. Consumer reporting agencies receive and assemble various types of consumer information and provide consumer reports to third parties for various purposes. Consumer reporting agencies are mostly contained within the NAICS industry “credit bureaus” (561450). However, not all entities within this NAICS code are consumer reporting agencies.[360] Additionally, some consumer reporting agencies specialize in providing consumer reports to facilitate other operations, such as employment screening, check and bank account screening, and insurance, and not for credit purposes.[361] Many small consumer reporting agencies will not be affected by the rule, either because they do not currently furnish consumer reports containing medical debt information or because, under the rule, consumer reports containing medical debt information may continue to be provided for purposes other than credit eligibility, such as employment screening or insurance.

    Creditors potentially directly affected by the rule are contained in multiple NAICS categories. These include depository institutions, such as commercial banks and credit unions, and non-depository institutions, such as mortgage and non-mortgage loan brokers, as well as firms that are primarily engaged in sales lending, consumer lending, or real estate credit. Creditors that currently use medical information related to debts, expenses, assets, and collateral in connection with a determination of a consumer's eligibility, or continued eligibility, for credit will be directly affected by the rule.

    Medical debt holders, which include health care providers and debt buyers, may also be indirectly affected by the rule. The rule will not affect these entities' ability to furnish information to consumer reporting agencies. However, because consumer reporting agencies will generally not be able to include medical debt on consumer reports provided to creditors for credit eligibility determinations, the rule may reduce the effectiveness of furnishing as a collection mechanism. Health care providers are broadly contained in the NAICS subsector 62. Debt collectors are contained in several NAICS categories, and include small entities such as debt buyers, collection agencies, and collection law firms.

    The SBA size standards use asset thresholds for depository institutions and revenue thresholds for non-depository institutions. Depository institutions are small if they have less than $850 million in assets. Consumer reporting agencies are small if they receive less than $47 million in annual revenues. Non-depository institutions in many industries are small if they receive less than $47 million in annual revenues, but the threshold is lower for some NAICS categories of non-depository institutions. The revenue thresholds for health care providers and debt collectors differ depending on the NAICS industry they belong to, ranging between $9 million in annual revenues and $47 million in annual revenues.

    Table 3 shows the number of small businesses within NAICS categories that may be subject to the rule according to the December 2023 NCUA and FFIEC Call Report data and the 2017 Economic Census data from the U.S. Census Bureau, which are the most recent sources of data available to the CFPB.

    ( print page 3348)

    Table 3—Number of Entities Within NAICS Industry Codes That May Be Subject to the Rule
    NAICS codes NAICS description Total number of entities Total number of small entities SBA size standard
    522110 Commercial Banking 4,248 3,170 <$850M (assets).
    522130 Credit Unions 4,702 4,202 <$850M (assets).
    522180 Savings Institutions and Other Depository Credit Intermediation 322 239 <$850M (assets).
    522210 Credit Card Issuing 6 1 <$850M (assets).
    522220 Sales Financing 2,367 2,124 <$47M.
    522291 Consumer Lending 3,037 2,915 <$47M.
    522292 Real Estate Credit 3,289 2,904 <$47M.
    522298 International, Secondary Market, and All Other Non-depository Credit Intermediation 5,422 128 <$47M.
    522310 Mortgage and Nonmortgage Loan Brokers 6,809 6,684 <$15M.
    522320 Financial Transactions Processing, Reserve, and Clearinghouse Activities 3,068 2,928 <$47M.
    522390 Other Activities Related to Credit Intermediation 3,772 3,621 <$28.5M.
    541110 Offices of Lawyers 163,725 833 <$15.5M.
    561440 Collection Agencies 3,224 3,050 <$19.5M.
    561450 Credit Bureaus 307 279 <$41M.
    621111 Offices of Physicians (except Mental Health Specialists) 161,286 158,262 <$16M.
    621112 Offices of Physicians, Mental Health Specialists 10,561 10,407 <$13.5M.
    621210 Offices of Dentists 125,329 124,787 <$9M.
    621310 Offices of Chiropractors 38,695 38,665 <$9M.
    621320 Offices of Optometrists 19,627 19,492 <$9M.
    621330 Offices of Mental Health Practitioners (except Physicians) 24,236 23,958 <$9M.
    621340 Offices of Physical, Occupational and Speech Therapists and Audiologists 26,722 26,217 <$12.5M.
    621391 Offices of Podiatrists 7,304 7,241 <$9M.
    621399 Offices of All Other Miscellaneous Health Practitioners 19,442 19,170 <$10M.
    621410 Family Planning Centers 1,472 1,398 <$19M.
    621420 Outpatient Mental Health and Substance Abuse Centers 6,523 5,879 <$19M.
    621491 HMO Medical Centers 27 3 <$44.5M.
    621492 Kidney Dialysis Centers 431 374 <$47M.
    621493 Freestanding Ambulatory Surgical and Emergency Centers 4,385 3,888 <$19M.
    621498 All Other Outpatient Care Centers 6,630 5,845 <$25.5M.
    621511 Medical Laboratories 3,365 3,106 <$41.5M.
    621512 Diagnostic Imaging Centers 4,272 3,898 <$19M.
    621610 Home Health Care Services 23,801 22,840 <$19M.
    621910 Ambulance Services 3,071 2,940 <$22.5M.
    621999 All Other Miscellaneous Ambulatory Health Care Services 3,557 3,332 <$20.5M.
    622110 General Medical and Surgical Hospitals 2,560 1,130 <$47M.
    622210 Psychiatric and Substance Abuse Hospitals 396 213 <$47M.
    622310 Specialty (except Psychiatric and Substance Abuse) Hospitals 332 131 <$47M.
    623110 Nursing Care Facilities (Skilled Nursing Facilities) 9,137 8,374 <$34M.
    623210 Residential Intellectual and Developmental Disability Facilities 6,885 6,322 <$19M.
    623220 Residential Mental Health and Substance Abuse Facilities 4,165 3,674 <$19M.
    623311 Continuing Care Retirement Communities 3,874 3,533 <$34M.
    623312 Assisted Living Facilities for the Elderly 14,338 13,885 <$23.5M.
    623990 Other Residential Care Facilities 3,194 2,931 <$16M.
    Total 739,915 554,973

    Table 4 provides the estimated number of small entities within the categories of credit bureaus, depository institutions, and non-depository institutions, debt collectors (including debt buyers), and health care providers as well as the NAICS codes these entities may fall within. Under the rule, small consumer reporting agencies will no longer be able to provide to creditors consumer reports that contain medical debt information under the financial information exception. The CFPB is not able to precisely estimate the number of small consumer reporting agencies whose activities will be affected by the rule. As discussed above, many consumer reporting agencies currently specialize in providing consumer reports for purposes that will not be affected by the rule. Additionally, consumer credit markets currently rely heavily on consumer reports from consumer reporting agencies which are not small entities.[362] For these reasons, the CFPB estimates that only a small fraction of the small consumer reporting agencies identified in Table 4 will be affected by the rule. The CFPB requested data to more precisely quantify the number of small consumer reporting agencies that will be affected by the rule, but did not receive relevant comments.

    Small creditors that will be directly affected by the rule are included in several NAICS categories that can be broadly divided into depository and non-depository institutions. Small creditors will be generally prohibited from considering medical information from consumer reports (and other sources) in credit eligibility determinations under the rule, unless a specific exception applies. However, some small creditors currently do not consider medical information that will be prohibited under the rule, and others only consider medical debt information if consumers disclose that they have ( print page 3349) made monthly payment arrangements with medical debt holders.[363]

    While all small creditors will be subject to the rule, the CFPB lacks the data to precisely quantify how many small creditors currently make credit decisions in ways that will be affected by the rule. Small creditors who are currently in compliance, whether in whole or in part, with the rule might not be impacted as much as small creditors who currently consider medical debt information (and certain other categories of medical information) from consumer reports or other sources. The CFPB requested data to precisely quantify the number of small creditors that may be directly affected by the rule, but did not receive relevant comments.

    Table 4—Estimated Number of Small Entities by Category 364
      NAICS code(s) Estimated number of small entities
    Consumer Reporting Agencies 561450 281
    Depository Institutions 522110, 522130, 522180, 522210 7,612
    Non-depository Institutions 522220, 522291, 522292, 522310, 522320, 522390 14,454
    Debt Collectors 522298, 541110, 561440 4,011
    Healthcare Providers 621111, 621112, 621210, 621310, 621320, 621330, 621340, 621391, 621399, 621410, 621420, 621491, 621492, 621493, 621498, 621511, 621512, 621610, 621910, 621999, 622110, 622210, 622310, 623110, 623210, 623220, 623311, 623312, 623990 521,895

    5. Projected Reporting, Recordkeeping, and Other Compliance Requirements of the Rule, Including an Estimate of the Classes of Small Entities Which Will Be Subject to the Requirement and the Type of Professional Skills Necessary for the Preparation of the Report

    The rule may impose reporting, recordkeeping, and other compliance requirements on small entities subject to the rule. These requirements generally differ for entities in two classes: credit bureaus that function as consumer reporting agencies, and depository or non-depository institutions that function as creditors. Based on Table 4, these requirements will be imposed on, at most, an estimated 281 small consumer reporting agencies and 22,006 small creditors. The CFPB does not expect that debt collectors and health care providers listed in Table 4 will have reporting, record keeping and other compliance requirements.

    Requirements for Consumer Reporting Agencies

    Under the rule, consumer reporting agencies will only be able to provide to creditors (in connection with credit eligibility determinations) consumer reports that contain medical debt information if they have reason to believe that the creditor intends to use the medical debt information in a manner that is not prohibited. Thus, if consumer reporting agencies continue to receive and record medical debt information from furnishers, consumer reporting agencies may need to devise policies and procedures to ensure that they appropriately restrict the provision of medical debt information to creditors. However, these compliance costs may only apply to consumer reporting agencies who, at baseline, provide consumer reports containing medical debt information to creditors based on the existing financial information exception. It is the CFPB's understanding that this task is mostly performed by the NCRAs (none of which are small entities), and the CFPB is not aware of any small consumer reporting agencies that provide consumer reports containing medical debt information to creditors at baseline. Compliance for affected small consumer reporting agencies will generally require professional skills related to software development, legal expertise, compliance, and customer support. The CFPB does not have the data to estimate the costs of reporting, recordkeeping, and other compliance requirements for small consumer reporting agencies, and requested but did not receive data to quantify these costs.

    Requirements for Creditors

    The rule will generally prohibit creditors from using information related to medical debt (among other categories of medical information) in credit eligibility decisions. The CFPB's final rule prohibits CRAs from furnishing medical debt information to creditors pulling a general report in order to underwrite a loan, which means small creditors would not have to incur the compliance costs associated with updating their underwriting procedures to exclude medical debt information. However, creditors using their own proprietary credit score may choose to change their underwriting procedures in response. Currently, many creditors use medical debt information from consumer reporting agencies that will no longer be included in consumer reports under the rule. The rule will not change any existing law or guidance regarding the information that creditors must request from applicants. Creditors may use (or continue to use) certain information, including information relating to medical debt, that consumers provide in response to questions in credit applications that do not specifically request medical information to satisfy ability-to-repay requirements. The rule may cause creditors to modify their underwriting procedures to rely more heavily on consumer information that they obtain from credit applications. These changes will generally require professional skills related to compliance, underwriting, and legal expertise. The CFPB requested data and evidence to estimate these costs, but did not receive relevant comments. ( print page 3350)

    Requirements for Debt Collectors

    One debt collector trade association commented that initial compliance costs will be at least $100,000 per debt collector. This estimate included costs such as updating software, hiring attorneys to ensure compliance with the rule, renegotiating contracts with vendors, and updating their business practices. However, the rule will not prohibit debt collectors from furnishing medical debt information to consumer reporting agencies or directly impose any other reporting, recordkeeping, or other compliance burdens on them. As discussed in part VII.E.4, the rule may make furnishing a less effective means of inducing payment of medical debts. This may reduce debt collectors' revenue; however, reductions in revenue will not be due to reporting, recordkeeping or compliance requirements that the rule imposes on debt collectors.

    Requirements for Health Care Providers

    The CFPB understands that at baseline, health care providers do not generally furnish medical debt information to consumer reporting agencies. But even if they do furnish debt collection information to consumer reporting agencies, as described above, the rule does not impose any requirements on furnishers, nor does the rule impose other requirements on health care providers. Accordingly, the CFPB has determined that the rule will not impose reporting, recordkeeping, and other compliance costs on health care providers. However, as discussed in part VII.E.4, the rule may make furnishing a less effective means of inducing payment of medical debts. This may impose costs on health care providers if they turn to other collection mechanisms that may be more costly or less effective than furnishing, such as debt collection litigation, or if the rule causes them to renegotiate contracts with medical debt buyers or debt collectors, but these costs will not be due to reporting, record keeping, or compliance requirements that the rule imposes on health care providers.

    6. Description of the Steps the Agency Has Taken To Minimize the Significant Economic Impact on Small Entities Consistent With the Stated Objectives of Applicable Statutes, Including a Statement of the Factual, Policy, and Legal Reasons for Selecting the Alternative Adopted in the Final Rule and Why Each One of the Other Significant Alternatives to the Rule Considered by the Agency Which Affect the Impact on Small Entities Was Rejected

    When developing the proposal, the CFPB decided to include a prohibition on consumer reporting agencies furnishing medical debt information to creditors who did not have a permissible purpose by virtue of the fact that this rule (or other laws) prohibit them from considering the information in credit underwriting. This provision was in large part proposed, and is now finalized, in order to minimize the economic impact on small entities. In the absence of such a prohibition, consumer reporting agencies might continue to include medical debt information on credit reports, in which case creditors would have to update underwriting models and credit scores to avoid giving it any weight. By prohibiting consumer reporting agencies from sending the data, creditors are able to forgo that substantial compliance cost.

    The CFPB considered exempting small entities from the rule, in whole or in part. Several commenters, including debt collectors, stated that the CFPB should consider limiting the scope of the rule to apply only to some forms of data, or to certain medical debts, such as those originating from emergency medical services. Another commenter stated that the CFPB should consider exempting small businesses below a certain size threshold. However, the CFPB has determined that such exemptions will not achieve the objective of FCRA section 604(g)(2) and the rule to protect consumer privacy with respect to sensitive medical information.

    The CFPB also considered several other alternatives to the rule that would possibly result in lower costs for small entities. These alternatives include: (1) alternative compliance timelines, (2) allowing creditors to consider specific types of medical information, (3) codifying and broadening the voluntary changes in medical collections reporting implemented by the NCRAs in 2022 and 2023, (4) requiring consumer reporting agencies to independently investigate the accuracy of furnished medical debt collections, and (5) defining when a furnisher must investigate the accuracy of furnished medical collections information.

    The CFPB considered making the rule effective more than 60 days after the issuance of a final rule. During the SBREFA process, several small creditors stated that they would need time to comply with the proposals discussed at the panel. One small creditor stated that their compliance department is already working at full capacity to comply with recently issued rules, and that they and others in the financial industry would need additional time to comply with further rules. A debt collector trade association stated in a comment that an implementation period of 60 days is too short for small businesses to comply with the rule, while Advocacy stated that stakeholders believe it will take 18 to 20 months to comply with the rule. The CFPB has determined that compliance with the rule would not impose significant compliance costs on small entities, and as a result the CFPB does not believe additional time for compliance is necessary. Further, allowing additional time for compliance would extend the period during which sensitive medical information may continue to be used for credit eligibility determinations.

    As described in the SBREFA Outline, the CFPB considered removing the financial information exception only with respect to medical information relating to debts, while continuing to allow creditors to consider medical information relating to expenses, assets, collateral, income, benefits, and the purpose of the loan. The CFPB has determined that a creditor's consideration of medical information relating to expenses, assets, and collateral is not warranted, and is therefore removing the financial information exception with respect to these additional categories of medical information.

    The final three alternatives considered may not achieve some of the objectives of the rule. These alternatives were included in the discussions with small entity representatives and the SBREFA Panel. As discussed in part VII.B, the NCRAs voluntarily implemented changes in the consumer reporting of medical debt. Because their changes were voluntary, codifying and broadening the changes may protect consumers from the possibility that NCRAs might choose to reverse their policies in the future. The last two alternatives would serve to increase the accuracy of medical collections information on credit reports. The CFPB has determined that these three alternatives would not achieve the objective of protecting consumer privacy with respect to sensitive medical information.

    After considering these significant alternatives, the CFPB declines to adopt them because none of the alternatives would achieve the objective of FCRA section 604(g)(2) to protect consumer privacy with respect to sensitive medical information, and thus are not appropriate methods for reducing the economic impact on small entities in the context of this rule. ( print page 3351)

    7. Description of the Steps the Agency Has Taken To Minimize Any Additional Cost of Credit for Small Entities

    Because the rule will only affect how small consumer reporting agencies report and small creditors obtain or use consumers' medical information, the CFPB does not expect that the rule will affect the business lending market. The CFPB concludes that the costs of credit for small creditors and small consumer reporting agencies will not be impacted by the rule. Commenters, including the small entity representatives cited by Advocacy in its comment, stated the possibility of credit score creep increasing underwriting standards more broadly. To the extent that this happens, the cost of credit may rise for small business owners who rely on personal credit. However, because the share of consumers with medical collections on their consumer reports is only 5 percent at baseline, the CFPB views this possibility as unlikely.

    IX. Paperwork Reduction Act

    The CFPB has determined that the final rule would not impose any new information collections or revise any existing recordkeeping, reporting, or disclosure requirements on covered entities or members of the public that would be collections of information requiring approval by the Office of Management and Budget under the Paperwork Reduction Act.[365] The existing information collections contained in Regulation V, which implements the FCRA, are approved by OMB under OMB Control Number 3170-0002 which currently has an expiration date of October 31, 2025.

    X. Congressional Review Act

    Pursuant to the Congressional Review Act (5 U.S.C. 801 et seq.), the CFPB will submit a report containing this rule and other required information to the U.S. Senate, the U.S. House of Representatives, and the Comptroller General of the United States at least 60 days prior to the rule's published effective date. The Office of Information and Regulatory Affairs has designated this rule as a “major rule” as defined by 5 U.S.C. 804(2).

    XI. Severability

    The CFPB intends that, if the consumer reporting agency prohibition on furnishing medical debt information finalized in § 1022.38 (or any provision or application of that section) is stayed or determined to be invalid, the amendments to § 1022.30 are severable and shall continue in effect. The CFPB also intends that if the amendments to § 1022.30 (or any provisions or applications of those amendments) were stayed or determined to be invalid, § 1022.38(b)(1) would not take (or continue in) effect, because it relies on the amendments to § 1022.30, but § 1022.38(b)(2) is severable and shall continue in effect. Furthermore, if the result of a stay or judicial determination is that creditors are generally able to obtain or use medical information in connection with determinations of consumers' eligibility, or continued eligibility, for credit, the CFPB intends the prior version of § 1022.30(d) to continue in effect.

    XII. Technical Appendix

    This appendix describes the technical details of the CFPB's analysis that aims to estimate how medical collection consumer reporting affects consumer access to credit, considering an “equilibrium” in which all medical collection tradelines are removed from consumer reports, as under the rule. The analysis also compares the performance of new credit accounts that can be traced to creditors' inquiries for consumers that have medical collections. The analysis exploits a change in consumer reporting practices that occurred in 2017 that has prevented medical collections that are less than 180 days past their date of first delinquency from appearing on consumer reports obtained from the nationwide consumer reporting agencies (NCRAs).[366] As a result of this change, when consumers applied for credit in the 180 days before a medical collection tradeline was added to their consumer report, they had an outstanding medical debt that was in collections, but creditors would not have seen evidence of those medical collections on consumer reports when making determinations about whether to extend credit to the consumers.[367]

    1. Data Used

    The data for this analysis are derived from the CFPB's Consumer Credit Information Panel (CCIP), a 1-in-50 de-identified nationally representative sample of credit records from one of the three NCRAs. The data include information on consumers' credit accounts, collections, public records, credit scores, and inquiries, which are creditor requests for consumer reports. Each credit account is described by a “tradeline,” which includes the account's product type, balance amount, initial credit limit or loan principal, date of origination, anonymized firm identifier, and delinquency status.[368] Collections are also described by tradelines, which include the collection's balance amount, the original creditor's industry classification, and the date that the collection was added to the consumer report. Each inquiry includes the product type for which the consumer applied and the date that the inquiry was made. The sample used in the analysis includes all inquiries made by creditors within 180 days of a medical collection tradeline's addition to a consumer report. In other words, the sample includes inquiries made within 180 days of the time each medical collection became visible to creditors.

    The CFPB created two datasets to estimate the effect of medical collection reporting on access to credit and credit account performance. The first dataset includes all inquiries made in the 180 days before and after each medical collection's addition to a consumer report (inquiry dataset). The second dataset includes the two-year performance of all credit account tradelines that can be traced back to an inquiry in the inquiry dataset (performance dataset).[369] Both datasets only include inquiries made and credit account tradelines opened in response to credit applications from consumers with medical collections. The analysis is limited to inquiries associated with medical collections first reported at least six months after the final implementation of the NCAP in September 2017, which ensured that all medical collections were identifiable as such and that all consumers with reported medical collections had a past-due medical bill for at least 180 days prior to the medical collection's ( print page 3352) appearance on their consumer report. [370] Given these constraints, the dataset includes inquiries associated with medical collections that were furnished to the NCRA that provides the CFPB's CCIP between March 2018 and July 2023.[371]

    Each dataset in the primary sample includes a subsample of inquiries and tradelines that were associated with medical collection tradelines having initial balances over $500 and that were made when any other medical collection tradelines on the consumer report had initial balances over $500. This specification is referred to as the “over-$500” sample and mimics the current reporting environment in which medical collections under $500 are not included on consumer reports.[372] The CFPB also created versions of the inquiry and performance datasets that do not make any restrictions on the dollar amount of medical collection tradelines and presents results for this “full sample” in parallel with those for the “over-$500 sample.”

    Creditors only observe the consumer reports of consumers that apply for credit, so the analysis is inherently limited to consumers that actively seek credit. The CFPB found in the proposed rule, and included in part VII.E.5, that there was a near-zero change in consumers' propensity to demand credit when medical collections under $500 were removed from consumer reports. The CFPB expects that the composition of consumers actively seeking credit will not be affected by the rule.

    When a consumer has multiple medical collection tradelines, the data contain duplicates of the inquiries and credit account tradelines if they occur within 180 days of different medical collection tradelines. For example, suppose a consumer has two medical collection tradelines that are first reported on May 1 and on September 1. Suppose a creditor makes an inquiry on August 1. This inquiry will appear in the inquiry dataset twice: once for the May 1 collection, and once for the September 1 collection. Inquiries and credit account tradelines are also duplicated when consumers have multiple medical collection tradelines reported on the same day.

    Three reporting changes occurred during the sample period that removed certain types of medical collections from consumer reports.[373] However, because the analysis exploits the date that a medical collection was added to a consumer report instead of the date it was removed from a consumer report, these changes do not undermine the general methodology of the analysis. The reporting changes do affect the types of medical collections that were on consumer reports when inquiries were made.[374] The CFPB first describes each of these three changes and their impact, before addressing the consequences for the analysis. First, all paid medical collections were removed from consumer reports in June 2022. Fewer than 2.5 percent of medical collections reported between January 2017 and March 2022 were ever marked as paid.[375] Second, medical collections that were between 180 days and 365 days past due were removed from consumer reports in June 2022, and the delay before medical collections could be added to consumer reports was permanently extended to one year. The CFPB does not have an estimate of how many medical collections were affected by this change, as the number of days that the medical debt is past due is not provided in the CCIP. Finally, all medical collections under $500 were removed from the NCRAs' consumer reports in April 2023. Combined, these reporting changes contributed to a large decline in the number of consumers with medical collection tradelines on their consumer report, from 14 percent of consumers in March 2022 to 5 percent of consumers in June 2023.[376]

    Because of these reporting changes for some inquiries that were made after a medical collection tradeline was first reported, the medical collection may not have been present on the consumer report by the date of the inquiry. For example, if a consumer had a medical collection with an initial balance less than $500 first reported in February 2023, and an inquiry in May 2023, the inquiry would be classified as occurring about three months after the collection but would not in fact have that collection tradeline included on the consumer report at the time of the inquiry. The CFPB expects this to attenuate the results, as inquiries made “with medical collection reporting” would have outcomes more similar to inquiries with the medical collection not yet reported. Medical collections reported before January 2022 would not have associated inquiries affected by any of these reporting changes.

    The analysis of the performance dataset is not affected by the recent reporting changes. Because the focus is on two-year performance, the performance analysis only included tradelines opened before January 2022, as they require sufficient time to measure two-year performance. Therefore, the performance regressions are not impacted by these medical collection removals.

    Commenters including a bank trade association commenter and a researcher stated that the time period considered in the proposal was not reflective of the ( print page 3353) current market because it was marked with instability in the medical debt collection environment, including pandemic-era changes and State policy changes. The CFPB acknowledges this limitation but finds it infeasible to study two-year delinquency risk without using accounts that were originated at least two years ago. Furthermore, because relatively few medical collections are included on consumer reports at baseline, the analysis needs to incorporate older data to have sufficient statistical power to identify statistically significant effects.[377] The CFPB expects that any differences in consumer behavior as a result of these changes, compared to the current baseline, may affect the magnitude of the results but not the direction. For example, if mortgage forbearance caused fewer consumers with medical collections to become delinquent on their mortgages, the estimated difference in mortgage performance between consumers with reported medical collections and consumers with unreported medical collections may be smaller than at the current baseline. However, there should be no difference in the coefficient's sign if consumers with unreported collections are more likely, in any time period, to be seriously delinquent than consumers with reported medical collections because creditors use medical collection information to avoid bad debt risks.

    Other commenters, including at least one researcher and at least one debt collector, stated that the analysis in the Technical Appendix to the proposed rule is subject to self-selection bias because only consumers actively seeking credit are included in the dataset. The inquiry and performance datasets are structured at the inquiry or credit account tradeline level, and not at the consumer or medical collection level. This means the analysis can be interpreted as modeling credit decisions and outcomes from creditors' perspective, rather than modeling the decisions of consumers or debt collectors.

    Commenters, including at least one debt collector, health care provider, researcher, and individual, stated that the results of the Technical Appendix were skewed or too narrow because they were limited to medical collections with initial balances over $500. As described above, the CFPB also presents results for the full sample, regardless of medical collection balance amount. The results from this sample are similar to those in the primary sample, as described below.

    Multiple researcher commenters stated that the results of the CFPB's analysis could not be validated or fully evaluated with the information included in the Notice of Proposed Rulemaking. Releasing the data would be a violation of the CFPB's contract with the NCRA that provides its CCIP, however, and courts have held that an agency can rely on confidential information in its rulemaking so long as the agency discloses information to allow interested parties to comment on the methodology and general data.[378] Here, the CFPB discussed its data set, provided information about its methodologies, and invited interested parties to comment. The CFPB considered the comments that addressed the analysis and has determined that the available evidence supports the choices made in the final rule. While some commenters also suggested that the CFPB erred in not obtaining peer review of its analysis, they did not articulate why peer review would be required in this rulemaking.

    2. Construction of the Inquiry Dataset

    Because inquiries in the dataset are made in the 180 days before and after a medical collection is reported, the inquiries in the dataset occurred between September 2017 and January 2024. The dataset includes the number and type of medical and nonmedical collection tradelines that were included on the consumer report at the time each inquiry was made.

    Identifying unique medical collections over time in the CCIP may be imprecise; the CFPB assumes that unique medical collections are characterized by their dollar amounts, dates of medical collection account opening (usually the date the medical collection was assigned to the debt collector or other furnisher), and dates of the account's addition to the consumer report. Medical collections are rarely consistently reported for the full seven-year period for reporting adverse information permitted by the Fair Credit Reporting Act.[379] This poses challenges in tracking the same medical debt over time, as debts can disappear and reappear. Medical debts in collections are often transferred between debt collectors ( e.g., reassigned to a different collector by the health care provider or sold to a debt buyer), and when this happens the dates and dollar amounts associated with the medical collection tradelines may change, making it difficult to link these records. While these may be experienced as unique collections by the consumer as a new debt collector attempts to make contact, they may not be representative of the number of unique medical debts that each consumer has, as many of the debts are reflected by multiple subsequent collections.[380]

    ( print page 3354)

    The inquiry dataset is used to estimate the impact of medical collection reporting on consumers' access to credit, as measured by inquiry success. The CFPB classifies an inquiry as “successful” if the inquiry leads to an open tradeline. This definition of “success” does not necessarily mean that the specific credit application that generated the inquiry was being approved. The CFPB cannot directly observe whether the specific credit application that generated the inquiry in question was approved, and it is challenging to infer approval for a specific inquiry for several reasons. First, the CCIP does not include inquiries made to other NCRAs, and creditors do not always make inquiries to all three NCRAs. The CCIP therefore includes credit account tradelines that cannot be matched to an inquiry. These tradelines cannot be included in the CFPB's analysis because the empirical strategy requires that one know the date of each tradeline's associated inquiry. Second, the CCIP does not include creditor names, but instead has an anonymized company identifier; however, a particular creditor often has a different identifier for inquiries and for opened credit account tradelines. Thus, even if the consumer opened a tradeline with the same creditor that pulled their consumer report, it may not be identifiable as such in the data. Therefore, the CFPB cannot be certain that the observed inquiry is associated with a specific opened tradeline. The CFPB instead follows approaches used in academic research and the CFPB's Consumer Credit Trends credit tightness series and assumes that a credit account is associated with an inquiry if it is opened within a certain number of days after the observed inquiry and is of the same credit account type.[381] The number of days varies for different account types because of differences in the typical length of time between an account application and origination.[382] Finally, when consumers shop for credit, multiple inquiries may be made in a narrow window of time, even though the consumer only intends to open one account. The CFPB assumes that multiple inquiries for one consumer within a certain shopping window indicate the consumer's shopping behavior, and therefore only the last of these inquiries is included in the datasets, where each credit account type's window length is equivalent to its maximum time-to-origination.[383] For example, if a consumer had inquiries from mortgage lenders on April 1 and May 1, these would be treated as one observation, dated May 1, and it would be counted as a successful inquiry if a mortgage account was opened by August 29.

    A researcher commenter restated the limitations described above, which were also described in the proposal, but characterized this discussion as indicating that the CFPB did not have a “clean standard” to identify inquiry success in the Technical Appendix to the proposed rule. As described above and in the rule, the CFPB's construction of inquiry success is the best available measure and has been used in academic research and the CFPB's policy research.

    3. Construction of the Performance Dataset

    The performance dataset includes all originated credit account tradelines that are associated with successful inquiries in the inquiry dataset. The match between credit account tradelines and inquiries is one-to-one: each tradeline is matched to one inquiry, and each inquiry is matched to, at most, one tradeline.[384] The CFPB calculated the two-year performance for each originated credit account tradeline, with performance success measured by whether the tradeline was ever 90 or more days delinquent (seriously delinquent) within the first two years of its origination date.[385] Because the CFPB focuses on two-year performance, credit account tradelines opened after January 2022 are not included in the analysis as the CFPB cannot observe a full two years after origination. The CFPB was able to identify the two-year performance of over 94 percent of the credit account tradelines opened before January 2022. The exceptions are accounts that stopped being reported by the furnisher before the end of two years.

    4. Inquiry Summary Statistics

    Table 5—Inquiry Summary Statistics 386
      (1) Credit cards (2) Mortgages (3) Other inq. type
    Panel A: Unsuccessful, Over $500 Sample:
    Shopping window (days) 0.47 16.87 0.89
    No. open mortgages 0.03 0.11 0.04
    No. open credit cards 0.73 1.18 0.68
    No. open other trades 0.61 0.82 0.64
    Any D90+ trades 0.30 0.29 0.29
    Credit score 563.89 613.81 566.76
    Obs. (Unique Inquiries) 259,532 44,524 218,127
    Panel B: Successful, Over $500 Sample:
    Shopping window (days) 1.00 42.74 1.11
    No. open mortgages 0.07 0.23 0.07
    No. open credit cards 1.36 1.85 1.11
    ( print page 3355)
    No. open other trades 0.71 0.99 1.08
    Any D90+ delinquent trades 0.26 0.20 0.29
    Credit score 624.44 673.12 602.45
    Credit amount 1,645.96 244,846.31 5,374.88
    Two-year D90+ 0.21 0.03 0.25
    Past due amount 145.19 304.43 661.84
    Obs. (Unique Inquiries) 117,147 11,188 13,160
    Panel C: Unsuccessful, Full Sample:
    Shopping window (days) 0.46 16.09 0.86
    No. open mortgages 0.03 0.12 0.04
    No. open credit cards 0.69 1.15 0.64
    No. open other trades 0.56 0.80 0.60
    Any D90+ trades 0.30 0.30 0.30
    Credit score 562.12 607.76 563.39
    Obs. (Unique Inquiries) 892,295 171,704 761,275
    Panel D: Successful, Full Sample:
    Shopping window (days) 0.97 40.69 1.06
    No. open mortgages 0.08 0.26 0.06
    No. open credit cards 1.32 1.84 0.98
    No. open other trades 0.70 0.96 1.04
    Any D90+ trades 0.27 0.20 0.30
    Credit score 621.08 670.13 597.12
    Credit amount 1,582.59 238,199.13 5,597.18
    Two-year D90+ 0.20 0.03 0.23
    Past due amount 125.17 201.84 598.32
    Obs. (Unique Inquiries) 409,209 42,138 52,669

    Table 5 provides summary statistics for the unique inquiries in the data. The summary statistics are provided separately for “unsuccessful” inquiries that do not result in originated credit account tradelines, which are provided in Panels A and C, and for “successful” inquiries that can be associated to originated tradelines, which are provided in Panels B and D. Panels A and B are limited to the over-$500 sample, while Panels C and D provide summary statistics for the full sample. Table 5 shows that successful inquiries are associated with stronger credit profiles for every inquiry type and for both considered samples. The average successful credit applicant has more open pre-existing credit account tradelines, fewer seriously delinquent pre-existing credit account tradelines, and a higher credit score in the month or quarter before inquiry was made than the average unsuccessful credit applicant.[387] The table also shows that successful credit applicants shop for longer than unsuccessful credit applicants in the sample. Panels B and D further include the average characteristics of credit accounts opened in response to successful inquiries, measuring the credit limit at time of origination, the past due amount, and serious delinquency status two years after origination, showing that credit cards are much more likely than mortgages to be seriously delinquent within two years from opening, perhaps in part because credit cards are unsecured. However, the average past due amount is lower for credit cards, perhaps because average credit card monthly minimum payments are much lower than mortgage monthly payment amounts.

    5. Consumer Summary Statistics

    ( print page 3356)

    Table 6—Consumer Summary Statistics 388
      (1) Mean (2) Median (3) Obs. (unique consumers)
    Panel A: Over $500 Sample:
    No. medical collections 2.24 1.00 266,147
    Months between date of last med. coll. and date of first med. coll 20.47 0.00 266,147
    No. credit card inquiries 1.42 1.00 266,147
    No. mortgage inquiries 0.21 0.00 266,147
    No. other inquiries 1.11 1.00 266,147
    Credit score at first inquiry 594.52 588.00 214,485
    Missing credit score at first inquiry 0.19 0.00 266,147
    Consumer age at first inquiry 40.29 38.00 261,488
    Northeastern share at first inquiry 0.08 0.00 266,147
    Midwestern share at first inquiry 0.15 0.00 266,147
    Southern share at first inquiry 0.61 1.00 266,147
    Western share at first inquiry 0.14 0.00 266,147
    Panel B: Full sample:
    No. medical collections 4.08 2.00 688,682
    Months between date of last med. coll. and date of first med. coll 35.77 10.92 688,682
    No. credit card inquiries 1.89 1.00 688,682
    No. mortgage inquiries 0.31 0.00 688,682
    No. other inquiries 1.52 1.00 688,682
    Credit score at first inquiry 596.10 590.00 558,362
    Missing credit score at first inquiry 0.19 0.00 688,682
    Consumer age at first inquiry 41.89 40.00 676,075
    Northeastern share at first inquiry 0.10 0.00 688,682
    Midwestern share at first inquiry 0.19 0.00 688,682
    Southern share at first inquiry 0.54 1.00 688,682
    Western share at first inquiry 0.16 0.00 688,682

    Table 6 provides summary statistics at the consumer level, using the first observation for each consumer observed in the inquiry dataset. On average, a consumer in the over-$500 sample experiences 2.24 medical collections that appear within 180 days of an inquiry. These medical collections are, on average, approximately 20 months apart from the earliest to the latest reported. Nineteen percent of the consumers in the sample do not have a credit score in the month before their first inclusion in the sample; for consumers who do have a credit score, it is most often subprime.[389] More than 60 percent of consumers in the sample are located in Southern States, reflecting the disproportionate share of consumers with medical debt in the South documented in prior research.[390] These summary statistics support the generalizability of the results, as the sample of consumers is generally similar to the overall population of consumers with medical collections during this time period.[391]

    One debt collector commenter stated that the CFPB should have instead considered all inquiries associated with medical collections over $500 instead of making the restriction, in the proposed rule, that these inquiries are not made when a medical collection under $500 is included on the consumer report. The CFPB chose not to change its construction of the over-$500 subsample because the relevant question is how inquiries would be evaluated under the rule, relative to the baseline, in which no medical collections under $500 are included on consumer reports. The over-$500 sample, as initially constructed, is the closest approximation for estimating the effects of the rule. Results are included for the full sample to show that the estimated effects are broadly similar for all inquiries associated with medical collections of any size. Furthermore, the summary statistics for consumers in the full sample are similar to those for the over-$500 sample, but consumers in the over-$500 have nearly two fewer medical collections reported within 180 days of an inquiry in the sample. Though this at first may seem counterintuitive, this is because consumers with several medical collections often have at least one medical collection under $500 which removes them from the over-$500 subsample. ( print page 3357)

    6. Empirical Strategy

    The CFPB used a regression discontinuity in time (RDiT) design to estimate the effect of reported medical collections on consumers' access to credit and the performance of credit account tradelines resulting from creditors' inquiries. Regression discontinuity is a quasi-experimental design that, under certain assumptions, allows estimation of the causal effect of a treatment or intervention where a treatment is assigned by a threshold value of that variable.[392] In the present context, inquiries are “treated” when a medical collection tradeline is added to the NCRA's database. The date that a medical collection is added to a consumer report is the “threshold” that potentially creates a discontinuous effect on the studied dependent variables: inquiry success and two-year serious delinquency. Before this date, creditors cannot observe the medical collection on the consumer report at the time an inquiry is made, but the CFPB can observe using the CCIP that the consumer did have a medical debt in collections that would eventually be reported. The proximity of each inquiry to the threshold, referred to as the “running variable” in regression discontinuity terminology, is equal to the number of days between the date that the collection was first included on the consumer report and the date that the inquiry was made. When the inquiry date occurred after the medical collection reported date (or in other words, the medical collection was included on the consumer report before the inquiry was made), this running variable is greater than or equal to the “threshold” zero; for values less than or equal to zero, the medical collection was not included on the consumer report when the inquiry was made. The key assumption of a regression discontinuity analysis is that nothing is changing discontinuously across the threshold besides the treatment.

    To analyze inquiry success, the CFPB estimated Equation 1 using the inquiry dataset:

    Yijk = α + γDijk + βZijk + δDijk × Zijk + εijk (1)

    Where i is a consumer, j is an inquiry, and k is the medical collection associated with the inquiry. Yijk is a binary variable equal to one if the inquiry is successful, i.e., if a tradeline is originated within 14 days for a credit card or auto loan, 120 days for a mortgage, or 30 days for other loans. Dijk is the running variable, i.e., the number of days after medical collection k was added to the consumer report that inquiry j was made. Dijk is negative if the inquiry was made before the medical collection was added, and positive if the inquiry was made after. Zijk is a binary variable equal to one if the inquiry j was made after the date when collection k was reported. The coefficient of interest, β, represents the difference in the likelihood that an inquiry is successful for inquiries made after a medical collection is added, relative to inquiries made before. The intercept α allows estimation of a more flexible linear form.

    The CFPB also estimated Equation 1 for the performance dataset, using the two-year performance of tradelines that can be traced to an inquiry included in the inquiry dataset as the dependent variable. The estimating equation is largely unchanged, though j is interpreted as a tradeline associated with an inquiry in the inquiry dataset (rather than the inquiry itself), and Yijk is a binary variable equal to one if the account is at least 90 days delinquent on the tradeline at any point within the first two years after the tradeline is originated (rather than if the inquiry is associated with a tradeline origination, as in the inquiry dataset regression).

    In the results described below, the CFPB estimated six specifications to estimate impacts on inquiry success and account performance. The first specification is limited to the over-$500 sample, as defined above. The second and third specifications separate the over-$500 sample into two groups: inquiries that were made when the consumer had no nonmedical collections on their consumer report, and inquiries made when consumers had nonmedical collections on their consumer report. These specifications test whether reported medical collections affect inquiry success and better predict account performance for consumers with fewer other signals of negative information. The hypothesis is that the effects of a reported medical collection should be larger for inquiries made without nonmedical collections on the consumer report. If a consumer already has nonmedical collections, the appearance of a medical collection likely implies a smaller marginal change in expected delinquency risk. Finally, the CFPB then estimated each of these three specifications for all inquiries in the sample.

    The CFPB only reports its estimates of the parameter b , which provides the effect of medical collection furnishing on inquiry success and account performance. Combined across the main results and balance tests described later, the CFPB estimated a total of 192 b coefficients, so the reported standard errors were adjusted using the Benjamini-Hochberg procedure, a method for accounting for multiple comparisons (under which it is more likely to find a statistically significant result by chance than in a one-off analysis).[393]

    To justify the robustness of the main specification, the CFPB considers the potential threats to identification that can arise from RDiT specifications. RDiT varies from a standard regression discontinuity design because the running variable is not generally continuous. As summarized by an academic paper, RDiT designs can be biased if observations far from the threshold time period are used for identification, as there may be autoregressive properties or unobservable confounders.[394] This is often required in RDiT designs that have little cross-sectional variation, as the sample size can only grow by adding observations further from the threshold, rather than by adding additional cross-sectional units. A researcher commenter cited this concern in their critique of the CFPB's analysis. However, the data underlying the analysis discussed in this document contains ample cross-sectional variation, with 663,678 unique inquiries in the inquiry dataset and 401,027 unique tradelines in the performance dataset for the over-$500 sample. Furthermore, the analysis considers observations that are no more than 180 days from the threshold, minimizing the extent of possible autoregression.

    In addition to these features of the datasets that limit the potential for bias arising from the RDiT design, the CFPB estimates the regressions using econometric best practices as implemented by a practitioner software package.[395] Standard errors are ( print page 3358) clustered by consumer to account for correlation within consumer observations over time. Additionally, the CFPB conducted several robustness checks to support the validity of the main design, described in detail after the discussion of the main results.

    A researcher commenter stated that a consumer may take steps to improve their credit profile near the threshold time period, introducing bias into the model if the effects of these changes are erroneously attributed to the medical collection report. The CFPB finds it implausible that a consumer would choose to improve markers of their financial wellbeing over the short amount of time near the appearance of a medical collection on their consumer report. It estimates balance tests to test for this phenomenon in Tables 9 and 10 and finds no supporting evidence. Even if a consumer did improve their credit profile near the date that the medical collection is added to their consumer report, this would only attenuate results, as consumers with reported medical collections would look like better risks than they would absent this behavior. This would shrink the difference, from a creditor's perspective, between consumers with reported and unreported medical collections.

    One researcher commenter stated that the CFPB should not have included the Dijk × Zijk term in its regression equation because it is highly correlated with other variables in the regression equation, leading to multicollinearity bias. In fact this term, which is standard in RDiT equations, does not lead to multicollinearity bias and instead increases the precision of the estimated parameters. The term allows the relationship between the running variable and the outcome variable to change across the reporting threshold. Because some medical collections appear on a consumer report for fewer than 180 days, the slope between the running variable Dijk and inquiry success Yijk may be positive for positive values of Dijk because inquiries made farther from the medical collection report date are less likely to occur when the medical collection appears on the consumer report, likely leading to a greater likelihood of inquiry success. There is no similar expectation of a positive relationship in the 180 days before the medical collection is reported, i.e., for negative values of Dijk. The estimated parameter γ would conflate these two relationships if the interaction term is omitted from the regression equation.

    A researcher commenter stated that the CFPB should have considered more heterogeneity between groups in the Technical Appendix of the proposed rule, such as a consumer's age, their number of medical collections, and whether their medical collection has been disputed. While specific effects on these groups may be of general interest, the rule is not limited to certain subpopulations or types of medical collections, so the parameter of primary interest to the CFPB is the average effect taken over the entire population that has medical collections over $500, which mimics the current reporting environment.

    One researcher commenter stated that the CFPB did not provide measures that can be used to assess model quality, primarily concerning a hypothetical in which consumers far from the regression discontinuity threshold receive too much weight in the analysis. The CFPB included just 180 days before and after the threshold to mitigate this concern, as well as using econometric best practices in its regression equation as described above. The commenter did not describe specific, actionable examples of the measures that would assuage their concern.

    One NCRA commenter stated that the CFPB should have studied differences between consumers with medical collections and consumers without medical collections in the Technical Appendix of the proposed rule instead of limiting its focus to consumers with reported and unreported medical collections. The commenter stated that it was important to distinguish between consumers with and without medical collections because these groups have different payment performance.

    The CFPB does not agree that a comparison between all consumers with medical collections and all consumers without medical collections is relevant to understanding the impacts of the rule. Although consumers with medical collections may have a different delinquency risk than consumers without medical collections, the rule will not change which consumers have outstanding medical collections. The rule instead changes whether medical collections appear on a consumer report that a creditor receives for the purpose of a credit eligibility determination. The analysis discussed in this part considers whether creditors use medical collection information that appears on a consumer report to deny consumers with medical collections access to credit and limit their delinquency risk. This provides the closest understanding of the environment that would be created by the rule: consumers with reported medical collections are like the baseline while consumers with unreported medical collections are like the post-rule environment, and the CFPB's analysis compares them.

    A researcher commenter stated that the CFPB should have used propensity score matching instead of a RDiT approach in the Technical Appendix of the proposed rule. The commenter suggested a design that would compare consumers with “hidden” medical debts, or consumers in the 180 days before their medical collection is added to their consumer report, to similar consumers without medical debt on their consumer reports.

    The CFPB does not agree that a propensity score matching approach as suggested by the commenter would be appropriate. The CFPB could control for information included on consumer reports, but not unobservable variables like outstanding medical debt, as most medical debt is not included on consumer reports. Therefore, the consumers with hidden medical debt would be compared to consumers who, for the most part, do not have medical debt. Differences between these groups would not be related to the inclusion of a medical collection on consumer reports but would instead be driven by the presence of medical debt. This analysis would not be as relevant to the rule as the CFPB's analysis.

    7. Results on Inquiry Success

    The CFPB first uses the inquiry dataset to consider how medical collection reporting affects inquiry success. Importantly, an unsuccessful inquiry does not necessarily imply that the lender denied the credit application. Consumers may be approved for credit with worse terms than they would have received absent medical collection reporting and decline the offer of credit as a result, or consumers may choose not to take up approved credit for idiosyncratic reasons. The CCIP does not include data on the terms of originated accounts or on credit approvals that do not lead to originated accounts. However, this is less likely to be an issue with credit cards because the CFPB understands that credit card accounts are generally issued automatically if the creditor approves an application, with little opportunity for a consumer to decline. The CFPB assumes that consumers' underlying demand for credit is unaffected by medical collection reporting, so changes in inquiry success across the reporting threshold can be attributed to creditors' denial of credit account applications or provision of worse terms, rather than ( print page 3359) changes in who applies. The CFPB justifies this assumption below.

    Table 7—The Effect of Medical Collection Reporting on Inquiry Success 396
      (1) Over $500 (2) Over $500, no NMC (3) Over $500, NMC (4) All (5) No NMC (6) NMC
    Panel A: Credit cards:
    RD Estimate ***−0.047 ***−0.072 ***−0.029 ***−0.033 ***−0.049 ***−0.022
      (0.006) (0.009) (0.006) (0.003) (0.005) (0.003)
      [−0.059, −0.036] [−0.090, −0.055] [−0.041, −0.018] [−0.038, −0.027] [−0.059, −0.040] [−0.028, −0.017]
    Avg. success 0.294 0.381 0.222 0.275 0.364 0.214
    Observations 601,230 267,276 333,954 3,026,355 1,233,571 1,792,784
    Panel B: Mortgages:
    RD Estimate *−0.026 *−0.040 −0.003 −0.014 −0.013 −0.005
      (0.012) (0.018) (0.012) (0.009) (0.015) (0.006)
      [−0.049, −0.004] [−0.074, −0.006] [−0.027, 0.022] [−0.031, 0.004] [−0.043, 0.017] [−0.016, 0.006]
    Avg. success 0.186 0.248 0.098 0.167 0.235 0.089
    Observations 79,372 46,003 33,369 439,685 237,413 202,272
    Panel C: Other credit accounts:
    RD Estimate *−0.014 *−0.020 −0.010 ***−0.015 ***−0.024 **−0.010
      (0.006) (0.009) (0.007) (0.003) (0.005) (0.004)
      [−0.026, −0.003] [−0.038, −0.002] [−0.024, 0.004] [−0.021, −0.009] [−0.033, −0.015] [−0.017, −0.003]
    Avg. success 0.242 0.307 0.197 0.246 0.316 0.205
    Observations 469,290 190,942 278,348 2,484,030 908,849 1,575,181
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    * p < 0.1, ** p < 0.05, *** p < 0.01.

    Table 7 provides the results of the main regression discontinuity analysis on inquiry success. Each panel represents a different loan type, as products generally have different underwriting procedures. At a high level, several summary observations can be made. First, just over half of the inquiries in the full sample of the inquiry dataset are for credit cards. Only 7.4 percent of the inquiries in this sample are for mortgages, compared to almost 17 percent of all inquiries in the CCIP. This likely reflects the fact that most consumers in the sample have thin credit files [397] and subprime credit scores, and therefore may be less likely to apply for mortgages than for other types of credit, given the higher underwriting standards of mortgages.[398] Inquiry success rates are higher for all loan types when inquiries are made without nonmedical collection tradelines on the consumer report than when nonmedical collection tradelines are present, with differences as large as 15.9 percentage points. This is expected because consumers with less negative information on their consumer reports are more likely to be approved for credit or receive favorable terms. Perhaps less intuitively, average success rates for credit cards and mortgages are also generally higher for the subsample of inquiries made by consumers who only have medical collection tradelines over $500, if they have any. As discussed above, inquiries made by consumers with many medical collection tradelines are often excluded from the over-$500 sample because at least one of those medical collection tradelines is under $500. The average number of medical collection tradelines on a consumer report when an inquiry is made in the full sample, in Column 4, across all loan types, is 5.03. Conversely, the average number of medical collection tradelines on a consumer report when an inquiry is made, for inquiries made with all medical collection tradelines greater than $500, in Column 1 is 1.08. Thus, the over-$500 sample is positively selected, i.e., consumers in this sample have less negative information than consumers in the full sample, at least as measured by the number of medical collection tradelines present on their consumer reports. Despite the positive selection into the over-$500 sample, the CFPB expects these results to most closely represent the effects of removing all medical collection tradelines from consumer reports given the parallel with the NCRAs' current practice for under-$500 medical collection tradelines.

    Turning to the regression estimates in Table 7, Column 1 of Panel A (credit cards) shows that a medical collection being reported causes a 4.7 percentage point decline in the likelihood of inquiry success for the over-$500 sample. This represents a 16.0 percent decline from relative to the average success rate for inquiries to the left of the regression discontinuity threshold ( i.e., inquiries made before the medical collection was reported). The effect is larger in absolute value for inquiries made when the consumer had no nonmedical collection tradelines on their consumer report, shown in Column 2, than when consumers had nonmedical collection tradelines on their consumer report, shown in Column 3. This supports the hypothesis that medical collection reporting has a larger effect on consumers without outstanding nonmedical collections. Columns 4 through 6 repeat the groups from Columns 1 through 3 but include the full sample. The regression result shown in Column 4 of Panel A describes a 3.3 percentage point, or 12.0 percent, ( print page 3360) decline in inquiry success for inquiries made with these larger medical collections reported relative to inquiries made without these medical collections reported. Again, effects are larger in absolute value for inquiries made when consumers did not have nonmedical collection tradelines on their consumer report than when nonmedical collection tradelines were present.

    The first three Columns of Panel B (mortgages) find relatively small and no more than marginally significant effects of medical collection reporting on mortgage inquiry success. Medical collection reporting reduces mortgage inquiry success by 2.6 percentage points, or 14.0 percent of its baseline level. The effect appears to be driven by inquiries made when there were no nonmedical collection tradelines on the consumer report, as the coefficient in Column 3 is statistically insignificant and small. However, the estimates in Columns 1 and 2 are only statistically significant at the 10 percent level.[399] All estimates for the full sample in Columns 4 through 6 are statistically insignificant. Using the 95 percent confidence interval for the coefficient in Column 4 of Panel B, it is possible to reject effects larger than a 3.1 percentage point, or 18.6 percent, decline in inquiry success for the full sample.[400]

    Panel C provides results for all other types of credit accounts. The estimated effects are all smaller in magnitude than the results for credit cards and vary in statistical significance. The coefficients imply that medical collection reporting causes a 1.4 percentage point decline in the likelihood of inquiry success for non-mortgage and non-credit-card credit accounts for the over-$500 sample, or a 5.8 percent decline from the baseline inquiry success rate. Estimated effects are similar for the full sample. As with the effects on credit cards and mortgage inquiries, effects for both samples are larger for consumers without nonmedical collection tradelines.

    8. Results on Account Performance

    The estimated effects on inquiry success show that the underwriting procedures for many credit types penalize consumers for having medical collection tradelines on their consumer reports, with generally larger effects for consumers with medical collection tradelines over $500. The CFPB next considered whether this use of medical collection tradelines protects creditors from delinquency risk. If creditors use medical collection information to accurately predict whether consumers have high delinquency risk and deny their applications, then originated accounts resulting from a successful inquiry for a consumer with an unreported medical collection at the time of the inquiry would be more likely to be seriously delinquent than those resulting from a successful inquiry for a consumer with a reported medical collection. However, to the extent that creditors provide worse credit terms to consumers with reported medical collections and such worse credit terms increase the likelihood of serious delinquency, one might expect the opposite: Originated accounts resulting from an inquiry for a consumer with an unreported medical collection could be less likely to be seriously delinquent (because they received more affordable credit terms) than those resulting from an inquiry for a consumer with a reported medical collection (because they received worse credit terms). These opposing effects make it impossible to determine how the underlying delinquency risk of consumers with and without unreported medical collections varies. However, the results of this analysis are still informative as to how two-year delinquency rates are affected by medical collection reporting, net of the effects of application denials and the provision of worse terms.

    Table 8—The Effect of Medical Collection Reporting on Two-Year Credit Account Performance 401
      (1) Over $500 (2) Over $500, no NMC (3) Over $500, NMC (4) All (5) No NMC (6) NMC
    Panel A: Credit cards:
    RD Estimate −0.000 0.002 −0.003 0.002 0.004 −0.005
      (0.012) (0.014) (0.021) (0.006) (0.007) (0.008)
      [−0.023, 0.023] [−0.026, 0.031] [−0.045, 0.038] [−0.009, 0.013] [−0.010, 0.018] [−0.021, 0.011]
    Avg. D90+ 0.231 0.190 0.293 0.223 0.171 0.284
    Observations 96,297 56,423 39,874 565,680 305,980 259,700
    Panel B: Mortgages:
    RD Estimate −0.011 −0.021 0.033 0.004 −0.006 0.034
      (0.014) (0.014) (0.034) (0.007) (0.006) (0.019)
      [−0.039, 0.017] [−0.049, 0.007] [−0.033, 0.100] [−0.009, 0.017] [−0.018, 0.007] [−0.003, 0.071]
    Avg. D90+ 0.035 0.025 0.069 0.038 0.029 0.065
    Observations 10,177 7,944 2,233 56,976 43,106 13,870
    Panel C: Other credit accounts:
    RD Estimate −0.012 −0.011 −0.009 −0.001 −0.002 −0.002
      (0.014) (0.015) (0.021) (0.006) (0.006) (0.009)
      [−0.040, 0.015] [−0.041, 0.019] [−0.050, 0.033] [−0.012, 0.011] [−0.014, 0.011] [−0.019, 0.016]
    Avg. D90+ 0.182 0.135 0.235 0.171 0.120 0.216
    Observations 71,760 36,951 34,809 459,094 213,481 245,613
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    * p < 0.1, ** p < 0.05, *** p < 0.01.
    ( print page 3361)

    Table 8 shows the results of the main regression discontinuity analysis in the performance dataset. Across all loan types and subsamples, the estimated effects of medical collection reporting on serious delinquency are small and statistically insignificant. Column 1 of Panel A shows that, in the over-$500 sample, the CFPB can reject effects larger in absolute value than 2.3 percentage points, or 10.0 percent of the baseline delinquency rate, with 95 percent confidence. That is, it would be highly unlikely to find an estimate as small as what is reported in Table 8 through chance alone if having an unreported medical collection was associated with an increase in the rate of serious delinquency by 10 percent or more. The confidence interval is tighter and the central estimate more positive ( i.e., unreported medical collections associated with less delinquency) for inquiries made when consumers did not have nonmedical collection tradelines on their consumer report than when these collection tradelines were present. This means that the true effects for inquiries made without nonmedical collection tradelines are more likely to be positive. Further, if there is a difference in delinquency rate for consumers with unreported medical collections, these consumers are less likely to be delinquent than consumers with reported medical collections. This also holds for the full subsample in Columns 4 through 6.

    These results broadly find that credit card lenders use medical collection information in underwriting, but do not reduce their two-year serious delinquency risk for originated credit account tradelines by doing so. Fewer accounts are originated to consumers with reported medical collections, but those that are originated are no less likely to be delinquent than accounts originated to consumers with unreported medical collections. This suggests that removing medical collections information from credit card underwriting would increase access to credit without negatively impacting the likelihood of serious delinquency for consumers with medical collections, all else equal.

    The results in Panel B show qualitatively similar estimates for mortgages, but with less precisely estimated effects. The effects are less precise because the average serious delinquency rate is much lower for mortgages than for credit cards: only 3.5 percent of mortgages in the over-$500 sample are seriously delinquent within two years, compared to 23.1 percent of credit cards. The lower frequency in the dependent variable as well as the smaller sample size will naturally lead to wider confidence intervals. Column 1 shows that the CFPB can only reject marginal reductions in mortgage delinquency rates with reported medical collections that are larger in absolute value than 3.9 percentage points, or 111.4 percent of the baseline delinquency rate, with 95 percent confidence. For the full sample, the CFPB can reject marginal reductions larger in absolute value than 0.9 percentage points, or 23.7 percent of baseline delinquency rate. Though these results are too imprecise to allow the rejection of large effects, their statistical insignificance can be interpreted as suggestive that removing larger medical collection tradelines from mortgage underwriting would not cause increases in serious delinquency risk.

    As for credit cards, the results for non-mortgage and non-credit-card accounts, shown in Table 8, are mostly statistically insignificant and small in magnitude. Again, the CFPB concludes that the use of medical collections information in underwriting does not reduce the delinquency risk of accounts originated to people with reported medical collections.

    These results suggest that, absent consumer reporting of medical collections, the additional credit accounts that creditors provide to consumers whose medical collections would no longer be reported would be no more likely to be delinquent than the credit accounts creditors provide at baseline. In line with economic theory, the CFPB expects that creditors only provide credit if the account's expected profit is positive. Under this expectation, creditors would not provide accounts to consumers with unreported medical collections at baseline if they were not profitable. However, it is possible that creditors currently provide those accounts not because they are profitable, but because they have no other mechanism for identifying and either denying the credit applications of, or changing the terms provided to, applicants with unreported medical collections. In this case, the rule would reduce profit for creditors by increasing the number of unprofitable loans in their portfolio.

    The CFPB illustrates this concern with a simple example. Suppose that a creditor's applicant pool is equally divided across three nonoverlapping groups of consumers, which are identical in all attributes except for the presence of collections and delinquency risk. Assume that applicants with no collections have a delinquency risk of 1 percent and applicants with medical collections have a delinquency risk of 1.25 percent. Suppose for simplicity that a lender seeks to minimize their delinquency risk and is unwilling to provide loans if the expected delinquency rate is 1.2 percent or higher. If half of consumers with medical collections (or one-sixth of the total population) have those medical collections included on their consumer report at baseline, the lender provides loans to consumers with no collections and those with unreported medical collections, for an overall delinquency risk of 1.08 percent. If no medical collections were included on consumer reports, creditors would provide accounts to all consumers with no collections and with medical collections, for an overall delinquency risk of 1.13 percent.

    Under this line of reasoning, the above results related to account performance would be unrelated to the consequences of the rule. It would be unsurprising that consumers with reported medical collections have the same underlying delinquency risk as consumers with unreported medical collections, because in the example delinquency risk is determined by the medical collection itself, and not as a consequence of consumer reporting. Instead, the relevant question would be whether consumers with medical collections have a higher delinquency risk than consumers without medical collections, holding all else equal.

    However, this example presupposes that delinquency risk is an inherent quality of consumers, rather than in part determined by the terms of credit extended to consumers. The dollar amounts and interest rates impact the likelihood of delinquency, as well as creditor revenue. These levers remain available to creditors under the rule and can be used to attenuate reductions in revenue that result from any increases in delinquency risk. Indeed, unlike in this simple example, the performance results in Table 8 show that creditors willingly provide accounts to people with reported medical collections at baseline. This requires that there exist terms for which credit accounts can be profitably, on expectation, provided to consumers with medical collections. Because creditors will not be able to differentiate between consumers with and without medical collections using information provided on consumer reports under the rule, any changes in terms of credit under the rule may impact all consumers, not just those with medical collections.

    Furthermore, if the credit extended to consumers with unreported medical collections were unprofitable at the pre-rule baseline, the CFPB expects that ( print page 3362) creditors could request this information on credit applications to ensure they do not provide loans to these applicants or provide different terms. Credit applications commonly request information from consumers that may not be available on their consumer reports, such as their employment status or income. Given the relatively small share of medical debt that is included on consumer reports, creditors could request this information from consumers directly if it were a key determinant of account profitability. At baseline, however, mortgage creditor applications, for example, ordinarily do not specifically request medical information.

    A researcher commenter described these results in the proposed rule, equivalent to Table 8, as showing that not having nonmedical debt, including products like student loans and auto loans, leads to higher rates of delinquency than having a product like a credit card. The commenter stated that these results suggested a problem with the CFPB's methodology for the Technical Appendix overall, as one would expect nonmedical debt to be associated with a greater rate of delinquency.

    While the CFPB agrees with the general principle that counterintuitive results of any statistical analysis may warrant additional scrutiny, the commenter does not accurately characterize the analysis above, and the results are not counterintuitive in the way the commenter suggests, much less indicating a problem with the CFPB's methodology. The CFPB's analysis does not compare delinquency rates between consumers with and without nonmedical debt in general. Rather, as discussed above, the results in Table 8 include versions with the sample split by the presence or absence of nonmedical collections tradelines. Nonmedical collections are not equivalent to nonmedical forms of debt such as student loans or auto loans, as a debt only goes to collections after it is seriously delinquent. Further, comparing Columns 2 and 3 or Columns 5 and 6 shows that credit products originated to people with nonmedical collections have higher delinquency rates, on average, than credit products originated to people without nonmedical collections, as would be expected.

    The researcher commenter also stated that the results in the proposed rule equivalent to Table 8 showed that there was a near-significant impact of nonmedical debt on mortgage delinquency. Again, this is an inaccurate characterization of the analysis presented above. The CFPB did not estimate the effect of nonmedical debt on delinquency at all. Instead, the CFPB found one subsample—consumers with non-medical collections tradelines and medical collections of any dollar amount—for which the effect of having a medical collections tradeline reported on mortgage delinquency is positive and close to being statistically significant at 95 percent. This is shown in Column 6 of Panel B of Table 8. If the effect were estimated with statistical significance, it would suggest that consumers with reported medical collections in this subsample are more likely to become seriously delinquent on mortgages. Differences in the terms provided to consumers with reported and unreported medical collections could lead to these higher delinquency rates, but the CFPB expects that this result is more likely attributed to statistical noise, given the inconsistency with the results from the other subsamples studied.

    The CFPB also considered whether to compare different credit scoring models, constructed with and without medical information, as a way to determine how well such models predict account performance. Such an approach, however, would call on the CFPB to design its own credit scoring models and determine what types and magnitude of differences between the results of the models were meaningful, and may depend more on the specifications of the models constructed than the actual rate of default in any studied population. The CFPB finds its regression discontinuity design and balance tests a more appropriate and reliable measure for how medical information improves creditors' ability to minimize their risk of default. The results of the CFPB's analysis of the performance of actual accounts indicate that creditors who use medical information do not reduce risk by doing so.[402]

    9. Results Related to Credit Demand and Selection

    The results described in the previous two subsections suggest that creditors use medical collections information in their underwriting procedures, but this information does not enable them to originate accounts that are less likely to become seriously delinquent. This interpretation of the regression discontinuity results relies on the identifying assumption discussed above: the only difference between the inquiries made before and after a medical collection tradeline is added to a consumer report is the medical collection reporting itself, rather than that the application delinquency risk (quality) is lower for consumers with reported medical collections. This section discusses evidence supporting this identifying assumption.

    Though the analysis benefits from ample observations near the threshold, as discussed above, RDiT specifications may still be affected by anticipation or selection effects if cross-sectional observations can sort themselves on either side of the threshold. In this setting, consumers may be less likely to apply for credit after a medical collection tradeline is added to their consumer report. If consumers with lower delinquency risk have more knowledge about when a medical collection tradeline will be added to their consumer report, they may be more likely to apply for credit immediately to the left of the threshold ( i.e., just before the medical collection tradeline is added to the consumer report). The CFPB first considered how the magnitude of credit demand changes across the reporting threshold by plotting the number of inquiries made in each week relative to the week of the medical collection tradeline's addition to the consumer report.

    ( print page 3363)

    Figure 1: Inquiry Distribution Across Weeks 403

    Figure 1 plots the number of inquiries made in each week relative to the week before the date a medical collection tradeline was added to a consumer report, represented as week zero. For all credit account products, credit demand is largely stable through the 25 weeks before the medical collection is reported, but there is an immediate reduction in the week that the medical collection is reported. Credit demand rebounds quickly from this initial drop but remains persistently lower for the 25 weeks after the medical collection is reported, only approaching its pre-report level by the final considered week for credit cards and mortgages. Though the reduction in credit demand is sharp around the week of the medical collection's first report, it is not large; at most, credit demand falls by 8 percent of the baseline (for mortgages).

    Any reduction in credit demand corresponding to medical collection reporting may appear to threaten the identifying assumption, which requires that applications for credit made by consumers with reported medical collections only differ from those made by consumers whose medical collections were not yet reported because of the medical collection reporting itself, and not because application quality differs. However, credit demand may fall for reasons that do not simultaneously affect credit application quality. For example, many NCRAs provide credit monitoring services that alert a consumer when a collection is added to their consumer report.[404] A consumer who planned to apply for credit may no longer do so if they are aware of a medical collection tradeline's negative effect on their credit score, which would affect their access to credit. The causality may also flow in the other direction if debt collectors track consumer reports and use “collection triggers” to focus their medical collection reporting after consumers apply for or open new credit accounts.[405] These mechanisms cannot be observed in the data but could explain the observed discontinuous decline in credit demand around medical collection reporting.

    To estimate if credit application quality changes across the threshold, the CFPB estimated balance tests using Equation 1, where Yijk is equal to one of several variables that describe the consumer report at the time of the inquiry j. This estimates how inquiries made with reported medical collections differ from inquiries made with unreported medical collections. If such differences are large in absolute value and statistically significant, one might be concerned that there are underlying differences in the types of credit applications made when medical collections are reported that could be driving the regression discontinuity results, instead of the medical collection reporting itself. Finding small or imprecise coefficients would support the identifying assumption that the only difference in inquiries across the regression discontinuity threshold is the addition of a medical collection tradeline to the consumer report.

    Table 9—Inquiry Balance Tests 406
      (1) Credit card (2) Mortgage (3) Other credit accounts
    Panel A: Over $500 sample:
    RD Estimate 0.117 0.257 0.118
      (0.172) (0.464) (0.172)
    Avg. consumer age 39.295 41.430 38.637
    RD Estimate **−3.208 4.034 −0.540
    ( print page 3364)
      (1.192) (3.572) (1.255)
    Avg. credit score 576.254 617.565 569.366
    RD Estimate ** 0.012 −0.001 0.008
      (0.005) (0.009) (0.005)
    Avg. missing credit score 0.197 0.074 0.151
    RD Estimate 0.032 0.050 0.026
      (0.035) (0.115) (0.039)
    Avg. num. open loans 1.328 1.997 1.275
    RD Estimate −0.001 −0.010 −0.008
      (0.005) (0.012) (0.006)
    Avg. any D90+ 0.265 0.256 0.268
    RD Estimate 49.549 *−259.894 29.122
      (63.234) (149.575) (72.823)
    Avg. tot. past due am 1,131.626 1,155.664 1,276.969
    Panel B: Full sample:
    RD Estimate 0.072 −0.111 −0.077
      (0.077) (0.235) (0.087)
    Avg. age 41.092 43.078 40.784
    RD Estimate *−1.472 1.868 −0.817
      (0.590) (1.990) (0.642)
    Avg. credit score 569.811 606.276 561.472
    RD Estimate ** 0.007 0.002 * 0.005
      (0.003) (0.004) (0.003)
    Avg. missing credit score 0.171 0.073 0.134
    RD Estimate −0.010 −0.092 −0.010
      (0.020) (0.047) (0.018)
    Avg. num. open loans 1.122 1.749 1.065
    RD Estimate 0.001 −0.000 0.000
      (0.003) (0.006) (0.004)
    Avg. any D90+ 0.262 0.260 0.267
    RD Estimate −33.152 −72.382 70.836
      (42.478) (76.899) (40.274)
    Avg. tot. past due am 1,073.628 1,135.919 1,190.611
    Standard errors in parentheses.
    *  p < 0.1, **  p < 0.05, ***  p < 0.01.

    Table 10—Performance Balance Tests 407
      (1) Credit card (2) Mortgage (3) Other credit accounts
    Panel A: Over $500 sample:
    RD Estimate 0.261 0.294 0.200
      (0.296) (0.894) (0.366)
    Avg. consumer age 41.404 42.692 40.184
    RD Estimate −3.694 7.807 0.502
      (2.012) (7.099) (2.608)
    Avg. credit score 618.329 668.427 601.025
    RD Estimate −0.005 0.005 0.002
      (0.006) (0.010) (0.007)
    Avg. missing credit score 0.078 0.014 0.099
    RD Estimate *** 0.286 * 0.564 0.089
      (0.092) (0.340) (0.092)
    Avg. num. open loans 1.884 2.834 1.804
    RD Estimate 0.017 −0.019 −0.002
      (0.009) (0.027) (0.013)
    Avg. any D90+ 0.248 0.191 0.268
    RD Estimate 175.228 −332.580 16.765
      (112.690) (302.978) (180.777)
    Avg. tot. past due am 1,034.492 673.171 1,220.532
    Panel B: Full sample:
    RD Estimate ** 0.411 0.871 0.068
      (0.154) (0.630) (0.200)
    Avg. consumer age 43.264 44.083 42.246
    RD Estimate −1.670 −0.602 −1.194
      (0.921) (3.340) (1.197)
    Avg. credit score 611.625 660.599 590.484
    RD Estimate −0.001 0.002 −0.000
      (0.003) (0.005) (0.004)
    ( print page 3365)
    Avg. missing credit score 0.057 0.016 0.087
    RD Estimate −0.027 −0.162 0.029
      (0.042) (0.157) (0.045)
    Avg. num. open loans 1.671 2.588 1.530
    RD Estimate 0.003 −0.028 0.007
      (0.005) (0.016) (0.007)
    Avg. any D90+ 0.256 0.189 0.274
    RD Estimate 82.685 −135.890 35.141
      (88.985) (138.828) (76.515)
    Avg. tot. past due am 1,005.487 609.676 1,191.860
    Standard errors in parentheses.
    *  p < 0.1, **  p < 0.05, ***  p < 0.01.

    Table 9 provides results for the inquiry dataset and Table 10 provides results for the performance dataset. Nearly all coefficients are not statistically significant, and where there is statistical significance, the magnitude of the coefficient is never larger than 20 percent of the mean value. This implies that credit applications submitted by consumers with reported medical collections are similar to those submitted by consumers whose medical collections are not yet on their consumer reports at the time of application, and differences in inquiry success and account performance can be attributed to the medical collection reporting itself.

    If all credit accounts were equivalent in their terms, and delinquency risk was an immutable characteristic of consumers, one may instead expect creditors to require applicants with reported medical collections to have credit profiles that reflect lower risk than those without reported medical collections, because consumers with reported medical collections have an additional, potentially negative, signal on their consumer report. Consider, under these assumptions, a simple example in which a creditor only provides credit to applicants whose expected delinquency risk is less than 10 percent. Suppose also that the presence of at least one medical collection increases an applicant's true delinquency risk by 1 percentage point. In this case, creditors will provide accounts to consumers whose true delinquency risk is between 0 and 11 percent for applicants with unreported medical collections, and between 0 and 10 percent for applicants with reported medical collections. If risk is equally distributed across the population, on average, a consumer offered credit with unreported medical collections would be 0.5 percentage points more likely to be delinquent than a consumer offered credit with reported medical collections.[408]

    To the contrary, the balance tests estimated in Table 10 show that there are no sizable and statistically significant differences between the credit profiles of consumers with reported or unreported medical collections that open credit accounts, for the considered possible differentiating variables. These balance tests suggest two possible explanations:

    First, some creditors could use medical collection information to deny all applicants with such information, while other creditors could disregard this information. In this case, creditors that ignore medical collections information would provide credit to the same types of consumers on either side of the regression discontinuity threshold, thus not causing a discontinuous change in the delinquency risk of approved consumers. The findings in Table 7 would be explained by creditors that deny all consumers with reported medical collections, but these creditors would not contribute to estimating the delinquency risk of consumers with reported medical collections; there is no delinquency rate to measure because these consumers did not open an account. These differences in creditors' understanding of the usefulness of medical collection information could explain the statistically insignificant differences in delinquency rates across the regression discontinuity threshold shown in Table 8.

    Second, creditors could provide different terms to consumers with reported medical collections, which may impact their delinquency risk. Consumers with reported medical collections may appear to be better credit risks than consumers with unreported medical collections (on a differentiating variable for which the balance tests were not estimated because the CFPB does not have the relevant data), but if they are provided worse terms, those terms may increase their delinquency risk above what it would have been had they received the terms provided to consumers with unreported medical collections. Additionally, consumers with lower delinquency risk may be less likely to take up an offered loan with worse terms. If, in the example above, consumers with a delinquency risk between 0 and 1 percent choose not to take up an offered credit account when their medical collection is reported and they are provided worse terms, the average delinquency rate would be 5.5 percent for consumers with reported medical collections, as in the sample of consumers with unreported medical collections.[409]

    The CFPB does not have information about the terms of credit provided to ( print page 3366) consumers with reported or unreported medical collections, or information about credit application approvals that are not taken up by consumers, and therefore cannot estimate the extent to which the delinquency results are driven by either possible explanation. Regardless of the underlying mechanism, the CFPB concludes that even when creditors, at baseline, use medical collection information, they do not reduce their underlying delinquency risk by doing so. This suggests that differences in inquiry success and account performance can be attributed to the medical collection reporting itself, rather than a change in the consumer's risk of default arising from the underlying medical debt. Therefore, removing this information under the rule will lead creditors to provide more credit accounts to consumers that are similar in delinquency risk to the credit accounts they already provide.

    Two researcher commenters stated that the CFPB needed to include control variables in its regressions, specifically suggesting State of residence, credit score, or credit balances. One of these commenters stated these variables may need to be controlled for if they are correlated with either inquiry success or account performance and change discontinuously around the medical collection reporting threshold date, citing academic literature.[410]

    The CFPB does not agree that including controls for State of residence, credit score, or credit balances is necessary or appropriate. If these control variables were correlated with inquiry success or account performance and changed discontinuously across the threshold date, estimating balance tests on these control variables would lead to statistically significant and large effects, but Tables 9 and 10 find no evidence in support of this hypothesis. The CFPB did not estimate balance tests for State of residence but finds it implausible that sufficiently many consumers would change States in response to a medical collection (so that a consumer's State correlated with the time between the inquiry and the medical collection report) that the move would discontinuously impact either inquiry success or account performance. Instead, the CFPB interprets its coefficients as an average of effects across all states, weighted by the number of inquiries included in the sample from each State. Additionally, the CFPB included in the Notice of Proposed Rulemaking, and reproduced below, a version of its results including control variables for day-of-week effects. These are the only effects mentioned as likely needed control variables in the academic literature cited by a commenter, but they do not meaningfully change the results.[411]

    To further test for the presence of anticipation or selection effects, the CFPB estimated a “donut” regression that removes from the sample all inquiries made within seven days of their associated medical collection's addition to the consumer report. If the regression estimates are driven by anticipation or selection, the effects would be much smaller when estimated without observations near the reporting threshold, as application quality would be less selected from the threshold. In addition, medical collections may not be reported to all three NCRAs on precisely the same date. The creditors that make inquiries to the NCRA that provides the CFPB's CCIP may observe a medical collection on an inquiry they make to a different NCRA and use this information, even though it appears in the CCIP that the medical collection was not reported. Additionally, the construction of inquiry shopping windows and inherent imprecision in connecting inquiries to opened tradelines may further limit the accuracy of calculating the running variable to a precise day. This is especially important near the reporting threshold because a one-day error in assigning the date a medical collection was reported or an inquiry was made could be sufficient to erroneously categorize the medical collection reporting status of an inquiry. The CFPB further considered variation in dates within inquiry shopping windows below.

    Table 11—The Effect of Medical Collection Reporting on Inquiry Success and Credit Account Performance, Using a 14-Day Donut 412
      (1) Over $500, success (2) Over $500, D90+ (3) All, success (4) All, D90+
    Panel A: Credit cards:
    RD Estimate ***−0.060 (0.0080) [−0.075, −0.045] −0.006 (0.015) [−0.036, 0.024] ***−0.041 (0.005) [−0.050, −0.032] 0.008 (0.008) [−0.009, 0.024]
    Avg. dep. var 0.294 0.232 0.275 0.223
    Observations 578,088 92,708 2,908,047 543,865
    Panel B: Mortgages:
    RD Estimate **−0.037 (0.017) [−0.071, −0.004] −0.022 (0.025) [−0.071] ***−0.043 (0.008) [−0.060, −0.027] −0.003 (0.011) [−0.026, 0.019]
    Avg. dep. var 0.186 0.035 0.167 0.038
    Observations 76,358 9,797 422,584 54,818
    Panel C: Other Credit Accounts:
    RD Estimate −0.009 (0.009) [−0.027, 0.009] −0.038 (0.025) [−0.087, 0.012] *−0.010 (0.004) [−0.018, −0.002] 0.008 (0.010) [−0.012, 0.027]
    ( print page 3367)
    Avg. dep. var 0.242 0.182 0.245 0.171
    Observations 451,474 69,159 2,387,333 441,523
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    *  p < 0.1, **  p < 0.05, ***  p < 0.01.

    Table 11 provides the “donut” specification regression results. By comparing Column 1 of Table 7 to Column 1 of Table 11 and comparing Column 4 of Table 7 to Column 3 of Table 11, one can observe that effects on inquiry success are larger in absolute magnitude and more statistically significant for credit cards and mortgages in the donut specification than in the main specification. This shows that the main results using the inquiry data are not driven by selection or anticipation effects. Instead, the results in the main specification may be attenuated by fuzziness in the date that the medical collection was reported or that the inquiry was made, as discussed above.

    Despite the modest differences between Table 11 and Table 7 for the inquiry dataset, there are no meaningful differences in the magnitude or statistical significance of effects for the performance datasets, as shown by comparing Column 1 of Table 8 to Column 2 of Table 11 and comparing Column 4 of Table 8 to Column 4 of Table 11. This provides further evidence that the use of medical collection reporting in underwriting does not improve account performance.

    A final concern is that it could be problematic if there is bunching at certain values of the running variable because the likelihood of a medical collection being reported, or an inquiry being made, differs across days of the week. For example, fewer than 4 percent of the medical collection tradelines associated with inquiries in the inquiry dataset were reported on a Sunday, compared to nearly 28 percent reported on a Tuesday. The distribution of inquiries in the inquiry dataset (across all inquiry product types) is more even, with a low of 8.5 percent on Sunday, just over 15 percent on Monday through Friday, and nearly 14 percent on Saturday. Combining these two features, an inquiry made on a Monday is more likely to correspond to a medical collection tradeline on the subsequent day than an inquiry made on a Saturday. If the types of inquiries made on Mondays differ from those made on Saturdays, there may be disproportionately more inquiries made on Monday for the running variable value immediately before the threshold (equal to -1), which could cause selection bias in the estimated effect. To test whether this selection biases the regression results, the CFPB estimated an additional specification that adds binary indicator variables to the main specification for the day of the week of each observation's inquiry date and date of the medical collection report.

    Table 12—The Effect of Medical Collection Reporting on Inquiry Success and Credit Account Performance, Controlling for Day-of-Week Effects 413
      (1) Over $500, success (2) Over $500, D90+ (3) All, success (4) All, D90+
    Panel A: Credit cards:
    RD Estimate ***−0.048 −0.002 ***−0.034 0.001
      (0.006) (0.012) (0.003) (0.006)
      [−0.059, −0.038] [−0.024, 0.021] [−0.039, −0.028] [−0.010, 0.012]
    Avg. dep. var 0.294 0.231 0.275 0.223
    Observations 601,230 96,297 3,026,355 565,680
    Panel B: Mortgages:
    RD Estimate *−0.027 −0.017 −0.014 0.005
      (0.011) (0.015) (0.009) (0.007)
      [−0.049, −0.004] [−0.045, 0.012] [−0.032, 0.003] [−0.008, 0.018]
    Avg. dep. var 0.186 0.035 0.167 0.038
    Observations 79,372 10,177 439,685 56,976
    Panel C: Other credit accounts:
    RD Estimate *−0.014 −0.015 ***−0.015 −0.002
      (0.006) (0.014) (0.003) (0.006)
      [−0.026, −0.003] [−0.042, 0.013] [−0.021, −0.010] [−0.013, 0.010]
    Avg. dep. var 0.242 0.182 0.246 0.171
    Observations 469,290 71,760 2,484,030 459,094
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    *  p < 0.1, **  p < 0.05, ***  p < 0.01.
    ( print page 3368)

    Table 12 provides the regression results for a version of Equation 1 that includes day-of-the-week controls. Results are very similar to the main specification, as can be seen by comparing Column 1 of Table 7 to Column 1 of Table 12, Column 4 of Table 7 to Column 3 of Table 12, Column 1 of Table 8 to Column 2 of Table 12 and comparing Column 4 of Table 8 to Column 4 of Table 12. The CFPB concluded that the main results are not caused by bias in the distribution of inquiry or medical collection timing across days of the week.

    10. Results Related to Credit Shopping

    As described above, the main specification defines the running variable using the date of the last inquiry observed within the inquiry shopping window. This creates imprecision in the measurement of the inquiry date for inquiry observations that reflect shopping windows with multiple inquiries if they were not made on the same date.[414] Because this imprecision could attenuate results, the CFPB estimated Equation 1 separately for inquiry observations that reflect multi-inquiry-date shopping windows (Shopping) and for inquiry observations that reflect shopping windows that only contain one inquiry date (No Shopping). The CFPB estimated this robustness check for the inquiry dataset first, and then for the performance dataset.

    Table 13—The Effect of Medical Collection Reporting on Inquiry Success, Separated by Shopping Behavior 415
      (1) Over $500, shopping (2) Over $500, no shopping (3) All, shopping (4) All, no shopping
    Panel A: Credit cards:
    RD Estimate −0.043 (0.020) [−0.082, −0.003] ***−0.050 (0.005) [−0.060, −0.039] 0.000 (0.013) [−0.025, 0.026] ***−0.035 (0.003) [−0.040, −0.030]
    Avg. success 0.445 0.279 0.422 0.262
    Observations 51,481 549,749 250,319 2,776,036
    Panel B: Mortgages:
    RD Estimate −0.019 (0.028) [−0.074, 0.037] −0.022 (0.011) [−0.043, −0.001] ***−0.041 (0.014) [−0.068, −0.014] −0.002 (0.011) [−0.024, 0.020]
    Avg. success 0.329 0.123 0.308 0.111
    Observations 24,266 55,106 126,393 313,292
    Panel C: Other credit accounts:
    RD Estimate 0.002 (0.015) [−0.030, 0.027] −0.016 (0.006) [−0.029, −0.004] −0.015 (0.007) [−0.029, −0.001] ***−0.015 (0.003) [−0.021, −0.008]
    Avg. success 0.391 0.213 0.394 0.217
    Observations 77,603 391,687 400,620 2,083,410
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    *  p < 0.1, **  p < 0.05, ***  p < 0.01.

    Table 13 shows results for inquiry success for inquiries associated with multi-date versus single-date shopping windows. For credit cards and other non-mortgage accounts, the results are only statistically significant for single-date shopping windows and are also larger in absolute magnitude. Fewer than 10 percent of credit card inquiries are associated with multi-date shopping windows, which is expected given the small average shopping windows for credit cards shown in Table 5. Alternatively, the only statistically significant result for mortgages appears for inquiries associated with multi-date shopping windows in the full sample. This limited ability to identify a precise effect is reflected in the main specification as well, as shown in Table 7. The CFPB concluded that, for non-mortgage products, the inability to observe the exact date that an inquiry was made may attenuate the results in the main specification, and the true effect of having a medical collection reported may be a larger decrease in inquiry success than what is reported in Table 7.

    ( print page 3369)

    Table 14—The Effect of Medical Collection Reporting on Two-Year Credit Account Performance, Separated by Shopping Behavior 416
      (1) Over $500, shopping (2) Over $500, no shopping (3) All, shopping (4) All, no shopping
    Panel A: Credit cards:
    RD Estimate −0.010 −0.000 0.023 −0.001
      (0.035) (0.013) (0.018) (0.006)
      [−0.079, 0.059] [−0.025, 0.025] [−0.013, 0.059] [−0.013, 0.011]
    Avg. D90+ 0.320 0.218 0.313 0.210
    Observations 12,288 84,009 70,222 495,458
    Panel B: Mortgages:
    RD Estimate −0.005 −0.025 0.009 0.001
      (0.020) (0.020) (0.011) (0.008)
      [−0.045, 0.036] [−0.063, 0.014] [−0.012, 0.030] [−0.015, 0.018]
    Avg. D90+ 0.041 0.027 0.046 0.030
    Observations 5,673 4,504 30,756 26,220
    Panel C: Other credit Accounts:
    RD Estimate −0.013 −0.003 −0.000 −0.001
      (0.026) (0.014) (0.012) (0.007)
      [−0.065, 0.039] [−0.030, 0.025] [−0.023, 0.023] [−0.014, 0.012]
    Avg. D90+ 0.216 0.170 0.207 0.158
    Observations 19,879 51,881 122,953 336,141
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    *  p < 0.1, **  p < 0.05, ***  p < 0.01.

    Table 14 provides the same robustness check as Table 13 but estimates effects on serious delinquency using the performance dataset. As in previous robustness checks, the estimated results on account performance are all statistically insignificant, and nearly all are small in comparison to the baseline average delinquency rate. The CFPB considers these results as evidence that imprecision in assigning inquiry dates does not drive the lack of statistical significance in the main specification.

    Finally, the CFPB tested whether classifying the timing of an inquiry shopping window using the last inquiry makes a difference to the results. Although it makes intuitive sense to focus on the last inquiry—a consumer finishes shopping, then either gets a new account or does not—this could impact whether a consumer is considered treated or not by having a medical collection reported or not. For example, if a consumer applied for accounts that created inquiries on March 5 and March 17, had an account opened on March 19, and had a medical collections tradeline reported on March 15, in the main specification described above, they would be considered to have a medical collection at the time of the inquiry. This may be accurate, if the March 17 inquiry (or another inquiry after March 15 that was made with a different NCRA) resulted in the open account, but it also may be inaccurate, and influence the results reported above. To further test how the definition of shopping windows may affect the main results, the CFPB estimated a version of the analysis using the first date of the shopping window instead of its last date to define the running variable.

    Table 15—The Effect of Medical Collection Reporting on Inquiry Success and Credit Account Performance, Classifying Shopping Windows by First Inquiry Date 417
      (1) Over $500, success (2) Over $500, D90+ (3) All, success (4) All, D90+
    Panel A: Credit cards:
    RD Estimate *** −0.049 0.002 *** −0.035 0.004
      (0.004) (0.012) (0.003) (0.006)
    ( print page 3370)
      [−0.058, −0.041] [−0.021, 0.025] [−0.040, −0.030] [−0.008, 0.016]
    Avg. dep. var 0.294 0.231 0.275 0.222
    Observations 600,209 95,973 3,021,234 563,942
    Panel B: Mortgages:
    RD Estimate −0.010 0.003 −0.010 0.003
      (0.012) (0.013) (0.008) (0.006)
      [−0.033, 0.014] [−0.022, 0.028] [−0.026, 0.006] [−0.009, 0.015]
    Avg. dep. var 0.182 0.033 0.163 0.035
    Observations 74674 8836 415412 49986
    Panel C: Other credit Accounts:
    RD Estimate −0.010 −0.020 ***−0.012 −0.003
      (0.006) (0.014) (0.003) (0.006)
      [−0.021, 0.002] [−0.048, 0.008] [−0.018, −0.006] [−0.015, 0.008]
    Avg. dep. var 0.242 0.182 0.246 0.171
    Observations 467,949 71,401 2,476,494 456,828
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    *  p < 0.1, **  p < 0.05, ***  p < 0.01.

    The results in Table 15 are very similar in size to those in the main specification, as seen by comparing Column 1 of Table 7 to Column 1 of Table 15, Column 4 of Table 7 to Column 3 of Table 15, Column 1 of Table 8 to Column 2 of Table 15 and comparing Column 4 of Table 8 to Column 4 of Table 15. The coefficients in Column 1 of Table 15, estimating the impact of medical collection reporting on inquiry success, are no longer marginally significant for mortgages and other credit accounts. This may be because the last inquiry observed within an inquiry shopping window is a better proxy for the date that the creditor observed the consumer report for these products, which is sensible if consumers continue to shop when they reject an earlier credit offer, or their application is rejected. The CFPB considers these results as evidence that, given the inherent challenges in assigning inquiry dates, the method of using the last date that an inquiry was observed within a shopping window is the best available classification.

    11. Results Related to Alternative Measures of Account Performance and Inquiry Success

    Moving on from statistical and data construction considerations, the CFPB returns to the applicability of the results to the considered equilibrium in which all medical collection tradelines are removed from consumer reports. Creditors may respond to reported medical collections by providing lower amounts of credit, especially for products whose applications do not typically request a certain amount of credit, such as credit cards (and unlike mortgages). The CCIP does not contain data on the dollar amount of credit that consumers were offered if consumers decided not to open an account, but it can observe credit limits and loan principals for originated accounts. Moreover, the CFPB understands that credit card accounts are typically opened automatically if approved by the creditor, such that consumers do not have an opportunity to decline an offer of credit with a lower limit than they prefer. The CFPB estimated Equation 1 using the account's credit limit (for revolving accounts) or loan principal (for installment accounts) as the dependent variable. This regression can only be run for the performance dataset because credit limits and loan principals cannot be observed for unsuccessful inquiries.

    Table 16—The Effect of Medical Collection Reporting on Credit Account Limits and Loan Principals 418
      (1) Over 500 (2) All
    Panel A: Credit cards:
    RD Estimate ***−384.312 ***−247.492
      (80.367) (33.855)
      [−541.829, −226.795] [−313.848, −181.137]
    Avg. credit am 1,481.169 1,312.252
    Observations 96,208 565,222
    Panel B: Mortgages:
    RD Estimate −12,746.532 −15,734.984
      (11,952.690)
      [−36173.374, 10680.309] [−33208.174, 1738.206]
    ( print page 3371)
    Avg. credit am 232,565.905 225,877.236
    Observations 10,163 56,918
    Panel C: Other credit accounts:
    RD Estimate 254.621 −195.017
      (398.877) (220.971)
      [−527.164, 1036.407] [−628.113, 238.078]
    Avg. credit am 20,994.097 20,380.048
    Observations 71,739 458,968
    Standard error in parentheses, 95 percent confidence intervals in brackets.
    * p < 0.1, ** p < 0.05, *** p < 0.01.

    Table 16 provides estimates for the effect of medical collection reporting on credit limits and loan principals. The results in Panel A show that medical collection reporting leads to lower credit limits for originated credit cards, with an average reduction in provided credit limits of $384 for the over-$500 sample and $247 for the full sample. This represents a meaningful reduction in consumer access to credit, as baseline average credit limits are lower than $1,500 for both samples. As expected, the CFPB does not find statistically significant effects for mortgages or other non-credit-card account types. Consumers generally apply for a specific dollar amount of credit for installment products, and the dollar amount of credit provided is not a margin that would generally be affected by medical collection reporting.

    The CFPB understands that the classification of serious delinquency is not the sole determinant of account performance. Three other measures of performance are considered in this final set of regressions, estimated on the performance dataset: whether the account is ever 30 days or more delinquent within two years of its origination, whether the account is 90 days or more delinquent at the end of its first two years after origination (instead of whether it was ever 90 days or more delinquent within that two-year period), and the dollar amount past due or charged off for accounts with nonzero past due or charged off amounts at the end of its first two years after origination. If the primary classification of serious delinquency is a good proxy for account performance, then results for the first two alternative measures should be similar to their counterparts in the main performance results in direction and statistical significance. The results for past due amounts may be more nuanced, as Table 16 above shows that medical collection reporting lowers the credit limits of credit cards. This may cause lower past due amounts in response to medical collection reporting because consumers cannot borrow as much as they can absent medical collection reporting.

    Table 17—The Effect of Medical Collection Reporting on Two-Year Credit Account Performance, Alternative Classifications 419
      (1) Over $500, D30+ (2) Over $500, D90+ alt. (3) Over $500, past due am. (4) All, D30+ (5) All, D90+ alt. (6) All, past due am.
    Panel A: Credit cards:
    RD Estimate 0.008 −0.006 **−215.199 0.002 −0.003 **−62.830
      (0.013) (0.011) (86.597) (0.006) (0.005) (29.197)
      [−0.017, 0.032] [−0.027, 0.015] [−384.926, −45.472] [−0.010, 0.015] [−0.013, 0.008] [−120.055, −5.604]
    Avg. dep. var 0.321 0.164 713.724 0.316 0.153 643.677
    Observations 96,297 96,297 19,945 565,680 565,680 111,342
    Panel B: Mortgages:
    RD Estimate −0.034 0.002 4,477.430 0.012 0.001 261.686
      (0.027) (0.010) (2,894.862) (0.012) (0.005) (1,682.921)
      [−0.087, 0.018] [−0.018, 0.022] [−1196.394, 10151.255] [−0.012, 0.036] [−0.009, 0012] [−3036.779, 3560.152]
    Avg. dep. var 0.125 0.021 7,511.005 0.118 0.019 6,018.840
    Observations 10,177 10,177 409 56,976 56,976 1,954
    Panel C: Other credit Accounts:
    RD Estimate −0.006 −0.002 −803.533 −0.000 0.000 −562.913
      (0.016) (0.013) (732.117) (0.008) (0.005) (301.400)
      [−0.037, 0.025] [−0.027, 0.023] [−2238.455, 631390] [−0.016. 0.015] [−0.009, 0.010] [−1153.647, 27.821]
    Avg. dep. var 0.322 0.156 7,012.189 0.316 0.145 6,510.499
    ( print page 3372)
    Observations 71,760 71,760 13,777 459,094 459,094 81,546
    Standard errors in parentheses, 95 percent confidence intervals in brackets.
    ** p < 0.1, ** p < 0.05, *** p < 0.01.
Visit the Official Version
Leave a Comment

Document Information

Effective Date:
3/17/2025
Published:
01/14/2025
Department:
Consumer Financial Protection Bureau
Entry Type:
Rule
Action:
Final rule.
Document Number:
2024-30824
Dates:
This final rule is effective March 17, 2024.
Pages:
3276-3374 (99 pages)
Docket Numbers:
Docket No. CFPB-2024-0023
RINs:
3170-AA54: Prohibition on Creditors and Consumer Reporting Agencies Concerning Medical Information (Regulation V)
RIN Links:
https://www.federalregister.gov/regulations/3170-AA54/prohibition-on-creditors-and-consumer-reporting-agencies-concerning-medical-information-regulation-v
Topics:
Banks, banking, Banks, banking, Banks, banking, Banks, banking, Consumer protection, Credit unions, Holding companies, National banks, Privacy, Reporting and recordkeeping requirements, Savings associations
PDF File:
2024-30824.pdf
CFR: (1)
12 CFR 1022