Use of Complex Algorithms for Credit Decisions CFPB Requirements
On May 26, the Consumer Financial Protection Bureau (the Bureau or CFPB) released its third circular, stressing that creditors must comply with the Equal Credit Opportunity Act (ECOA) and Regulation B, even when using complex algorithms, sometimes called uninterpretable or “black box” models, to make decisions credit. The circular explains that companies must provide the applicant with specific reasons for the denial of a credit application or adverse action, even if the creditor company uses complex credit algorithm models that do not even allow the creditor himself to “identify with precision the specific reasons for refusing credit or taking other adverse action. Thus, the CFPB will not accept a generic statement that a consumer has failed to meet a lender’s proprietary lending standards model as a sufficiently specific and consistent explanation for an adverse credit decision.
The circular makes it clear that federal consumer protection laws apply and are enforced regardless of the technology employed by a creditor. It also states that failure to comply with ECOA and Regulation B – which require a creditor to state the primary reason(s) for an adverse action – cannot be justified when the technology a creditor “uses to assess requests is too complicated or opaque to understand”. CFPB Director Rohit Chopra explained:[c]companies are not exempt from their legal responsibilities when letting a black box model make lending decisions[.]Thus, according to the CFPB, a creditor violates Rule B when it uses a system that fails (on purpose or inadvertently) to produce an accurate and specific explanation for each credit decision.
The press release announcing the Circular highlights third parties who assist the Bureau in its enforcement efforts. The Bureau explains that whistleblowers play a “central role” in providing information about companies using these so-called “black box” models and encourages company employees, especially tech workers, to provide information. Government partners are also, the CFPB notes, “essential” to help with law enforcement. To that end, the CFPB says it continues to monitor the work of government entities, including the National Institutes of Standards and Technology, to assess the risks and benefits of emerging technologies.
This latest development aligns with other recent proactive measures taken by the Bureau. Mere explanation by a creditor that their “system” or “model” made the decision is not acceptable to the CFPB, particularly where the creditor themselves cannot identify the specific reason(s) that caused its technology to make a particular unfavorable decision. Further, the CFPB’s warning that it will not accept a company’s explanation that the technology is “too new” shows that a creditor defending violations of the Lending Act because that its technology is immature and has not worked as expected or expected is unlikely to find a receptive audience. Fintechs and other relevant businesses may wish to assess whether their technology and lending models adequately inform them and consumers about specific reasons for credit decisions.
*Special thanks to Summer Associate Miranda Carnes, 3L at American University Washington College of Law, for her valuable contributions to this GT Alert.
©2022 Greenberg Traurig, LLP. All rights reserved. National Law Review, Volume XII, Number 158