The Consumer Financial Protection Bureau (the "CFPB" or the "Bureau") recently announced the issuance of its first no-action letter ("NAL") to Upstart Network, Inc. ("Upstart"), a San Carlos, California-based online lending platform that uses alternative data to model consumer credit decisioning and pricing. The letter signifies that the CFPB has no present intention to recommend an enforcement or supervisory action against Upstart for violation of the Equal Credit Opportunity Act. As stated in the Bureau's press release announcing this action, the NAL comes as the Bureau "continues to explore the use of alternative data to help make credit more accessible and affordable for consumers who are credit invisible or lack sufficient credit history." In addition, the NAL is issued in the midst of heightened regulatory interest in and scrutiny of alternative credit data and modeling techniques. The Bureau issued two related requests for information, one in November 2016 on data aggregation services and the other in February 2017 on the use of alternative data, modeling techniques, and machine learning techniques in consumer lending. The NAL tends to suggest that companies may have some flexibility in the use of alternative underwriting modeling to offer consumer credit; however, as noted below, the practical utility of such NALs may be limited.

THE REQUEST AND THE NAL

As explained in the company's Request for a No-Action Letter ("Request"),1 Upstart provides a platform for borrowers to obtain unsecured, fixed-rate personal loans. Upstart explains that its credit underwriting model for these loans has the ability to underwrite risk for "thin file" applicants. In addition to traditional credit information, the company uses non-traditional criteria such as an applicant's school, area of study, academic performance, and employment history to make underwriting and pricing decisions. These techniques allow the company to identify credit-worthy individuals who have short or limited credit histories. Upstart's lending platform was launched in 2014. According to the company's submission, Upstart has enabled its bank partner to offer more favorable interest rates to borrowers with limited credit or work history than would be the case without consideration of these factors.

The CFPB initiated its NAL program in February 2016 with the intent to promote "innovative financial products or services that promise substantial consumer benefit where there is substantial uncertainty whether or how specific provisions of statutes implemented or regulations issued by the Bureau would be applied."2

The Bureau previously has shown concern that the use of some non-traditional factors could have a disparate impact due to their correlation with race, ethnicity, and/or gender. The agency's first use of its NAL authority nevertheless addresses alternative credit decisioning and pricing models identified in the Bureau's request for information on the impact of alternative data on credit access for consumers who are so-called credit invisible.3 In this case, Upstart provided the CFPB with confidential credit criteria, credit modeling factors, and objective data showing that thin file applicants are underserved by credit-score-based decisioning. Upstart also demonstrated that against a test model its credit model resulted in comparatively favorable loan pricing and presumably more affordable and beneficial credit being made available to its credit-challenged applicants.

Under the terms of the NAL, the CFPB reserves the right to conduct supervisory activities or engage in an enforcement investigation to enforce Upstart's compliance with the NAL. The letter is valid for three years from the date of issuance. For the letter to remain in effect, Upstart must regularly report lending and compliance information to the CFPB to aid in the agency's understanding of the impact of using alternative data on protected populations.

TAKE HOME LESSONS

For alternative credit providers, particularly those using online product delivery (FinTech), who want to understand the precedential value of the NAL, we believe the following observations are pertinent:

  • The overall utilization by consumer financial services providers of the CFPB's NAL process remains unclear. On the surface, the agency's NAL process, in the 18 months since it was announced, has resulted in only one final action. On the other hand, it is not known just how many other NAL requests are in process at the Bureau. By the same token, it is not known how many financial services providers have considered this path and elected not to embark on it.
  • Upstart's efforts, in particular its Request, may well serve as a template for other financial parties to follow in the CFPB's NAL process, particularly as to credit modeling. Observers of the Bureau's NAL Policy of February 2016 noted that the Bureau requires detailed explanations of business processes of submitters, some of which could be quite proprietary. However, the company, in preparing and submitting its request, evidently believed the risks associated with this level of disclosure were outweighed by the benefit of obtaining the NAL.
  • Upstart is to conduct fair lending testing and is obligated to provide to the Bureau, upon request, information including "the data underlying [its] test results."4 To the extent the Bureau will be monitoring borrower-level performance of Upstart loans, it is conceivable that the CFPB could raise questions later about the effectiveness of Upstart's modeling of alternative credit criteria in serving borrowers' interests. That is, the CFPB's interest may well go beyond fair lending testing outcomes.
  • The NAL in this case did not define "alternative [credit] criteria," and it appears that the criteria utilized by Upstart, other than "traditional" criteria, were quite limited.5 Also, the CFPB constrained the NAL to the  automated underwriting model described in the Request, stating that the NAL does not apply to any other products or services or underwriting models or the application of other laws and regulations to Upstart. Should Upstart wish to adjust its NAL-disclosed credit model to add other alternative credit criteria, presumably it would need to consult with the Bureau in advance to preserve the efficacy of the NAL.
  • The NAL is for Upstart's benefit only and does not pertain to anyone other than Upstart.

Footnotes

1 Upstart's submission was "by the book" for CFPB NALs. That is, the company responded to the questions posed by the Bureau in its previous NAL guidance.

2 See Bureau of Consumer Financial Protection Policy on No-Action Letters, 81 Fed. Reg. 8,686 (Feb. 22, 2016).  

3 See Bureau of Consumer Financial Protection "Request for Information Regarding Use of Alternative Data and Modeling Techniques in the Credit Process," 82 Fed. Reg. 11,183 (Feb. 21, 2017).

4 The Request stated, "Upstart commits to sharing the results of its fair lending and access-to-credit test results with the Bureau." The Request also stated, "On a routine basis, Upstart will compare applicant outcomes from its underwriting model against outcomes that would result under a model without non-traditional variables. This will include an analysis of any different outcomes for specific applicant groups, including groups defined by race/ethnicity, sex, age, income, credit history, educational background, and other non-credit based variables."

5 In this regard, it is helpful to refer to the CFPB's February 2017 http://files.consumerfinance.gov/f/documents/20170214_cfpb_Alt-Data-RFI.pdf request for information, wherein the agency posited that that "alternative data" is any data that are not "traditional" and "alternative modeling techniques" are modeling techniques that are not "traditional" modeling techniques. In describing the latter, the CFPB stated that such techniques include but are not limited to "decision trees, random forests, artificial neural networks, k-nearest neighbor, genetic programming, 'boosting' algorithms, etc." Here the Bureau stated, "We use 'alternative' in a descriptive rather than normative sense and recognize that there may not be an easily definable line between traditional and alternative modeling techniques."

Because of the generality of this update, the information provided herein may not be applicable in all situations and should not be acted upon without specific legal advice based on particular situations.

© Morrison & Foerster LLP. All rights reserved