Recent fair lending developments: OneWest settlement; ‘no action’ on alternative data and machine learning insights; and a proposed ruling on disparate impact
Scott Grotewold and Stephanie Wormack, fair lending risk specialists at the Federal Reserve Bank of Chicago, discussed a recent fair lending settlement, alternative data and machine learning insights, and a proposed rule that interprets a recent U.S. Supreme Court ruling on disparate impact at an economic development forum in Milwaukee. Key takeaways included:
- In response to allegations of discriminatory practices, OneWest Bank has agreed to invest more than $7M to provide better information about, and access to, loans in predominately minority neighborhoods in Los Angeles. They have also agreed to open a full service branch dedicated to these neighborhoods, and originate $100M in home purchase loans in these areas.
- The CFPB provided a “No Action Letter” allowing Upstart – a lender that uses alternative data to assess the credit worthiness of applicants – to test a new approach to underwriting and pricing loans in an attempt to expand credit (especially to those with sparse credit records). Initial evidence suggests this program expanded credit, reduced interest rates, and resulted in no significant race, ethnicity, gender, or age-based disparities in access to credit.
- The Department of Housing and Urban Development (HUD) proposes a new rule to alter the way regulators and plaintiffs can show discrimination under the Fair Housing Act (FHA). These changes generally increase the burden plaintiffs must meet to claim such discrimination may have occurred, and provide defendants additional ways in which they can rebut these claims.
Additional detail is provided below.
HUD announces settlement with OneWest Bank
In July, the HUD announced that it had approved a settlement between the California Reinvestment Coalition (CRC) and CIT Group, Inc., and CIT Bank, N.A., dba OneWest Bank, Pasadena, CA, resolving allegations that the bank engaged in lending discrimination by “redlining” in the Los Angeles region.2 The finding stems from a complaint the CRC filed with HUD that alleged the bank’s marketing and origination practices discriminated on the basis of race and national origin, in violation of the FHA. The complaint alleged the bank discriminated in marketing and origination of home mortgage loans, as evidenced by the low number of mortgages made to African American and Latino borrowers relative to the demographics of the area and industry peers, and the bank located and maintained branches in areas that do not serve minority neighborhoods and borrowers. The complaint also alleged the bank maintained and marketed bank-owned foreclosed homes in predominantly white neighborhoods better than neighborhoods with higher concentrations of minority residents. Foreclosed homes owned by the bank in neighborhoods of color contained trash in the yard, boarded up windows, and properties were not clearly marked as being for sale. By comparison, bank-owned foreclosed homes in white neighborhoods were well maintained and clearly marked as being for sale. As part of the settlement, OneWest Bank will invest $5 million in a loan subsidy fund to increase credit opportunities for residents of majority-minority neighborhoods; devote $1.3 million toward advertising and community outreach; and provide $1 million in grants for homebuyer education, credit counseling, community revitalization, and homeless programs. OneWest Bank is also committing to originate $100,000,000 in home purchase, home improvement, and home refinance loans to borrowers in majority-minority areas, and to open a full-service branch serving the banking and credit needs of residents in a majority-minority and low- and moderate-income neighborhood.
CFPB issues No-Action Letter to Upstart
In August, the Consumer Financial Protection Bureau (CFPB) issued a No-Action Letter (NAL) to Upstart Network, Inc., a company that uses alternative data and machine learning, including information related to borrower’s education and employment history in making credit underwriting and pricing decisions.3 The NAL program was created to facilitate development of innovative products or services that offer the potential for significant consumer benefit where there is regulatory uncertainty for emerging products and services. The NAL program provides a recipient reassurances that the CFPB will not bring a supervisory or enforcement action against a company for providing a product or service under the covered facts and circumstances in an effort to support marketplace innovation. Pursuant to the NAL, Upstart provides the CFPB with information comparing outcomes from its underwriting and pricing model (tested model) against outcomes from a hypothetical model that uses traditional application and credit file variables and does not employ machine learning (traditional model), which is independently validated through fair lending testing to ensure that it did not violate antidiscrimination laws.
The program addresses two key questions associated with alternative data and machine learning:
- Does the model’s use of alternative data and machine learning expand access to credit?
- Does the model’s underwriting and pricing outcomes result in greater disparities than the traditional model with respect to prohibited bases?
The results provided from the access to credit comparisons show that the tested model approves 27 percent more applicants than the traditional model and yields 16 percent lower average annual percentage rates (APRs) for approved loans. The results provided show that the tested model significantly expands access to credit compared to the traditional model. With regard to fair lending testing, which compared the tested model with the traditional model, the approval rate and APR analysis results provided for race, ethnicity, sex, and age show no disparities that require further fair lending analysis under the compliance plan.
HUD releases proposed rule interpreting the FHA
In August, HUD released a proposed rule to amend the HUD interpretation of the FHA’s disparate impact standard.4 The proposal provides a burden shifting framework that better reflects the U.S. Supreme Court’s 2015 ruling in Texas Department of Housing and Community Affairs v. Inclusive Communities Project, Inc. The HUD guidance adds five elements that a plaintiff must plead to support allegations of disparate impact discrimination. The five elements would require a plaintiff to adequately allege:
- The challenged policy or practice is “arbitrary, artificial, and unnecessary” to achieve a valid interest or legitimate objective;
- A “robust causal link” between the challenged policy or practice and a disparate impact on members of a protected class;
- The challenged policy or practice has an adverse effect on members of a protected class;
- The disparity caused by the policy or practice is significant; and
- The complaining party’s alleged injury is directly caused by the challenged policy or practice.
Unique to this proposal are three methods for defendants to rebut a disparate impact claim, including:5
- Material factors used in the model do not rely on factors which are substitutes or close proxies for protected classes under the FHA, and that the model is predictive of credit risk or other similar valid objective;
- The model is produced, maintained, or distributed by a recognized third party that determines industry standards, the model inputs and methods are not determined by the defendant, and the defendant is using the model as intended by the third party; and
- A neutral third party has analyzed the model and determined it is a demonstrably and statistically sound algorithm, and that none of the factors are substitutes or close proxies for protected classes.
Fair lending summary
To mitigate fair lending risk and to properly comply with the matters discussed above, an effective fair lending risk management program should evaluate treatment and impact on protected classes within strategic decisions, lending guidance, and policies and procedures compared to actual implementation of practices for fair lending risk. Opportunities for innovation in lending must coincide with commensurate fair lending risk management oversight and testing. Noncompliance with fair lending laws and regulations have significant consequences such as prohibitions on expansion along with reputational, legal, and financial implications. Fair lending considerations should be a factor used to determine strategic expansion, promotion of loan growth, marketing distribution, and lending outcomes. Application of loan programs, guidelines, terms, and conditions should include evaluation of fair lending risk along with a determination for the adequacy of controls to mitigate fair lending risk.
1 Steve Kuehl is a senior advisor for community and economic development for the Federal Reserve Bank of Chicago.
2 For further information on the HUD announced settlement with OneWest Bank, visit https://www.hud.gov/press/press_releases_media_advisories/HUD_No_19_113.
3 For further information on the CFPB issued No-Action Letter to Upstart Network, Inc., visit https://www.consumerfinance.gov/about-us/blog/update-credit-access-and-no-action-letter.
4 For further information on the proposed rule to amend the HUD interpretation of the Fair Housing Act’s disparate impact standard, visit https://www.hud.gov/press/press_releases_media_advisories/HUD_No_19_122.
5 The proposed rule does not define a “substitute,” “close proxy,” or “industry standard,” and the assessment to predict credit risk is on the model as a whole and not on any individual variable.