Upstart, NAACP Legal Defense and Educational Fund, and Student Borrower Protection Center announce second in a series of reports on fair lending
SAN MATEO, CA (November 10, 2021) — Upstart (NASDAQ: UPST), a leading artificial intelligence (AI) lending platform, the NAACP Legal Defense and Educational Fund, Inc. (LDF), and the Student Borrower Protection Center (SBPC) today announced the release of the second report from the independent auditor under its groundbreaking agreement to transparently review the fair lending outcomes of Upstart’s model and assess best practices in the use and testing of alternative data in AI lending platforms.
The agreement among Upstart, SBPC, and LDF sets a standard for cooperation, transparency, and independent oversight of automated underwriting models. As more companies experiment with and deploy alternative data and machine learning algorithms in underwriting, the work accomplished under this agreement will serve as a guidepost to align industry underwriting and fair lending testing practices with the goals of advancing racial equity in the credit marketplace.
Among the key initial findings of the analysis provided by the independent auditor, Relman Colfax reported:
- No significant price disparities exist for any protected class groups, suggesting protected classes did not, for the time period studied, receive unfair pricing offers from Upstart’s model.
- Statistically significant approval disparities existed for Black applicants for the time period studied when compared to non-Hispanic white applicants, which warrants further steps of fair lending analysis to confirm if there are any alternatives or modifications available to Upstart. Relman Colfax reported that these disparities, which did not take into account differences in legitimate creditworthiness metrics, do not, standing alone, demonstrate a fair lending violation.
- The variables used in Upstart’s model, standing alone, do not appear to be proxies for race and national origin. This is a key finding, as Upstart’s use of nontraditional variables such as education in underwriting was a topic that gave rise to the agreement. On this point, the report also highlighted that there are inherent limitations in analyzing variables to determine if the interaction of variables within the AI/ML model function as proxies for protected classes, and as such, cannot technically eliminate the possibility that proxies exist.
The parties offered the following public statements on the process to-date and subsequent steps and goals:
“Upstart has proven over the years that consumers are much more than a three-digit credit score and is committed to using its AI models to improve lending outcomes for all protected classes. Working transparently with advocates like LDF and SBPC on the next phase of this process will help Upstart build the most equitable, inclusive, and accessible underwriting platform available. More broadly, I hope that this series of reports from Relman Colfax is a guide that can help all lenders better understand the obligation to test transparently and to improve on the status quo by relentlessly optimizing models for fairness and inclusion, as well as accuracy.” – Nat Hoopes, Vice President and Head of Public Policy and Regulatory Affairs at Upstart.
“We applaud Upstart for participating in this transparent and rigorous fair lending testing and independent monitorship of their lending platform. We share the belief that the introduction of AI into the credit marketplace should disrupt and not replicate histories of racial discrimination and economic injustice directed at Black communities and other protected classes. Both traditional underwriting approaches and newer AI models can reproduce—or exacerbate—underlying inequality in society that uniquely burdens Black credit applicants. The question is whether lenders work to uncover disparities in their platforms and seek out alternative approaches. We look forward to working further with Upstart and the Monitor to uncover disparities in their platform and seek out alternative approaches—something we hope all lenders will commit to doing.”- Ashok Chandran, Assistant Counsel at LDF.
“The progress made under this agreement shows that all lenders should be transparent and rigorous about testing their models with independent third parties. We want to make sure that complex algorithms generate fair outcomes for borrowers, millions of whom remain heavily burdened by student debt. The process we have chosen to work on with Upstart could help guide the lending industry to set high standards when using new technology and data sources.” – Mike Pierce, Executive Director of the Student Borrower Protection Center.
Upstart (NASDAQ: UPST) is a leading AI lending platform partnering with banks and credit unions to expand access to affordable credit. By leveraging Upstart's AI platform, Upstart-powered banks and credit unions can have higher approval rates and lower loss rates, while simultaneously delivering the exceptional digital-first lending experience their customers demand. More than two-thirds of Upstart loans are approved instantly and are fully automated. Upstart was founded by ex-Googlers in 2012 and is based in San Mateo, California and Columbus, Ohio.