How AI Drives More Affordable Credit Access
By: Allison Huber, Head of Content Marketing at Upstart
For credit unions focused on improving financial inclusion, relying on traditional credit models can inadvertently shut the door on creditworthy borrowers. These outdated approaches not only limit consumers' access to capital, but also deny credit unions the chance to expand their member base.
Embracing more accurate credit assessment methods is the key to unveiling the untapped potential within the communities and regions that credit unions serve. AI-powered underwriting models not only discern a borrower’s true creditworthiness, but ensure that underserved communities are no longer overlooked and instead provided with equitable opportunities for financial growth. Both community development financial institutions (CDFIs) and low-income designated (LID) institutions benefit by partnering with a trusted AI lending partner to increase affordable credit access to underserved populations.
Who is underserved?
Since Hispanic consumers and Black consumers have lower credit scores on average, the current system makes it difficult for these groups to obtain credit at equal rates as Asian borrowers and white borrowers. According to average credit scores by race, Asian borrowers were at the top of the 850-perfect score with 745, considered Very Good. White borrowers were ranked in the Good category at 734, followed by Hispanic borrowers at 701 and Black borrowers at 677.
Traditional methods also favor mortgage holders over renters. This means that historic credit calculations rarely track consistent payment of rent and utilities - so a perfect rent payment history may not boost scores at all. Further racial divides include the fact that Black American college graduates owe an average of $25,000 more in student loan debt than White college graduates, making payment of other debts more difficult. In addition to race, younger consumers have lower scores on average due to a shorter history of loan repayment.1
In contrast to traditional credit scoring methods, the Upstart model approves 35 percent more Black borrowers and 46 percent more Hispanic borrowers. In addition, both underserved communities benefit from APRs below the credit score only model, with Black and Hispanic borrowers receiving 28.70 and 34 percent lower APRs, respectively.2
Upstart-powered lenders can also expect higher approval rates and lower loss rates, making financial inclusion beneficial to both low-to-moderate communities and lenders. Overall, the Upstart AI model approves 44.28 percent more borrowers than a traditional model at 36 percent lower APRs.3 In addition, the model drives more inclusive lending with 28.8 percent of Upstart Powered Loans going to LMI communities.4
A trusted AI Lending partner can expand inclusive lending
When one Tucson, Arizona, credit union sought to personalize its loan experience and grow membership, it began a search for a fintech partner. Vantage West wanted to improve credit decisioning accuracy, with an end goal of automatically deciding 70-80 percent of consumer loans. The Upstart Referral Network provided a way for Vantage West to diversify its geography beyond Arizona and grow members outside of its branch network.
Prior to Upstart, the credit union was not lending heavily to near-prime borrowers, but thanks to improved credit decisioning enabled by AI, the credit union could lend deeper down the credit spectrum without increasing losses.
“That’s what I love about our partnership – we have Upstart’s experience and partnership to dip into that near prime category so we can learn as an organization and build that capability,” said Jeremy Pinard, Vantage West, Chief Lending Officer. Pinard emphasized that understanding risk appetite was key in building the partnership, and after six months, loss performance performed better than the original 5 percent target.5
AI enables fairer decision-making
The “computerized brain” makes decisions based on logic versus human brains, which are subject to unconscious biases. The path forward, driven by statistical analysis, is to work toward better serving historically underserved communities, with closer collaboration between regulators, fintechs and special interest groups. Working toward a common goal of enabling better access to opportunity for more Americans is achievable with AI being the tool that banks and credit unions employ to prevent bias and to expand opportunity.6
A partnership with Upstart is integral to using AI with its additional data points to improve credit decisioning and uncover more qualified borrowers. Moving forward, partners have continued optionality as market conditions change and provide the revenue to invest back into better experiences for their members.
Access to credit equals access to the American dream. Partnering with Upstart enables CDFIs and LID institutions the opportunity to offer greater access to capital to LMI communities, improving more inclusive decisioning.
- Hanson, M., & Checked, F. (2023a, December 8). Student loan debt by race [2023]: Analysis of statistics. Education Data Initiative.
- As of October 2023, and based on a comparison between the Upstart model and a hypothetical traditional model. Upstart does not collect demographic data on borrowers. Upstart uses standard industry methodology to estimate borrower demographic status to conduct access-to-credit analysis comparing Upstart to traditional credit model outcomes. For more information on the methodology behind this study, please see Upstart’s Annual Access to Credit results here.
- Ibid
- Based loans originated on the Upstart platform from Jan 2017 to January 2024. LMI categorization is based on comparing median income in customers zip code vs median income within the MSA of that zip code
- Vantage West Credit Union partners with upstart to offer a more streamlined, digital-first borrowing experience. Upstart Network, Inc.
- Expanding credit access compliantly with AI