Two CUNA Strategic Services alliance providers are empowering credit unions to improve members’ financial well-being by expanding access to low-dollar loans and using artificial intelligence (AI) to make more socially equitable loans through alternative credit scoring models.
QCash Financial, a credit union service organization formed by $4.6 billion asset Washington State Employees Credit Union (WSECU) in Olympia, Wash., allows credit unions to make small dollar, unsecured loans in less than 60 seconds without the use of a credit score.
QCash has fine-tuned its decision engine after more than a decade of lending experience, data analytics, automated underwriting, and real-time funding. This gives struggling members an alternative to high-cost predatory lenders.
“What we’re really talking about is when a member is in a life event; when an emergency happens,” says Denise Wymore, QCash marketing manager.
In a 2021 survey, WSECU learned that 59% of its members have used QCash loans for living expenses or family emergencies.
QCash uses relational credit scoring model that considers member standing and eligibility, deposit history, and credit union relationship.
It’s based on the model credit unions relied on for years with traditional select employee groups. “This is what a loan officer in the 1980s did,” Wymore says. “It was about character, capacity, and collateral. This is the ‘80s automated with no FICO score at all.”
WSECU learned it could make loans to members in low credit tiers that may not have been approved through traditional scoring methods, she says.
Members use the loans for auto repairs, bill consolidation, unemployment, the birth of a child, and other life events, Wymore says. “Without those loans we know where they would be: at a payday lender.”
Zest AI uses machine learning and AI to produce more accurate and equitable credit scores in providing loans to members, says Teddy Flo, chief legal officer.
“We use compliant data sources to produce a risk score that’s fairer to women and people of color,” he says. “This can transform credit union technology, but more importantly the lives of credit union members.”
Traditional credit scoring models essentially divide loan applicants into two groups—“good” and “bad” applicants—with a straight line based on their ability to repay.
With more advanced algorithms, machine learning provides a more nuanced line that considers more factors and makes fewer errors than the traditional linear system, Flo explains.
Not only do machine learning and AI approve more borrowers, they provide more statistical accuracy regarding their ability to repay, he says. In addition, machine learning and AI scores are a more accurate predictor of a borrower’s ability to pay during economic downturns.
“By replacing high-risk borrowers with low-risk ones, you can approve significantly more borrowers without increasing risk,” Flo says.
Perhaps most important, Zest AI has developed scoring models that are more accurate and equitable in approving loans for underserved communities, including, Black, Hispanic, Native American, and elderly consumers.
Overall, the final model can increase approvals by 35% across those population segments.
Flo highlights a credit union that saw a 38% increase in loan approvals without any change in its risk profile.
“Credit unions now have more choices,” he says. “They can get an enormous jump in statistical accuracy and become a lot more fair in their lending practices. This is critical in allowing credit unions to serve their mission.”
Both Flo and Wymore note the unintended zero-sum effect of declining a loan application: Members who are denied loans are highly unlikely to return for another opportunity—a reality that’s not lost on any experienced loan officer.
“But when you approve them, especially when they’re denied by other providers, they remember that for life,” Flo says. “So when you think about the life cycle benefit, the returns for both the member and the credit union—using advanced technology—are astounding.”