Disparity in Underwriting Is Everyone’s Problem
Jan 14, 2024
By John San Filippo
On December 14, 2023, CNN published a report detailing how its analysis of publicly available information from the Consumer Financial Protection Bureau (CFPB) showed that $168 billion Navy Federal Credit Union approved more than 75% of white borrowers who applied for a new conventional home loan in 2022, but approved less than 50% of Black borrowers who applied for the same type of loan. This was the broadest disparity among major lenders for that period.
The report was followed five days later by the filing a class-action suit against the Vienna, Va.-based credit union based on the CNN report. Then on January 11, 2024, dozens of lawmakers signed two different letters to federal regulators calling for an investigation into Navy FCU’s lending practices, again based on the noted CNN report.
However, as Andre M. Perry, senior fellow at Brookings Metro, pointed out in an MSNBC opinion piece on December 22, 2023, this incident lays bare a systemic problem that plagues the entire lending industry. “The problem with calling one lender on the carpet is that it often lets industry practices off the hook,” noted Perry. “Where there’s smoke, there’s fire, and when it comes to discrimination in lending, racial disparities represent the smoke not the fire. The fire is burning in foundational elements of underwriting, servicing and regulations that not just Navy Federal Credit Union abides by, but that other institutions abide by, too.”
Navy FCU echoed this thought in a press release, stating in part, “The statistics in the [CNN] article do not appear to have considered several key credit criteria that all financial institutions, including Navy Federal, rely on to assess mortgage applications.” And while the credit union declined further comment to Finopotamus, we did speak with three industry experts who are proponents of various alternative underwriting models that they claim can solve for at least part of this systemic disparity.
“This is not a new issue; it's an industry-wide phenomenon,” Pankaj Jain, president of the New York City -based artificial intelligence (AI)-based underwriting provider Scienaptic AI, told Finopotamus. “That’s one of the reasons why there 45 million people who are excluded from the financial ecosystem. The FICO score or the Vantage score or any of those standard scores are built to help a super prime, prime type of population.” He noted that those borrowers represent at best only 25% of the U.S. population.
“Legacy credit reports and scores are widely understood to contain inherent biases stemming from various historical and structural aspects of the traditional consumer credit system,” added Brian Reshefsky, CEO of the Chicago-based Edge, a company that scores potential borrowers based on a detailed cash-flow analysis. “Much of the problem with the traditional credit bureau paradigm is that it only considers past balances and payments on tradelines reported to the national bureaus.”
“It’s well-known that these legacy technologies have disparate impact, that is, they have unintentional targeting of protected groups,” observed Jeff LoCastro, CEO of the Sunnyvale, Calif.-based Neener Analytics. His company can determine a borrower’s credit worthiness based on their social media behavior.
“In the past the credit bureaus have been able to get special dispensation from governmental regulators because they argue that they cover a large percentage of the market – but do they? About 56% of the U.S. population alone is thin-file, no-file, or credit challenged,” LoCastro said. “The bureaus will admit that 35% are thin or no-file, but they completely forget about the 21% who bounce back and forth between visible and invisible consumers.”
New, Better Technology
“The future of risk decisioning is human data – data that is created, owned, and controlled by the consumer, and shared by that consumer directly through an opt-in with whomever they feel is offering a good value exchange,” continued LoCastro.
He added that this is in stark contrast to the current system wherein consumers have very little control over their own data and are subject to the shortcomings of the credit bureaus. “With human data, there is nothing to hack or steal and the data is 100% portable. The consumer doesn’t have to ‘game’ anything or wait 10 years to obtain even basic credit.”
According to Jain, contributing to the problem is the fact that credit bureaus only look at a snapshot of the borrower’s financial situation as it stands right now rather than evaluating trending data (i.e., whether the borrower’s financial situation is improving or declining over time). The bureaus also only consider a handful of available attributes.
“We use the underlying data and not the snapshot, but the trended information and how people are trending upwards or downwards,” said Jain. “This gives a lot more information so that you can really evaluate people. And then we append that with the alternate data which represents about 400 to 500 attributes around their economic story.” The result is a much fairer, unbiased evaluation, he offered.
“Cash-flow underwriting introduces a more complete view of an individual’s financial health, taking into account not only payments on obligations not reported to the bureaus like rent, utilities and buy now pay later (BNPL), but also income and balance behaviors,” added Reshefsky. “Taken together, these insights are both more predictive of performance for individuals below the prime credit score threshold and, importantly, free of traditional credit report biases.”
A Better Tomorrow
“A person can neither fundamentally improve their standard of living nor can a society create a sustained middle class without access to financial goods and services,” stated LoCastro. “The only way to get access to those is if the providers of those goods and services understand the risk of the engagement. Period. There is no other way. Bad risk decisioning destroys people and communities.”
Jain added that the Navy FCU situation should serve as a “wake-up call” for all community financial institutions. “There’s a conversation that needs to happen,” he said. “Put together a plan – how to deploy it, how to implement it, what are the timelines and all that. A lot more industry-wide education needs to happen. If you’re not going to change anything, nothing is going to change.”
Empowering Lenders: Scienaptic AI Integrates with Loan Director from Fiserv to Redefine Credit Decisioning
Loan Director clients can now seamlessly access Scienaptic’s AI-powered underwriting