Undocumented immigrants and credit invisibles represent both an enormous opportunity and responsibility for lenders. Recent advances in AI have made this dual-track to financial inclusion and profitability even more compelling. Daniel Chu, CEO and founder of Tricolor Holdings, examines AI-powered underwriting in the subprime auto industry as a glimpse into how to responsibly serve these customers.
The global pandemic has helped draw attention to undocumented immigrants as an essential front-line worker in modern American society. From hospital staff to grocery employees to restaurant delivery drivers, these individuals emerged as perhaps some of the most resilient contributors to the U.S. economy over the past five months.
However, these financial and workplace contributions stretch much further back. In 2018, undocumented immigrants paid $31.9 billion in federal, state, and local taxes while exercising a total spending power approaching $220 billion. And it’s been estimated that the agriculture industry would shrink by $30-60 billion annually without undocumented workers.
Looking ahead, this impact and importance will continue to grow far into the future. According to the New American Economy, 89% of some 11 million undocumented immigrants in the U.S. are of working age. In fact, the overall U.S. Hispanic market represents the eighth largest economy in the world – larger than Brazil and almost twice the size of Mexico. It’s growing 28% faster than the broader American economy, with Hispanics comprising 82% of U.S. workforce growth since 2008.
A Persistent Lack of Access to Affordable Credit
Despite its size, impact, and opportunity, the Hispanic population continues to struggle with access to affordable credit. A large percentage of this population is unbanked and part of an overall population of as many as 100 million people in America who don’t have a credit score or have a limited credit history.
This is because traditional lenders use FICO-based models to decide who is approved for credit and at what interest rate. While easy and intuitive, these “scorecard†methods are limited in their ability to quantify risk for everyone because they cannot generate sufficient segmentation power for credit invisibles or those with little or damaged credit history. A FICO-based model simply will not work for an undocumented population, showing high risk and unreliability when the opposite is true.
A situation made even worse because lenders are tightening credit requirements even further in reaction to the pandemic. According to Equifax, the percentage of auto loans originated to subprime borrowers is contracting and had dropped to 50% of pre-pandemic levels in August. Furthermore, while the credit score of subprime loan originations over the last couple of years has consistently averaged 550, it has increased sharply to approximately 575 over the most recent five-month span.
Without a FICO score, undocumented workers are unable to obtain reasonable credit, so they often turn to predatory lenders. Those impacted negatively by COVID-19 or who experience income fluctuations rely on high-interest-rate loans to set off a downward financial spiral.
Learn More: Fund Management: Leveraging AI To Streamline Data Operations
AI Unlocks Profitable Responsible Lending
Lenders and bureaus must make an effort to better score these credit invisible populations. Given the role of undocumented workers in society, the responsibility to our essential workers, and the opportunity to lend to such an enormous population is undeniable. Better tools have been proven effective at segmenting credit customers that exist below traditional FICO thresholds, but few have embraced them so far, choosing instead to remain inside a more conventional box.
In particular, leveraging artificial intelligence (AI) allows lenders to go beyond the FICO score and use non-conventional variables at scale to attain superior loan performance while also improving consumers’ access to credit. Consider that 85% of the auto borrowers in our portfolio at Tricolor would qualify for the Freddie Mac affordable lending program with income at or below 80% of their county’s median income. However, we’ve been able to effectively offer interest rates far below other well-known subprime auto competitors like Santander Consumer and with loan losses that are still 50% less than our industry average.
This is possible because AI and machine learning can more effectively segment borrowers on nontraditional attributes, producing reliable indications of the ability to repay even for those without a credit history. Our use of AI allows us to identify patterns among over 100 non-traditional attributes and their relationship to default, a competency that makes up for lack of credit history. This scoring system has helped us carve out six distinct credit categories (ranging from A+ to E) within an entire category of subprime customers that other underwriters lump into one credit grade.
Lenders that can bifurcate lower risk from higher risk applicants within this normally muddy segment stand to create adverse selection relative to their competitors. To succeed, lenders should first lower margins across all credit grades. As Jeff Bezos puts it, “There are two kinds of companies——those that work to raise their prices and those that work to lower them.â€Â While that sounds crazy at first, it actually helps attract higher-quality borrowers that will come to them for the more favorable terms and lower rates they offer.
Then, lenders can slope their financing terms sharply across grades to offer the most attractive terms to the highest-grade applicants within that segment. Conversely, this sloping will result in the poor conversion of lower grade applicants and drive even greater adverse selection for the competition.
In our case, this strategy has helped to create a defensible moat around our responsible lending product in the subprime auto category. Consumers know they can tap into affordable credit for high-quality cars. At the same time, capital market investors have learned they can trust the precision of our credit rating system within their portfolio.
Learn More: Is Cybersecurity the Biggest Challenge for Today’s Fintech Companies?
The Path To Financial Inclusion
Direct-to-consumer fintech models have a distinct advantage in leveraging AI and machine learning models to more cost-effectively and accurately underwrite credit for the unbanked and underbanked consumer. But this capability can also be more widely adopted by mainstream banks and lenders to supplement their prime customer lending operations.
This embrace of decision science can significantly advance social efforts to further financial inclusion and create a path to affordable credit access for a segment of the population that rightly deserves it. But more importantly, for these lenders, the responsibility to serve this demographic also comes with an enormous opportunity to tap into one of the largest and fastest-growing consumer markets in the U.S.
With the power of AI, it has never been more possible to do well by doing good.
Let us know if you liked this article on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We would love to hear from you!