‘Debiasing’ Debt with Data
Inspired by his father’s struggles, Seke Ballard’s BetaBank digital-banking venture uses technology to give minority entrepreneurs equal access to capital
Re: Seke Ballard (MBA 2010); By: Ralph Ranalli
Topics: Finance-Banks and BankingFinance-Financing and LoansDemographics-Diversity
Photo by Danielle Scruggs
Seke Ballard (MBA 2010) says he owes the inspiration for BetaBank, his small business-focused digital-banking venture, to his father—and not just as a person, but also as an archetype for every minority entrepreneur who has struggled to find capital in a banking system that has historically favored white men.
When Ballard was a boy growing up in southeastern North Carolina, his father started a pulpwood logging company. “People would pay him to clear commercial and residential property, and he would take the lumber and sell it to the local paper mill,” Ballard says. From a single truck, his father built the venture into a successful business; but his plan to expand to neighboring states hit a major roadblock when local bankers rejected his loan application 13 times.
As an adult, Ballard discussed with his father the deeper significance of that experience. “He said African Americans, as a matter of pragmatism, cannot expect to sustain social and political advances that aren’t built on a firm economic footing,” Ballard says. “He believed this was the unfinished business of Martin Luther King Jr., and that it remains unfinished to this day.”
Digging into the problem, Ballard learned that owners of minority businesses are not only rejected for loans more often than white borrowers, but they also are charged significantly higher interest rates, on average. Overall, more than half of small business owners say their financial needs are not being met by current financial institutions, he says, and small-business lending is one of the least automated parts of the U.S. banking system. That means it relies mostly on commercial bankers making personal judgments about borrowers.
Ballard says when his father went to those banks, he was sitting across the table from loan officers who weren’t evaluating merely what was written on paper. “They were looking at him and applying subjective character tests. Do you look trustworthy? Do you fit the image in my mind of a successful businessperson?”
The solution, Ballard thought, was to “remove the human from the equation” and create a lending process driven solely by data. Working for Amazon in Seattle at the time, he went to some data-scientist friends armed with 1.2 million Small Business Association (SBA) loan records obtained through the Freedom of Information Act. They filtered out everything but the financial and operating records on borrowers, and put the data through a random forest regression, a type of analysis known for producing good predictions that can be readily understood.
When they tested the resulting model, the results were astounding. They found they could accurately predict when an SBA loan would be repaid in 98.2 percent of instances. “That’s an implied default rate of about 1.8 percent,” Ballard says. For comparison, one SBA loan program included in the data had a trailing 10-year default rate of 17.6 percent. “That’s pretty night and day, in terms of performance. It was a eureka moment for us.”
HBS gave Ballard the tools needed to transform his idea from an altruistic impulse to a real venture, starting with an emphasis on good data, he says. “I remember my first-year finance professor would always say ‘garbage in, garbage out,’ If you don’t really interrogate that data and understand what aspects of that data might create bias, then you absolutely will not build your algorithm with intentionality.”
Ballard decided to field-test his idea on another underserved business population: the fledgling cannabis industry. Good Tree Capital launched in 2017, making loans to cannabis growers and dispensaries in six states using the new predictive credit-risk model. And as it had in early tests, the model for evaluating the loan application was a success, achieving a remarkable 100 percent repayment rate.
“It was clearly time to take to the next step in evolution and create a whole bank around it,” he says.
The new company, Beta Financial Services, filed an application with the FDIC in 2021 to create BetaBank, a “digitally native” institution created to provide fair, accessible, and cost-effective deposit and lending services to small- and mid-size businesses. The application quickly caught the eye of Google, which forged a partnership to provide the estimated 66,000 software-development hours required to build a digital bank from the virtual ground up. Now called BetaBank Powered by Google, Ballard’s venture will be the first bank built on the Google Cloud Platform, a suite of cloud computing services that runs on the same infrastructure that Google uses internally for its own end-user products.
Ballard says the technology rollout and capitalization process will take place over the coming year; his plan is to be operating a $250 million bank in five years. But he also believes BetaBank can have a long-term impact well beyond the boundaries of its balance sheet. The goal, he says, is to provide fair access to capital. This means not only businesses owned by minorities and women but also the rest of the underserved small-business community.
“While we ourselves are a relative drop in the banking bucket, my strong hope is that we become a catalyst for other larger organizations to rethink their systems around lending,” Ballard allows. “In 30 to 40, years we’d like to look back and say BetaBank played a role in retooling the ways in which major banks lend.”