Automation may reduce racial bias.
That is according to a recent study of Paycheck Protection Program (PPP) loans — a finding that runs counter to commonly held concerns that automation could encode common bias.

The study, “Racial Disparities in Access to Small Business Credit: Evidence from the Paycheck Protection Program,” was conducted by researchers at New York University, including the National Bureau of Economic Research at the Stern School of Business, with assistance from David Snitkof, head of analytics at automation platform Ocrolus.
The study has received recent press for highlighting the disparity in lending between fintechs and small banks, however, not for the role automation plays in the findings.
The New York-based Ocrolus processed 12.9 million Paycheck Protection Program (PPP) loan documents and contributed its data to the study, Snitkof told Bank Automation News.
“PPP was this incredibly interesting use case,” Snitkof said. “Ultimately, two and a half million businesses that got funding through PPP did so with documents that went through us and went through our partners. So we had a lot of data to look at.”
The 86-page study combines that data with additional findings from the Small Business Administration and Ocrolus business partners in the space, Snitkof said. Researchers started with several hypotheses that they tested against the data.
The first finding: Fintechs made more loans to Black small business owners than did small banks.
“One thing we noticed was that small banks made 3% of their loans to Black small business owners. Fintech companies made 26.5% of their loans to Black small business owners,” Snitkof said.
At big banks, Black-owned businesses represented between 4% and 8% of loan recipients. The study specifically mentions Wells Fargo, which made 7.8% of its PPP loans to Black-owned businesses, as an example on the higher end.
That led researchers to question their findings. Their first step was to control for factors such as race, gender, geography, industry, business size, checking account balances, revenue and payroll size, and existing banking relationships. In short, they controlled for anything that might explain the significant difference in lending.
“Even if you account for all of those fixed effects, still, there was a huge disparity,” Snitkof told BAN. “If you were a Black business owner, you were way more likely to get funding from a fintech company than a bank.”
Snitkof theorized that automation played a role, and researchers explored a possible correlation. The data bore out his suspicion.
“What we found was that the degree of automation was what was most correlated with the diversity of the population that ended up getting funding,” he said. “So even for small banks, there are many small banks that started PPP in a very manual way and then they automated, and when they automated, the diversity of those getting funded went up.”
The study identified banks that added automation by pinpointing situations in which the difference between application and funding times decreased. The study did not differentiate between types of automation. Snitkof noted there were many ways to automate aspects of the PPP lending process.
Why would automation be the differentiator?
It levels the playing field, Snitkof said, and this finding runs contrary to common concerns about automation increases bias.
“The human mind is the ultimate black box, and so who knows what a loan officer, a banker, a processor, whoever, is thinking — conscious or unconscious,” he said. “But if you automate, then everything’s on a much more level playing field. You’re just looking at the data and you’re evaluating a business on the merits of the financial conditions in the business.”
The findings are just the “tip of the iceberg,” Snitkof said, adding that he hopes the research will be widely read by policymakers.
“PPP is this great natural experiment on something that’s typically been very hard to study, in terms of racial disparity and access to credit,” he told BAN. “Fintech has a very big role to play, in increasing access to credit, increasing the diversity of financial products, [and] increasing the diversity of those who are obtaining credit.”