A major goal of the U.S. Small Business Administration? To encourage lenders to make more small loans and to include more minority, veteran and woman business owners in that borrower pool. On the face of it, it might seem that getting a small loan – often categorized as one of less than $500,000 – would be easier than aiming to borrow more.
Not so, said Brett Caines, CEO and co-founder of Lumos Technologies, a Wilmington financial technology firm.
“In 2023, just over 2 million small businesses looked for financing,” he said. “Eighty percent of them were looking for less than $250,000, 14% were looking for between $250,000 and $1 million. Less than 2% were looking for more than $1 million. Lenders want to focus on the $1 million loans because from an internal standpoint, they’re putting as much effort into processing a small loan as a large loan, and there’s a lot of work required. They are going for the $1 million loan every time.”
A large loan offers the lender a greater reward than a small loan, but the promise of reward isn’t the only factor making lenders hesitate to make smaller loans: They often regard potential loans of less than $500,000 as posing a greater risk.
Better calculations of a loan’s true risk lie at the heart of Lumos’ mission. Since its launch almost two years ago, the company has worked to develop predictive models that lenders can use to gauge a borrower’s risk more accurately. Its primary product, the Lumos PRIME+ scoring model, “improves underwriting efficiency and provides a more accurate and fairer risk assessment for small business loans less than $500,000,” according to the company’s website.
In building their predictive model, the Lumos team analyzed almost 30 years of data, starting with 1995, of small business loan performance.
“There were 2 million business loans of various types, various industries, in various economic cycles,” Caines said, explaining that its clients can explore different categories of businesses using multiple filters to gain a micro view of a potential borrower’s environment.
A lender could narrow the field to, say, the performance of restaurants in North Carolina and Virginia within a specific timeframe, looking at such factors as their creditworthiness, historical default rate, losses and lender recoveries.
“The purpose is to help banks develop strategies around lending,” Caines said.
After several months of collecting data, Lumos expanded its sources, drawing statistics from the Federal Reserve’s Economic Data Base, the Bureau of Labor Statistics and the U.S. Census Bureau. The team also began using machine learning to identify correlations among macro data. The broader data set and improved tech capabilities enable further slicing and dicing: Now, says Caines, lenders can look at the performance of, say, limited-service restaurants during periods of high inflation and high versus low unemployment.
“A 30-year data set is important,” he continued, noting that the years since 1995 have included boom-and-bust cycles in various sectors, as well as the recent impact of the COVID pandemic.
“The takeaway from all this is we are able to drill down and identify correlations with local factors on a state, county, even city level,” Caines said.
Sometimes lenders apply the wrong filters when they evaluate applications for small loans, Caines said, giving two contrasting examples.
One is a small business owner with a personal credit score of 800. He applies for a small business loan to start a restaurant in a city with 10% unemployment. He’ll likely get the loan, but the environment is not conducive to the restaurant’s success, Caines said.
The second is a small business owner with limited credit or a low credit score. But she’s in a city with 5% population growth, 2% unemployment and 4% wage growth. She could be the better risk, said Caines, noting that the Lumos predictive model takes that kind of information into account.
“Our PRIME + scoring model is specially built to help lenders quickly assess and decide about loans of less than $500K,” he said, adding that Lumos has run small business loans through the model to test its accuracy. Potential clients can do the same, sending their approved small loan applications to the company to be run through the model and its risk assessment compared with what actually happened with the loan.
The intent of Lumos’ PRIME + scoring model and its corollary products is to help lenders “say yes more often with confidence,” according to Caines. “Our goal as a company is to expand small business lending. Lenders turn down more good loans than necessary. Sadly, because many small business owners can’t get a loan through a bank, they put the money on a credit card at an insane rate.”
Lumos isn’t the only fintech in the neighborhood developing ways to make small loan processing more efficient and, therefore, more appealing. Live Oak Bank, one of the nation’s top SBA lenders, is developing technology that will streamline its own systems.
In Live Oak’s year-end earnings call last January, President BJ Losch said that the bank is working on “new products, like technology solutions,” making it easier for Live Oak to make smaller-dollar SBA 7(a) loans.
For its part, the SBA is working to smooth some wrinkles that impede smaller-loan lending. New policies for 2024 include waived origination fees for loans under $1 million, increased loan limits and a 90% SBA guaranty for small businesses that earn international revenues. It’s also simplifying the approvals process and expanding eligibility.
The agency also has recognized the so-called Silver Tsunami: the rising number of baby boomers reaching retirement age and looking to sell their successful businesses. New rules aim to make it easier for a would-be owner to acquire an existing business from a retiring owner.
In the January earnings call, Live Oak Chair and CEO James “Chip” Mahan said that he and his colleagues are excited at the prospects the revised SBA rules offer.
He noted that there are 5.3 million businesses that generate between $100,000 and $500,000 in annual revenues.
