ATLANTA — A new analysis has found that more than 15% of Paycheck Protection Program loans — worth about $76 billion — may have been obtained fraudulently, with many of the loans handed out by online lenders.

So far, few of these “dubious” loans appeared to have been detected by authorities or repaid, according to the study by researchers at the McCombs School of Business at the University of Texas at Austin.

Online lenders, known as fintechs, have streamlined processes that were used to issue pandemic relief loans to applicants quickly, satisfying one of the program’s top objectives by rapidly putting federal dollars into circulation. But that may have been a boon for bad players looking to avoid the more rigorous underwriting standards used by traditional banks, which are in place to help detect fraud, the study says.

“I see very broadly that there’s a trade-off between quick and easy access to this government money and susceptibility to abuse‚” said Sam Kruger, an assistant professor of finance and one of the study’s authors. “And I think one of the things that our research sheds light on is a potential cost of that ready access.”

The federal Paycheck Protection Program was enacted to help small businesses stay afloat during the pandemic. The forgivable loans were expected to cover payroll, rent, and utility expenses while state and local governments ordered the businesses to close or when they had to reduce operations to fend off the spread of COVID-19.

At the time, businesses from beauty parlors to dentists and restaurants were forced to lay off employees.

To look at the potential for fraud in the program, the researchers analyzed more than 10 million PPP loans that provided more than $780 billion, using various indicators that loan information may be suspect.

One measure was whether multiple loans were granted at a residential address. Other primary indicators were whether loans went to businesses that weren’t registered or registered after the cutoff date of Feb. 15, 2020, to qualify for loans; whether reported pay to workers appeared high relative to the industry and business location; and whether businesses reported different job numbers on applications for another pandemic relief loan program.

In one example cited in the study, 14 loans totaling nearly $800,000 — all but one of them approved by Atlanta-based Kabbage — went to 14 businesses that all used the same address, a modest single-family home in the Chicago suburbs. The companies had “colorful business names” and all claimed 10 employees. Eleven of the loans were for identical amounts, $53,229. Only one of the businesses was registered by Feb. 15, 2020. The 13 other businesses registered only shortly before the loans were approved.

In another case, Kabbage approved four separate $20,833 loans, all at another “modest suburban Chicago home,” in July 2020. Two of the businesses were listed as lawn and garden equipment manufacturers, one did automotive repair, and one was a nail salon.

Photos of the property showed no evidence of any such businesses, the report says, and the borrower in the salon industry didn’t appear to have a nail technician license.

Particularly high percentages of flagged loans were clustered near Atlanta and New Orleans and surrounding areas, the report says.

The University of Texas researchers found plenty of suspicious loans issued by traditional banks, as it analyzed the loans made in three waves. But they found fintech loans to be “highly suspicious” at almost five times the rate of traditional lenders, with FinTechs making up nine of the 10 lenders with the highest rates of questionable loans.

Of the more than 1.8 million questionable loans, fintechs originated 52%, while their market share of the loans was just under 29%. Overall, the researchers flagged more than 31% of fintech loans as potentially suspicious, compared with 11.6% of loans by traditional banks.

“Not only did fintechs have higher rates of suspicious lending, but those rates of suspicious lending grow quite a lot over time, when you look at and compare round one to round two to round three,” Kruger said.

Kabbage was purchased last year by American Express and is doing business as K Servicing. Monday, a spokesperson for American Express referred questions to K Servicing, which did not respond to a request for comment.

The company’s website boasts of the PPP loans it has made and refers to a report stating it “served the most vulnerable of businesses represented as over 92% of all loans were under $50,000.” It also credits itself for saving 945,000 jobs.

Fintechs have some important distinctions from traditional banks that may explain in part their disproportionate share of suspect loans.

The researchers cited an independent study that found online lenders increased access to PPP loans by lending in more ZIP codes with fewer traditional banks, lower incomes, and higher minority percentages. Before the PPP program, another study the researchers cited found that fintechs fill gaps in lending to small businesses that are left by traditional banks.

“Online lending does not appear to be the problem in and of itself,” the researchers wrote. It noted that two fintech lenders, Square and Intuit, had the lowest suspicious loan rate among all lenders.

But those two online lenders have established relationships with customers, the report says.

Researchers also noted a potential incentive for all lenders under the program: the profits they could make, while they did not bear any credit risk themselves if loans were bad. Lenders were explicitly allowed to rely on the borrowers’ information.

The University of Texas report says Kabbage earned an estimated $188.8 million in fees from issuing more than 180,000 PPP loans worth $3.3 billion.

Kabbage has come under scrutiny previously by news organizations. One news report said that the company sent at least 378 PPP loans worth $7 million to likely nonexistent farms.