Online lenders portray themselves in virtuous terms, arguing that digital innovation has enabled underserved U.S. consumers to refinance expensive debt at lower interest rates, while also boosting their credit scores.

But provocative new research from the Federal Reserve Bank of Cleveland casts doubt on that rosy outlook. Indeed, the study finds that borrowers who use online loan products end up in deeper debt than similarly situated consumers who do not.

“Consumers in the at-risk category — those with lower incomes, less education, and higher existing debt — may be the most vulnerable,” the research states.

The study, published Thursday, is likely to spark intense debate. One of its findings is that consumers who take out online loans, sometimes called peer-to-peer loans or marketplace loans, likely have access to traditional banking services.

That conclusion is in tension with another recent study by a different team of Fed economists. Those researchers, based at the Philadelphia Fed and the Chicago Fed, found evidence consistent with the argument that online lenders have played a role in filling the credit gap that has been left by banks.

“We are scratching our heads,” Nathaniel Hoopes, executive director of the Marketplace Lending Association, an industry trade group, said in an email, “because the Philadelphia and Chicago Federal Reserve recently conducted a more granular study and reached the opposite conclusion as this Cleveland Fed research.”

Online consumer lending has enjoyed tremendous growth throughout this decade. In 2010, digital lenders originated $249 million in unsecured personal loans, according to a recent study by the credit bureau TransUnion. By last year, the annual loan volume had grown more than ninetyfold.

peer to peer lending chart

The fledgling industry has generally drawn encouragement from policymakers. Near the end of the Obama administration, the Treasury Department declared that online lending has the potential to broaden access to affordable credit for underserved consumers.

The Cleveland Fed study reaches an unhappier conclusion. It finds the online industry’s loans “do not go to the markets underserved by the traditional banking system” and “resemble predatory loans in terms of the segment of the consumer market they serve and their impact on consumers’ finances.”

To conduct the study, the researchers relied on data from TransUnion. They identified roughly 90,000 individuals who took out an online loan between 2007 and 2012. Then they used a statistical technique to match those borrowers with other consumers who had similar financial profiles, but did not take out online loans.

The study finds that the consumers who took out online loans grew their other debts by about 35% more over the next two years than did their counterparts who did not take out the loans.

It also found that consumers who borrowed online had lower credit scores, more delinquent accounts and more total debt outstanding two years later than the similarly situated consumers who abstained.

The findings suggest that online loans — which are often three-year to five-year installment loans of up to $30,000 to $40,000 — are enabling some U.S. consumers to overspend. Even if borrowers use the loans to pay off existing credit card debt, there is nothing to stop them from running up large new tabs on those same cards.

Yuliya Demyanyk, a Cleveland Fed economist who co-authored the research, said in an interview that folks who apply for online loans are also getting credit from the traditional banking system.

“They’re not underbanked, they’re sort of overbanked,” she said.

Demyanyk also expressed alarm about rising late-payment rates in the online lending sector. Her research found that delinquency rates for online loans originated in 2013 were higher than for those originated in 2012, which were higher than for those originated in 2011. A similar pattern of rising delinquency rates emerged in the subprime mortgage market between 2001 and 2007.

“I actually had a déjà vu moment,” Demyanyk said.

One potential critique of the Cleveland Fed study is that it does not distinguish between various online lenders that target different consumers and charge widely diverging interest rates.

For example, Social Finance, or SoFi, refinances student loans for consumers with strong credit, while LendingClub offers personal loans to borrowers who often have somewhat lower credit scores, and certain other companies charge annual percentage rates well above 36% to customers who might otherwise turn to a payday lender.

The Cleveland Fed research looked at loans classified by TransUnion as online loans. A list of the specific companies included in that category was not available.