No traditional credit history? No problem. At least that’s how Lendbuzz, a startup that uses artificial intelligence to underwrite auto loans, sees it. The Boston-based software company announced Monday it has raised $150 million in new funding to expand its team and issue more financing backed by its machine learning algorithms.
Most of the cash—$130 million—for Lendbuzz is in the form of debt financing from four banks and an insurance company to fund car loans through the startup’s platform. The rest is an equity investment led by venture capital investor 83North.
“Banks and traditional lenders, the way they assess credit risk is based on FICO scores,” Lendbuzz co-founder and CEO Amitay Kalmar tells Xconomy. “If you’re 780 or above, you’ll get 3.9 percent” as your interest rate, he says as an example. “Then, if you do not have credit, either they do not serve you or your going to be priced very high at the highest [interest rate], similar to someone who has bad credit. The way we do it is, instead of punishing the entire population, we are trying to find other signals and other relevant data that allows us to assess their credit worthiness.”
Kalmar, a former Deutsche Bank investor, and chief technology officer Dan Raviv, a postdoc at the MIT Media Lab, founded the company in 2015 on the idea that there might be a better way to assess credit risk than a FICO score. Kalmar had his own experience with the you-need-it-to-get-it hurdle of the credit system when he came to the US 11 years ago from Israel.
“We moved to US, and I went to Bank of America. I opened first account but was denied for a $500 credit card,” he says.
The company initially focused on writing auto loans for people moving who had moved to the US. It soon realized there’s a broader pool of people with little to no credit history who also want access to financing, Kalmar says.
Currently, people with a limited credit history—or none at all—have few options if they want to finance a car. A 2015 study by the Consumer Financial Protection Bureau found 29 million American adults have no credit history and 19 million more have such limited credit histories it would be difficult to make a credit decision.
Lendbuzz starts by gathering data needed to feed into the company’s algorithm to make a credit decision. The data starts with a person’s employment status and history (including overseas work if they are a foreign national), and also bank account data. The accounts show payment of utility bills, payment of cellphone bills, also late fees paid, bounced checks, and average account balances, among other information.
“A lot of the things you can see in bank accounts enables us to build what we think is a very comprehensive profile of each borrower,” Kalmar says. “All the things you can see in a bank account are as, if not more, useful than a FICO score.”
The algorithms then rate the profile based on the data, essentially giving the person an independent credit score. When the algorithms recommend turning down a loan applicant, the machine learning system provides a rationale, which Lendbuzz then shares with the applicant.
Lendbuzz has originated thousands of loans since late 2016. About half of its business comes directly from consumers and the other half is arranged through car dealerships that use Lendbuzz.
Amitay points to data he says show “the underlying segment being served has good quality borrowers.” “Our algorithms are able to pick that up and approve loans for the ones we assess are good credit risk,” he says.
Most of Lendbuzz’s 40 employees work from its downtown Boston offices. The company plans to hire an additional 30 workers as part of the fundraising. Half of the positions will be focused on data science or engineering the company’s applications for consumers and dealerships, while the other half will be focused on sales.
