Alternative lenders are turning to homegrown scoring algorithms to evaluate ‘thin-credit’ customers
If you’ve ever sought a mortgage, a car loan or a credit card, you’ve likely been advised that the best deals go to those with the best credit scores.
But as a growing number of lenders and credit score developers invent new ways to measure creditworthiness, that advice may soon be out-of-date.
Some lenders are now rejecting the traditional credit score — which is typically packaged and sold by a major analytics company such as FICO or VantageScore — and are creating their own scoring algorithms that take into account a much different set of factors, such as a borrower’s education, work history, personal savings and investments.
Others are partnering with traditional analytics companies such as FICO and testing new underwriting models that fuse conventional credit information with alternative data sources, such as public records and utility payments.
“We just think we can do a better job by doing it ourselves,” says Louis Beryl, CEO and co-founder of alternative lender Earnest, which collects FICO scores for reporting purposes, but doesn’t use them to evaluate potential borrowers.
A different way to look at borrowers
Like other alternative lenders, such as Affirm, Avant, Vouch, Upstart, Basix and Social Finance (better known as SoFi), Earnest has come up with its own unconventional underwriting system for evaluating potential borrowers. “We build our own scores and what FICO does is build scores for other people,” says Beryl.
We just think we can do a better job by doing it ourselves.
|— Louis Beryl|
CEO, co-founder, Earnest
Instead of only looking at credit history, many of these lenders focus on alternate indicators that a borrower is a good credit risk, such as employment history, how much money a person has in the bank and how well they’ve managed their savings — in other words, “all of the things you think about a financially responsible person,” says Beryl. Other lenders don’t look at credit history at all or only use it in certain circumstances.
The alternative lender SoFi made news in early January 2016 when it announced it had officially become a “FICO-free zone” and had decided to stop using traditional credit scores to help evaluate its applicants.
“We found that the FICO score was anything but transparent, so we threw it out,” said SoFi CEO and co-founder Mike Cagney in a Jan. 12 news release. “Instead of relying on a three-digit number to tell us who’s qualified, we look for applicants who have historically paid their bills on time and make more money than they spend.”
SoFi co-founder Dan Macklin said that even before the announcement, FICO scores had never been a big factor in SoFi’s underwriting process — in part because so many of SoFi’s customers are young professionals with relatively short credit histories.
“For many of them, they just didn’t have the time to build up a credit history,” says Macklin. As a result, “We had a whole population of people who didn’t necessarily have great credit scores,” he says, but a deeper dive into their financial records indicated they were better credit risks than their FICO scores predicted.
For millennials, traditional credit not a priority
According to Brad Selby, a vice president at alternative lender Affirm, many young people also seem to be less interested these days in building up their credit histories, so it’s harder to evaluate them fairly using conventional credit metrics. With millennials, “We’ve found that building a traditional credit history is not the critical activity that it has been in previous generations,” said Selby in an email.
Millennials, for example, are significantly less likely than other generations were at the same age to own a home, and so are less focused on cultivating their credit. They’re also more likely to spurn traditional credit cards, which help build credit, in favor of prepaid cards or debit cards, which don’t.
To help fill in the missing information about millennials and other “thin-file borrowers,” some lenders such as Affirm are scooping up an even wider trail of data points they can use to size up applicants.
“Affirm considers thousands of data points about our applicants when assessing their ability to repay,” said Selby in an email. “Over time, our algorithm is able to determine which of these pieces of information correlate with repayment risk.”
We found that the FICO score was anything but transparent, so we threw it out.
|— Mike Cagney|
SoFI CEO and co-founder
For example, depending on the applicant, Affirm’s algorithms might take into account available social media signals, says Selby, or it might crunch other types of digital information, such as how a borrower previously interacted with the lender. “We are definitely in a new era for this type of underwriting. The hardware and software options that birthed ‘big data’ have paved the way, as has the profusion of data available.”
A fairer, less costly system?
By evaluating consumers more broadly than conventional lenders do, alternative lenders argue that they’re giving borrowers the opportunity to receive better rates than they would otherwise. “By actually increasing this thoroughness of underwriting, we’ll decrease the cost of credit to millions of people globally,” says Louis Beryl of Earnest, which offers rates as low as 5.25 percent.
Beryl says that if he were applying for a loan these days, he’d be upset if lenders didn’t give him credit for good financial behaviors that didn’t appear on his credit reports. “That would make me angry. Why aren’t they looking at this?”
Focusing on nontraditional information, such as work history and personal savings, also makes the financial system healthier by cutting out borrowers’ incentive to take out credit purely for the sake of increasing credit scores, argue lenders.
“The current system encourages people to develop a FICO score,” says SoFi’s Macklin, even if taking on more credit isn’t in their financial interest. For example, to earn a high score, borrowers are encouraged to take out a variety of loans, rather than pay for big-ticket purchases in cash, and to ask for credit limit increases. “That kind of FICO environment encourages people to do things that don’t make sense or are bad for their finances,” says Macklin. “Some people will be tempted by that higher limit on their credit cards and go spend that money.”
More data not always better data
Alternative lenders say their nontraditional approach to underwriting is an effective tool for identifying creditworthy borrowers missed by conventional credit scores. But not everyone’s convinced these new algorithms are as predictive as lenders claim.
According to Todd Baker, managing principal at Broadmoor Consulting, alternative lenders are taking a big risk relying on algorithms they developed and built themselves. The problem, says Baker, is that many of the lenders that tout new models for evaluating potential borrowers haven’t been around long enough to properly test them. To really know whether a new model can accurately predict a person’s creditworthiness, he says, lenders need to know how their scores perform during a downturn when many borrowers lose their jobs and struggle with their bills.
“The world is evolving toward using new criteria,” says Baker. But “the problem is not the new criteria.” The problem, he says, is that credit scoring models need to be tested over time before they can be trusted.
Traditional credit data is a solid source. It provides a common denominator to understand the creditworthiness of a consumer.
|— David Shellenberger|
FICO still has a leg up in incorporating new types of data into credit scores, he says. “Because they do have this long history, they can do some back-testing of any scoring algorithm they’re doing and have a higher degree of confidence than somebody doing this for the first time.”
New lenders say they have tested their underwriting methods and are confident in their models. “We’ve stress-tested them for different economic scenarios and believe that they will stand up to that kind of pressure,” says SoFi’s Macklin.
FICO gets into the game
In 2015, FICO announced it was teaming up with LexisNexis Risk Solutions and Equifax to produce its own alternative scoring model. The FICO score XD, which combines traditional credit data with new sources of information, such as property records and cable payments, won’t be available until later in 2016, says FICO’s David Shellenberger. But it’s currently undergoing testing by a dozen major credit card lenders and has so far received strong interest from new lenders.
According to Shellenberger, traditional lenders are keenly interested in incorporating nontraditional data into their credit decisions. But because they’re so heavily regulated, they’re limited by what types of data they can look at. “That knocked out a number of potential data sources for us,” he says.
As competition for new borrowers increases, lenders are becoming increasingly eager to use nontraditional data to augment conventional credit data and identify more creditworthy borrowers, says Ankush Tewari, a senior director at LexisNexis Riskview.
Use as a complement
But instead of replacing traditional credit scores, many conventional lenders simply want to use alternative scores as a complement to the scores they’re already using.
According to FICO’s Shellenberger, traditional credit scores are unlikely to be supplanted by alternative scores anytime soon. “Traditional credit data is a solid source,” he says. “It provides a common denominator to understand the creditworthiness of a consumer. It’s widely available, and I think that helps create a much more fluid and liquid credit market.”
That said, traditional lenders have been supplementing credit information with other data points, such as internal “behavioral score” data about what customers buy and how they’ve utilized their accounts, for years, he says. “There are different ways to augment credit data. Any smart lender who can find a way to do that, there’s benefit to them.”
“It’s not an either/or,” adds Shellenberger. “The smartest lenders are doing both.”