Ever since last month’s news that Facebook had secured a patent that would allow lenders to assess a borrower’s creditworthiness by checking their friends’ credit scores, there’s been buzz around the possibility of a new social network credit score. Facebook has been mum on its plans to enter the credit lending space (a spokesperson did not return multiple requests for comment and it’s possible it won’t ever use the patent), but it wouldn’t be the first company to use social networks to evaluate someone’s creditworthiness. Startups have been looking for ways to shake up the credit lending industry and make loans more accessible (and affordable) for borrowers who find themselves shut out of traditional lending institutions.
Hong Kong-based Lenddo is already fine-tuning the practice of social network scoring in countries like the Philippines, Colombia and Brazil — countries where underbanked middle-class borrowers and small-business owners struggle to find credit because they have no traditional credit scoring institutions. Lenddo developed an algorithm that allows the company to predict the likelihood a borrower will repay a loan based on hundreds of data points gathered from their activity on social media sites like Facebook and Twitter. Borrowers are rated on a scale of 1 to 1,000. Part of this data comes from collecting financial information from users’ social media friends, who voluntarily agree to offer this information. The logic is that if someone associates with financially responsible people, they’re more likely to be responsible as well. For the last four years Lenddo has been testing its model and the company claims it does an even better job of predicting whether a borrower will repay a loan than traditional underwriting.
“Artificial intelligence is simply better at administering credit in a fair way,” says Lenddo CEO Jeff Stewart, who co-founded the company in 2011. “I think what you’re going to see globally is a move away from rules-based loan officer administered credit, which is expensive and uses a very small amount of data.”
In China, where 80% of adults have no credit history, peer-to-peer lending platform China Rapid Finance is now underwriting loans for customers based on social media data. The company’s source is Tencent, the tech giant that owns popular gaming and social media sites like WeChat and the Chinese version of Candy Crush.
Consumer protection groups in the U.S. are looking at companies that are experimenting with social media scoring overseas with trepidation. The use of “big data” in marketing to target specific consumer groups is already a controversial practice, mostly because few consumers ever realize they are being tracked. Ed Mierzwinski, of the U.S. Public Interest Research Group, says the idea of any social media company having access to users’ financial data — even if that information is handed over voluntarily — is troubling.
“I think Facebook and all the other powerful social media and Internet companies are throwing spaghetti at the wall trying to look for new ways to monetize the information they’re collecting,” Mierzwinski says. “But this is one I don’t think is going to work.”
If social network scoring is working so well for Lenddo overseas, however, who’s to say the U.S. won’t be next in line to adopt similar scoring models? There are some 45 million adults in the U.S. who are considered “credit invisible” — meaning they either have no credit history at all or their credit files are so thin that they’re all but unscorable. A blank credit history makes it difficult to qualify for even the most basic forms of credit.
Companies like Lenddo would have a hard time entering the U.S. market, though, because of our strict credit underwriting regulations. In order to administer a credit scoring system of any kind, companies have to prove the efficacy of their risk models and follow laws dictating how they gather information, how they distribute it and how they protect consumers’ privacy. The Equal Opportunity Credit Act would also pose a significant hurdle. The EOCA is meant to ensure that credit scoring models don’t inadvertently discriminate based on age, sex, race, or religion. Depending on how social network scores are used, discrimination is a real risk factor.
“[Dealing with regulators] is a beast in and of itself,” says credit expert John Ulzheimer. “That is something that any social media company will have to consider before actually selling their data for risk assessment.”
Stewart says Lenddo has no immediate plans to launch its services in the U.S., but not just because of the tough regulatory environment. From a profitability standpoint, there are far more people without access to credit in emerging economies than in the U.S. “I think in the U.S. there’s a lot of hesitation to try new things, which is why you’ll see innovation much slower here than in the emerging markets,” so for now the U.S. is lower priority, he says.
If Lenddo were to be used by American lenders, Stewart says it would be used as a way to give access to credit to people who have already been denied based on traditional scoring models. “We have to be very careful as a society to make sure we’re not using technologies to find new ways to exclude people from access to credit,” he says. “From the lenders we’ve talked to, they’re interested in … using these additional means to essentially offer credit that they wouldn’t be able to offer otherwise.”
As far as Facebook goes, the company may have been granted a patent for technology that could create a social media credit score but that doesn't necessarily mean it's going to use it. The patent came along with a bundle of other patents it inherited from the now-defunct Friendster a few years ago.
Given the way that credit lending works in the U.S., however, it’s unlikely even a social networking score would stop lenders from relying on co-signers for borrowers who have little credit or bad credit. Vouch, another alternative lending startup, combines the two models in this way. Users apply for loans and ask friends and family members to “vouch” for them. Their connections decide how much money they are willing to pony up in case their friend defaults on his or her loan and essentially co-sign for that amount.
University of Pennsylvania researcher Pinar Yildirim co-authored a 2014 paper on Lenddo and was positive about its potential in developing nations where credit access is limited. Yildirim and co-author Yanhao Wei liken social network scoring to employee referrals — an employer is likely to follow a recommendation made by an employee they deem trustworthy and make favorable assumptions about that candidate based on its relationship with the employee.
“Consumers who may have been otherwise denied loans for various reasons might actually start to qualify for loans,” says Yildirim. “That’s the upside of this practice.”
Have a question for us? E-mail us at firstname.lastname@example.org.