If you have a lot of friends on Facebook, it means you are popular in certain circles. Now, some lenders think you may also be creditworthy.
Lenddo, Neo Finance and Affirm are among a growing number of credit companies that use personal data found on social networking sites such as Facebook, LinkedIn and Twitter to assess a consumer’s credit risk, according to an April 1 story in The Wall Street Journal. They believe that a person’s social standing, online reputation and/or professional connections are factors that should be considered when extending credit, especially to someone with a scant or spotty credit history who might otherwise have trouble getting a loan.
These start-ups hope to exploit a perceived shortcoming in traditional loan criteria based on FICO credit scores, in which people who have missed payments or lack borrowing experience would automatically be considered risky bets and penalized with higher interest rates on their loans. Or they could be turned down altogether, regardless of any mitigating circumstances such as a medical emergency or recent immigration to America. Lenddo, Neo Finance and Affirm make money principally through fees or commissions charged for each transaction. But whether or not their business models will last in the long run is another matter.
Take Neo Finance in Palo Alto, Calif. It targets auto loan borrowers with short credit histories who mostly qualify for high interest rate loans from conventional lenders. Neo can offer lower interest rate loans after considering information such as an applicant’s job history and the quality of connections on his or her LinkedIn profile to gauge future earning potential and employment stability. It does consider FICO, but only to look for any red flags. Neo does not use the score to calculate risk, the article said.
Hong Kong-based Lenddo takes it one step further by using a debtor’s social connections to exert pressure if he or she defaults on payments, according to the Journal. For example, the start-up will tell customers’ Facebook friends if they haven’t paid, and the friends’ Lenddo scores could suffer if the customer fails to repay the loan. Lenddo calculates its own credit score of 1 to 1,000 after looking through 100 databases and social networks for such things as an applicant’s location and number of connections. San Francisco-based start-up Affirm, which is led by PayPal co-founder Max Levchin, makes it easy for consumers to pay for goods and services using their smartphones — with two taps on the screen — and gives them 30 days to settle their bills without fees.
Some companies use data gathered from social networks to assess a borrower’s credit risk in countries where standardized consumer credit scores may not be available, and where loans are granted based on one’s reputation. Lenddo gives small loans to borrowers in developing nations to help improve their quality of life after scouring the online profiles of applicants. To explain the rationale behind its business model, the company’s website quotes Wall Street banker John Pierpont “J.P.” Morgan, who once said that character is more important in gauging one’s creditworthiness than money or property.
In an age where there is an explosion of personal information on the Internet, a phenomenon dubbed “Big Data,” incorporating social data for credit-scoring purposes is perhaps inevitable. But exactly which pieces of information will prove useful, and which won’t, will require time to figure out. “It’s going to take years to understand what measures are truly valid,” says Peter Fader, co-director of the Wharton Customer Analytics Initiative and a marketing professor. “It’s the Wild West … like the early days of FICO.”
Another Tech Era Begins
In the 1950s, engineer Bill Fair and mathematician Earl Isaac, both working at the Stanford Research Institute in Menlo Park, Calif., came to believe that applying advanced mathematics and statistics to analyze complex operational business processes would enable a company to make better decisions. In 1956, with $400 each, they founded Fair, Isaac & Co. and developed the FICO credit scoring system, which became the bedrock of credit-risk modeling in scores of industries. They believed a company’s operations contained a treasure trove of information that can be mined and analyzed, so decisions can be “methodical and data-driven — not just guided by gut feelings and consensus,” according to The Deciding Factor, a book written by former CEO Larry Rosenberger, vice president of corporate strategy John Nash and journalist Ann Graham. Fair, Isaac was renamed FICO in 2009.
Around the same time, another revolution serendipitously began to take shape. IBM, under president Thomas Watson, Jr., introduced the IBM 701, its first commercially available, large-scale mainframe computer to be manufactured in quantity. As computers became more ubiquitous, companies computerized their business operations. For the first time, more corporations were able to electronically capture and store data about customers, such as their purchases, on a massive scale. It led to an explosion in computing power and data mining. Indeed, Watson’s “decision to commit IBM’s business machine vision to computers would kick-start the information technology revolution in business and the beginnings of decision management in large corporations,” Rosenberger, Nash and Graham wrote.
Today, a similar tectonic shift is taking place in technology and data. Computers have become smaller and more powerful, data storage is getting cheaper and cloud computing makes information accessible from anywhere. Digital consumer data has never been this voluminous: 2.8 trillion gigabytes were created, replicated or consumed in 2012, according to research firm IDC. The figure is expected to double every two years until 2020 as people around the globe share information on social networks and elsewhere, using millions of connected devices.
“In the 1950s and 1960s, there was a huge paradigm shift. All of a sudden, computers became commonplace,” says Robert Stine, professor of statistics at Wharton who researches credit scoring. “Now, we’re seeing a new leap in the kind of data [accessible], and the technology that is available to manipulate that data.”
But like FICO in the early days, companies are grappling with which social data will become predictive of credit behavior over time. Is making racist comments on Facebook correlated with lack of creditworthiness? Neo founder Navin Bathija told The Economist in a February 9 story that he believes enough data would be available within a year to determine whether there is a link. In the same article, however, ZestFinance founder and former Google CIO Douglas Merrill said people who type only in lower-case, or upper-case, letters are more likely to be deadbeats, all other things being equal. ZestFinance uses Google-like search algorithms to assess a person’s credit risk by scouring thousands of potential credit variables. Traditional underwriting methods look at only a “handful,” the company said.
Meanwhile, FICO and social data have a distinct difference: FICO mainly uses quantitative data. According to the company, 35% of the FICO credit score is based on a consumer’s credit history, 30% from amounts owed, 15% from length of credit history, and 10% each for new credit and types of credit used. The score rises and falls based on consumer behavior. Stine says the score also could be customized to fit the needs of different industries. But using a person’s payment history to predict future behavior makes sense; correlating the number of Facebook friends to trustworthiness is another matter. “Just because you have a lot of friends doesn’t mean you have a high standing in the community,” notes Stine, pointing out that criminals are not exactly friendless.
Where social data is most useful is when it is applied to people with little or no credit history. “It’s an additional, valuable data source that could be quite predictive of someone’s behavior,” says Eric Bradlow, co-director of the Wharton Customer Analytics Initiative and a professor of marketing. “It’s going to be especially valuable when there is sparse data on an individual.” Looking at new variables is standard practice anyway when building predictive models in credit. “They’re constantly looking for variables that add predictive power to their score,” he notes.
Many lenders already use qualitative data in conjunction with FICO scores. For example, when it comes to people with little or no experience with credit — such as college graduates — the firms will try to find people who are similar to the loan applicant to forecast the kind of repayment behavior to expect. “If you’re like other people who match these characteristics, I have an idea of your paying characteristics as well,” Stine notes. But the applicant most likely will be offered credit under more stringent terms until he or she is proven to be trustworthy, such as a credit card with a lower credit limit and higher interest rate.
As long as data culled from social media is used to supplement FICO, which has been proven to work for nearly 60 years, credit companies can more easily steer clear of trouble. “FICO is a very, very good baseline,” Fader says. “The tires have been kicked many, many times.” What is folly is throwing away FICO altogether and relying only on data gathered from social media, without a mountain of evidence proving a strong correlation with repayment behavior. “Throwing away the FICO score and reinventing the wheel? I’m not a big fan of that,” Fader adds. “I wouldn’t bet the farm on it.”
Social Data Hits a Wall
A big challenge to using social data for credit assessment comes from consumer protection laws. According to the Equal Credit Opportunity Act, credit must be extended to all creditworthy applicants regardless of race, religion, gender, marital status, age and other personal characteristics. Most of that information can be gathered from a Facebook page. “It’s a very different world these days … but the laws still apply,” says David Musto, a Wharton finance professor who studies consumer credit. He noted that some companies have sparked controversy by looking through job applicants’ social media postings. “I won’t ask you if you’re married, but I could figure it out,” Musto notes. “It reduces verifiability of discrimination.”
Privacy is another issue. Last year, Germany’s largest credit agency, Schufa, sparked an outcry after several news outlets reported that it was planning to scrape data from social networks to gauge a consumer’s creditworthiness. It has since backed down. In the U.S., credit agencies are “tightly regulated,” Musto says. “They’re very careful to comply with the law…. I can’t see any of the big credit agencies getting into something like this.” At least one major lender, Citibank, does not see itself using social data to assess borrowers. Frank Eliason, head of social media for the company, told The Economist that the bank monitors social media for marketing but to use it to measure creditworthiness is a “dangerous game.”
Then there is the natural reluctance of lenders to switch to a new metric when FICO has worked well. Other companies have introduced new scoring in the past, including the three largest U.S. credit bureaus, but these have been slow to gain ground. Credit start-ups using social data could face the same obstacles. “That’s going to be the hard part — prying customers away from FICO,” Stine notes. Challengers to FICO have to prove “they’re building a better mouse trap.” But companies will continue to try to wrest market share, especially now. “It’s a lucrative market … and the barriers to entry have gotten lower because of technology,” he says. “You don’t have to have a massive computer filling a house in order to do these kinds of things anymore. You can use a cloud server. The algorithms have gotten more sophisticated and do not need as much fine-tuning as they once did.”
Another problem with social data: Unlike actual payment history, it is easier to manipulate one’s social media profile. “[Consumers] can buy Twitter followers” and attempt to boost their credit scores that way, Stine notes. “At some point, it becomes advantageous to manipulate these things.” It then becomes a face-off between borrowers who manipulate their profiles to get credit and lenders who try to discern truly creditworthy consumers, he adds.
But despite the hurdles, the age of “Big Data” is here and the market is adapting. Experian, one of the three big U.S. credit bureaus, recently launched the “Extended View” score that includes rental information as part of its calculation. The company, along with TransUnion and Equifax, also tweaked their FICO-challenger called “VantageScore” to improve credit assessment. For example, debts that go to a collection agency used to be factored into one’s credit score for about seven years. The new version of VantageScore will not do that as long as the debt is paid or settled and the balance is zero, according to a March 11 story in CNNMoney. The scoring system also will consider utility payments as part of a consumer’s credit profile. For its part, FICO said it would look into incorporating alternative data into the scores of people with limited or no credit histories.
Right now, however, FICO has no plans to use data gathered from social media, the Journal said. But Bradlow thinks the scoring criteria could change, especially after more correlations are established. “Eventually, social network data may become what goes into a FICO score,” he says. “I don’t see a diminished desire for summarizing someone’s credit history with a single score. What I see changing is what goes in it.”
As for credit companies that mine social data for credit assessment, “I don’t know if it’s a good or bad idea. I think it’s going to be inevitable,” Stine notes. “They’re trying to carve out a niche in a market where people haven’t used this kind of information. We’ll find out in a few years if this was a good idea.”