Friends Don’t Let Friends Have Bad Credit

Economics Ph.D. candidate Yanhao Wei is researching the concept of "social credit scoring."

Tuesday, August 18, 2015

By Susan Ahlborn

Would you be willing to give up a friend to get a loan? A new approach to calculating credit ratings might put that question on the table.

“If you apply for a credit card, they check your credit rating,” says Yanhao Wei, a doctoral candidate in economics. “But some companies might also ask for your phone record and Facebook page and then look at your friends’ credit scores, because they believe that will increase the accuracy of your score. So if you have a bad friend in terms of credit, that’s going to discount your credit score.”

Wei is the lead author of the article “Credit Scoring With Social Network Data,” which will be published in Marketing Science, the leading journal in quantitative marketing. He and his coauthors—Professors Pinar Yildirim and Christophe Van den Bulte of the Wharton School and Boston University Professor Chrysanthos Dellarocas—wanted to see how much adding friends’ information improved credit score accuracy, but also look at the longer-term effects: How would people react, and would the social credit scoring then continue to be effective? They also considered which aspects of the social network could be used to improve the accuracy of the credit score.

The first answer is yes, says Wei—credit scores are made more accurate by looking at the scores of friends. “If you are using one person’s credit score, that’s one signal. And if you’re getting a friend’s credit score, that’s another signal,” Wei says. When many signals are put together it can substantially increase the precision of the credit score.

This gives would-be borrowers an incentive to distance themselves from friends with bad credit, and Wei theorizes that will be the initial reaction. “That may sound a little bit exaggerated, but for some people who need credit, it’s something they’re going to do. So just the fact that companies are going to use credit scoring is going to change the social network structure to some extent.”

The authors predict two kinds of changes: People will have fewer connections, but be more similar to the friends they keep. “Only people with comparable credit scores are going to maintain their friendships,” says Wei. This will make the number of friends a person has another signal that lenders may take into account.

The use of social network credit scoring is limited in the United States—there are legal obstacles, and a lot of financial information on borrowers is already available. In developing nations like India or Mexico, however, social network credit scoring can provide necessary data when many people may be first-time borrowers with no credit history. In practice, lenders might give such borrowers a choice: Give us access to your social network or we’ll have to think twice before granting you a loan.

Wei compares the approach with group lending, in which small groups borrow collectively and members encourage each other to repay, as a way to overcome the lack of information on potential new borrowers in growing economies. “Before, these borrowers couldn’t get any money. Now they can.”