Many startups, backed by major technology companies, are using big-data techniques to offer short-term, small-dollar loans. These young companies want to reach the 68 million Americans who have either no credit history or a poor one, according to the Federal Deposit Insurance Corp. In lending, "big data" involves gathering numbers on everything from an applicant's number of Facebook friends to how regularly consumers pay their cellphone bills. Users say big data lets them offer loans that are more affordable than payday alternatives, the Pew Charitable Trusts reports.
The National Consumer Law Center, however, suggests that big data may not make much of a difference. The group's research found that loans based on underwriting from big-data startups that include LendUp, ZestFinance Inc., and Think Finance Inc. offer effective annual interest rates that range between 134 percent and 749 percent. Persis Yu -- an NCLC attorney and the report's author -- examined interest rates, loans terms, and fees published on the companies' websites and compared them to payday lenders. She concluded, "The big-data algorithms do not appear to lead to the development of better loan products."
Executives of the startup lenders acknowledge that default rates on their loans are high, the risk justifying their triple-digit interest rates. At the same time, they believe that the thousands of new variables can better predict creditworthiness. The new variables include borrowers' rent records, past payday loans, transactions with pawn shops, and collections. They also look at social media posts and whether borrowers fill out an application in capital letters. Some companies say that part of their algorithms and data sources are not used to make credit decisions, but simply to screen out fraud. The Federal Trade Commission plans to discuss whether the algorithms are discriminatory or in violation of borrower privacy.