Studies of fintech lending have shown that their digital access and ability to leverage alternative data have increased accessibility in underserved areas, enabled consumers with thin credit files to obtain credit, and provided a lower cost alternative to long-term credit card financing. This paper exams three questions: (1) Do proprietary loan rating systems accurately predict the likelihood of default? (2) Can a proprietary loan rating system, leveraging alternative data, that was developed in a favorable economic period continue to perform well under adverse economic conditions (such as the COVID-19 pandemic)? (3) Have fintechs been “cream skimming,” i.e., underpricing the cost of credit to top-tier customers? This study uses data from LendingClub, one of the largest fintech lenders in the personal loan market. We find that LendingClub’s loan rating system is superior to traditional measures of credit risk when predicting the likelihood of default and that the loan rating system continued to perform well during the pandemic period. Finally, we find no evidence of cream skimming.

View the Full Working Paper