The use of alternative data in credit decisions could make a significant difference in the cost and availability of credit for consumers, according to a new blog post published yesterday on the CFPB’s website. The blog post highlights data submitted to the bureau regarding outcomes generated by using an underwriting and pricing model incorporating alternative data—such as information about borrowers’ education and employment history—created by the Upstart Network. Upstart has been operating under a no-action letter from the CFPB to develop and test the model since 2017.
When comparing outcomes between a traditional model and the one that incorporated alternative data, Upstart found that using alternative data increased acceptance rates by 23% to 29% across all tested race, ethnicity and sex segments, while decreasing average APRs by 15% to 17%. Individuals that were considered “near prime” (with FICO scores between 620 and 660), younger applicants under age 25 and consumers with incomes of under $50,000 were significantly more likely to be approved under the tested model.
Fair lending testing showed no disparities between the traditional model and the tested model with regard to the approval rates and APRs provided for minority, female and senior borrowers, the CFPB added.