Abstract: More than 45 million U.S. adults lack traditional credit histories, creating a gap that alternative financial service data, such as payday lending records, could potentially fill. Using the staggered adoption of the largest alternative credit database, we examine the data's impact on automotive lenders in the subprime auto loan market. Alternative credit scores predict loan performance, leading lenders to offer better loan terms to higher-scoring borrowers. However, a history of using alternative financial services, even with relatively high alternative credit scores, comes with significant downsides: borrowers with payday loans histories experience higher delinquency rates, face higher interest rates, and reduced loan origination rates after the adoption of the alternative credit data. A flexible machine learning model indicates that only 6% of alternative financial service users possess sufficiently strong credit histories to offset the stigma of using these services. Consequently, alternative credit data limits credit availability and raises traditional loan costs for most users of alternative financial services. Alternative financial services are more commonly used in lower-income areas and communities with higher shares of Black residents, raising concerns that the adoption of alternative credit data may have disproportionate negative impacts on these populations. Our results contribute to the policy debate on credit data, consumer privacy, and financial inclusion.
Discussant: Benedict Guttman-Kenney, Rice University
Yunzhi Hu, University of North Carolina-Chapel Hill
Pavel Zryumov, University of Rochester
Abstract: We examine competition and collaboration between banks and fintech firms in a market with adverse selection. Banks have cheaper funding, while fintechs have better screening technology. Our innovation is to allow the bank to lend to the fintech, i.e., to finance its competitors. This partnership lowers fintech funding costs and reduces bank competition incentives. Lenders collaborate when average borrower quality is low but compete when it's high. While partnership funding always benefits the fintech, it increases the bank's profits only when the average borrower quality is low and benefits the borrowers only when the average quality is high.
Abstract: This paper examines the role of artificial intelligence (AI) in facilitating the non-judicial collection process of delinquent consumer debt. Leveraging a randomized field experiment conducted by a debt collection agency, we show that algorithmic calling decisions achieve higher repayment rates with fewer collection calls compared with human collection officers. Uncovering the black box of AI, we find that it extracts predictive signals from unstructured notes compiled by collectors. These signals not only predict whether the delinquent borrowers would repay during the non-judicial collection process, but also shed light on the underlying motivations or impediments of delinquent borrowers' repayment behavior.
Discussant: Benjamin Iverson, Brigham Young University
