On the fringe of credit visibility: the value of alternative data for assessing the credit risk of subprime underbanked consumers

扫码查看

原文链接

NETL
NSTL

外文摘要：In a modern economy, prospering without credit is difficult. Yet, Geraldes et al. (2022) report, for instance, that as many as 2.5 billion individuals in the world have little to no bank relationships. Referred to as underbanked consumers, they are unable to obtain credit due to their limited or non-existent credit history. Alternative data refers to data sources that are not traditionally used in credit scoring. Current research suggests that alternative data may contain predictive information helpful in assessing the creditworthiness of underbanked consumers. We use statistical and machine learning models to examine the value of alternative data for assessing the creditworthiness of the USA's subprime underbanked consumers. We use a proprietary dataset of automobile loans that includes both traditional and alternative data to compare the predictive value of each data type. Our main finding is that the informational content of alternative data is not subsumed by traditional data. In addition, we find that alternative data alone have value that can help lenders extend credit to subprime underbanked consumers, enabling them to fully participate in the mainstream economy.

外文关键词：

alternative datacredit scoringunderbanked consumerspersonal bankruptcyauto loans

作者：

Edwin Baidoo、Stefano Mazzotta

展开 >

作者单位：

Department of Decision Sciences and Management, Tennessee Tech University, 1105 North Peachtree Street, Cookeville, TN 38505, USA

Department of Economics, Finance, and Quantitative Analysis, Michael J. Coles College of Business, Kennesaw State University, 560 Parliament Garden Way, MD 0403, Kennesaw, GA 30144, USA||School of Data Science and Analytics, Kennesaw State University, 560 Parliament Garden Way, MD 0403, Kennesaw, GA 30144, USA

出版年：

2025

International journal of applied decision sciences

ISSN：1755-8077

年,卷(期)：2025.18(4)