What do a million observations have to say about loan defaults? Opening the black box of relationships
© 2017 Elsevier Inc. Using a unique dataset of more than 1 million loans made by 296 German banks, we evaluate the impact of many aspects of customer–bank relationships on loan default rates. Our research suggests a practical solution to reducing loan defaults for new customers: Have the customer open a simple transactions account – savings or checking account. Observe for some time and then decide whether to make a loan. Loans made under this model have lower default, as banks can use historical data about their borrowers to establish a baseline against which new client-related information can be evaluated. Banks assemble this historical information through relationships of different forms. We define relationships in many different ways to capture non-credit relationships, transaction accounts, as well as the depth and intensity of relationships, and find each of these can provide information that helps reduce default – even establishing a simple savings or checking account and observing the activity prior to loan granting can help reduce loan defaults. Our results show that banks with relationship-specific information act differently compared with banks that do not have this information both in screening and subsequent monitoring borrowers which helps reduce loan defaults.
Puri, M; Rocholl, J; Steffen, S
Volume / Issue
Start / End Page
Electronic International Standard Serial Number (EISSN)
International Standard Serial Number (ISSN)
Digital Object Identifier (DOI)