The Implications of Credit Risk Modeling for Banks’ Loan Loss Provisions and Loan-Origination Procyclicality

• Published in: Management science 2019-05, Vol.65 (5), p.2116-2141, Article mnsc.2018.3041
• Main Authors: Bhat, Gauri, Ryan, Stephen G, Vyas, Dushyantkumar
• Format: Article
• Language: English
• Subjects: Bad debts; Bank loans; Bank management; Banks (Finance); Credit risk; credit risk modeling; disclosure; financial crisis; Financial disclosure; Financial reporting; loan loss provisions; Loan losses; Loan originations; Loans; Policy making; procyclicality; Provisions; Risk assessment; timeliness
• ISSN: 0025-1909; 1526-5501
• DOI: 10.1287/mnsc.2018.3041

Economic policymakers express concern that procyclical lending by banks imperils financial stability. Prior research finds that banks that record timelier loan loss provisions originate more loans during downturns, consistent with loan loss–provision timeliness mitigating loan-origination procyclicality. Motivated by this concern and research, we examine whether banks’ credit risk modeling disciplines both their loan loss provisions and loan origination. We identify two forms of credit risk modeling from banks’ financial report disclosures: statistical modeling of the drivers of past loan losses and stress testing of future loan losses to adverse scenarios. We show that banks’ credit risk–modeling disclosures are positively associated with their loan loss–provision timeliness, with the ability of their provisions to predict future loan charge-offs, and with their loan origination during downturns. We further show that these associations vary in predictable ways across the two forms of credit risk modeling when we distinguish homogeneous from heterogeneous loans and stable periods from downturns. This paper was accepted by Mary Barth, accounting.