Nonlinear Loan Loss Provisioning

48 Pages Posted: 1 Aug 2018

See all articles by Sudipta Basu

Sudipta Basu

Temple University - Department of Accounting

Justin Vitanza

Board of Governors of the Federal Reserve System

Wei Wang

Temple University - Department of Accounting

Date Written: July 31, 2018

Abstract

The extant banking literature often models loan loss provisioning as a linear function of changes in loan portfolio quality. Large sample data suggest that this linearity assumption is invalid. Using a piecewise linear specification, we find that while loan loss provisions increase almost proportionally with contemporaneous increases in nonperforming loans, they change less when nonperforming loans decrease. This asymmetric loan loss provisioning is not driven by the mechanical effects of loan charge-offs on nonperforming loans and allowance for loan losses. We show that loan loss provisioning asymmetry is greater for banks with shorter-maturity loan portfolios and more volatile interest income, and is more pronounced during economic downturns. Our proposed piecewise linear specification has a (moderately) higher power for detecting earnings management than standard linear models.

Keywords: loan collectibility, loan duration, conditional conservatism

JEL Classification: G21, G28, M41, M48

Suggested Citation

Basu, Sudipta and Vitanza, Justin and Wang, Wei, Nonlinear Loan Loss Provisioning (July 31, 2018). Available at SSRN: https://ssrn.com/abstract=3223910

Sudipta Basu (Contact Author)

Temple University - Department of Accounting ( email )

Philadelphia, PA 19122
United States
215.204.0489 (Phone)
215.204.5587 (Fax)

Justin Vitanza

Board of Governors of the Federal Reserve System ( email )

20th Street and Constitution Avenue NW
Washington, DC 20551
United States

Wei Wang

Temple University - Department of Accounting ( email )

Alter Hall 450
1801 Liacouras Walk
Philadelphia, PA 19122
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
4
Abstract Views
72
PlumX Metrics