Information Noise and Credit Risk: Evidence from Corporate Bankruptcy

41 Pages Posted: 31 Oct 2013 Last revised: 4 Jun 2017

See all articles by Xin Xu

Xin Xu

Unisys Machine Learning and Advanced Analytics Services

Date Written: May 31, 2017

Abstract

Theory predicts that information noise induces interactions between the degree of noise and credit risk determinants (Duffie and Lando [2001, Econometrica 69 633-664]). Using well-known bankruptcy hazard models and over two million firm-months of data during 1979-2012, we demonstrate the existence of the noise-induced interaction effects, and strong evidence supporting their implications important to empirical credit risk research. The interactions significantly improve out-of-sample forecasting accuracy, with improvements persistent over time, robust to empirical choices, and more substantial when information quality is poorer. Higher degree of noise, while markedly reducing predictability, entails ambiguous changes in credit risk, which might reconcile inconsistent empirical findings in the literature.

Keywords: Credit Risk, Bankruptcy Forecasting, Information Noise, Hazard Models, Survival Analysis, Probability of Default

JEL Classification: C41, G17, G33

Suggested Citation

Xu, Xin, Information Noise and Credit Risk: Evidence from Corporate Bankruptcy (May 31, 2017). Available at SSRN: https://ssrn.com/abstract=2347079 or http://dx.doi.org/10.2139/ssrn.2347079

Xin Xu (Contact Author)

Unisys Machine Learning and Advanced Analytics Services ( email )

PO Box 288
Concord West, NSW 2138
Australia

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

Paper statistics

Downloads
190
Abstract Views
1,506
rank
197,861
PlumX Metrics