Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
Business failure prediction in manufacturing: a robust bayesian approach to discriminant scoring
Maurizio Baussola, Eleonora Bartoloni, Aldo Corbellini

Last modified: 2013-06-14

Abstract


This paper provides a methodological analysis of credit risk in manufacturing firms. By using a representative sample of both healthy and bankrupted firms during the period 2003-2009 we provide an in-depth comparison of the standard discriminant approach (DA) for bankruptcy prediction based on a logistic regression model and a Robust Bayesian Approach (RBA).We conclude that the use of a robust GLM regression methodology enables us to provide a more accurate separationbetween sound and unsound firms. However, we underline that the need for a firm’s credit worthiness measures is not ruled out and, therefore, more work has to be done even in such a Bayesian framework in order to achieve such a goal.

Full Text: PDF