Economic networks to predict corporate default

File pdf 603KB

This paper investigates the role of industry-specific effects and structural properties of intersectoral customer-supplier relations on the corporate default prediction of individual firms. We focus on a large sample of US exchange-listed companies over the period 1997-2015 and show that default prediction models that account for sectoral network effects have better in-sample and out-of-sample accuracy compared to benchmark models that focus only on firm-specific and macroeconomic attributes. We find that companies' default intensities are related to the aggregate financial health and competition level of the industry in which they operate. Moreover, companies' role as main commodity suppliers in the aggregate economy, as well as the importance of their customers (or suppliers) as main commodity buyers (or providers) improve model's in-sample fit and prediction accuracy.

Author(s):

Andreea Constantin    
University of Lausanne
Switzerland