Research Project: Credir Risk Modeling

Project leader
Moudud Alam
Project Members
Kenneth Carling
Lars Rönnegård
Project Period
Project Status
A major problem with today’s credit risk models is that they focus solely on the status of the individual company, ignoring the condition of the company’s suppliers and customers. Theoretical studies have pointed out that it is reasonable to regard the companies as mutual or indirectly dependent and that this dependence should be incorporated into credit risk models. That this is not just a theoretical or academic issue is clear from a study (Carling, Rönnegård and Roszbach, 2004) where we showed that if dependence was incorporated the estimated credit risk for the two studied the large Swedish banks would be doubled.

When it is clear that the dependence between the companies have to be modelled, so the question arises: How? This research project aims at exploring the possibility of using the SNI industry classification (see for details) to provide a structure to the dependence between companies and hence develop a better and parsimonious credit risk model. This research project also aims to verify if the dependence structure can be approximated through some well known measure of inter-industry business flows such as the Statistics Sweden’s input-output matrices. The above goals are to be fulfilled through the development of new computational techniques which enable to fit the class of sophisticated statistical models needed for modelling mutually dependent companies’ credit risk.
Research Profile
Complex Systems - Microdata Analysis
Riksbankens FOA