Abstract:

This thesis presents a model that measures credit risk in the Norwegian business sector, using firm bankruptcy as proxy for credit risk. Probit analysis, a discrete response model, is applied to micro level financial information from more than 500.000 observations from the period 1989-1998. Bankruptcies in the period 1995-1998 are used to develop the model, and bankruptcies in the period 1991-1993 are used for out of sample testing. A set of time-consistent indicators of bankruptcy is found by combining ideas from both the theory of industrial organisation and financial statement analysis. The results support the idea of a learning effect in companies. This effect is recognised with reduced risk of bankruptcy when observations are subject to age. Furthermore, the results indicate that debt and interest burden increase risk of bankruptcy, while equity decrease risk of bankruptcy. Real-estate companies generally have a lower risk, while restaurants generally have a higher risk.

JEL Classification: C13, G33.
Keywords: Bankruptcy, probit estimation, credit risk.

Working Paper 1999/9 (pdf, 323 Kb)