Modelling Probability of Default in the Nordics

Abstract: Credit risk is one of the greatest risks facing financial institutions, and it is therefore very important that models with good predictive power are used in order to etter capture this risk. This thesis proposes logistic regression models for modelling risk-drivers of the probability of default in a financial institution active in he Nordics. The thesis focuses on handling and analysing large amounts of data in an attempt to find significant risk-drivers for each of the countries in the Nordics, as well as for a pan-Nordic model. The conclusion is that a pan-Nordic model performs very poorly, while some individual countries can be modelled quite well.