Variables Important for Bankruptcy Prediction - A Logit Binary Approach

Sammanfattning: The purpose of this bachelor thesis is to estimate our own bankruptcy prediction model using logit binary data. Our choice of variables is based on Altman’s Z-score model 1968. A comparison is then done between results in Altman and our findings. We perform our estimates on 114 listed Nordic companies, where 37 of them went bankrupt during 2002-2012. We find that our estimated model can categorize defaulting and non-defaulting firms best, two years prior to the event of bankruptcy. This is done with a 76,8 per cent accuracy. Finally, we show that our model can predict bankruptcy of Nordic firms better than Altman’s Z-score model.