A Comparison of Alternative Methodologies for Credit Risk Evaluation

Credit risk refers to the likelihood that a firm or individual borrower will fail to meet a debt obligation. Credit risk evaluation is a very challenging and important problem in the domain of financial risk management. There are different methods and approaches for constructing credit risk assessment rating systems. The aim of this paper is to perform an empirical comparison of different popular techniques using a data set of Greek companies from the commercial sector. For this purpose three different methodologies are used, namely logistic regression, support vector machines, and the UTADIS (UTilités Additives DIScriminantes) multicriteria method. The results show that even with a considerable imbalanced data set with a small number of defaults, all methods provide good results. The UTADIS multicriteria method outperforms the two other techniques. Ensemble models are also tested, but are found to provide only marginal improvements