Towards an improved credit scoring system: the Greek case

During the development of credit risk assessment models, it is very important to find variables that allow the evaluation of a company's credit risk accurately, as the classification results depend on the appropriate characteristics for a selected data set. In this paper, new credit risk models tested on real data, which evaluate credit risk of Greek companies are introduced. These models use a combination of financial and credit behaviour data. The credit risk models, which are introduced in this paper, do have some important additional advantages: 1) they contain a relatively small number of variables; 2) their stability is tested on samples after the time-period of the time period of data-collection; 3) the characterisation of 'good' and 'bad' credit behaviour is strictly defined.