Ratingverfahren: Diskriminanzanalyse versus Logistische Regression

The credit business is an essential part of each banks' activities. With regard to the increasing regulatory obligations, the risk management and the associated minimum capital requirements gain in importance. Banks have got the possibility to determine their credit risk by an internal rating system. It is current practice to make use of quantitative methods to develop the sys-tem. This paper will focus on rating systems calibrated using the discriminant analysis and the logistic regression. Both methods will be presented on a theoretical basis and will be applied in a practical testing. In the first step, the borrowers are divided into rating classes. In the second step, a default probability is determined for each of them. The paper will explore the two approaches and reveal the differences.