The performance of credit risk evaluation models Empirical aspects and metricsLaura POPESCUAcademy of Economic Studies Bucharest Romania laura.popescu@brd.ro Credit risk evaluation has become a topic of great importance in the context of the current economic crisis which has brought the financial institutions on the point of dealing with an increased risk of granting credits to defaulted clients. Consequently, the need to use efficient evaluation methods is an increased one. Researchers have designed and studied a large variety of analysis models based on expertise, mathematical algorithms or artifficial intelligence techniques. All the results have concluded that one cannot determine a unique evaluation tool that is able to win the competition with the other methods under a large variety of circumstances. In order to obtain a result that can be associated with a high confidence level it is compulsory to take into consideration several characteristics. The current paper aims to present several techniques that can be used in order to evaluate the performance of credit risk evaluation models. Some important empirical aspects for performance estimation are described: the data sets’ analysis, discrimination and stability. At the same time the article introduces some performance metrics: the execution time, overloading, efficiency, acceleration and costs. The importance of these tools is emphasized in the process of correctly estimating the advantages and disadvantages of each available methodology. Keywords:performance, risk, evaluation, metrics, credits |