Redesigning Ratings: Assessing the Discriminatory Power of Credit Scores under Censoring

Abstract

Redesigning rating systems has been becoming an important issue for banks and other financial institutions in processing the implementation of the Basel II and III accords. The available data base for this task typically contains only the accepted credit applicants and is thus censored. To evaluate existing and alternative rating systems, we would actually need the full data base of all past credit applicants. In this paper we discuss how to assess the performance of credit ratings under the assumption that for credit data only a part of the defaults and nondefaults is observed. The paper investigates criteria that are based on the difference of the score distributions under default and nondefault such as the Kolmogorov – Smirnov statistic, the accuracy ratio and the area under curve. We show how to estimate bounds for these criteria in the usual situation that the bank storages only data of the accepted credit applicants. It is shown that these bounds can be helpful to assess discriminatory power even when a part of the data is not available.