We generalize an empirical likelihood approach to missing data to the case of consumer credit scoring and provide a Hausman test for nonignorability of the missings. An application to recent consumer credit data shows that our model yields parameter estimates which are significantly different (both statistically and economically) from the case where customers who were refused credit are ignored.
Zusammenfassung Die statistische Qualität von Kreditausfallprognosen lässt sich auf verschiedene Weise messen und vergleichen. Der vorliegende Übersichtsartikel fasst die in der Literatur gemachten Vorschläge zusammen und diskutiert deren Eignung für Kreditausfallprognosen im Privatkundengeschäft. Es zeigt sich, dass nicht alle Qualitätskriterien hier gleichermaßen sinnvoll sind. Insbesondere scheinen die in der Meteorologie beliebten Brier Scores und verwandte Kriterien für diese Anwendungen eher schlecht geeignet.
Schlüsselwörter Kreditausfälle · Wahrscheinlichkeitsprognosen · Scorekarten
JEL Klassifikationen C53 · G24Abstract The statistical quality of credit default forecasts can be measured and compared in different ways. This article surveys the various approaches that have been suggested in the literature and discusses their respective properties. For the particular case of credit scoring in the retail business, it is shown that some quality criteria are more useful than others. In particular, various measures that are popular in, e.g. meteorology, such as the Brier score have to be applied with caution.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.