Measuring Ratings Accuracy Using the Average Default Position

Abstract

Credit risk is often summarized by ordinal rating systems employed by credit rating agencies or bank credit departments. Assessing the power and utility of those rating systems is obviously important but can be quite challenging to demonstrate and to explain.

This brief note contributes a new measure to the existing literature on the accuracy of ordinal rating systems, the average default position (“ADP.”) Under this metric, a rating system would be considered more accurate the lower is its ADP; that is, the lower is the rating of a typical defaulter within the system’s overall distribution of outstanding ratings. This metric is mathematically simpler and considerably more intuitive than the more commonly used “accuracy ratio” (“AR”), from which it can, in fact, be derived.

Keywords: credit, performance

JEL Classification: G21, G24