A note on performance measures for business failure prediction models

Abstract

This note briefly describes three performance measures that can be used in business failure prediction models: the unweighted error rate (UER), D-max and the Gini-coefficient. The use of these measures (and the mathematical relationship between them) is illustrated with numerical examples. We hope that this note may help the reader to better understand (and possibly use) these classification criteria.

Notes

¹ See Balcaen and Ooghe (2006) for an overview of much-used statistical methodologies.

² This can also be the other way round, depending on the construction of the model.

³ Altman (1980) mentions different components of type I and type II costs in the context of commercial bank lending.

⁴ Consequently, we do not consider other types of measures, such as R ²-measures or entropy-based measures.

⁵ Hsieh (1993) and Koh (1992) discuss some of the difficulties of determining an optimal cut-off point.