Selection of variables in small business failure analysis : mean selection vs. median selection

This paper focuses on one of the most determinant processes in business failure assessment: Variable selection. After a preselection of variables based on previous empirical literature, we perform a statistical variable selection on a sample of small firms using both mean and median differences. As the resulting variables differ in each test, we have performed a varied group of business failure assessment methods (linear discriminant analysis, quadratic discriminant analysis, logistic discriminant analysis, k-th nearest-neighbor discriminant analysis, logit, and probit) to identify the implications of using one test or the other. Our results show that the nature of the sample determines not only the statistical variable selection test, but the most appropriate methods to assess business failure, which constitutes our main contribution. Additionally, we contribute new evidence on the addition of qualitative information (payment incidents), with previous evidence for SMEs being scarce.