The Impact of Methodology on the Effectiveness of Bankruptcy Modeling

The prevailing opinion in literature is that the accuracy of bankruptcy models cannot be appreciably improved by the choice of classification algorithm. A reflection of this conviction is the frequent usage of parametric methods. However, the nature of financial data places a limitation on the accuracy of these methods. An analysis of 1908 Czech industrial enterprises from 2004 to 2011 reveals that a nonparametric method, if used for the selection of model variables as well as the actual classification, can yield significantly better results than the traditional parametric approach.