Abstract:
Recently for the firm failure modeling academics started to use sophisticated techniques such as logistic regression, rough sets, neural networks, fuzzy logic, etc. But, practitioners often employ risk index models, less sophisticated technique which is very easy for implementation. Therefore, it is very interesting to evaluate weather sophisticated modeling techniques (logistic regression and neural networks) significantly outperform risk index models. Empirical testing on the sample of Croatian manufacturing firms has shown that designed risk index model performed pretty well in the segment of solvent firms, were it had classification accuracy of 86.4%, which was higher in comparison with logistic regression model (82.8%) but lower in comparison with artificial neural network (89.4%). However, in the segment of insolvent firms risk index model has shown moderate result in comparison with sophisticated techniques. Namely, in this segment of firms’ classification accuracy reached only 63.5%, which was lower in comparison with logistic regression model (78.9%) and artificial neural network (84.4%). Empirical finding suggests that bankruptcy prediction users from business should invest time and money in developing more sophisticated models in order to achieve more accurate predictions.
