THE ACCURACY OF BANKRUPTCY PREDICTION MODELS: A COMPARATIVE ANALYSIS OF MULTIVARIATE DISCRIMINANT MODELS IN THE ITALIAN CONTEXT

  • Salvatore Madonna Department of Economics and Management, University of Pisa, Italy
  • Greta Cestari Department of Economics and Business Science, University of Cagliari, Italy

Abstract

This research is driven by the conclusions of Bellovary, Giacomino and Akers (2007), who stated at that time that future research, rather than aiming to develop new bankruptcy prediction models (to add to the considerable body) should focus more on the use of existing models. In this regard, we aimed to verify the accuracy of three bankruptcy prediction models, all based on multivariate discriminant analysis, in predicting the fate of firms operating in a different business context and all located in EmiliaRomagna region of Italy. The models tested were: Altman's Z'-score ( 1993) , Alberici's Z-score ( 1975) , and Bottani, Cipriani and Serao's discriminant function (2004). In particular, we conducted a two-phase analysis, the first to determine the capacity of the three models to predict the fate of firms known to have gone bankrupt between 2012 and 2014, and the second to distinguish between bankrupt and buoyant firms in a mixed sample from the same period. The analysis was performed according to ex-post reasoning, and the investigated models were tested on retrospective data pertaining to two distinct samples of firms of known status. Specifically, the first sample comprised firms that already met the condition the models were designed to detect, i.e., bankruptcy, and the second comprised equal numbers of operational and failed firms. The models were applied to the annual financial statements pertaining to the last five years of activity of bankrupt firms, and the most recent five years of activity of the solvent firms. The predictive efficacy of each model was determined by comparing the results furnished by each model with the real world status of the investigated firms. The results obtained were: 1) Altman's model, applied with a single cut-off, is well able to detect signs of failure and to discriminate between failing and flourishing firm, even if taken out of its original context and applied to in a heterogeneous sample of firms; 2) Altman's model appears to meet the demand for generalizability, and is therefore suitable for large-scale investigations.

Downloads

Download data is not yet available.

Metrics

Metrics Loading ...
Published
2015-12-29
How to Cite
Madonna, S., & Cestari, G. (2015). THE ACCURACY OF BANKRUPTCY PREDICTION MODELS: A COMPARATIVE ANALYSIS OF MULTIVARIATE DISCRIMINANT MODELS IN THE ITALIAN CONTEXT. European Scientific Journal, ESJ, 11(34). Retrieved from https://eujournal.org/index.php/esj/article/view/6723