Original scientific paper
https://doi.org/10.17535/crorr.2018.0021
Firm Failure Prediction: Financial Distress Model vs Traditional Models
Ivica Pervan
orcid.org/0000-0003-1987-7702
; Faculty of Economics, University of Split, Split, Croatia
Maja Pervan
orcid.org/0000-0001-6895-5258
; Faculty of Economics, University of Split, Split, Croatia
Tamara Kuvek
; Faculty of Economics, University of Split, Split, Croatia
Abstract
Modelling firm failure, classically defined as bankruptcy, is problematic in the Croatian business environment since the bankruptcy procedure starts at a very late stage of crisis, when a firm liabilities are higher than assets. In order to overcome this problem, we propose an alternative definition of firm failure which is based on a firm's financial status, meaning financial distress, rather than its legal status. A firm is characterised as financially distressed when its EBITDA is lower than its interest expenses for two consecutive periods. Accordingly, we developed models based on three different failed firm statuses: (i) bankruptcy, (ii) rescue plan and (iii) financial distress. The application of logistic regression on a sample of Croatian firms has shown that a financial distress model has a high level of predictive power. Moreover, for the whole sample this model outperformed bankruptcy and rescue plan models in terms of overall accuracy and the prediction of failure status. Additional analysis has revealed that it is useful to develop a model for non-micro firms because such an estimation results in improved prediction power in comparison with a generic, one-size firm model. In the case of non-micro firms, the financial distress model outperformed the rescue plan model, while the hit rate was similar to the hit rate of the bankruptcy model. A developed financial distress model can be applied by investors and creditors in order to timely evaluate firm failure risks and undertake required business decisions.
Keywords
firm failure; financial distress; bankruptcy; rescue plan; Croatia
Hrčak ID:
212393
URI
Publication date:
13.12.2018.
Visits: 2.138 *