Predicting bankruptcy under alternative conditions: The effect industry and time period change on the accuracy of the model

Michal Karas, Mária Režňáková


Purpose. The possibilities of predicting a company’s bankruptcy or rather serious financial problems is a very up to date issue, as, due to current economic conditions, there is a considerable number of companies facing financial problems, that might results in theirs bankruptcy. An early identification of business partner’s financial problems could avoid serious economic loses. Bankruptcy prediction models represent a useful and favourite tools used to predict such problems. However, according to literature, application of such models in different time or industry should be resulting in a lower accuracy of these models, compared to conditions under the model was built. In our previous research we created our own bankruptcy prediction model. When creating the model we tried to applicate an approach different to previous ones. For creating the model we used the traditional method of linear discrimination analysis, but we employed only transformed variables with approximately normal distribution. What is more, the variables pairs are mostly negatively correlated. According to literature such factors should positively influence the model accuracy. However there is a very limited literacy how such application affects the stability of model’s accuracy.

The aim of this paper is to analyse the stability of model’s accuracy in application in different time period or different line of business. Moreover, we aim to examine and discuss the effectiveness of the procedure which was used to create the model.

Methodology. We test our bankruptcy model in four different lines of business, namely in the field of manufacturing, construction, agriculture and transportation branch. Moreover we test the accuracy of the model on the sample of companies that experienced bankruptcy in time period from 2008 to 2013. We employ the chi-square test and the Fisher exact test to analyse the difference between the tested and original accuracy.

Results. The original accuracy of the model was 93.89% of correctly classified active companies and 89.12% of correctly classified bankrupt companies (a year before bankruptcy). The results reached on alternative sample do not meet these values.

The theoretical contribution. We aim to contribute to the theoretical discussion about factors influencing the accuracy model and to the efforts of creating more robust bankruptcy prediction model, i.e. model not industry specific.

Practical implications: The model could be practically useful to SME for evaluation counterparty risk and to avoid loses due to counterparty financial problems.

Keywords: bankruptcy prediction, model robustness, credit risk

Paper type: Research paper


  • There are currently no refbacks.

KTU_school_economics and business Riga Technical University
Brno University of Technology Tallinn University of Technology