The actual predictive power of bankruptcy models in terms of time use

Abstract

The quality of bankruptcy models is habitually based on how well they can predict the financial health of a business a year in advance. However, corporate accounting data is publicly available usually with one year delay. So, the research question is not how accurate the prediction is one year in advance, but two years in advance. Therefore, an analysis of 7 models was performed to compare the change in their accuracy depending on time. The results showed a decline in prediction accuracy over time. The found total success rate of prediction is from 67.38% to 80.64%, which confirms that some of these models are also usable for two-year prediction of the financial health of the company.