Data reduction and univariate splitting – do they together provide better corporate bankruptcy prediction?

Abstract

For many years the number of companies becoming insolvent has been increasing in the majority of Central and Eastern European countries recently accessed to the European Union1, and the crisis has substantially boosted this tendency. Accordingly an escalating interest can be noticed towards multivariate statistical bankruptcy prediction models in business life. Statistical based solvency prediction is most extensively carried out by financial institutions. Since the Basel II Capital Accord has been in forth for financial institutions, it is not the question whether to apply statistical forecast methods in credit appraisal and probability of default prediction, but it is a great problem, which methods should be applied and how. Therefore the discussion on methodological problems of corporate survival and solvency prediction is living its renaissance in many countries.