Modele predykcji upadłości MŚP w Polsce – analiza z wykorzystaniem modelu przeżycia Coxa i modelu regresji logistycznej
Prediction models of SME bankruptcy in Poland – analysis using Cox survival model and logistic regression model

Summary/Abstract: Credit risk is associated with the banking activity and is the most important type of the risk to which banks are exposed. Bankruptcy risk assessment is based on models using the discriminant analysis and logistic regression. However, the requirements for the models are changing and methods used in the banking sector are insufficient. Newer and more sophisticated techniques are developed. One of these methods is survival analysis the popularity of which has increased in the recent years. The aim of this study was to compare new technique (survival analysis) used in the credit models with the traditional ones, like logistic regression. The paper analyses the strengths and weaknesses of both methods and the possibility to use time-dependent variables in survival models (macro-variables) and nominal variables in logistic regression model. The inclusion of macro-variables increases the prediction power of such models. The number of models was built on the basis of the Polish SMEs (turnover above 2 million Euro) data set consisting of 1,561 enterprises including 807 defaults (FS 2002-2010).

• Details
• Contents