Banking Crises and Early Warning Systems: A Model Suggestion for Turkish Banking Sector

This study aims to evolve a model used in foreseeing possible banking crises in Turkey. In the light of many empirical studies’ findings, a specific model for Turkey is developed. This model is estimated by using multivariate adaptive regression splines (MARS) which is a non-linear and non-parametric estimation method. Estimation results show that significance and explanation levels of model are strongly high. According to model’s findings, banking crises can be predominantly attributed to external factors. Systemic financial crises, exchange rate open position, and terms of trade are observed to be main determinants. Besides these factors, the study shows that capital adequacy, interest rate risk, and market risk are the other important factors.