Early Warning Systems for Banking Crises in Montenegro: Combination of Signal Approach and Logit Model

Abstract

Very high costs of systemic banking crises impose the necessity for early warning systems. Purpose of this research is to create an adequate early warning system for systemic banking crises in Montenegro using combination of signal approach and logit model. Beside benchmark model with signal horizon of 24 months two more models with signal horizons of 18 and 12 months are estimated. Although in the literature signal horizon usually implies period of 12 months before and 12 months after beginning of the crisis, here benchmark model implies 24 months before beginning of the crisis. On the basis of indicators with the best performances composite indices are created. Their robustness is checked using simple logit regressions that are compared by Bayesian model averaging. Indicators of credit expansion proved to have dominant role in early warning models for systemic banking crises in Montenegro. Also, indicators showed that Montenegrin banking system is significantly exposed to trends on the global level.