Evaluating and forecasting banking crises through neural network models: An application for Turkish banking sector

The performance of neural networks in evaluating and forecasting banking crises have been examined in this paper. An artificial neural network model which works with the banking data belonging to the same date and another artificial neural network model which works with cross sectional banking data have been formed and tested. The optimal topologies of these models have been determined by Taguchi approach which is a design of experiments method. Both models can forecast the values of the output neurons consisting of Non-performing Loans/Total loans, Capital/Assets, Profits/Assets and Equity/Assets ratios by using 25 input neurons consisting of macroeconomic variables, the variables related to the external balanced financial system's structure, and time with very small errors. Consequently, it has been seen that artificial neural networks which are capable of producing successful solutions for semi-structural and non-structural problems, can be used effectively in evaluating and forecasting banking crises. (c) 2006 Published by Elsevier Ltd.