ASSESSMENT OF RISK OF BANKRUPTCY OF BANKS OF UKRAINE BY A MODERN METHOD OF ARTIFICIAL NEURAL NETWORKS

Abstract

The article examines modern methods of bankruptcy risk assessment of Ukrainian banks. Bankruptcy of any bank has negative consequences for a wide range of subjects and can lead to negative processes in all economy. Diagnostics of bankruptcy is a timely detection of insolvency, unprofitability, financial dependence on external sources of financing, low business activity. As a rule, in classical models of diagnostics of bankruptcy use indicators of profitability, financial stability, liquidity and business activity. For diagnostics of bankruptcy in the world it is used the models constructed on the basis of financial coefficients today. In article the comparative analysis of such most known models is carried out. From a set of foreign methods of management of risk of the simplest and widespread method of the analysis the GАР-management method is. But as practice shows, one method for exact assessment of probability of bankruptcy is not enough. And uses of models and methods focused on the developed countries is it is not quite relevant to economy of Ukraine. Today there is an urgent need of development of modern model of forecasting of bankruptcy of banks in the conditions of uncertainty and doubtful data for realities of the Ukrainian bank sphere, but which would be simple and convenient in use. As an alternative to statistical methods, for forecasting of risk of bankruptcy of the Ukrainian banks, the modern neural network model can be used. This model will be useful to clients of banks which want to define insolvent banks in the nearest future (1–1,5 years). Therefore in this article the modern method of modeling of assessment of probability of bankruptcy of banks ₋ a method of artificial neural networks is offered. Direct testing of a possibility of application of neural network for definition of bankruptcy of banks of Ukraine was held on a set from 126 test data sets which were selected from quarterly financial statements of 5 banks of Ukraine, three of which are solvent, and two are in an elimination stage. The research showed that it is expedient to use a method of artificial neural networks at assessment of risk of bankruptcy of the banking sector in general. It will allow to allocate, in the presence of necessary data, banks which are solvent and insolvent, that is, have rather high percent of risk of bankruptcy.