Machine learning techniques in Kazakhstan banks’ sustainability assessment

Abstract

The research aims to find new approaches in bank sustainability assessment. During the last 15 years banks in Kazakhstan were affected by several global crises. Those events led to significant problems within the financial sector, which resulted in serious consequences. Untimely and incomplete identification of problems required governmental financial support or even liquidation of those banks. One basic problem appears to be the lack of an appropriate mechanism for preventive reaction. This thesis uses an approach of estimating bank’s sustainability via invented Bank Sustainability Index (BSI) and Weighted Average Bank Sustainability Index (WABSI). BSI reflects liquidity, capital adequacy and credit portfolio nature of the bank. Such indicators help to determine the current position of a particular bank as well as the overall performance of the bank sector at a given point of time. Machine learning techniques were used as practical tool for models development. Those models allow one to make predictions and reveal which economic indicators most affect the financial system. The research provides comprehensive overview of the banking sector of Kazakhstan and analysis of reasons of default cases.