| Abstract [eng] |
The master's final project identifies the areas of activity risks for commercial banks, focusing exclusively on the 21st century, the most severely emerged financial risk and its sorts which made the most harmful effect for markets, in order to assess the sensitivity and importance of the risk parameters in relation to changes in the sector and to create a mathematical - programmatic model that optimizes the volume of credit, thereby maintaining a predefined level of financial risk to the sector. The project consists of three parts, one theoretical presentation of the risk areas of commercial banks' activities and financial risk assessment methods. Two practical ones - the elaboration of the relationship between the experiencing financial risks of commercial banks registered in the Republic of Lithuania and the generated returns and also between the nature of the new crediting, and the last part – the optimization of the crediting volumes of commercial banks registered in the Republic of Lithuania, with financial risks, modeling. The theoretical part presents the main aspects of the activity risk areas of commercial banks related to the significance of risk in commercial banking, systemicity of risks, factors and actions that can cause risks, grouping of risks and the occurrence of different groups of risks in commercial banking. The emphasis is also placed on the presentation of the risks of a financial nature and its valuation methods, distinguishing three risk components - market, credit and liquidity risks, which are describing through the prism of risks occurrence, manifestation or measurement, possible partial control, or even the elimination of it at individual cases. Also, the most popular mathematical methods of financial risk assessment, their application features and specificity are presented, the problems of modern commercial banking are identified in optimizing the volume of crediting in the context of a financial risk. The second part of the work briefly presents the research structure and methods also detailing the tendencies of return on capital of commercial banks registered in the Republic of Lithuania in the context of financial risks indicators, revealing the trends of mutual dynamics characteristic of indicators, giving the logical explanation for discovered trends. Also, the impact assessment of the new crediting of commercial banks registered in the Republic of Lithuania on the financial risk profile, identifying the impact of 10 different types of new crediting and the trends of mutual distribution, in terms of market, credit and liquidity risks separately, determines and justifies the existence of statistically reliable links between the commercial banking sector financial risk factors and the specific type of new lending. By showing the occurrence of risk factors, taking into account the tendencies of distribution, the degree of manifestation and strength, by describing the dependencies inherent in the scale of new loans, by detailing the types of newly granted loans that would significantly increase the market, credit and liquidity risks separately. In the last part of project, the artificial neural network model is developed for the purpose of optimizing the most significant risks of commercial banks registered in the Republic of Lithuania - liquidity and credit risks. The model is based on year 2008 Q I– year 2017 Q III period of independent and addictive data, it uses an algorithm BFGS 49, has 90% learning efficiency, 84% testing efficiency and 90% prediction of meaning - optimization reliability. Model functions on the basis of quarterly input and output data values, 3 input variables, 3 artificial neurons in the same hidden layer of artificial neurons, input and output bias coefficient weights, and 6 output variables. The model is most severely affected by the capital adequacy ratio of commercial banks registered in the Republic of Lithuania, the weakest impact is from the net interest margins of commercial banks registered in the Republic of Lithuania. The values of input variables demonstrating the lowest optimization accuracy in terms of new long-term legal entities crediting in small amount, and the highest accuracy for new short-term lending for acquire accommodation, as well as new short-term corporate lending in large amount. At the period of year 2011 Q I - year 2017 Q II, commercial banks registered in the Republic of Lithuania, at the different quarters of activity have experienced different levels of financial risk – they took advantage of the leverage offered by the opportunities but the sector experienced market risk did not result in obvious threats. However, at the Republic of Lithuania registered commercial banks experienced credit risk and liquidity risks, had made the core risk for the sector and was depending from the volumes of new small (large) legal entities short-term (short-term) and long-term financing, financing long-term consumption of individuals, financing short-term acquire of accommodation and the financing other short-term individuals’ goals. Based on artificial neural network model, which was made from year 2008 Q I - year 2017 Q III data sample, and imitation scripts the largest new lending volumes of commercial banks registered in the Republics of Lithuanian recorded, immediately after a long period of economic stagnation (when the capital adequacy ratio – 8.1 – 8.5%, non-performing debt instruments index – 2 – 1.6%, the net interest margin – 2 – 1.2%), the lowest new lending volumes recorded – at the very slow economic growth period (when the capital adequacy ratio – 15 – 17%, non-performing debt instruments index – 5 – 3%, the net interest margin – 2 – 1.6%). |
