Thesis for scientific degree of Doctor of Technical Sciences on the specialty 01.05.04 – system analysis and the theory of optimal decisions. – National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Ministry of Education and Science of Ukraine, Kyiv, 2018.
The thesis is devoted to solving an important scientific and applied problem of creating a systemic methodology for evaluating the risks of financial systems and developing a unified information technology for monitoring and risk management to ensure their minimization, increasing the efficiency for financial risk analysis, estimation and forecasting the risks of financial systems.
Risk assessment of financial systems is usually carried out on the basis of large statistical data volumes with some uncertainty both of the data by themselves and of their collection and processing means. There were also imposed very strict requirements regarding the data quality and volumes for making any forecasts. The requirements for speed, quality, and correct risks’ processing in the current competitive environment are constantly increasing, so the task of new methods, models and technologies developing is urgent in order to support the adoption of managers’ decisions in dealing with risks.
The scientific novelty of the work is determined by the following theoretical and practical results obtained by the author. The developed system methodology of financial risks investigation and evaluation is based on the basic principles and procedures of system analysis and takes into account the basic principles of risk management. The system methodology combines static and dynamic evaluation. The peculiarity of the static evaluation is the expansion of the use of a scorecard to the risks’ analysis of various types, developed neural-fuzzy method of incorporating previously rejected data, the application of the developed criteria for taking into account information risk as a manifestation of various types of uncertainties and external influences and information threats. The use of the own combined method of processing incomplete data for financial risks modeling allowed to carry out a deep analysis of the completeness, informativeness and quality of data.
The dynamic evaluation is based on the proposed dynamic assessment and adaptive risk management principles. The probabilistic-statistical method of estimating the risk of financial losses based on the combined use of optimal filter, regression and Bayesian analysis. Using of the Bayesian dynamic network allows estimate dynamically the losses from the financial risk realization.
In the manuscript a dynamic method for risk assessment was developed which allows the construction of different survival models types (parametric, nonparametric), with the ability to predict the risk level and the degree, and the moment which characterize the permissible, critical or catastrophic level of risk. Algorithms for determining the moment of risk occurrence by the given critical risk degree or level were developed.
The adaptive principle of the financial risks’ management and structural and parametric adaptation method were developed. They allow take into account the existence of uncertainties, non-stationary features, season effects, nonlinearity for financial processes. The method of structural and parametric adaptation contains the original quadratic criteria of the risk quality processing.
In the dissertation the extended information technology and information decision support system (IDSS) were created on the basis of the proposed system methodology and developed methods. The proposed information technology can be implemented in the form of client-server architecture, microservices and clouds; it is flexible and adaptive both to the practical tasks and to the integration into the existing enterprise information system.
The practical significance of the obtained results are: development of the methodology for the financial risks systemic analysis in the presence of uncertainties using the methods of dynamic financial risks estimation and forecasting, taking into account the time and evaluation of the moment of risk transition to a higher degree. An important practical result is development of information technology and IDSS based on the proposed models and methods, which allows improve the efficiency of processing various types of financial data, to perform an analysis of financial and information risks. The developed IDSS components for behavioral models have allowed identify the customers with the outflow risk, rank them for expected losses, and evaluate financial position of a company as a whole due to the total amount of possible losses.
Results of the thesis fulfillment were used and implemented at the National Bank of Ukraine, companies SAS, ELSIKO LLC, LLC Lifecell, foreign company Artcom Venture GmbH and in the educational process of NTUU “Igor Sikorsky KPI”.
