The Copula-Based Measurement Of The Market Risk And The Operational Risk Under Mixed Operation

Mixed operation is a special mode of operation that different financial bodies which belong to different kinds of industries operate all the businesses both from their own industry and other industries.Under this operation mode,every financial body operates more businesses and provides more kinds of service than they do under general situation.Though we pay less attention on the operation mode of our country's financial bodies than other countries,the situation that the marketing environment is more and more complicated as the progress of the science and technology and the development of social economy leads to the consideration about it for financial bodies.After becoming a member of the World Trade Organization,our country's economy develops much more faster and the competition between the financial bodies is more and more intense.Under the above background,many financial bodies begin to adopt diversified operation mode both for their existence in the competition and their better development.This phenomenon makes the strictly separate operation mode which is always implemented by the financial bodies in our country changed and the restrictions about the scope of operating businesses for these financial bodies are relaxed.From another perspective,the mixed operation mode has been the development trend of the operation mode for the international financial industry and it also the main operation mode for many developed countries such as America,Japanese,England,Germany and France etc.Therefore from the current development trend of domestic and international financial industry,the mixed operation mode must be the inexorable trend for the operation mode of our country and under this trend the study about the mixed operation naturally becomes the necessary and inevitable subject for the scholars.By concluding and analyzing the available research related to the mixed operation,we found that the study about this subject mostly focuses on the bank efficiency and financial stability under mixed operation but the research related to the risk faced by financial bodies under mixed operation is very few.And almost all the research about the risk faced by financial bodies under mixed operation is the qualitative analysis about the risk but for the quantitative study about it is almost no as far as I know.Market risk,operational risk as well as the credit risk are there main financial risks in financial industry.The analysis and measurement as well as other research for every kind of financial risk are always the main subject studied by risk managers and scholars especially after the economic crisis with different scope and different scale occurs.As the precondition for effective management of financial risks is the accurate measurement of them,the research about risk measurement is the key and difficult point for other research about risks.Based on the above background,this thesis put forward the research about the measurement of market risk and operational risk which are the two main financial risks in financial industry.Both the theoretical study and the empirical study of the quantitative risk measurement of these two main risks are given in the thesis.The main content in every part of this thesis is as follows:In the first chapter,the background of the subject in this thesis,the research significance and the research overview for every part of the thesis are given.Besides the research content as well as the structure arrangement and innovation point etc.of the thesis are stated too.In the second chapter,the copula-based grouped risk aggregation model is built for the the measurement of the market risk faced by the financial body under mixed operation based on the characteristics of the market risk under mixed operation.In this model,all the basic risks are grouped according to the type of financial products they come from and different copulas are used to describe the dependency relationships for different groups of basic risks.By this model,the high-dimensional copula in traditional measuring methods is substituted by the lower-dimensional copulas by the above way which is not only more consistent with the actual situation but also avoids the dimension disaster in the applications of copula.In addition,the simulation algorithm and steps based on this model are given in the thesis and the convergence of this algorithm is proved in the thesis too.In the third chapter,the study on the market risk measurement faced by the finan-cial body under mixed operation based on the vine copula model is given.According to the theory and methods of different vine copulas,the value of VaR is calculated based on the C-vine copula,D-vine copula and R-vine copula model respectively.The algorithm and steps for fitting the dependency structure of multidimensional data by C-vine copula model as well as the steps and method for calculating the VaR of portfolio by Monte Carlo simulation are described in this chapter.Additionally,the comparison and analysis for the VaR of market risk under different vine copulas are represented too.In the fourth chapter,the theory and method for measuring the operational risk for the commercial bank under mixed operation is studied based on the binary copula model.The operational risk caused by inner fraud and external fraud and the total operational risk are both studied.Based on the extreme value theory and the property of the exponential distribution,the marginal distribution functions are discussed and the second-order approximated analytic expressions for these distribution functions are given.The total operation risk measurement is studied in the case that the lost caused by inner fraud and the lost caused by external fraud are independent and the case that the above-mentioned two kinds of lost are not independent.In the independent case,the distribution function of the total lost is given.In the dependent case,the dependency structure between the above-mentioned two kinds of lost is described by different binary copula models whose tail behavior differs from each other and the analytic expressions for the joint distribution function of these two kinds of lost based on different models are given.Meanwhile,the algorithm and steps for calculating the VaR of total operational risk as well as the method and procedure for estimating the parameters of every binaxy copula model are elaborated in this chapter.In the fifth chapter,the empirical study about the operational risk measurement based on the GPD-copula model is given.Firstly the parameters of the marginal distribution functions and every binary copula are estimated and then the VaR of the sample data is calculated based on Monte Carlo simulation.During the process of parameter estimation for every binary copula model,different estimation methods are used and the final estimated value of each parameter of these two-dimensional copulas is substituted by the weighted average of all the estimated value under different method.The processed value of the corresponding standard error of every estimated parameter under different method is the weight of the above weighted average.Finally the value of VaR under different models is compared and analyzed.After estimating the parameters of every binary copula model,the goodness of fit for every copula based on parameter bootstrap is tested and the dependent degree and dependency structure between inner fraud risk and external fraud risk are discussed based on the tested result.In the sixth chapter,the conclusion of the whole study in the thesis is given.Additionally,the inadequacy in this study is analyzed and the further research about this subject is prospected.The main research results in this thesis are as follows:1.The copula-based grouped model for the measurement of market risk under mixed operation is built.2.Based on the above model,the simulation algorithm and steps are given for calculating the VaR of market risk and the convergence of the algorithm is proved in the thesis too.Meanwhile,the example of the simulation and the analysis on the simulation results which proved the effectiveness of the algorithm are provided.3.The algorithm and steps for fitting the dependency structure of multidimensional data by C-vine copula model as well as the steps and procedure for calculating the value of VaR for portfolio by monte carlo simulation are given.4.The value of VaR for sample data of market risk under different vine copula models is obtained and all the value of VaR is back-tested by Monte Carlo simulation.Based on the back-testing results,the performance in empirical study of market risk measurement of every vine copula model is analyzed and compared and the conclusion that C-vine copula model performs best is drawn.5.In the study of operational risk measurement,the second-order approximated analytic expressions for the distribution functions of the lost caused by inner fraud and the lost caused by external fraud based on the extreme value theory and the property of the exponential distribution are given.6.The algorithm and steps for estimating the parameters of marginal distribution functions and different binary copula models axe given.7.The algorithm and steps for calculating the value of VaR for the total operational risk based on Monte Carlo simulation are given.8.The value of VaR for sample data of operational risk under mixed operation based on GPD-copula model is calculated and the value under different models is compared and analyzed.The compare and analysis find that using the binary Gumbel copula to describe the dependency relationship between inner fraud risk and external fraud risk can not only measure the operational risk under mixed operation more accurate,but also reduce the regulatory capital for commercial bank.9.The goodness-of-fit for every binary copula model is tested by the method of parameter bootstrap and based on the testing result,the conclusion that the binary Gumbel copula describes the dependency relationship between inner fraud risk and external fraud risk better than the binary t copula and the binary Clayton copula describes the relationship worst is drawn.10.The dependent degree and dependency structure between inner fraud risk and external fraud risk in the total operational risk are discussed based on the testing results and it concludes that the inner fraud risk and external fraud risk owns a significant right tail relationship.