Research On Credit Risk Dependence Of National Economic Industry System

The integration of industries with capital closely linked is an important feature of the modern economy.Because of the complementary or similar business scope of the national economy,it is obvious that the degree of correlation and the transmission of the default become more serious.The risk of one industry will affect the development of other related industries and even the entire market,triggering the system crisis.As the benchmark of international banking supervision practice,"Basel Agreement III" put forward suggestion of effective prevention of systemic risk from the perspective of macro prudential,To implement effective macro prudential regulation,we need to understand the possibility of a series of continuous losses caused by an event among the system consisting of a series of institutions and markets,i.e.the systemic risk with feature of overflow and contagion..In fact,it is the systemic risk that causes the systematic risk,Therefore,in order to prevent systematic risk,we need to understand the mechanism of risk transmission,such as the measure of risk,risk dependence,trigger events and the contagion path.At present,the research on multivariable portfolio credit risk is far from mature,especially when there exists complex nonlinear dependence relationship between individual variables,the measurement of the portfolio credit risk will become more complex.The existing risk dependence model about multivariate often leads to the mismeasure of the credit risk,misjudge of risk contagion mechanism and the misrecognition of systemically important department,so as to mislead formulating measures of macro prudential supervision.Based on the analysis on the measurement of credit risk and risk dependence,we empirically analyse the influence of macroeconomic variables on the credit risk of a single industry and inter industry dependence in different market,to investigate the existence of periodic default dependencies.In this paper,a quantitative analysis framework is established to describe the complex dependence structure of multiple variables and to systematically analyze the characteristics,effects and path of credit risk contagion and to provide a basis for the prevention and control of systematic risk.The main work and conclusions are as follows:(1)From the perspective of external macro factors and internal infectious factors of credit risk,this paper discusses the generation mechanism of credit risk dependence and contagion mechanism of credit risk.Through the analysis and comparison of the existing literature,the conclusion is drawn as follows: The structural model based on endogenous default mechanism is the most suitable method to measure credit risk of current China.Besides,the copula function,which can describe the complex nonlinear correlation between random variables,has the advantage of measuring the dependence on default.(2)In this paper,the risk measurement model(CCA)is modified to GARCH-CCA model from the perspective of time varying volatility of assets.and is applied to measure credit risk of nine industries of national economy industry system.The main empirical results are as follows: Consistent with the macro economic development trend,the overall trend of every industry's default risk experience are roughly the same.In the virtual economy,the credit risk of the financial industry is the highest,followed by the real estate industry.In the real economy,the credit risk of the mining industry is the highest one,followed by the manufacturing industry.(3)BALQR model is constructed based on credit risk of the industry and the Gibbs sampling algorithm for quantile regression is put forward to examine the importance of macro factors affecting the credit risk of the industry under different market conditions.Using binary copula function to measure the dependence of credit risk,and combining the macro factors,we put forward the periodic dependence criterion of industry credit risk.Empirical evidence shows that there are significant differences in the impact of macroeconomic factors on credit risk among different industries and among different quantile,There exists periodic default dependency between some industries on upper and lower tail,but it is not the main form of credit risk dependence and the performance of periodic defaults dependency is more obvious in the presence of lower tail relative to the upper tail.(4)The R-vine copula model is constructed to describe the dependence structure of credit risk in industry,and the contagion dependence characteristics of industry credit risk are found.The suitable vine copula structure type is selected by the model evaluation standard,all "pair-copula" types and parameters are identified and estimated.Through the empirical research,nonlinear correlation between each industry's credit risk is found,catalytic industry and conditional isolation industry are also found out in the process of credit risk contagion.(5)The general algorithm for simulation of response to extreme value under R-vine structure is proposed and is applied in the dependency structure of credit risk of nine industries.The risk contagion effect of one industry crisis in the whole system is investigated and systematic important industry are recognized.The simulation results show that the real estate industry has the greatest impact on the whole industry system followed by the information industry and transportation industry,the importance of the financial industry is of the minimum.In addition,the credit risk contagion within the entity economy is obvious,the virtual industry has strong risk immunity and strong risk infectivity,the financial industry is neither easy to be infected nor easy to infect other industries.(6)PCBN model is constructed based on credit risk of national economy industry using learning algorithm of network structure to study the joint distribution of credit risk of nine industries and the path of credit risk contagion.Further more,the R-vine copula model and PCBN model are comparatively analysed in theory and practice.The industries that has an important role in the hub and the industries with conditional independent relationships and the path of credit risk contagion are found out from the empirical result.No matter in the number of parameters to be estimated or whether it reflects the conditional independence relations or whether it reflects the view of causal relationship,PCBN model is superior to R-vine copula model by comparison.But as for the model accuracy,R-vine copula model has a slight advantage.