Research On Credit Risk Measurement And Contagion Effect Based On Defaults Dependency

In recent years, the continuous financial crisis bring huge attacks to global economy. During the period of crisis, it usually appears a phenomenon that one company default causes the related companies default successively and even go bankrupt, which have a domino contagion effect. Along with the financial market opens and the contact increases, the joint default events and contagion effects caused by default dependency grows quickly, which makes a large amount of loan capital lost and brings big challenge to credit risk management of bank. So how to join the default dependency into credit risk management system become anxious to be solved.However, because of the laggard risk management technology, the domestic bank cannot meet the needs when they measure the complicated default dependent risk, which severely restrict the development of Chinese banking. Therefore, through studying and absorbing advanced risk management technologies abroad, it is necessary to correctly measure the default dependency of correlated companies so as to effectively prevent and control the dependent credit default risk, which is important essential to the development of financial market and credit risk management of commercial bank of China. The studies on dependent default risk measurement, contagion effect and risk precaution measures are becoming urgent and important problems.This paper centers on the two questions of dependent default risk measurement and contagion. On the basis of theoretical analysis of credit risk quantification and generation theory of default dependency, we establish a quantitative analysis framework for dependent credit default risk measurement by combining copula theory. Then we use Chinese capital correlated listed companies as sample to do empirical study on dependent default risk measurement and contagion effect test. And we put forward some useful measures on credit portfolio risk measurement management, making risk limitation and contagion immunity plan to help bank' s dependent default risk precaution and controlling. The main works and conclusions are as follows:(1) We state and compare the credit risk theory, models and methods, and give analysis to the default dependency' s principle, characteristic, classification and the influence factors, which provide theoretical and technical support for the construction of credit risk model with default dependency. We summarize that the default dependency with four characteristics can realize in four ways. The positive default dependency is the root cause of the wide default event happened. Cyclical default dependency originates from the volatility of macroeconomic factors, and the default contagion comes from the close relationship between enterprises. Capital correlation, credit quality, strategic alliance and relationship between bank, government and enterprise are the most important influencing factors of default dependency.(2) We propose a quantitative analysis framework of credit risk with default dependency based on Copula function. That is, using Copula theory method to solve the description of dependency structure, implanting Copula function in the structure model or the intensity model to transform a single risk measure into dependency risk measure. Where, the choice of Copula function is the critical problem. We compare the choice methods of Copula function, and find that using the maximum likelihood estimation for parameter estimation of multiple Copula function, together with the Kernel density nonparametric estimation directly fitting the joint distribution, can effectively avoid the dependency structure distorted in describing due to the improper hypothesis of marginal distribution and Copula families. This paper further put forward a goodness-of-fit test method based on the kernel density estimation and minimum distance test, which can effectively solve the problem of multiple Copula choice. Thus, the goodness-of-fit test of Copula extending to multiple conditions has realized in practice.(3) Based on the quantitative framework on credit risk with default dependency, we choose capital groups with capital correlated members in Chinese stock market as samples to do empirical study. According to the selected optimal Copula to describe the dependency structure, we measure the joint default probability of capital groups and give credit rating to them. Based on this, through the Monte Carlo simulation of loss distribution, we separately calculate the losses of credit portfolio, and its VaR and CVaR, in the condition of secured loan and unsecured loan. Then we use Kupiec failure test and Christoffersen interval prediction test to verify the effectiveness of the model. Finally, we compare the default dependency risk in different influencing factors. Results show that:the credit risk of capital group companies in China is with five features, which are internal frequent related transactions, universal chain guarantees, distorted financial statements, high systemic risk and difficulty in risk identification. The Copula functions applied in capital groups with same scales are relatively concentrated. The credit assets in smaller capital groups have obvious peak and fat tail, but the dependency structure tends to symmetry increasingly in large-scale capital groups. The effective rate of measurement model of credit risk with default dependency is more than 90 percent. There are 41.67% of capital group companies above A credit-level. Only 5% of trade time in a year may have huge losses, which shows the elusive, sudden and huge loss features of default dependency. The credit assets losses do not obey the normal distribution, presenting a peak and deflection. In the condition of guarantee loan, the potential loss of extreme value event is small. The default dependency of capital group companies is smallest under horizontal merger, and specialized operation, but is biggest under government leading.(4) Based on the results of value at risk, according to the optimization principle, we establish a linear programming model in confidence level with CVaR as target function and the expected return as constraint conditions, which can optimize distribution of unify credit line of the capital correlated company within the members. It helps the banks to realize the minimum credit portfolio risk with expected returns, in issuing and managing loans. Then we do empirical analysis based on Chinese capital group companies. Results show that, given different expected returns, the bank can obtain corresponding optimal credit combination to make the credit portfolio risk to the minimum. And by adjusting the expected returns, we can get the effective boundary of credit portfolio.(5) In the quantitative analysis framework of credit risk, the credit risk contagion effect generated by default dependency is studied. From the two aspects of Causal effect and information effect, we give explanation to the mechanism of default contagion, and describe the timely marketing effect and delayed marketing impact. Then we do empirical studies on default risk contagion of listed companies with correlated capitals. By using granger causality test, we verify the existence of risk contagion. Then we use the time-varying structure Copula of time series and the sequential diagnosis of Z test to do contagion point diagnosis and the risk of spillover effects test. Results show that the risk dependency changes a lot before and after the contagion point, some of companies have obvious and frequent fluctuation spillover effect, which declare that the risk fluctuation spillover is the result of severe contagion evolution. With the positive default dependencies between companies are more likely to infect the credit risk, and risk transmit in one-way or infect cross the members. We summarize that cyclical factors control the average of default risk and the default contagion fluctuate around the average risk increasing the extra risk losses.(6) On the basis of credit default risk measurement and risk contagion test, we propose three-step strategies on default dependency risk precaution and control from advance evaluation, concurrent control and afterwards tracking to improve the prevention and control effect. Namely: first, by carrying out the quantitative management of credit portfolio risk effectively, we can master customers' credit status and related information before loaning; secondly, by imposing credit ceiling mechanism of default dependency risk, we can prevent the excessive credit financing and multi-head acquisitions of credit loans. Thirdly, by setting credit risk contagion immune program, we can timely track and feedback default signals and master the risk spillover effects, so as to prevent huge losses caused by default risk contagion.