基于Copula函数的整合风险度量研究

发布时间：2018-06-24 17:52

本文选题：相依结构Copula + 整合风险度量　；参考：《湖南师范大学》2014年硕士论文

【摘要】：随着经济金融全球化的快速推进,金融市场的风险日趋复杂化,已从原来的仅仅考虑单一风险转向综合考虑多元化风险,因此对风险进行整合度量已成为必然。现在,整合风险管理模式已引起了业界和学界的高度重视,整合风险管理模式的核心是对市场风险、操作风险和信用风险等不同类别风险的整合管理。目前金融机构面临的风险主要包括信用风险、市场风险和操作风险,各风险之间存在相关性,并且具有非线性和尾部相关性等特点。传统的在正态分布假设下的Pearson相关系数分析方法已不再适用,而Copula函数不受边际分布选择的限制,更好的描述风险间的相关结构。所以,本文系统全面地研究了Copula在金融整合风险度量中的应用。理论部分,首先概括了Copula相关理论知识,然后对常用Copula的相关性进行深入的分析,并探讨了如何对Copula模型进行参数估计与模型选择,最后详细总结了风险度量指标及其测量方法。实证部分,鉴于度量整合风险必须在充分考虑各风险间相关性的基础上进行,本文分别从风险间的相关性和整合度量风险值这两个方面对整合风险度量进行实证研究。第一个实证,是分析信用利差与市场风险联动性,利用基于相关系数的Copula参数估计法研究相关结构,实证结果表明：两者存在一定的正相关,相关结构可以较好地用Frank Copula函数来刻画。在第二个实证中,我们以12家中国上市商业银行为研究对象,首先确定各风险收益率的分布,然后利用Copula构建相依结构,并选择最优模型,最后用蒙特卡罗模拟算法和重要性抽样算法计算不同风险组合的VaR和CVaR,并用返回测试比较两种算法的优劣,研究结果表明当损失较小时,Monte Carlo计算的风险值更有效。以上工作为我国研究整合风险管理模型提供了较好地理论和技术支持。
[Abstract]:With the rapid development of economic and financial globalization, the risk of financial market is becoming more and more complicated. At present, the integrated risk management model has attracted great attention from the industry and academic circles. The core of the integrated risk management model is the integration management of different types of risks such as market risk, operational risk and credit risk. At present, the risks faced by financial institutions mainly include credit risk, market risk and operational risk, each risk has a correlation, and has the characteristics of nonlinear and tail correlation. The traditional Pearson correlation coefficient analysis method under the normal distribution hypothesis is no longer applicable, while the Copula function is not restricted by the marginal distribution, and it is better to describe the correlation structure between risks. Therefore, this paper systematically studies the application of Copula in financial integration risk measurement. In the theoretical part, we summarize the theory of Copula, then analyze the correlation of copula, and discuss how to estimate and choose the parameters of Copula model. Finally, the risk measurement index and its measurement method are summarized in detail. In the empirical part, in view of the fact that the integration risk must be measured on the basis of fully considering the correlation among the risks, this paper makes an empirical study on the integration risk measurement from the two aspects of the risk correlation and the integrated risk measurement. The first is to analyze the linkage between credit spread and market risk, and to study the correlation structure by Copula parameter estimation method based on correlation coefficient. The empirical results show that there is a certain positive correlation between them. The related structure can be well characterized by Frank Copula function. In the second empirical study, we take 12 Chinese listed commercial banks as the research object, first determine the distribution of the risk return rate, then use Copula to construct the dependent structure, and select the optimal model. Finally, the Monte Carlo simulation algorithm and importance sampling algorithm are used to calculate the VaR and Cvar of different risk combinations, and the advantages and disadvantages of the two algorithms are compared with the return test. The results show that the Monte Carlo method is more effective when the loss is small. The above work provides better theoretical and technical support for the study of integrated risk management model in China.
【学位授予单位】：湖南师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F830.9

【参考文献】