基于修正KMV-藤Copula的中小企业信用风险的研究

发布时间：2018-03-05 10:40

本文选题：KMV模型　切入点：Copula技术　出处：《西南交通大学》2015年硕士论文　论文类型：学位论文

【摘要】：作为金融领域重要组成部分之一的商业银行,其能否健康有序的发展,关系着整个国民经济的发展,而作为商业银行最重要风险之一的信用风险,对其进行度量就成为商业银行所面临的一个重要课题。本文从一维和多维的角度对企业的信用状况进行了探究。首先,文章介绍了KMV模型的基础理论及其结构；然后,利用修正的KMV模型对单资产的信用风险做了实证研究。其中,基于股权价值影响因素的分析,利用非流通股和流通股计算了股权价值,并直接利用股权价值的对数收益率来分析波动特性。由于大的波动紧跟大的波动,小的波动紧跟小的波动,构建了以波动变化为门限变量的GARCH模型,来刻画一般的GARCH模型所不能描述的非对称性；最后在多维角度上,本文引入藤Copula,结合修正的KMV模型,分析了基于同一板块的四家公司的联合违约概率,得出公司间是否违约是会相互影响的。接着基于修正的KMV-Copula模型,探究了这种影响程度,求解了当效益相对较好的公司违约时对其它家的条件违约概率和效益相对较差的公司违约时对其它家的条件违约概率,最终得出效益相对较好的公司违约时对其他家的影响要比效益相对较差的公司违约时对其他家的影响大。其中,还尝试了加入藤结构下的Pair Copula来描述高维相关结构,将信用风险的研究,由二维拓展到高维,并且使用Pair Copula分解技术,不仅保留Copula函数的优势,而且模型在建立多元Copula函数时更加灵活,条件限制更少。本文挑选Copula函数时,挑选了二元Frank Copula函数、Guassion Copula函数、Gumbel Copula函数、Clayton Copula函数,然后根据AIC准则选取最优的Copula,使其在衡量商业银行信用风险方面可以更加准确。最后,基于文章的实证结果及社会形势,提出了建议和期望。
[Abstract]:As one of the important parts of the financial field, whether the commercial bank can develop healthily and orderly is related to the development of the whole national economy, and as one of the most important risks of the commercial bank, the credit risk. It is an important subject for commercial banks to measure it. This paper probes into the credit status of enterprises from the perspective of one-dimensional and multi-dimensional. Firstly, the paper introduces the basic theory and structure of KMV model, and then, An empirical study on credit risk of a single asset is made by using the modified KMV model. Based on the analysis of the influencing factors of equity value, the equity value is calculated by using non-tradable shares and tradable shares. And directly use the logarithmic return of equity value to analyze the volatility characteristics. Because the large fluctuation follows the big fluctuation, the small fluctuation follows the small fluctuation, the GARCH model with the fluctuation change as the threshold variable is constructed. Finally, we introduce Fujiao Copula, combined with the modified KMV model, to analyze the joint default probability of four companies based on the same plate. Then, based on the modified KMV-Copula model, we explore the extent of this impact. The probabilities of conditional default to other countries and the conditional default probability to other countries when the company with relatively good benefit is in default and the company with relatively poor benefit are solved. In the end, it is concluded that the company with relatively good benefits will have a greater impact on other households when it defaults than the less effective one. Among them, the Pair Copula under the rattan structure is also tried to describe the high-dimensional correlation structure. The study of credit risk is extended from two dimension to high dimension, and the Pair Copula decomposition technique is used, which not only preserves the advantage of Copula function, but also makes the model more flexible and less restricted in the establishment of multivariate Copula function. This paper selects the binary Frank Copula function and its Gumbel Copula function, and then selects the optimal Copula according to the AIC criterion to make it more accurate in measuring the credit risk of commercial banks. Finally, based on the empirical results and the social situation of the paper, Suggestions and expectations are put forward.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F276.3;F832.4;F224

【参考文献】