基于Pair-Copula函数的商业银行操作风险度量

发布时间：2018-03-08 09:36

本文选题：操作风险　切入点：Bootstrap抽样　出处：《河南师范大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着一些因操作风险而导致的损失事件的发生,如何准确而有效地度量操作风险已经成为广大学者和银行业重点关注的问题.在我国,对操作风险的研究和度量尚处在摸索阶段,大多属于定性分析,因此,对操作风险行之有效的度量方法的研究是十分有意义的.本文以商业银行的损失数据为样本,利用Bootstrap抽样方法,在损失分布法的基础上得到损失缺口,考虑到操作风险各类损失事件之间存在一定的相关性,利用Copul a理论整合四类操作风险的损失数据,由于多元随机变量两两之间适用的Copula函数不一定相同,基于此,本文提出了Pair-Copula的方法,它可以选择不同的Copula函数,进而更加准确地刻画随机变量之间的相关性,而且简化了参数估计的过程.在此基础上,利用Monte Carlo模拟法计算单类操作风险的在险价值VaR,最后,计算出四类操作风险的整体损失.研究结果表明:Bootstrap抽样方法有效克服了操作风险数据缺乏的问题;基于Pair-Copula方法的模型构造,能够有效捕捉到损失事件之间的相关性;引入的VaR可以准确度量操作风险.
[Abstract]:With the occurrence of some loss events caused by operational risk, how to measure the operational risk accurately and effectively has become the focus of attention of scholars and banks. The research and measurement of operational risk are still in the exploratory stage, most of them belong to qualitative analysis. Therefore, it is very meaningful to study the effective measurement methods of operational risk. This paper takes the loss data of commercial banks as a sample. Using Bootstrap sampling method, the loss gap is obtained on the basis of loss distribution method. Considering that there is a certain correlation between various loss events of operational risk, Copul a theory is used to integrate the loss data of four kinds of operational risks. Because the suitable Copula function between two pairs of multivariate random variables is not necessarily the same, this paper proposes a method of Pair-Copula, which can select different Copula functions and more accurately depict the correlation between random variables. Furthermore, the process of parameter estimation is simplified. On this basis, the Monte Carlo simulation method is used to calculate the risk value of a single type of operational risk. The overall loss of four kinds of operational risks is calculated. The results show that the problem of lack of operational risk data is effectively overcome by using the method of "1: bootstrap sampling", and the model structure based on Pair-Copula method can effectively capture the correlation between loss events. The introduced VaR can accurately measure the operational risk.
【学位授予单位】：河南师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F224;F831.2

【相似文献】