基于Copula函数的债市相关性与风险价值分析

发布时间：2018-02-24 20:30

本文关键词： Copula函数尾部相关系数广义Pareto分布 VaR　出处：《天津财经大学》2012年硕士论文　论文类型：学位论文

【摘要】：当前,随着我国经济持续稳定的发展,财政收入快速增长,财政政策的宏观调控作用也不断增强,国家的各项基础设施建设取得了重大成就。但同时,我国的财政风险问题也逐渐显现出来。国家财政支出压力越来越大,国债发行规模不断扩大,尤其是地方政府债务负担越来越重,这使得我国的财政风险进一步扩大,如何防范财政风险,保持财政的稳定性和可持续性,已经成为一个重要性的课题。当前财政风险问题的研究以定性分析为主,定量分析极少,并且大多数以财政风险的总体度量指标来衡量中央政府的财政风险,缺少对微观债券流通市场风险的准确度量。即使有这方面的一些研究,大部分都是用传统的线性方法对有关变量的风险相关性做了分析。然而这样就导致了一些问题,如正态分布假设是否合理?传统的线性相关系数是否能较好地反映变量之间的相依关系?怎样才能准确度量出组合变量的风险价值?这些问题的解决对于我国债券市场的风险管理具有重大的意义,进而能够更加全面的认识我国的财政风险,为财政风险的防范提供一些有用的措施和建议。本文利用极值理论中的Copula函数度量了国债市场和企债市场的尾部相关性风险,并将其与传统的线性方法得出的相关系数进行了对比。得出：传统的线性相关系数低估了市场的非系统性风险,市场中存在着大量的极值风险。接着,将两个市场收益率序列用广义Pareto分布进行拟合,采用基于Copula函数的蒙特卡罗模拟法,计算出不同的资产权重和置信水平下的市场组合风险价值,即组合VaR (Value at risk).最后,针对我国目前债务风险的现状,给出了一些政策建议。
[Abstract]:At present, with the sustained and stable development of China's economy, the rapid growth of fiscal revenue and the increasing macro-control role of fiscal policies, major achievements have been made in the construction of various infrastructure facilities in the country. The problem of fiscal risk in our country is also gradually emerging. The pressure of national financial expenditure is increasing, the scale of national debt issuance is expanding, especially the local government debt burden is becoming heavier and heavier, which makes the financial risk of our country further expand. How to prevent financial risks and maintain financial stability and sustainability has become an important issue. At present, the research of financial risk is mainly qualitative analysis, but the quantitative analysis is rare, and most of them measure the central government's fiscal risk by the overall measure of fiscal risk. There is a lack of accuracy in the market risk of micro-bond circulation. Even with some research in this area, the traditional linear method is used to analyze the risk correlation of the relevant variables. However, this leads to some problems. Is the assumption of normal distribution reasonable? Can the traditional linear correlation coefficient better reflect the dependent relationship between variables? How can we accurately measure the risk value of portfolio variables? The solution of these problems is of great significance to the risk management of our country's bond market, and then we can fully understand the financial risk of our country, and provide some useful measures and suggestions for the prevention of the financial risk. In this paper, we use the Copula function in the extreme value theory to measure the tail correlation risk between the bond market and the enterprise bond market. By comparing it with the correlation coefficient obtained by traditional linear method, it is concluded that the traditional linear correlation coefficient underestimates the non-systemic risk of the market, and there are a lot of extreme risk in the market. Two market yield sequences are fitted with generalized Pareto distribution, and Monte Carlo simulation method based on Copula function is used to calculate the market portfolio risk value under different asset weights and confidence levels, that is, portfolio VaR value at risk.Finally, In view of the present situation of debt risk in China, some policy suggestions are given.
【学位授予单位】：天津财经大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F832.51;F224

【参考文献】