风险评价的VaR方法及应用

发布时间：2018-01-12 07:10

本文关键词：风险评价的VaR方法及应用　出处：《青岛大学》2017年硕士论文　论文类型：学位论文

更多相关文章： VaR GARCH Cornish-Fisher Chebyshev-Markov

【摘要】：经济全球化带来的全球竞争,使我国的金融市场在这几年内风起云涌,变化不断,全球各个国家的金融市场大都出现波动,使得各个国家都把风险管理放在了比较首要的位置。一个主要的风险度量方法就是VaR,即风险价值,VaR方法使得金融产品或组合的风险能够用一个可与其收益匹配的简单的数字来表示,方便地度量金融市场的风险水平。本文介绍了VaR的一些基本概念和性质,建立了VaR的点估计以及区间估计。本文用VaR与GARCH类模型相结合,建立正态分布、t分布以及GED分布下GARCH类模型,得到VaR的点估计和区间估计,并将其与Cornish-Fisher展开式和Chebyshev-Markov近似分位数得到的VaR的点估计和区间估计进行比较,其中点估计利用Kupiec提出的返回检验的方法作为比较准侧,区间估计利用覆盖率(CP)、参数落在区间上侧的错误概率(Upper)、参数落在区间下侧的错误概率(Lower)以及区间长度的中位数(Med)来作为比较准则,最后发现利用在GED分布的EGARCH模型下,Chebyshev-Markov(CM)近似分位数得到的VaR的估计效果较好。
[Abstract]:The global competition brought by economic globalization makes the financial market of our country surge and change in the past few years, the financial market of every country of the whole world appears the fluctuation mostly. Make each country put risk management in a more important position. One of the main risk measurement method is VaR, that is, the value of risk. The VaR approach enables the risk of a financial product or portfolio to be represented by a simple number that can match its earnings. This paper introduces some basic concepts and properties of VaR, establishes the point estimation and interval estimation of VaR. In this paper, we combine VaR with GARCH model. The model of GARCH class under normal distribution and GED distribution is established, and the point and interval estimates of VaR are obtained. It is compared with the point and interval estimates of VaR obtained by Cornish-Fisher expansion and Chebyshev-Markov approximate quantiles. The point estimation uses the method of return test proposed by Kupiec as the relative quasi-side, and the interval estimation uses the coverage rate to estimate the error probability of the parameter falling on the upper side of the interval (Upper). The error probability of the lower side of the interval and the median Meder of the interval length are used as the comparison criteria. Finally, it is found that the parameters are used in the EGARCH model of the GED distribution. The VaR estimated by Chebyshev-Markov-CM) approximate quantiles is effective.
【学位授予单位】：青岛大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F832.5

【参考文献】