风险价值度VaR在风险控制领域的应用

发布时间：2018-04-15 21:22

本文选题：风险度量 + 风险价值度VaR　；参考：《山东大学》2014年硕士论文

【摘要】：随着世界经济的不断发展,经济全球化与金融一体化趋势变得越来越明显,金融技术不断地进步,世界金融市场与金融环境发生了翻天覆地的变化,金融风险不断增加,金融风险管理便伴随着金融市场风险的增加应运而生,成为了金融企业与工商管理机构的主要内容。为了更好地度量风险,经过各国金融界的大量的分析研究,基于市场价值测量法(mark-to-market)的风险价值方法(Value at Risk,简称VaR),成为了当前金融界进行风险度量的主要方法。风险价值方法(VaR方法)是近年来国际上比较流行的一种风险管理工具。它在金融风险的计量、预测和控制等领域已得到十分广泛的应用。它的核心内容涉及分布的拟合及尾部分布的处理等问题,当然这些问题也是当今国外金融领域研究的最热门问题之一。传统的研究认为金融市场的变量的收益率的变化应当服从高斯分布(也就是正态分布),但是近些年的研究表明传统的正态分布模型假设严重低估了风险,因为现实中的经济变量的收益率分布并非是严格意义上的正态分布,变量收益率的分布十分的类似正态分布,但却具有厚尾效应,即分布的尾部比正态分布的尾部更加的肥厚。关于如何处理厚尾效应,不同的人有不同的处理方法,有人采用基于极值理论的广义帕雷托(Pareto)分布(即幂律分布)来代替正态分布,有人对传统的历史模拟法进行改进来计算VaR,总而言之,目前对VaR值的计算并没有一套统一的标准方法。本文中首先会对VaR的起源、定义等相关的背景知识做详细的介绍,然后给大家介绍一下目前学术界以及应用界进行VaR值计算的常用的主要方法,并且会对这些方法进行详尽的分析对比,给出每一种方法优点与不足,提出关于VaR值计算的改进方案,保证该方法可以准确的计算出组合的风险价值度的大小,进而准确的度量出组合的风险。当然,在对VaR的改进的过程中,不可避免的会涉及到一些其他的数学模型,此时也会对这些模型做出详细的介绍,以保证文中对VaR值计算方法的改进的严谨性与准确性。例如,在对VaR进行更新的过程中会涉及到对标的变量波动率的更新,此时需要引进波动率更新的常用的指数加权移动平均模型(EWMA模型)与广义自回归异条件异方差模型(GARCH模型)等数学模型,到时候文中会对相关模型做出详细的介绍。最后进行实证分析,通过VaR系统或VaR模型来计算投资组合的VaR值,并结合组合的历史数据来进行回顾测试,对比组合的每天的实际损失是否超出了通过模型计算的得到的VaR值,此时实际损失超出组合的VaR值的情形称之为例外。假如组合的VaR值的置信度为99%,那么在回顾测试中就需要检验例外发生的天数与检测的整体天数的比例是否达到1%,如果例外发生的比例小于等于1%,此时得到结论：VaR模型得到的结果是准确的。当然若例外发生的比例大于1%,此时文中计算VaR的方法是有问题的,需要调整计算VaR值的方法来提高VaR模型的精确度。回顾测试为VaR系统的重要组成部分之一,在进行回顾测试时可能会遇到一些问题,文中会对遇到的问题做详细的分析,以确保VaR系统的准确性。
[Abstract]:With the continuous development of world economy, economic globalization and financial integration trend has become more and more obvious, financial technology continues to progress, the change of the world happened to turn the world upside down the financial market and the financial environment, financial risks increase, financial risk management is accompanied by an increased risk of financial market emerged, become the main content of financial institutions and business enterprises management. In order to measure the risk, through the analysis of a large number of financial circles all over the world, the market value measurement method based on the value at risk (mark-to-market) method (Value at Risk, referred to as VaR), has become the main method of the financial risk measurement.
The risk value method (VaR method) is a risk management tool is more popular in recent years in the world. It is in the financial risk measurement, prediction and control in the field has been widely used. Its core content relates to the fitting and the tail of the distribution problems such as distribution, of course, these problems are the most popular problem today study abroad in the financial field. The traditional research that the change of financial market variables yields should obey the Gauss distribution (i.e. normal distribution), but recent studies show that the traditional assumptions of normal distribution model seriously underestimated the risk, because the reality of the economic variables in the distribution of return rate is not normal distribution in the strict sense, similar to normal distribution is the distribution of variable rate of return, but with a thick tail effect, namely the tail of the distribution tail than the normal distribution. The more fat thickness On how to deal with the heavy tail effect, different people have different processing methods, one using generalized Pareto based on extreme value theory (Pareto) distribution (i.e. power-law distribution) instead of the normal distribution, some of the traditional historical simulation method to calculate the VaR, in short, there is not a uniform standard method for calculation of VaR value.
First the origin of VaR in this paper, the definition of the background knowledge related to do a detailed introduction, and then introduce the current academic circles and the application value of VaR methods of calculation, and will carry out a detailed analysis and comparison of these methods, gives each method advantages and disadvantages, put forward a VaR the value calculation of the improved scheme, to ensure that the method can accurately calculate the value of portfolio risk degree, and then measure the portfolio risk. Of course, in the process of improvement of the VaR, inevitably involves some other mathematical models, this will also make a detailed description of these models. In order to ensure the value of the improvement method for preciseness and accuracy of VaR in this paper. For example, in the process of VaR update involves variable underlying rate update, this requires the introduction of fluctuation Exponential weighted rate update moving average model (EWMA model) and generalized autoregressive conditional heteroskedasticity model (ISO GARCH model) and other mathematical model, and this paper will make a detailed introduction to the related model. Finally, the empirical analysis, through the VaR system or VaR model to calculate the VaR value of investment portfolio, and review test combined with historical data, the actual loss of contrast combination of every day is beyond the model calculated by VaR value, then the actual loss exceeds the combined VaR value of the call exception. The VaR value of the combination if confidence is 99%, then the overall review test will need in the days of days with the exception of the test whether the proportion reached 1%, if the exception occurs the proportion of less than or equal to 1%, then get the conclusion: the results of the VaR model is accurate. Of course, if the cases occurred outside The ratio is greater than 1%, the calculation method of VaR at this time is a problem, need to adjust the VaR value calculation method to improve the accuracy of VaR model. The test review is one of the most important part of VaR system, in the review of the test may encounter some problems, this paper will do a detailed analysis of the problems encountered. In order to ensure the accuracy of the VaR system.

【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F830.91

【相似文献】