【作者】 王雷；

【导师】 秦永松；

【作者基本信息】 广西师范大学 ， 应用统计， 2015， 硕士

【摘要】 最近三十年,受到经济全球化、信息技术以及金融理论等因素的影响,全球金融市场得到了迅速发展。这使得全球的金融市场变得更加开放,全球范围内的资本流通速度加快而且更加自由化。在全球金融市场中不同风险特性的资本得到重新配置和组合,这导致全球金融市场的运作方式和风险的表现产生了很大程度的改变,因而金融市场出现了前所未有过的波动。与此同时,金融机构为了规避金融风险,提升市场竞争力,开展了一系列的金融创新活动,在技术进步和管制放松的刺激情况下,这些活动就显得异常活跃[2]。随着全球经济的快速发展,金融全球化、自由化逐渐加强。随之而产生的市场风险也成为人们的关注热点,全球范围内的专家学者都对其展开了研究。怎样去量化市场风险,也即怎样去测定市场风险,就成为摆在我们面前急需解决的一个大问题。标准差(Standard Deviation)、绝对离差(Absolute Deviation、偏差(Deviation)、下端部分矩(Lower Partial Moment)、风险价值VaR、条件风险价值CVaR、条件期望损失CES等度量方法是目前正在使用或已经被提出来作为度量风险的主要工具。其中,由于巴塞尔委员会对于VaR的认可,VaR开始受到全球金融分析方面专家的青睐,被选用作为金融机构风险管理的国际统一标准。随着VaR模型及其计算方法的不断发展和优化,以及Artzner在1999年初次提出一致性公理后,将VaR作为风险度量的标准遭到了质疑,因为有研究者在理论和实证分析这两方面都证实了VaR对于次可加性并不满足,因而得出VaR不是一致性风险度量的结论[3]。在此情况下,人们为了弥补VaR的不足,于是就开始构造和设计一个既容易估计和计算又满足一致性公理的风险度量。Rockafellar[5]和Uryasev提出了CVaR;Scaillet[6]提出了CES的非参数估计；Artzner[7]等提出了最坏条件期望WCE；Acerbi[4]提出了谱风险度量等等,并证明了它们既是一致性风险度量又能够方便计算。其中CES和CVaR应用的较为广泛,因为它们相对于VaR度量更有优势。本文介绍了国内外对VaR、CES以及CVaR的研究情况,并且简单介绍了四种窗宽选择的方法：主观选择法、参照标准分布法、经验法则和无偏最小平方交叉实证法。在模拟研究中,窗宽通过经验法则选出,使用R软件编程,生成随机数,运用局部线性估计模拟出CVaR和CES的估计值,列出不同分位数p下两者的数值,并进行相应的比较,分析CVaR和CES的变化趋势。然后,以深成指数和上证指数为研究样本,用Eviews软件画出两只股票的日收益率图像,同时对股票日收益率进行ADF检验。从ADF检验和日收益率图像可以得出,深成指数序列和上证指数日收益率序列都是平稳的。然后运用模拟中的局部线性估计对条件风险价值CVaR和条件期望损失CES进行估计,计算出股票市场数据的(:VaR和CES值,统计分析出超过真实VaR的百分比,以此来验证模拟中的结论。更多还原

【Abstract】 By the impact of economic globalization, information technology and financial theory and other factors, global financial markets have been developing rapidly the last three decades. This makes the global financial markets have become more open, and capital flows faster and more liberalization around the world. Capital of different characteristic risks reconfigure and combine around the global financial markets, which leads to the mode of operation and performance of the risk of the global financial markets generate a great deal of changes.so there has been an unprecedented volatility over the financial market. At the same time, financial institutions launch a series of financial innovation in order to avoid financial risks and enhance the market competitiveness, and these activities appear to be extremely active under the conditions of the technological advances and the simulation of deregulation. With the rapid development of the global economy, financial globalization and financial liberalization gradually strengthened. Market risks generating from which have become a hot attention, experts have carried out research in this area all over the world. How to quantify market risks, that is how to measuring market risks, has become a big problem that needing to be resolved in front of us.The main tools that have being used or have been proposed to measure the risk are Standard Deviation, Absolute Deviation, Deviation, Lower Partial Moment, Value at Risk(VaR), Conditional Value at Risk(CVaR), Conditional Expected Shortfall(CES) and so on. Due to the recognition of the Basel Committee on VaR, global financial analysts have been in favored of VaR, and VaR was chosen as the international standard for risk management of financial institutions. With the growing development of VaR model and its calculation and optimization, after the Artzner proposed the consistency axiom in 1999, the standard that taking VaR as a risk measurement tool has been questioned, because that some researchers confirmed that the VaR don’t satisfied the sub-additivity property in both theoretical and empirical analysis, and thus draw a conclusion that VaR is not a coherent risk measurement tool. In this case, in order to compensate for the lack of VaR, people began to construct and design a risk measurement tool that satisfy both estimate easily and the consistency axiom. Rockafellar and Uryasev proposed CVaR; Scaillet proposed the nonparametric estimation of CES; Artzner put forward the worst conditional expectation(WCE); Acerbi proposed spectral risk measurement, etc. and proved that they are both coherent risk measurement tools and can calculate easily. Due to the advantages relate to VaR, CES and CVaR is widely used as a risk measurement tool.This paper describes research conditions on VaR、CES and CVaR at home and abroad, and introduced four methods of selecting windows width:subjective selection method and the reference standard distribution method, rule of thumb and unbiased least squares cross-empirical method. Here I use the local linear estimate to estimate the conditional expected shortfall and conditional value at risk. The window width was selected by rule of thumb in the simulation. I use the R software program to generate random numbers to simulate CVaR and CES estimates, and display the values at different quantiles and compare to each other, then analyze the changing trends at CVaR and CES. Then, I select the Shanghai Composite Index and Shenzhen Component Index for the study, then I use Eviews to draw the graph of the daily return rate of stock and take ADF test for the daily return rate of stock. and found that the Shanghai Composite Index and Shenzhen Component sequence diagram fluctuate around zero, so that the series is stationary. then I calculate the CVaR and CES values of stock market data by using the Local linear estimation, and calculate the percentage of which exceed the real VaR to verify the simulation conclusions.更多还原