基于GARCH模型的VaR及CVaR在金融风险度量中的应用

发布时间：2018-07-06 15:21

本文选题：金融风险 + GARCH模型　；参考：《复旦大学》2014年硕士论文

【摘要】：风险是指未来收益的不确定性,金融风险是指金融变量的变动所引起的资产组合未来收益的不确定性。通常我们关注的是风险可能带来的损失,因此可以将风险的概念表述为“由于结果的不确定性而带来损失的可能性”。随着经济全球化以及金融市场一体化,金融市场变得愈加复杂。如何有效地管理金融风险是国内外金融理论界和实物界关注的重中之重。金融市场风险度量是度量由于市场因子的变化而致使金融资产产生的损失。金融风险度量是风险管理的核心部分。目前,金融市场风险度量的主流方法主要有：均值-方差分析、灵敏度方法、波动性方法、VaR方法、压力试验以及极值理论。研究表明,金融资产收益率的时间序列具有尖峰后尾的特性,其并不服从正态分布。为了正确估计金融风险,本文利用GARCH族模型,选取2004年1月2日至2013年12月31日的上证指数的日收盘价数据,比较分析了假设收益率序列服从正态分布、t分布以及GED分布下的GARCH基于族模型的CVaR以及VAR的计算方法,并且和传统的VaR方法进行比较,得到了以下结论：上海股票市场收益率具有尖峰后尾的特点,并且具有明显的GARCH效应。收益率序列基于GED分布下的VaR以及CVaR的计算结果要好于基于正态分布以及t分布的计算结果。这是由于正态分布的尾部较薄,随着置信水平的增加,基于正态分布的风险度量容易低估风险。而t分布的尾部太厚,会造成高估风险的后果。CVaR可以在VaR失效时比较准确的度量出极端损失,相对于VaR而言,CVaR是一种能覆盖更大范围尾部风险的风险测度指标。当金融资产收益率相关性不显著时,VaR及CVaR同时满足次可加性,但是CVaR使得次可加性的效果更为明显,可以更好的体现出风险分散化效应。本文的创新之处是将基于GARCH族模型下的CVaR以及VaR模型应用到我国金融市场的风险度量中(本文利用上证指数),辅以定性分析,对我国金融市场风险进行度量和研究。
[Abstract]:Risk refers to the uncertainty of future income, and financial risk is the uncertainty of future income of portfolio caused by the change of financial variables. Usually we focus on the possible loss of risk, so we can express the concept of risk as "the possibility of loss due to uncertainty of the result". With the economic globalization and the integration of financial markets, financial markets have become more and more complex. How to manage financial risk effectively is the most important concern of domestic and foreign financial theorists and physical circles. Financial market risk measurement is to measure the loss of financial assets caused by the change of market factors. Financial risk measurement is the core part of risk management. At present, the main methods of financial market risk measurement are: mean-variance analysis, sensitivity method, volatility method and VaR method, pressure test and extreme value theory. The results show that the time series of financial asset return has the characteristics of peak and tail, and it is not obedient to normal distribution. In order to estimate financial risk correctly, this paper selects the daily closing price data of Shanghai Stock Exchange Index from January 2, 2004 to December 31, 2013 by using the GARCH family model. The calculation methods of CVaR and VAR based on family model for GARCH under normal distribution and GED distribution are compared and compared with the traditional VaR method. The following conclusions are obtained: the yield of Shanghai stock market has the characteristics of peak and tail and has obvious GARCH effect. The results of VaR and Cvar based on GED distribution are better than those based on normal distribution and t distribution. This is because the tail of the normal distribution is thin. With the increase of the confidence level, the risk measurement based on the normal distribution is easy to underestimate the risk. The tail of t distribution is too thick, which will result in overestimated risk. CVaR can accurately measure extreme loss when VaR fails. Compared with VaR, CVaR is a kind of risk measure index which can cover a larger range of tail risk. When the correlation of financial asset returns is not significant, VaR and Cvar satisfy the subadditivity simultaneously, but Cvar makes the effect of sub-additivity more obvious, which can better reflect the risk diversification effect. The innovation of this paper is to apply CVaR and VaR model based on GARCH family model to the risk measurement of financial market in our country (using Shanghai Stock Exchange Index) and to measure and study the risk of financial market in our country with qualitative analysis.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.5

【参考文献】