节点文献
基于极值分布VaR模型的中国股指期货风险评估
The Application of VaR Model to Risk Management of Index Futures
【作者】 周敏;
【作者基本信息】 上海师范大学 , 应用统计(专业学位), 2013, 硕士
【摘要】 近二十多年来,世界经济的快速发展,全球一体化特别是经济全球一体化的趋势越来越明显,伴随着世界经济的快速发展,特别是以布雷顿森林体系的崩溃为标志,全球金融市场的稳定性,变得越来越脆弱,金融市场风险一天都不曾消失,如同“达摩克利斯之剑”一样悬在上空。金融风险将影响作为一个整体的世界经济格局,而不是一个单一的金融机构,它的影响,就可能会影响世界经济的稳定发展。金融风险度量已成为一个非常重要的课题,研究它具有十分重要的理论和现实意义。风险测度理论在风险管理实践过程中不断发展,VaR风险度量方法使金融风险的定量研究得到进一步的发展。我国金融市场目前尚不成熟,套期保值工具品种还较为缺乏,伴随着金融市场的快速发展,其所承担的风险也越来越大。股指期货作为基础性风险管理工具,具有价格发现、套期保值的功能,沪深300股指期货合约正式上市,标志着我国股指期货的诞生,它的推出填补了我国股市缺乏针对系统性风险的管理手段这一空白,将改变股票市场缺乏规避系统性风险工具的现状,本文用极值分布VaR模型对股指期货的数据进行了实证研究,分析VaR值对股指期货风险评价研究具有重要意义。VaR值的准确度量不仅与资产收益率的分布有关,还与资产收益率的波动性有关,准确估计资产的波动率对于VaR的度量显得十分必要。本文对VaR模型进行了深入研究,包括极值波动下的VaR风险度量及其高频数据波动的VaR风险估计。介绍了极值分布中阈值和特征参数的估算方法,同时用两种方法,即分为广义极值模型和广义Pareto模型,对风险价值VaR作了实证研究,由于正态分布与金融时间序列的分布不符,其收益率具有“尖峰厚尾”的特性,从分布来看,五分钟损失序列的分布显然偏离正态分布,它表现出一个粗大的尾巴,带有负偏态和相对较大的峰度,并呈现波动集聚性。其次,广义帕累托分布是最有用和最实用的极值理论模型的研究成果,在分析金融市场风险时,由于样本数据具有厚尾分布,厚尾分布具有显著的优势。实证研究结果表明,极值理论可以准确度量尾部分布的极端事件的风险价值。
【Abstract】 These days, financial risks exist in financial market everywhere. Once thefinancial risks happen, they often bring huge economical losses. The riskmeasurement has become a very important issue to risk management in financialmarket. It has great theoretical and practical significance to study on it. The riskmeasure theory has been improved and developed continuously in the practice.With the VaR method widely used in the financial modeling, the accuracy is verycrucial to VaR measurement. The characteristics of financial return rate are the corecontent to all financial models, the traditional VaR estimation methods generallymake some assumptions that the return obey the assumed distribution in advance,but a large number of empirical studies show that the actual rate of returndistribution is not obey assumptions to distribution. We make use of the extremevalue theory and trimmed linear moment to describe the characteristics of the returndistributions, it solves problem that moment estimation cannot describe thecharacteristics of the thick tail, and apply it to study the risk in Chinese stock market.In addition, from the perspective of the High frequency data, We also make use ofthe Cornish-Fisher method to fit the real distribution, and we apply realized volatilityto Cornish-Fisher VaR estimation.This paper applies the GEV model and GPD model separately to carry on theanalysis of real example to the risk in the futures market of index of our country andcarries on the data analysis with R and EViews software. Through the analysis of realexample, the paper draws a conclusion that the fitting result of GPD model isobviously superior to that of GEV model.
【Key words】 VaR Model; Index Futures; High Frequency Data; Extreme ValueTheory;
- 【网络出版投稿人】 上海师范大学 【网络出版年期】2014年 01期
- 【分类号】F832.51;F224
- 【下载频次】452