基于garch-evt的中国原油价格风险研究

发布时间：2018-04-20 01:33

本文选题：原油 + 价格风险　；参考：《华侨大学》2012年硕士论文

【摘要】：本文从原油的金融属性、油价风险的产生和控制，以及风险的定量分析等方面入手，对中国原油价格风险进行了研究。在我国的各大油田之中，大庆油田是产油量最大的，国际上也将大庆原油价格作为中国原油价格的代表，因此本文选取大庆原油价格作为中国原油价格的代表进行实证研究，从而为我国的原油价格风险管理提供一定的参考作用。在险价值（VaR）是当今风险量化的主要方法，是应用最为广泛的风险度量工具。该方法具有简洁、明了的特点。但是，传统的VaR计算方法一般都要对金融收益序列的分布做出假设，某种程度上降低了模型的可信度，而且广泛采用的正态分布假设也不能完全反映实际金融收益的尖峰厚尾特征。作为关注风险管理的人，他们最需要得到的信息又恰恰是关于较高分位数的尾部。因此，本文利用GARCH模型对于大庆原油收益率序列建模，描述其随时间的波动性；选用极值理论中的POT模型对于标准残差序列建模，描述其尾部特征。两者结合起来可以计算出原油收益率序列的VaR值。通过后验测试与其它传统度量VaR的模型比对，得出以下结论：第一，GARCH-EVT模型在95%置信水平下与GARCH-normal出错率一致，并且小于GARCH-t模型和GARCH-GED模型；第二，，97.5%和99%置信水平下，GARCH-EVT模型出错率小于GARCH-normal模型、 GARCH-t模型和GARCH-GED模型，显示出在极端事件发生时的优势；第三，尽管落后于GARCH-EVT模型，但GARCH-GED模型在97.5%和99%置信水平下出错率小于GARCH-normal模型和GARCH-t模型。最后进行了Kupiec似然比率检验，结果显示了模型的有效性和准确性。总体说来，GARCH-EVT模型在度量中国原油价格风险时更精确，出错率更小，更适合作为中国原油市场的风险管理的工具。
[Abstract]:In this paper, the financial property of crude oil, the production and control of oil price risk, and the quantitative analysis of oil price risk are studied in this paper. Among the major oil fields in China, Daqing oil field is the largest oil production, and the international Daqing crude oil price is also regarded as the representative of China crude oil price. Therefore, this paper selects Daqing crude oil price as the representative of China crude oil price to carry on the empirical research. So as to provide a certain reference for China's crude oil price risk management. Value-at-risk (VaR) is the main method of risk quantification and the most widely used risk measurement tool. The method is simple and clear. However, the traditional VaR calculation methods generally assume the distribution of the financial return series, which reduces the credibility of the model to some extent. Moreover, the widely used hypothesis of normal distribution can not completely reflect the peak and thick tail characteristics of actual financial returns. As people who care about risk management, what they need most is the end of the higher quartile. Therefore, the GARCH model is used to model Daqing crude oil yield series, which describes the volatility with time, and the POT model in extreme value theory is used to model the standard residual series to describe the tail characteristics. Combined, the VaR value of crude oil yield series can be calculated. By comparing the posteriori test with other traditional VaR models, the following conclusions are drawn: first, the GARCH-EVT model is consistent with the GARCH-normal error rate at 95% confidence level and is smaller than the GARCH-t model and GARCH-GED model; The error rate of GARCH-EVT model is lower than that of GARCH-normal model, GARCH-t model and GARCH-GED model at the second, 97.5% and 99% confidence levels. Third, although it lags behind the GARCH-EVT model, the GARCH-EVT model shows its advantages in the event of extreme events. But the error rate of GARCH-GED model is smaller than that of GARCH-normal model and GARCH-t model at 97.5% and 99% confidence level. Finally, the Kupiec likelihood ratio test is carried out, and the results show the validity and accuracy of the model. Generally speaking, GARCH-EVT model is more accurate and less error rate in measuring the risk of Chinese crude oil price, so it is more suitable as a tool for risk management in Chinese crude oil market.
【学位授予单位】：华侨大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F426.22;F764.1;F832

【参考文献】