基于GARCH-MIDAS模型的国际原油价格波动影响因素研究

发布时间：2018-03-26 10:46

本文选题：原油价格波动　切入点：GARCH-MIDAS模型　出处：《中国矿业大学》2017年硕士论文

【摘要】：随着经济的发展、全球化的深入,国际石油市场更加国际化、金融化,石油作为经济发展的动能来源,其价格波动愈加受到广泛关注。同时,我国作为发展中大国,石油的对外依存度却在不断增加。在这一背景下,研究油价长期影响因素和各影响因素对原油价格波动的影响特点具有重要的现实意义。本文阐述国际原油市场的主要概况、行业结构及国际油价的体系,并从理论角度上探讨了能够影响国际油价波动的因素,包括供求、库存、美元汇率、投机因素及替代能源,同时,也探讨了地缘政治、突发事件、天气等不可度量的因素对油价波动的影响。在此基础上,构建了广义自回归条件异方差混频数据抽样模型(GARCH-MIDAS),从水平层面和波动层面分别研究各因素对原油价格波动造成的影响。通过实证研究发现:(1)长期来看,供求因素是原油价格波动的基本影响因素;库存和美元汇率大致在相同的程度上对油价波动造成影响;投机因素对油价波动的影响也是显著的,但作用的程度远小于供求、库存和美元汇率;另外,作为一定程度上可以替代石油的另外两大化石能源——煤炭和天然气,仅天然气的价格可以对油价波动产生影响,但这种影响相对来说很小。(2)在对各因素影响油价波动的水平层面和波动层面分析中发现,水平层面上,各因素对油价波动的影响方向是不同的。其中,原油需求、替代能源中的天然气和投机因素这三方面对原油价格波动起到助推的作用;而美元汇率、原油供给和原油库存水平值的增长则会稳定油价波动;煤炭价格的水平值则起不到显著作用。波动层面上,除了替代能源价格的波动率不会对油价波动产生显著影响,其他各因素的波动率均会推动原油价格的波动。(3)在使用经典GARCH-MIDAS模型进行实证的同时,也考虑尝试加入其他低频外生变量和高频外生变量,构建多因素混频模型,观察各因素组合一起之后对原油价格波动的影响情况,结果表明,各因素对原油价格波动的影响情况与单因素模型的情况基本一致,更加印证了供求因素、美元汇率和库存作为影响原油价格波动的重要地位。另外,本文观察GARCH-MIDAS模型在估计多因素模型参数的效果,发现添加外生变量有利于提高GARCH-MIDAS模型的刻画能力、预测估计效果;同时,引入高频外生变量有助于刻画的长期成分更好地识别原油价格的波动。根据以上研究结论,本文就我国实际情况,提出以下政策建议,主要包括:走节约型的发展道路,从需求角度提高我国的“能源安全”性;加强国际间的能源交流与合作,拓展海外石油业务市场;完善石油战略储备促进能源安全;建立石油外汇储备;完善我国的石油交易市场,发展石油期货市场;积极发展新能源,减少对石油的依赖,同时减少对环境的污染。
[Abstract]:With the development of economy, the deepening of globalization, the international oil market more international, financial, oil as a source of economic development momentum, its price fluctuations more attention. At the same time, China is a developing country, the dependence on foreign oil has been increasing. In this background, study on the long-term impact of oil prices the influence factors and influence characteristics has important practical significance to oil price fluctuations. This paper describes the main status of the international crude oil market, the industry structure and the international oil price system, and discusses the factors to influence, the international oil price volatility from a theoretical point of view including supply and demand, inventory, the dollar, speculative factors and alternative energy. At the same time, also discusses the geopolitical factors, unexpected events, weather is not the measure of oil price fluctuation. On this basis, construct the generalized autoregressive conditional heteroskedasticity mixing The data sampling model (GARCH-MIDAS), study the influence of various factors on the crude oil price fluctuations resulting from the level and fluctuation level respectively. The empirical study found that: (1) long term, supply and demand factors are the basic factors affecting the oil price fluctuation; inventory and the dollar exchange rate at roughly the same level of oil price volatility impact effect; speculative factors of oil price fluctuation is significant, but the effect is far less than the supply and demand, inventory and the dollar exchange rate; in addition, as to a certain extent, can replace petroleum two other fossil fuels - coal and natural gas, only the price of natural gas can affect the fluctuations in oil prices, but the impact is relatively small. (2) found in the level of oil price fluctuation analysis of the impact of various factors and the fluctuation level, level and direction of various factors on the oil price fluctuation is different. Among them, the original Demand for oil, natural gas and alternative energy speculation in these three aspects to boost the role of oil price fluctuations; while the dollar exchange rate, crude oil supply and oil inventory level of growth will stabilize oil prices fluctuation; coal price level is not significant. The wave level, in addition to alternative energy the price volatility will not have a significant impact on oil price fluctuation, the other factors will promote the volatility of crude oil price fluctuations. (3) the use of the classical GARCH-MIDAS model for empirical at the same time, also consider trying to join the other low frequency and high frequency variable exogenous variables, constructs a multi factor mixing model, observing the various factors together after the impact of oil price fluctuations, results show that the influencing factors of oil price fluctuation and the single factor model is basically consistent, more proof of the supply and demand factors, the dollar Exchange rate and stock status as an important impact of oil price fluctuation. In addition, we observe the GARCH-MIDAS model in the estimation of model parameters the effect of multiple factors, found that adding exogenous variables can improve GARCH-MIDAS model description ability, estimate the effect; at the same time, the introduction of high frequency exogenous variables contribute to a better characterization of long-term component identification of crude oil prices fluctuation. According to the above conclusions, this paper on China's actual situation, put forward the following policy recommendations, including: the path of development of economy, improve our country's "energy security" from the perspective of demand; energy strengthen international exchanges and cooperation, expand overseas oil business market; improve the strategic petroleum reserve to promote energy security the establishment of foreign exchange reserves; oil; improve the oil trading market in our country, the development of the oil futures market; actively developing new energy, reduce dependence on oil, at the same time Reduce the pollution of the environment.

【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F764.1;F713.35

【参考文献】