石油市场的内外部联系、价格发现与风险管理研究

【作者】 陈磊；

【导师】 曾勇； 杜化宇；

【作者基本信息】 电子科技大学 ， 管理科学与工程， 2012， 博士

【摘要】 近年来，石油市场迅猛发展，吸引了大批商业机构和金融机构参与。石油期货市场与现货市场、石油市场与关联市场的联系愈加紧密。次贷危机、能源市场事件等严重冲击了石油市场，造成油价剧烈波动。这些发展态势给石油市场内外部联系、价格发现、风险管理的研究带来新的挑战。而现有研究未能契合石油市场发展动态，存在角度单一、方法局限、结论矛盾等不足。本文立足石油市场发展动态，关注次贷危机、能源市场事件的影响，发展并改进多种计量经济模型与方法，从多角度分析石油市场内外部联系的新特征、石油期货市场价格发现功能的变化与机理、石油市场风险值预测与分位数建模、石油期货对冲比率的模型选择等问题。主要研究内容与结论如下：首先，考虑次贷危机、能源市场事件的影响，本文分别从价格、收益率、分位数等层面，采用随机协整检验、非线性协整检验、异方差识别法、风险的Granger因果关系检验等方法考察石油期货价格与现货价格的协整关系、石油价格与天然气价格的协整关系、油价变动对股票市场的影响、石油市场与美元市场的极端风险溢出效应等。研究表明，次贷危机期间石油期货价格与现货价格间存在随机协整关系；石油价格与天然气价格间存在结构变化协整关系和机制转换协整关系；油价变动与股价变动间存在交互作用；次贷危机加强了市场间的风险传导，石油市场与美元市场间存在双向极端风险溢出效应。其次，针对现有研究关于石油期货市场价格发现功能的不同观点，本文提出了改进的机制转换协整检验方法，采用该方法并结合永久短暂模型分析石油期货市场价格发现功能的变化与机理。研究表明，石油期货市场的价格发现功能在油价平稳（低波动）时较强，在油价波动（高波动）时较弱；石油期货市场的价格发现功能与市场参与者的交易行为有关，投机交易导致价格发现功能弱而套利交易和套期保值交易导致价格发现功能强。再次，针对CAViaR模型在估计方法和模型形式上的不足，本文提出了贝叶斯CAViaR模型和门限CAViaR模型，并采用这两种新模型分析了石油现货市场的油价VaR以及石油期货市场的收益率的分位数特征。研究表明，贝叶斯CAViaR模型在参数估计和模型检验上具有优势，门限CAViaR模型在刻画分位数动态变化模式方面具有优势；石油现货市场上油价VaR存在自回归特征，并受油价涨跌的不对称影响，且油价下跌的作用更强；石油期货市场上收益率的左尾分位数受油价涨跌影响，而右尾分位数仅受油价下跌影响。最后，为克服石油期货对冲比率模型选择的困难，考虑石油期货收益与现货收益间存在交互作用以及次贷危机可能造成油价特征变化并影响模型的预测能力，本文采用异方差识别法分析OLS对冲比率的估计偏误，发展结构BEKK模型并估计石油期货对冲比率，并从交互作用和模型风险的角度考察石油期货对冲比率的模型选择问题。研究表明，现货收益对期货收益的反馈效应导致OLS对冲比率的估计偏误；石油期货市场与现货市场间存在收益的交互作用和波动溢出效应，可采用结构BEKK模型刻画；石油期货对冲比率的估计应综合考虑油价走势以及次贷危机的影响，在OLS模型和结构BEKK模型间做出选择。更多还原

【Abstract】 With the rapid development of international oil markets in recent years,commercial organizations and financial institutions have become major investors in oilmarkets. The relationships between oil spot and futures markets as well as between oilmarkets and linked markets are much stronger. Besides, due to subprime crisis andenergy market events, oil price always fluctuates dramatically. All these trends bringnew challenges to research on the internal and external relations, price discovery andrisk management in oil markets. However, existing researches ignore such trends. Inaddition, these researches often adopt simple methods and obtain inconsistentconclusions from single perspectives.Based on the oil market trends and influences of subprime crisis and energy marketevents, this dissertation improves and develops econometric methods. This dissertationalso analyzes the internal and external relations in oil markets, price discovery functionof oil futures market and its change, VaR forecasting, quantile modeling and modelselection of oil futures hedge ratio from several perspectives. The conclusions are asfollows.Firstly, using stochastic cointegration test, nonlinear cointegration test, the methodof Identification through Heteroskedasticity and Granger causality in risk, thisdissertation examines the cointegration relationships between oil spot and futures pricesas well as between oil price and natural gas price, the influence of oil price fluctuationson stock market, the extreme risk spillover effect between oil market and US dollarmarket from price, return and quantile. It is shown that, there is a stochasticcointegration between oil spot and futures prices during subprime crisis, and regimeshifts cointegration and regime switching cointegration between oil price and naturalgas price. The interaction exists between oil price change and stock price change.Subprime crisis strengthens risk contagion, and causes bidirectional extreme riskspillover effect between oil market and US dollar market.Secondly, according to different opinions of price discovery function in oil futuresmarket, this dissertation proposes an improved regime switching cointegration test. With this method and Permanent-Transitory model, this dissertation analyzes the changeof price discovery function in oil futures market and its reason. It is shown that, theprice discovery function of oil futures market is strong when the volatility of oil price islow and weak when the volatility of oil price is high. Price discovery function of oilfutures market is related to trading behavior of market participants. Speculation tradingwill weaken price discovery function, while arbitrage trading and hedge trading willstrengthen price discovery function.Thirdly, considering the shortages of CAViaR model in estimating method andmodel specification, this dissertation develops Bayesian CAViaR model and ThresholdCAViaR model. Based on these models, this dissertation analyzes oil price VaR in oilspot market and quantile characteristics of return in oil futures market. It is shown that,Bayesian CAViaR model is easier to estimate parameter and test model specification.And Threshold CAViaR model is better to describe the quantile dynamics. In oil spotmarket, VaR has autoregressive effect and is affected by oil price fluctuation, and theinfluences of oil price’s reducing are stronger than oil price’s increasing. In oil futuresmarket, the left tail quantile of return is affected by oil price fluctuation, but the righttail quantile is only affected by oil price’s reducing.Finally, existing researches can not provide the effective model selection strategyof oil futures hedge ratio estimation. Considering the interaction of oil spot and futuresreturns and the change of model forecasting ability affected by subprime crisis, thisdissertation analyzes the estimation bias of OLS hedge ratio and develops structuralBEKK model. And this dissertation discusses the model selection of oil futures hedgeratio estimation. It is shown that, due to feedback effect of spot return to futures return,the estimation bias of OLS hedge ratio exists. The return interaction and volatilityspillover effect between oil spot and futures markets can be described by structuralBEKK model. The estimation of oil futures hedge ratio should consider oil pricetendency and the influence of subprime crisis, and select the proper model between OLSmodel and structural BEKK model.更多还原