【作者】 陈秋雨；

【导师】 周生春；

【作者基本信息】 浙江大学 ， 政治经济学， 2011， 博士

【摘要】 2008年1月9日,作为股指期货探路先锋的黄金期货合约正式挂牌上市,这是中国第一个真正意义上的金融期货品种,上市之初就强烈的吸引了投资者的目光,但仅仅几个月,大部分投资者就出现了亏损并绝望离场,市场流动性极差。为了改善这种状况,2008年3月,银监会正式批准商业银行进入期货市场从事期货交易。然而金融海啸席卷全球后,包括商业银行在内的投资者对金融衍生品谈虎色变,流动性依然没有得到改善。2010年7月,中国人民银行等六部门联合发文要求大力发展黄金市场,提高金融市场的竞争力和应对危机的能力。文件强调了要深入研究黄金市场的发展规律,切实做好黄金期货风险管理。因此研究黄金期货的市场特征及风险控制有着非常重要的理论和现实意义。然而黄金期货在学术研究领域并没得到学者们太多的垂青,相关的学术研究相当少,国内黄金期货上市时间很短,定量方面的研究更是凤毛麟角,且普遍存在样本量小、模型简单、没有运用经济学和金融学理论进行深入分析等缺点。本研究利用金融时间序列方法对中国黄金期货首次构建了完整的计量体系,结束了黄金期货研究零散、局部、模型简单的局面。本研究对其他期货品种有很强的示范作用,同时还有助于提高期货从业人员和投资者的分析手段和风险控制能力。本研究主体分为三部分：首先,运用了单位根、自相关和非线性独立同分布检验方法对黄金期货市场的鞅式弱有效进行了检验,结果无法判断其是否弱式有效,笔者认为应采取分形理论方法进行深层次检验,并将其区分为均值方程和波动率方程分别进行。其次,为了探讨影响黄金期货市场分形特征的外部因素,笔者使用了协整方法、误差修正模型、格兰杰因果检验、脉冲响应函数等方法分析与中国黄金期货市场关系密切的因素之间的协整关系,这些因素包括有黄金现货、美元指数、国内股票、纽约金、伦敦金和黄金租赁。最后,由于黄金期货市场为分形市场,因此基于有效市场理论的风险控制方法无效,本研究利用极值方法和期望不足理论对其风险进行研究,专注于处理尾部的巨额损失,并构建了黄金期货风险控制体系。本研究的结论：(1)就市场本身而言,黄金期货市场并非有效市场,而是一个具有分形特征的市场,且均值方程和波动率方程均存在长记忆特征。(2)从影响市场的外部因素来看,黄金现货引领黄金期货价格,因此黄金期货市场并未发挥价格发现功能；美元指数、上证指数、深证成指和黄金租赁与中国黄金期货均不存在长期均衡关系；纽约金单向引导中国黄金期货；伦敦金和中国黄金期货存在双向引领关系。这再次验证了黄金期货市场的分形特征。(3)市场的分形特征决定了传统VaR风险控制方法无效,基于极值的BMM方法和POT方法可作为普通风险值,而期望不足可作为谨慎风险值,同时还应考虑重现水平风险值,对黄金期货实行三限管制为最科学的风险控制体系。本研究的创新点主要有以下几点：一是将分形分析方法半参数模型SEMIFAR运用到中国黄金期货。它允许序列中可能存在确定性趋势外加一个随机趋势,同时考虑长记忆和短记忆,这跟以往人们只考虑其中的一种趋势有所不同。二是首次将重现水平运用到期货风险控制领域。重现水平主要应用在水文学、气候学和股票市场上,期货领域的应用研究近乎空白。三是将期望不足ES和Mean-ES模型运用到黄金期货领域,弥补了VaR风险评估方法对超越VaR后的风险无能为力的缺陷,并首次提出将极值方法、期望不足和重现水平相结合,对黄金期货实行三限管制。四是将黄金租赁的核心指标作为中国黄金期货的主要影响因素进行定量分析。另外,研究发现美元指数、中国股市和黄金租赁均与黄金期货价格不存在长期均衡关系。更多还原

【Abstract】 January 9,2008, as pioneer of the stock index futures, gold futures contracts were formally listed on the market, It was the first real financial futures in China and drew the attention of the investors as soon as it was listed. However, several months later, most investors lost their money and left the market hopelessly. The market liquidity was extremely low. To change the situation, in March 2008, the China Banking Regulatory Commission approved commercial banks to enter the futures market dealing in futures transactions. Nevertheless, after the financial crisis swept the world, investors including commercial banks were frightened by financial derivatives. The liquidity was not improved. In July 2010, six departments including the People’s Bank of China issued a file jointly requesting to develop the gold market to improve the competitiveness of the financial market and the ability to deal with crisis. In the file, it is requested to study the development rule of the gold market in depth and control the risk of gold futures. Therefore, it is practically significant to study the gold futures market characteristics and risk control.However, gold futures do not drew much attention from scholars in the academic area. There are very few related academic researches. Because of the short listing time of gold futures in China, there are seldom quantitative researches. Moreover, they have the defects of small amount of samples, simple models and not using the economic and financial theories to make in-depth analysis. The study adopted the financial time series method to create a complete measuring system for Chinese gold futures, putting an end to the scattered, partly and simple-modeled gold futures study. The study is a good model for other futures and will help improve the analysis approaches and risk control ability of the employees in the futures company and the investors.The main body of the article consists of three parts:Firstly, it applied unit root test, autocorrelation test and non-linear independent identically distribution test to verify the martingale weak-form market efficiency of the gold futures market. The result was unable to determine if the gold futures market was weak-form efficient. The author suggested adopting fractal theory for an in-depth verification and adopting mean equation and volatility equation respectively.Secondly, to discuss the external factors affecting the fractal character of gold futures market, the author adopted cointegration analysis, Error Correction Model, Granger Causality analysis and Impulse Response Functions to analyze the cointegration relation among the factors closely related to Chinese gold futures market. These factors included gold spot, USDX, Chinese stock market, COMEX Gold Futures, London Gold and Gold Lease Market.In the end, since the gold futures market is a fractal market, the risk control method based on Efficient Market Hypothesis is inefficient. The author adopted Extreme Value method and Expected Shortfall (ES) Theory to study the risk, focused on the huge loss of tail and created a risk control system for gold futures.Conclusions of the study:(1) In terms of the market itself, the gold futures market is not an efficient market but a market having fractal character and both mean equation and volatility equation have long memory character.(2) In terms of the external factors affecting the market, gold spot lead the price of gold futures. Therefore, the gold futures market does not play the price discovery function. USDX, Shanghai Stock Exchange Composite Index, Shenzhen Stock Exchange Component Index and Gold Lease all have no long-term equilibrium relations with Chinese gold futures. COMEX Gold Futures unilateral lead Chinese gold futures. London Gold and Chinese gold futures have bidirectional Granger Causality relations. The results further verified the fractal character of the gold futures market.(3) The market fractal character determines that the traditional VaR control method is ineffective. The article argued that the most scientific risk management method was to set the VaR calculated by the Extreme-Value-based BMM method and POT mehtod as ordinary risk value and the Expected Shortfall value as prudential risk value while taking into account the Return Level risk value to realize the control over these three lines. The innovations of the article included:firstly, it used the fractal analysis method semiparametric model SEMIFAR in Chinese gold futures study area, which allowed deterministic trend plus a random trend to exist in the time series. Meanwhile, long memory and short memory were taken into account, which was different from the past practice that only one trend was taken into account. Secondly, it introduced Return Level into the futures’ risk control area. The Return Level is mainly used in Hydrology, Climatology and stock market. It was hardly applied in the area of futures. Thirdly, ES and mean-ES models were used in the area of Gold Futures, making up the defect that VaR assessment method was unable to assess the risk exceeding VaR. The author raised the idea of integrating the Extreme Value, ES and Return Level methods to control Chinese Gold Futures risk over these three lines for the first time. Fourthly, the core indicators of Gold Lease Market were set as the main influential factors on Chinese Gold Futures for a quantitative analysis. In addition, the research found out that USDX, Chinese stock market and Gold Lease Market did not have long-term equilibrium relation with the price of Chinese Gold Futures.更多还原