【作者】 黄荣兵；

【导师】 龚朴；

【作者基本信息】 华中科技大学 ， 管理科学与工程， 2010， 博士

【摘要】 金融衍生产品市场的功能在于规避、转移和管理风险,然而由于市场中的交易是以保证金方式进行的,因而存在较大的杠杆效应,这使得衍生产品市场蕴含着巨大的风险。作为衍生产品市场风险控制核心工具的保证金,其设置的正确与否对市场是否成功具有重要的作用。对于保证金大小的设置,当前国际上采用基于组合风险的保证金设置方法,其典型代表是标准资产组合风险分析(SPAN)系统,其得到的保证金既能有效控制风险,又能提高交易者资金的利用效率,降低交易成本。而我国交易所目前仍然采用基于策略性的方法,此种方法设置的保证金使得在大部分时间内,投资者保证金被过多占用,机会成本高昂,资金的使用效率低下,保证金的收取不能很好地反映真实的市场风险,难以实现对风险的有效控制。为了对市场风险进行有效的监控,并提高市场的活跃性,我国保证金的设置需要转换到基于组合风险的动态保证金设置上。本文在当前得到广泛使用的风险计量技术的框架内,对实用性和可操作性已得到国内外金融行业普遍认同的SPAN系统进行深入研究,把当前风险管理领域内的VaR风险计量技术和Copula技术的丰硕研究成果融入到SPAN系统中,解决SPAN系统的输入参数的设置问题。考虑到金融市场的时变属性,本文把重点放在时变风险计量技术上,在对时变VaR风险计量技术和时变Copula技术在应用中的适用性进行论证之后,给出基于时变VaR方法和时变Copula技术的SPAN系统的输入参数的设置方案。另外,考虑到风险计量的目的是为了更好地进行风险管理,因此本文也对风险资本的配置方法进行了介绍。对于本研究的贡献,主要体现在以下几个方面：首先,本文改进了时变t-Copula模型中的时变相依参数的演化方程。现有国内对Copula的研究大多数停留在静态Copula上,考虑到时变Copula模型能更好地揭示金融市场中随机变量之间的动态本质,因此本文采用时变Copula模型对随机变量之间的相依性进行描述。时变Copula模型的难点在于如何确定时变相依参数的演化方程。本文在前人研究的基础上,采用包含自相关和两个变量累积概率的历史项之差的绝对值的演化方程作为时变t-Copula参数的演化方程,建立了时变t-Copula的新的演化方程,克服了前人所给时变t-Copula演化方程所存在的缺点。实证结果表明,本文提出的新的演化方程对数据的拟合程度优于前人所给的演化方程对数据的拟合程度。其次,本文解决了SPAN保证金系统输入参数设置的开放性。现有对SPAN保证金系统的研究并没有给出一套比较完整的参数设置方案及其技术实现细节。为了解决SPAN保证金系统的开放性问题,本文采用测度风险的参数、半参数和非参数时变VaR方法和描述相关结构的时变Copula技术,对SPAN系统中的主要输入参数进行设定,给出VaR-SPAN系统中主要输入参数设定方案和设置的具体步骤,解决了其在可操作性方面存在的技术障碍。实例结果表明,本文给出的SPAN系统中与期货组合有关的主要输入参数的设置方案所得的结果是准确而合理的。接下来,本文把SPAN系统应用于国内期货组合的保证金计算中,实证分析了国内期货组合保证金设置的合理性。国内期货交易所现阶段对期货组合保证金的设置采用静态的方法收取,并没有考虑到合约与合约之间存在的相关性,必然导致对风险的高估。本文通过文中给出的SPAN保证金输入参数设置方案对国内的期货组合的参数进行设置,在此基础上把SPAN系统计算流程运用于我国衍生品期货组合保证金的计算中,通过SPAN系统计算的仿真程序,采用国内期货交易所商品期货组合的实际数据,实证检验了我国以静态方式设定的组合保证金远高于通过SPAN计算流程所得的组合保证金大小。此外,除了上述三个贡献,本文也对国际上近期的风险配置理论和方法进行了介绍。风险计量的目的在于对风险进行有效管理,而对风险的有效管理则涉及到对基于风险得到的风险资本进行有效的配置。鉴于国内对风险资本配置理论与方法研究的文献的匮乏,本文对风险资本配置近年来的理论研究成果进行了梳理,并对当前最新的理论成果进行了介绍,特别对新近的基于Copula的尾部风险值的资本配置方法进行了介绍,以期为国内资本配置的研究提供参考。更多还原

【Abstract】 The function of financial derivative market is to avoid, transfer and manage risk. But the transactions in the market are carried out by ways of margin, there is a big leverage effect which makes the derivative market have enormous risk. Whether the design of margin which is the core instrument to control the derivative market risk is reasonable or not is important for the successful of the market. Internationally, there use the margin setting method which based on portfolio risk to set the size of the margin at present. The typical representative is Standard Portfolio Analysis of Risk (SPAN) system. On one hand, there set margin so that it can effectively control risk; on the other hand, it raises the efficiency of usage of traders’fund and reduce transaction costs. In China, exchanges are still use the approach that based on strategic. By this method, the margin of the investment is excessive occupied for most of the time so that it leads to high opportunity cost and inefficient use of funds. The collection of margin can’t well reflect the risk of market and it is difficult to achieve effective control of risk. In order to effectively control and monitor the market risk and increase the activity of the market, the setting of margin needs to be converted to a dynamic margin setting based on portfolio risk.This paper made a deep research on SPAN system, whose applicability and maneuverability generally recognized by financial industry at home and abroad, within the framework of risk measurement techniques which is widely used at present. The paper put the fruitful research achievements of value-at-risk techniques and copula techniques in the field of risk management into the SPAN system, and solve the problem of setting the input parameters of SPAN system. Taking into account the time-varying properties of financial market, the paper focus on the time-varying risk measurement techniques. The paper have demonstrated the applicability in the use of time-varying risk measurement techniques and time-varying copula techniques, and then gave the proposal of setting input parameters of SPAN system based on time-varying value-at-risk method and time-varying copula techniques. In addition, taking into account that the risk measurement is aimed at better risk management, we also introduced some risk capital allocation methods. The contribution of the study is mainly in the following aspect: First of all, this paper improved the evolution equation of the time-varying dependent parameters in the time-varying t-copula model. The majority of current domestic studies about copula stay in a static copula. Taking into account that time-varying copula model can better reveal the dynamic nature between random variables in the financial markets; this article uses time-varying copula model describing dependencies between the random variables. The difficulty of time-varying copula model is how to determine the evolution equation of the time-varying dependent parameters. Based on previous studies, this paper uses evolution equation which includes autocorrelation and absolute value of the difference between the historical items of two variables’cumulative probability as the evolution equation of the time-dependent t-copula’s parameters, sets up a new evolution equation of the time-varying t-copula, and overcomes the shortcomings of the time-varying t-copula’s evolution equations given by our predecessors. The empirical result shows that the proposed new evolution equation fits the data better than the evolution equations given by our predecessors.Secondly, this paper solved openness of the setting of SPAN margin system’s input parameters. Existing research on the SPAN margin system did not give a relative complete set program of parameters and its technical implementation details. To address the SPAN margin system’s openness questions, this paper adopts parameters, semi-parametric and non-parametric time-varying value-at-risk methods measuring risk and time-varying copula technique describing dependence structure to set SPAN system’s main input parameters, gives the setting program and the concrete setting procedures of the VaR-SPAN system’s main input parameters, and addresses the technical barriers existing in its operability. Example results show that results from setting program of the major input parameters in the SPAN system relative to futures combination given by this paper are accurate and reasonable.Next, this paper applied the SPAN system in the margin calculation of domestic futures portfolio and empirically analyzed the reasonability of domestic futures portfolio’s margin setting. Domestic futures exchanges at this stage use static method to collect the futures portfolio’s margin settings, and don’t take into account the correlation existing between contracts which will inevitably lead to overestimation of risk. This paper set domestic futures combinations’parameters by the setting programs of the SPAN margin’s input parameters given in the text, on this basis, used the SPAN system’s calculating processes in our domestic derivatives, futures portfolios’margin calculations, and using actual data of domestic futures exchanges’commodity futures portfolios, empirically tested the portfolios’margin set by static way is much higher than that calculated by the SPAN processes in our country through the simulation program calculated by the SPAN system.In addition to the above three contributions, this paper introduced recent theories and methods of risk allocation in the world. Risk measurement is aimed at effective risk management, which is related to effective allocation of risk capital that we get based on risk. Considering the lack of domestic literature about risk capital allocation theories and methods, this paper sorted out the theoretical study literature of risk capital allocation in recent years, and introduced the latest theoretical achievements. Especially, this paper introduced the capital allocation method based on tail value-at-risk and copula which is newly developed.更多还原