【作者】 王锋；

【导师】 刘传哲；

【作者基本信息】 中国矿业大学 ， 金融工程与风险管理， 2011， 博士

【摘要】 久期是重要的微观结构信号,反映了市场上交易者的决策过程和动态性变化,久期问题的研究有助于对金融市场行为和微观结构的认识。本文对我国燃料油期货市场的不同久期特征、微观影响因素以及久期的应用进行了全面综合的研究。对于燃料油期货高频数据而言,以日交易量最大原则形成的时间频率为5分钟的连续数据是最优的。在对燃料油期货市场不同久期的特征与影响因素进行分析之前,运用了多种定量评价准则对各久期的不同ACD模型分别进行了估计和对比,并从中选择出各种久期所对应的优选模型。其中用高频数据构建的价格久期、交易量久期和持仓量久期模型中LOG-WACD模型相对最优,而用超高频数据构建的交易久期和报价久期模型中LOG-EACD模型相对更好。对不同久期的特征进行研究发现:各久期均具有明显的持续性和集聚性特征;日内特征方面,价格久期、交易量久期和持仓量久期均表现为上下午的“双N”型模式,交易久期具有上午“N”、下午“倒U”型变化模式,而买方报价久期与卖方报价久期具有全天“Г”型特征。然后在各优选模型基础上加入各种微观结构变量进行实证研究,找出了各自的微观影响因素,其中绝大多数微观结构变量对对应久期有着显著的解释作用。从久期视角研究了国外期货交易对国内燃料油期货市场的影响,结果表明:新加坡交易所燃料油期货开市交易前后国内期货市场的交易久期与价格久期特征存在差异,说明新加坡期货市场对国内市场存在一定的信息溢出效应;相对而言,国内与国外同步交易时的交易久期和价格久期更短,交易密度更大、价格波动更频繁,但由于流动性效应的限制使得市场行为间的绝对差异程度并不太明显。结合久期模型和波动模型对燃料油期货市场的波动性进行了研究,结果表明:燃料油期货市场没有明显的风险溢价和非对称效应;久期、成交量变动和持仓变动对期货收益率没有显著的影响;成交量、持仓量和交易的到达率都与波动性正相关。这些结论验证了Easley和O’Hara的市场微观结构假说,说明此理论在我国期货市场具有更好的适合性。将久期模型应用于燃料油期货市场日VaR与日内VaR的测度中,并对不同类型数据和模型在日VaR与日内VaR测度中的有效性进行了检验。结果表明:基于超高频数据和久期模型的日VaR测度方法更优;结合久期模型、波动模型和Monte Carlo模拟方法的综合日内VaR测度模型相对于基于固定时间间隔的传统VaR测度方法具有更好的预测能力;利用IVAR模型对预测时间窗口内的日内VaR进行预测,发现燃料油期货市场的日内VaR具有在开盘时最大,随后在一小段时间内迅速衰减并逐渐趋于稳定的特征。构建了反映价格、成交量和持仓量共同变化的共同久期,并在此框架下通过构建二元选择模型对期货市场量价分析法的短期预测力进行了检验,结果表明:共同久期框架下的二元选择模型具有较好的短期预测能力;量价分析法中一半以上的经验法则得到了验证,说明量价分析法在短期价格预测中仍然是有效的,是值得信赖的重要技术分析方法之一。最后从燃料油期货标准合约设计、交易制度、交易模式、交易者数量和结构等五个方面提出了完善燃料油期货市场微观机制的对策与建议。该论文有图98幅,表68个,参考文献104篇。更多还原

【Abstract】 Duration is an important microstructure signal, reflects the process of makingdecision and dynamic changes. The study of duration can help us to understand thefinancial market behavior and its microstructure better. This paper made acomprehensive study on different duration characteristic, the microinfluenceelement,and duration application of our fuel oil futures market.The high frequency data of five time frequency is optimum for the fuel oilfutures market which is generated by the rule of daily max trading volume. This paperestimated and contrasted different ACD model with each duration by using diversequantitative evaluation criterions, then selected the best model of each durationaccordingly, before analyzed the characteristic and influence elements of differentduration in fuel oil futures market. The LOGWACD model is relatively superiorwhen built price duration, trading volume duration and open interest duration modelwith high frequency data, while the LOGEACD model is relatively better when builttrading volume duration and quote price duration model with ultrahigh frequencydata. This paper studied on different duration feature, it could be find that eachduration has obvious persistence, volatility clustering. In aspect of intraday character,price duration, trading volume duration and open interest duration all present“doubleN”pattern in a.m. and p.m., trade duration presents“N”pattern in a.m.,but“∩”pattern in p.m., besides quote price duration are provied With“Г”pattern all day.Then, various microstructure variations were added into on the basis of each superiormodel to do the empirical research in order to find out the microinfluenceelement accordingly, which most of the variations have obvious explanation effect toaccording duration.This paper studied the effect of foreign future trading on the domestic fuel oilfutures market from the perspective of duration. Results indicate that domestic futuresmarket behavior has a certain difference between the SGX during trading section ornot, which shows the SGX has a certain extent of information spillovers effect on thedomestic market. What’s more, synchronous trading section with the foreign markethas relatively shorter trade duration and price duration, and greater trade density andmore frequent price volatility, but the absolute difference of market behaviors isinconspicuous because of the limitation of liquidity effect. The study on the volatility of fuel oil futures market with the duration model andvolatility model shows that the market hasn’t obvious risk premium and volatilityasymmetry effect. There is no obvious effect on futures return by duration, tradingvolume and open interest volume. Positive correlation exists among trading volume,open interest volume and trading reach ratio with volatility. All these results attest tothe Easley and O’Hara’s market microstructure theory, proving that the theory haspreferable applicability in our fuel oil futures market.This paper applied the duration model to the measurement of daily VaR andintraday VaR in the fuel oil futures market, and the efficiency test was made withdifferent kinds of data and model. Results indicate that the measurement based onultrahigh frequency data and duration model better. The comprehensive intraday VaRmeasurement model based on duration model, volatility model and Monte Carlosimulation method predicts better than traditional methods with fix time interval. It isdiscovered that the intraday VaR will maximize at opening in the fuel oil futuresmarket, decreasing rapidly in a short later period, approaching stability gradually,when the IVAR model was used to measure the intraday VaR of forecasting timewindow.This paper built joint duration reflecting the joint variation among price, tradingvolume and open interest volume, and the binary choice model was built under thisframe in order to empirically test the shortterm predictability of volume and priceanalytical method. The research results indicate that the binary choice model predictsbetter in the short term under the frame of joint duration, over half of the empiricalrules have been verified, which illuminates that volume and price analytical method iseffect in shortterm price forecasting, and it is one of important and worth trustingtechnical analysis method.Finally, this paper came up with the solutions and suggestions in improving themicromechanism of fuel oil futures market in five aspects, that is the standardcontrast design, exchange mechanism, trading pattern, traders’quantity and structure,and so on.These are 98 figures, 68 tables and 104 references.更多还原