中国期货市场日内效应分析  

Intraday effects analysis of Chinese futures markets

作　　者：刘向丽[1] 程刚[2] 成思危[3] 汪寿阳[3] 洪永淼 

机构地区：[1]北京信息科技大学理学院,北京100085 [2]中国科学院研究生院数学科学学院,北京100190 [3]中国科学院研究生院管理学院,北京100191 [4]美国康乃尔大学经济学系与统计科学系,纽约14850

出　　处：《系统工程理论与实践》2008年第8期63-80,共18页Systems Engineering-Theory & Practice

基　　金：北京市教委科技创新平台基金;国家自然科学基金委员会创新研究群体基金(70221001)

摘　　要：运用1分钟高频数据对我国三个市场、六个品种的商品期货的收益率和交易量的日内变动模式进行研究,得出了日内绝对收益率及交易量的＂L＂型变化模式.这跟证券市场的＂U＂型日内特征不同,我们根据金融市场微观结构理论、交易机制及交易者心理给予解释.在此基础上,利用Granger因果关系检验和向量自回归模型（VAR）,研究了影响收益波动性的各种因素.结果表明绝对收益率与交易量、持仓量之间两两存在双向Granger因果关系,这是与股市的只存在由交易量到绝对收益率的单向Granger因果关系不一样的结论,原因在于期货市场的做空机制.通过对VAR模型进行方差分解和脉冲响应分析,实证分析了三者之间的动态关系及影响程度.结论表明：当以绝对价格波动作为被解释变量时,其自身的滞后项可以解释90%左右的残差扰动,交易量可以解释10%左右的的残差扰动.当以交易量作为被解释变量时,其自身的滞后项可以解释80%左右的残差扰动,绝对价格波动可以解释20%左右的残差扰动.当以持仓量作为被解释变量时,其自身的滞后阶数解释了45%-70%的残差扰动,交易量解释了25%-45%的残差扰动,绝对价格波动解释了5%-10%.各方程变量解释基本都稳定在20-30分钟后.实证结果还表明持仓量对绝对价格波动和交易量有微弱的影响,而绝对价格波动与交易量有较强的互动影响,并且就此给投资者以相关建议.This paper adopts minutely high-frequency data of yields, volumes and open interests of the six cormnodities futures in three exchanges of China to probe into their intraday change patterns, and discovers the ‘L＇ pattern intraday movement of absolute yield and volume. We exert financial market microstructure theory, traders＇ psychology and trading mechanism to explain the different intraday phenomenon between the futures market and the stock market, which has a distinctive‘U＇ pattern. Basing on this, by the help of Granger causality test theories and Vector Autoregressive models （VAR）, we study the factors that influence the volatility of yields and the lagged orders. The results show that there are two-way Granger causality among each two of the absolute yield, volmne and open interest, which is different from the empirical result of the stock market that there is a one-way Granger causality from volmnc to abolute yield, and the difference comes from the short mechanism of futures market. Through the decomposition of variance of VAR models and the impulse response analysis, we empirically analyze the dynamic relationship among the three factors. The conclusion shows that： when we treat the volatility of absolute yield as an explained variable, the residual disturbance can be explained about 90% by its lagged terms, and volume can explain about 10%. When volume is interpreted as an explained variable, about 80% of the residual disturbance can beexplained by its lagged tetras and the volatility of the absolute price explains about 20%. When open interest is considered as an explained variable, lagged tetras of its own interprets about 45% to 70% of the residual dis, the volume explains 2.5% - 45%, and the volatility of absolute price explains 5% to 10%. All the explanatory variables reach stable about 20 - 30 minutes later. Empirical result tells the weak impact from open interest to volatility of absolute yield and volume and the strong correlation between volatility of absolute yield and volume.

关 键 词：高频数据 日内特征 市场微观结构 GRANGER因果关系 向量自回归 

分 类 号：F830.9[经济管理—金融学]