中国股票市场的交易与信息——基于自回归条件持续时间标值模型的实证研究

Trading and Information in China Stock Market——An Empirical Analysis with Autoregressive Conditional Duration Marked Models

构建多元线性和半参数单指数自回归条件持续时间标值模型及其估计方法,基于分笔交易数据研究中国股票市场交易与信息之间的线性与非线性动态关系。实证结果表明：（1） 交易持续时间存在明显正自相关性、过度分散性和聚集效应;（2） 滞后收益率、成交量、买卖价差对交易持续时间有显著线性正影响,滞后波动率对交易持续时间有显著线性负影响,各滞后市场微观结构特征变量对交易持续时间的影响普遍支持Easley和O＇Hara （1992）“无交易预示着无消息＂的发现;（3） 滞后收益率、波动率、成交量和买卖价差对交易持续时间的非线性正、负影响有差异,各滞后市场微观结构特征变量对交易持续时间的影响没有一致性的结论,Diamond和Verrechia（1987）的“无交易预示着坏消息＂以及Easley和O＇Hara（1992）的“无交易预示着无消息＂的结论同时成立。

By constructing multivariate linear and semi - parametric single - index Autoregressive Conditional Duration marked model and its estimation methods, this article studies the linear and nonlinear dynamic rela- tionship between trading and information in China stock market. Evidence from the empirical analysis con- firms： （1） there are clustering effect, over-dispersed property and positive autocorrelation for transaction duration; （2） the lagged return, volume, and bid- ask spread have linear significant positive effects on transaction duration, whereas the lagged volatility has a linear significant negative effect on transaction dura- tion, which is generally supported the findings of Easley and O＇ Hara （ 1992 ） ; （3） the lagged return, vol- ume, bid- ask spread, and volatility has different nonlinear effect on transaction duration. The findings of Diamond and Verrechia （1987） and Easley and O＇ Hara （1992） are proved at the same time.