信息性交易概率分解与买卖价差研究

本文首先将信息性交易概率（PIN）分解为个股信息性交易概率（PINID）和市场信息性交易概率（PINM）：利用2003年7月至2004年6月高频分笔数据，计算出上证50样本股的周PINID．发现PINID在PIN中占有大部分的比重，得到样本股的平均PINID为0．1052。其次，本文运用非参数检验验证了随着成交量增加，PINID显著减小。最后，在考虑PIN和PINID下，对买卖价差的影响因素进行研究。研究表明，虽然PIN对买卖价差有着显著的正向影响，但仅其中的PINID部分才是决定因素。这也验证了PIN分解的意义。此外，波动率和流通股比例均对买卖价差有着显著的正向影响，而股票价格、成交量和换手率的影响则不显著。

The Decomposition of Probability of Informed Trading and Bid-Ask Spread

In this paper, we first decompose the probability of informed trading (PIN) into individual stock's probability (PINID) and market's probability (PINM). With our model and high-frequency data from July 2003 to June 2004, we calculate weekly PINID for 50 stocks in the SSE. We find that the PINID accounts for majority of PIN and the average PINID for all stocks is 0.1052. A no-parametric test indicates that the PINID will decline as the daily volume increases. And finally we investigate the determination of bid-ask spread by taking into account PIN and the PINID. Although the regressions indicate that PIN has significantly positive effect on the spread, PINID plays an essential role in the spread,which verifies the rationality of PIN's decomposition. At the same time, volatility and the percentage of circulating shares also have significantly positive effect on the spread while the effect of PINM, price, volume and turnover rate are not significant based on our findings.