基于高频数据的连续Beta、跳跃Beta与低Beta异象

Continuous and Jump Beta Based on High-frequency Data and Low Beta Anomaly

本文基于二次变差和二次幂变差的方法，将股票价格按平滑程度划分为交易日内连续部分、日内跳跃部分和非交易时间隔夜部分，采用中国A股市场2001年1月至2016年7月的高频数据，分别估计个股的日内连续、日内跳跃和隔夜跳跃等三个类型的Beta，考察不同类型Beta的风险溢价情况。经验分析发现，采用单因素和双因素排序分组时，我国A股市场存在较强的低Beta异象，低Beta异象主要是日内收益跳跃部分引起的。控制了反转异象、特质波动率异象、最大日收益效应后，低Beta异象依然是显著的。但是，控制了流动性和换手率后，Beta的系数变得不再显著，说明在一定程度上我国股票市场的低Beta异象可以被流动性异象和换手率异象解释。

Based on the method of quadratic variation and quadratic power variation, this paper divides the stock price into continuous parts, the intraday jump and the non-trading time part by the degree of smoothness. Using the Chinese A-share market high-frequency data from January 2001 to 2016 July, we estimate intraday continuous Beta, intraday jump Beta and overnight jump, and examine the different types of Beta risk premiums in the cross-section. Empirical analysis shows that there is strong low Beta anomaly in the A-share market when using single-factor and two-factor sorting. The low-Beta anomaly is mainly caused by the intraday jump Beta. The low Beta anomaly is still significant after controlling reversal, idiosyncratic volatility and maximum effects. However, after controlling of the liquidity and turnover rate, Beta coefficient becomes no longer significant, indicating that to some extent, low Beta anomaly in China＇s stock market can be explained by the liquidity and turnover rate.