中国股市风格溢价双长记忆性研究

Research on the Double Long Memory Properties of Style Premium in Chinese Stock Market

利用中国股市风格资产在日、周、月和季频率上的数据(2000-2013),本文引入偏t分布作为新息分布来描述中国股市规模溢价和价值溢价分布存在的"尖峰厚尾"特征,根据最小化信息准则构建ARFIMA-YHGARCH-skt模型,以刻画中国股市风格溢价序列收益过程和波动过程的双长记忆性。研究表明中国股市并不存在显著的规模溢价,只存在显著的价值溢价:在收益过程方面,规模溢价具有收益长记忆性,但并不显著;而价值溢价在日、月和季度频率上的序列具有显著的收益长记忆性;在波动过程方面,规模溢价和价值溢价均在日、周频率上的序列具有显著的波动长记忆性。

Using different frequency data （such as day, week, month, quarter） from 2000 to 2013 in Chinese stock market, the paper introduces partial t distribution as a new income distribution to describe the ＂fat tail＂ feature in size and value premium distribution in the Chinese stock market. And it buihs ARFIMA - HYGARCH - skt model according to minimized information criterion to capture double long memory properties of return process and volatility process of size premium and value premium in Chinese stock market. The results show that： there is a significant value premium in Chinese stock market, but not in size premium. The size premium return process has a long memory, but is not significant. However, the value premium based on the day, month and quarter frequency sequence has significant long memory in return process. Size premium and value premium based on day and week frequency sequence have significant long memory properties in volatility process.