论文部分内容阅读
在指令不均衡与股票收益关系研究中,常常遇到两个困难:第一,不同市场环境下,前者对后者存在异质影响;第二,往往涉及大规模数据处理。为此,运用大规模数据分位数回归的方法,一方面揭示不同分位点处指令不均衡对股票收益的异质影响,细致刻画两者之间关系;另一方面适应大规模数据建模要求,得到更为可靠的结果。以上证A股和深证A股为研究对象,通过大规模数据分位数回归方法,得到了比均值回归更多有用信息。实证结果表明:第一,在高分位点处,滞后1期指令不均衡对股票收益具有正向影响且呈现上升趋势,而在低分位点却具有负向影响;第二,控制当期指令不均衡后,滞后期指令不均衡对股票收益具有负向影响,且随着分位点的增加呈现下降趋势。这些结果意味着,指令不均衡对股票收益具有一定的解释能力和预测能力。
There are often two difficulties in the study of the relationship between instruction imbalances and stock returns: first, the heterogeneity of the latter affects the latter in different market environments; and secondly, large-scale data processing is often involved. Therefore, the use of large-scale data quantile regression method, on the one hand to reveal the heterogeneous impact of instruction imbalances on stock returns at different sub-sites, to carefully describe the relationship between the two; on the other hand to adapt to large-scale data modeling Request, get more reliable results. Taking the above A shares and the A shares of Shenzhen Stock Exchange as the research objects, we get more useful information than the average regression through large-scale data quantile regression. The empirical results show that: first, at the high score point, the lagged one-period instruction imbalance has a positive effect on the stock returns and shows an upward trend, but has a negative impact on the low-quantile point; secondly, After being unbalanced, the unbalanced lag instruction has a negative impact on stock returns, and shows a downward trend with the increase of sub-sites. These results imply that unbalanced instructions have some explanatory power and predictive power on the stock returns.