关键词: 投资者高频情绪, 股市成交量, 分位数Granger因果关系检验, 分位数向量自回归模型, 脉冲响应分析

Abstract: According to “Research Report on Individual Investors in 2019” released by Shenzhen Stock Exchange, the structure of Chinese stock market investors is mainly retail investors. Compared with institutions, retail investors are at a relative disadvantage in terms of information acquisition, screening and analysis ability, and are more susceptible to market sentiment and irrational behaviors, which aggravate stock market turbulence. In theory, the trading behavior of investors is the basis of the capital market. The traditional financial theory takes the assumption of rational man as the premise, focuses on the fundamental information of the market, and ignores the impact of individual investor behavior on the stock market. Behavioral finance theory holds that investor psychology and other micro factors will influence the choice of investor behavior, thus affecting the stock market.
Considering that the stock market trading volume represents the active degree of capital trading in the stock market and reflects the conversion of market trading sentiment, it is an important index for studying the stock market. In addition, under different market conditions, the influence of investor sentiment of different polarity on the stock market may be heterogeneous. Therefore, based on the high-frequency perspective, this paper analyzes the heterogeneous characteristics and mechanism of investor sentiment on stock market trading volume, which is of great significance for our financial risk management and control.
This paper selects the time from January 29, 2018 to September 30, 2019 as the sample period, divides the stock market into three market states: bull market, bear market and volatile market according to the price trend, and tests the robustness of the samples taken during the COVID-19 outbreak. The trading volume of Shanghai Composite Index is taken as the research object, and the data comes from Wind database. Using crawler technology, based on the stock forum post of “East Money Information” as the information source, text analysis is used to construct the investor high-frequency sentiment index with half an hour as the unit, and the data is divided into optimistic and pessimistic emotional polarity, so as to study the intra-day effect of different polar emotions on the stock market, and analyze the characteristics and mechanism of the emotion effect in the stock market from the perspective of high frequency. Furthermore, this paper uses the quantile vector autoregressive model to study the asymmetry between variables, especially the heterogeneity of the tail extreme quantile, and uses the quantile impulse response analysis to study the heterogeneity of the impact effect of investor sentiment of different polarity on stock market trading volume in terms of impact intensity, reaction speed, response time, etc.
The results show that:(1)The influence of investors' high-frequency emotions with different polarity on stock market trading volume is obviously different, and the pulse intensity of investors' pessimism on stock market trading volume is significantly greater than that of investors' optimism, and the attenuation is relatively slow.(2)The influence of investors' high-frequency sentiment on the trading volume of the stock market at the same point varies with the change of market state.(3)In the same market state, the influence of different sub-points on trading volume also has heterogeneity, the pulse intensity of high-frequency investors on trading volume at the lower sub-point is significantly greater than that at the upper sub-point, and the pulse intensity at the middle point is the weakest.
Through the research of investors' high-frequency sentiment and stock market, it is found that the stock market under extreme conditions is more likely to be “dominated” by investors' sentiment, which is not only reflected in the intensity of impact, but also more durable in duration. The research conclusion provides new evidence for the study of stock market changes and affirms the important value of incorporating investor sentiment information into the influencing factors of stock market changes; The use of quantile regression technology can effectively capture the tail information of skewed data, and also provide decision-making reference for regulatory authorities to control extreme risk.

Key words: high-frequency investor sentiment, stock market trading volume, quantile Granger causality test, quantile vector autoregression model, impulse response