Abstract:

The contagion of investor sentiment among industries will accelerate the spread of financial risks in the stock market. Therefore, with the rapid spread of information today, it is of great significance to study the contagion effects of investor sentiment among industries. This paper first extracts the low-frequency part (long-term trend) and high-frequency part (short-term fluctuations) of industry investor sentiment based on the VMD-WA model, and then establishes a VAR model for the impulse response analysis of the low-frequency part of the inter-industry sentiment to test the industry investor sentiment, the mean spillover effect of the industry. The DCC-GARCH model is established for the high-frequency part to analyze the volatility spillover effect of industry sentiment. The research results show that for the industry sentiment trend item, when the sentiment of an industry changes positively, the sentiment of the infected industry is always positively contagious first, and then reverses to negative contagion after a period of time. Only the sentiment of the pharmaceutical industry affects the information. Industry sentiment has no significant contagious effect. For industry sentiment fluctuation items, inter-industry emotional contagion presents different relationships in different periods, and the change trend of emotional contagion between different types of industries is different. Based on the above content, this paper proposes policy recommendations that should focus on the risk contagion of upstream and downstream industries in the industrial chain and should do a good job in risk control at the time of policy announcement.

Key words: investor sentiment contagion, multiple time scales, variational mode decomposition, industry

摘要：

研究不同行业间在不同时间尺度下的投资者情绪传染效应。首先基于VMD-WA模型提取行业投资者情绪的长期趋势项（低频）与短期波动项（高频），然后对行业间情绪的长期趋势项进行脉冲响应分析，以检验行业投资者情绪间的均值溢出效应；对短期波动项建立DCC-GARCH模型，以分析行业情绪间的波动溢出效应。研究结果表明：对行业情绪趋势项而言，当一行业情绪正向变化时，总是对被传染行业的情绪先产生正向传染，一段时间后逆转为负向传染，只有医药行业情绪对信息行业情绪无显著传染效应；对行业情绪波动项而言，行业间情绪传染性在不同时期呈现不同的大小关系，不同类型行业间的情绪传染性的变化趋势不同，发生显著改变时总是伴随着重大事件。

关键词: 投资者情绪传染, 多时间尺度, 变分模态分解, 行业