摘 要
在传统金融学理论中，投资者都被假定为理性的，但无论是发达国家成熟市场
还是发展中国家新兴市场，投资者并非完全理性，存在大量非理性的投资者情绪。
此外，在我国不同的行业板块之间也存在投资者的情绪差别，因此，不同于以往研
究针对整个股票市场，本文聚焦近年来受资金关注度较高、国家重点发展的半导体
板块来构建投资者情绪指数，研究其对半导体板块股指收益率的影响。
本文梳理国内外学术界关于投资者情绪定义、构建及与股票收益率之间关系的
相关文献，在此基础上选取了半导体板块月度成交量、换手率、市盈率、市净率、
波动率以及消费者信心指数这六个指标，通过两次主成分分析法构造半导体板块投
资者情绪综合指数 SENT，并通过考察 SENT与 BW指数及半导体板块股指的契合度
来验证指数构建的有效性。实证部分，本文选取半导体板块月度股指收益率作为为
因变量，SENT为自变量，构建 VAR模型。考察在剔除宏观控制变量因素影响之
后，两者具备着怎样的关系。
本文结论如下：文通过采用半导体板块代理指标构建的半导体板块的投资者情
绪综合指数 SENT在现实意义上是有效的，其与板块股指走势相关度较高，同时与
上证 BW指数在不同阶段也具有一定的相关性，皆符合市场现实情况。在 VAR模型
的估计结果中，半导体板块收益率受滞后 1期的投资者情绪影响最大。格兰杰因果
检验结果表明，在滞后 5阶的情况下，投资者情绪与板块收益率存在双向因果关
系，其中投资者情绪对板块收益率的影响更为强烈。从脉冲响应函数图来看，投资
者情绪的高涨会较为快速的作用于收益率的变化，使得其也产生较为强烈的影响，
但同时影响也迅速冷却，并且产生略微的反向抑制作用。根据方差分解，可知投资
者情绪对行业收益率的解释力度较强，而行业收益率对投资者情绪对解释力度较
弱。在得出半导体行业投资者情绪综合指数 SENT之后，基于此构建的投资策略是
有助于提高收益率的，也让本文的研究结果有一定的现实意义。
关键词：投资者情绪；半导体；主成分分析；VAR模型
I

ABSTRACT
In traditional financial theory, investors are assumed to be rational, but whether it is a
mature market in a developed country or an emerging market in a developing country,
investors are not completely rational, and there are a lot of irrational investor sentiments. In
addition, there are also differences in investor sentiment between different industry sectors
in China. Therefore, unlike previous studies on the entire stock market, this article focuses
on the semiconductor sector that has attracted a high degree of capital attention in recent
years and is developed by the country to build investors. Sentiment index, to study its impact
on the yield of the semiconductor stock index.
This article combs the domestic and foreign academic circles on the definition and
construction of investor sentiment and the relationship with stock returns. On this basis, the
semiconductor sector is selected for monthly trading volume, turnover ratio, price-earnings
ratio, price-to-book ratio, volatility and The six indicators of the consumer confidence index
are used to construct the comprehensive index of investor sentiment in the semiconductor
sector through two principal component analysis methods, and the validity of the index
construction is verified by examining the fit between SENT and the BW index and the
semiconductor sector stock index. In the empirical part, this article selects the monthly stock
index return rate of the semiconductor sector as the dependent variable and SENT as the
independent variable to construct a VAR model. Examine the relationship between the two
after excluding the influence of macro-control variables.
The conclusions of this article are as follows: The comprehensive investor sentiment
index SENT of the semiconductor sector constructed by the use of the semiconductor sector
proxy indicators is effective in a practical sense. It has a high correlation with the trend of the
sector’s stock indexes, and at the same time, it also has a strong correlation with the Shanghai
BW index at different stages. Certain correlations are in line with market reality. In the
estimation results of the VAR model, the yield of the semiconductor sector is most affected
by investor sentiment lagging one period. Granger causality test results show that, in the case
of lagging 5 steps, there is a two-way causal relationship between investor sentiment and
sector returns, and investor sentiment has a stronger impact on sector returns. Judging from
the impulse response function graph, the upsurge of investor sentiment will act on the change
of yield relatively quickly, making it also have a relatively strong impact, but at the same
time the impact will also quickly cool down and produce a slight reverse
III

inhibitory effect. According to the decomposition of variance, it can be seen that investor
sentiment has a strong interpretation of industry yield, while industry yield has a weak
interpretation of investor sentiment. After deriving the comprehensive index of investor
sentiment in the semiconductor industry, SENT, the investment strategy constructed based
on this is helpful to increase the rate of return, and also makes the research results of this
article have certain practical significance.
Key words: Investor sentiment; semiconductor; principal component analysis; VAR
model
IV
。。。以下略