主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2019, Vol. 27 ›› Issue (1): 22-33.doi: 10.16381/j.cnki.issn1003-207x.2019.01.003

• 论文 • 上一篇    下一篇

中国股市投资者情绪测度指标的优选研究

刘学文   

  1. 烟台大学经济管理学院, 山东 烟台 264005
  • 收稿日期:2016-04-14 修回日期:2018-08-20 出版日期:2019-01-20 发布日期:2019-03-25
  • 通讯作者: 刘学文(1976-),男(汉族),山东烟台人,烟台大学经济管理学院副教授,硕士生导师,研究方向:公司财务、行为金融,E-mail:thanksytu@163.com. E-mail:thanksytu@163.com
  • 基金资助:

    山东省社会科学规划研究项目(18CKJJ13)

Study on the Optimization of Investor Sentiment Measure Indicators in Chinese Stock Market

LIU Xue-wen   

  1. School of Economics and Management, Yantai University, Yantai 264005, China
  • Received:2016-04-14 Revised:2018-08-20 Online:2019-01-20 Published:2019-03-25

摘要: 股市投资者情绪单项测度指标众多、良莠不齐,Baker和Wurgler基于主成分分析构建综合投资者情绪测度指数的方法虽被广泛采用,但其所用指标在选用时具有主观随意性,故需对这些指标进行优选。可构建一套开放式的指标优选方法系统即倒金字塔滤网模型,首先对海选到的指标进行初选,剔除数据不可得不连续的指标,再通过相关性分析剔除无关项、通过聚类分析剔除高度相关项,然后通过主成分分析确定情绪测度维度和指标个数,最终用变异系数法优选出各维度内的最显著指标。然后,用优选后的指标基于主成分分析构建综合情绪测度指数。实证分析发现,不同时段应优选出不同的指标来测度投资者情绪,倒金字塔滤网模型适用于任何时段的指标优选,用优选后的指标构建的优化综合投资者情绪指数OISI比用未经优选的指标构建的CICSI指数可更好地解释中国股市投资者情绪。

关键词: 投资者情绪测度指标, 优选, 相关性分析, 聚类分析, 主成分分析

Abstract: There are many indicators to measure investor sentiment in stock market.The method of constructing comprehensive investor sentiment index based on principal component analysis by Baker and Wurgler (2006) is widely used,the indicators used by them are subjective and arbitrary,so it is necessary to optimize these indicators.An open indicator optimization method system called Inverted Pyramid Filter can be constructed.First of all, primary indicators need to be selected,and the indispensable indicators are eliminated.Then,the irrelevant items are eliminated by correlation analysis,and the highly correlated items are eliminated by clustering analysis.Furthmore, the dimension of investor sentiment measurement and the sum of indicators are determined by principal component analysis.Additionally the principal component analysis is used to determine the dimensions of investor sentiment measures and the number of indicators.Finally,the most significant indicators in each dimension is selected by using the coefficient of variation method.The empirical analysis shows that the optimization of sentiment measurement indicators is a dynamic process,and different indicators should be selected to measure investor sentiment in different periods.Inverted Pyramid Filter model is suitable for any period of indicator optimization.The optimized investor sentiment index OISI constructed with the optimized index is compared with the unoptimized index CICSI,the OISI can better explain investor sentiment.This paper provides a method reference for the indicators optimization before the construction of the investor sentiment composite index,and prepares the objective index for the construction of the index,so that the index based on the optimized indicators can also be optimized,so as to better explain the overall investor sentiment in the stock market.Moreover,other research and practice related to investor sentiment can be better done.

Key words: investor sentiment measure indicators, optimization, correlation analysis, cluster analysis, principal component analysis

中图分类号: