A Principal-Component Approach to Measuring Investor Sentiment

Haiqiang Chen,Terence Tai-Leung Chong,Xin Duan

DOI: https://doi.org/10.1080/14697680903193389

IF: 1.986

2010-01-01

Quantitative Finance

Abstract:The usefulness of the investor sentiment measure in the stock market has received increasing attention in recent years. Various measures of investor sentiment have been proposed over the last two decades. Early studies, such as Lee et al. (1991), Swaminathan (1996) and Neal and Wheatley (1998), use closed-end fund discounts as a proxy for market sentiment. Since both closed-end funds and small stocks are mainly owned by individual investors, the sentiment affects the returns of small stocks in the same way as it influences closed-end funds. Otoo (1999) studies the survey-based Michigan consumer confidence index and finds a strong positive relationship between consumer sentiment and stock prices. Kumar and Lee (2002) construct a measure of individual investor sentiment based on the buy–sell imbalance and show that it has incremental explanatory power for small stocks, value stocks and stocks with low institutional ownership. Fan and Yao (2003) propose a composite index which contains the information of various market indicators.x Brown and Cliff (2004) use a vector autoregressive framework to explain the mutual influence between the sentiment measure, market returns and volatility. It is shown that investor sentiment measures have little explanatory power for short-term stock returns. Baker and Wurgler (2006) construct a measure of stock market sentiment based on the common variation of the closed-end fund discount, NYSE share turnover, the number of IPOs, the average first-day returns on IPOs, equity share in new issues and dividend premium. They show that stocks with highly subjective valuations and