본 연구는 국내 금융시장의 개별기업에 대한 정보를 투자자 심리지수의 대용변수로 사용하는 투자심리지수를 조사한다. 개별기업주식의 일별 거래정보와 해당종목에 대한 개인투자자의 매도매수정보를 바탕으로 생성된 투자심리지수와 개별기업의 주가와의 관계를 살펴보고, 기업 고유특성에 따라 투자자 심리지수가 어떠한 영향을 미치는지 살펴보았다. 2000년부터 2015년까지 KOSPI 유가증권시장에 상장된 제조업을 대상으로 분석한 결과, Fama-French의 3요인 변수에 모멘텀 요인변수를 추가한 Carhart의 4요인 위험변수를 통제하고도 투자자 심리는 주가의 수익률을 설명하는 유의한 변수임을 보였다. 투자자 심리지수는 규모가 작고, 주가가 낮을수록, 장부가치 대 시장가치비율이 높을수록, 초과수익률이 높을수록, 과거수익률의 변동성이 큰 기업에 더 큰 영향을 미치는 것으로 나타나 기업 특성별로 투자심리가 다르게 영향을 미치는 것으로 나타났다. 또한 투자자 심리와 수익률간의 유의한 관계가 투자자 거래비중에 영향을 받음을 확인하였다. 즉, 개인투자자가 선호하고 실제 거래비중이 높은 기업일수록, 기관이 주를 이루는 외국인투자자의 주식 보유비중이 낮은 기업일수록 투자심리에 큰 영향을 받는 것으로 나타났다. 이는 기관투자자에 비해 개인투자자가 상대적으로 정보열위에 있고 비합리성과 심리편의를 더 가짐을 간접적으로 뒷받침한다.
This study suggests an investment sentiment index that exploits daily information on individual firm characteristics and individual investors’ trading behavior in the Korean stock market. We empirically examine the explanatory power of our sentiment indicator for the cross-sectional returns of individual stocks and portfolios after controlling for appropriate market risk factors. While previous studies use a single variable or multiple market-wide variables to measure investor sentiment, we efficiently measure sentiment by extracting common factors from various individual firm characteristic variables and trades by domestic individual investors, who are normally regarded as noisy and behaviorally biased. By extending methodologies suggested by Ryu, Kim, and Yang (2017), Yang, Ryu, and Ryu (2017), and Yang and Zhou (2015, 2016), we construct a composite sentiment indicator based on principal component analysis using five key variables: the relative strength index, psychological line index, adjusted turnover rate, logarithm of trading volume, and individual investor buy-sell imbalance. We analyze how the sentiment effect varies depending on firm and stock characteristics by constructing several portfolio groups classified according to firm size, stock price, book-to-market ratio, excess return, volatility of past returns, individual trading ratio, and foreign investors’ holding for individual companies. We categorize the portfolios into five quintile groups based on each criterion and generate an investor sentiment index for each portfolio group. Our sample data comprise a daily stock trading dataset for all available manufacturing companies listed on the KOSPI stock market from 2000 to 2015. This sample mitigates possible industry effects and biases and enables us to investigate the uncontaminated results for sentiment effects by maintaining homogeneity among the sample firms. To ensure consistency and clarity, we exclude companies facing trading suspension and/or administrative issues. We extend the literature on investor sentiment and contribute to research by constructing a composite sentiment indicator that includes information on various stock and firm characteristics. We use buy-sell order imbalance information on individual investors, as their investment strategies and trading patterns are likely to be affected by psychology, sentiment, and mood. Our sentiment indicator exhibits more robust explanatory power than existing sentiment measures do, as it successfully explains the cross-sectional asset returns after controlling for the four risk factors. Our empirical analyses using this sentiment indicator provide several important findings and implications. We find that our investor sentiment indicator may explain cross-sectional stock and portfolio returns, even after controlling for Fama and French factors (market factor, size factor, and book-to-market factor) and the additional momentum factor suggested by Carhart (1997). It is particularly interesting that the sentiment indicator maintains its explanatory powers after considering and controlling for the momentum effect, consisting of representative time-series stock price patterns driven by investor psychology and behavioral biases. This result indicates that the sentiment indicator may be an important factor in explaining asset price movements. The analyses considering firm and stock characteristics show that the sentiment effect varies significantly according to stock and firm characteristics, being more prominent for smaller firms as well as stocks that are lower priced, have higher book-to-market ratios, have greater excess returns, and are more volatile. The sentiment effect also increases for stocks exhibiting greater individual investor participations, while its effect is lower when stocks are mostly owned by foreign investors, who are mostly professional institutional investors. These results tend to suggest that domestic individual investors are uninformed and noisy traders, while foreign institutional investors are better informed and more sophisticated. Considering that our sentiment indicator captures firm and stock characteristics reflecting individual stock trading on a daily basis, our methodology can be easily applied to analyze sentiment issues in various industry sectors at a relatively high-frequency level. The sentiment indicator can also be used to examine the transmission and spillover effect of market sentiment and behavior across financial markets and countries.