本研究以1996年3月至2017年12月於臺灣股市上市公司為樣本，探討臺灣個股股票報酬如何受到投資人情緒的影響。首先，本研究採用主成分分析建構投資人情緒指標，以市場周轉率、資券比與股利溢酬作為情緒的代理變數，為捕捉投資人不理性的部分，利用總體經濟因子分別對不同的情緒代理變數進行迴歸分析，以降低景氣循環對於情緒的影響。接著，本研究以前期投資人情緒對個股股票溢酬進行迴歸分析，結果顯示當前期投資人情緒越高時，則將有較低的股票溢酬，且因樣本期間曾經歷過金融危機，故將金融危機納入考量，探討投資人情緒於金融危機時，對於個股股票溢酬是否有加強的影響效果，實驗結果顯示其並無加強的影響效果。再者，本研究也檢驗股票溢酬於投資人情緒轉變的變化情形，實證結果顯示兩者間並無關聯，而為驗證投資人情緒的代表性，利用情緒指標與情緒變化量兩因子交叉分析下期個股股票溢酬，結果顯示交乘項對於臺灣個股股票溢酬具有解釋能力，最後，本研究將樣本公司依其公司特徵進行分類，檢驗不同公司特徵的個股股票報酬對於情緒的敏感程度，研究結果顯示股票波動性大與有發放現金股利的公司較易受情緒影響。;This study explores how investor sentiment affects the individual stock return in Taiwan. We use monthly data for the stocks listed on the Taiwan stock exchange (TSE) from March 1996 to December 2017. First, an investor sentiment index is the first principal component of three different investor sentiment proxies: the market turnover, margin buying/short selling ratio and dividend premium. Each of them is orthogonalized to the macroeconomic indicators to purge the index of sentiment that reflects business cycle conditions. Second, we regress individual stock return on prior investor sentiment. The result reveal that prior investor sentiment has negative effect on individual stock return. Plus, we take subprime mortgage crisis into consideration to examine if there are the different results. Our findings uncover that the impact of investor sentiment does not reinforce during the subprime mortgage crisis. Third, we examine how the changes in sentiment affects the subsequent individual stock return. The result reveal that the predictive power for return is insignificant. Then, we interact the investor sentiment index with the change in sentiment and investigate the impact of interaction terms on subsequent individual stock return. The evidence show that there is the explanatory power of interaction terms on return. Finally, we classify the sample with different firm characteristics and examine whether the sentiment sensitivity varies with firm characteristics. We find that dividend-paying stocks and high volatility stocks are more sensitive to sentiment.
