本研究採用主成份分析，取得市場週轉率、初次上市櫃 (initial public offerings, IPO) 家數、IPO報酬、新股發行比率、上漲家數除以下跌家數、ARMS指標、創新高的股票家數除以創新低的股票家數、融資變動比率、融券變動比率與放空比率等情緒代理變數共同解釋的變異量，合成一個情緒指標，旨在探討投資者情緒對於股票報酬之影響。此外，本研究採用追蹤資料 (panel data) 分析法，將情緒指標納入迴歸模型中，檢測投資人情緒是否可解釋台灣股票報酬。本研究的實證結果發現，高投資者情緒之投資組合報酬率高於低投資者情緒之投資組合。再者，在投資者情緒高低差異下，公司特徵對於股票報酬的影響亦不相同。當期情緒指標對當期股票報酬具有顯著的正向影響；前期情緒指標對當期股票報酬具有顯著的負向影響。

This study examines how investor sentiment influences stock returns. This study employs principal component analysis to form a composite index of sentiment. The sentiment proxies include the market turnover, the number of initial public offerings, the returns of initial public offerings, the equity share in new issues, the ratio of the number of advancing issues to declining issues, the ARMS index, the ratio of the number of new highs to new lows, the percent change in margin borrowing, the percent change in short interest, and the ratio of short sales to total sales. Furthermore, the investor sentiment variable is included in the panel data regression model to examine whether investor sentiment can explain stock returns in Taiwan. The empirical results show that the portfolio returns of high investor sentiment are higher than low investor sentiment. Moreover, the effect of firm characteristics on stock returns is diverse under different investor sentiments. Finally, the current sentiment indexes have positive influence on stock returns at the current period while the lagged sentiment indexes have negative impacts on stock returns at the current period.