臺灣股票市場的橫斷面期望報酬率

Abstract

本研究建立三個分別包含3、6及13個公司特徵因子的模型，模擬一個投資人在真實世界中如何運用當下能取得的公開資訊進行Fama-Macbeth迴歸分析，並以迴歸計算之係數預測股價的期望報酬率。研究結果顯示，本研究之模型所計算的期望報酬率能在橫斷面上顯著的解釋股價的真實報酬率。同時，期望報酬率投資策略所建構的做多及多空避險投組在常用來衡量績效的平均月報酬率及Sharpe ratio等面向上皆有非常良好的表現。在臺股市場實際應用上，即使加入交易稅及手續費的衝擊，投資組合仍有遠勝大盤的績效表現，且能透過交疊數個月的投資組合進一步縮減策略的換股率及交易成本。最後，本研究的結果在不同子樣本及迴歸期間下普遍不受太大影響，且在Fama-French三因子迴歸的檢定下有非常優異的超額報酬，顯示本研究結果的穩健性。

This study establishes three models, each containing 3, 6, and 13 firm characteristics or factors respectively, to simulate how an investor can utilize publicly available information in the real world for Fama-Macbeth regression analysis. The regression coefficients are then used to predict the expected returns of stocks. The results of the study show that the expected returns calculated by the models can significantly explain the real returns of stocks in the cross-section. Furthermore, the expected-return-sorted long-short portfolios exhibit excellent performance in various aspects, such as average monthly return and Sharpe ratio. In the actual application in the Taiwan stock market, even with the impact of taxes and transaction costs, the portfolios still outperform the market index, and the turnover rate and transaction costs of the strategy can be further reduced by overlapping. Finally, the results of this study are robust across different subsamples and regression rolling years, and exhibit excellent excess returns based on the Fama-French three-factor regression, indicating the robustness of the study findings.