價值股與小型股在90年代的表現不如預期，顯示這些股票風格並不能帶給投資人過去文獻所顯現的報酬。近年來，有關風格擇時策略的研究開始興起，在美國、英國、日本皆發現了相當可觀的潛在報酬。此篇論文的目的，是要檢驗風格輪動策略在台灣股票市場的執行效果，以對國外的風格投資實證結果做延伸應用。首先，此篇論文探討風格輪動策略的潛在利益。接著，建立模型預測未來風格溢酬，並與消極策略比較績效結果。這裡使用的預測模型，調整自Bauer et al. (2004)所使用的動態模型方法，並增加適合度統計量的選擇條件，以確保模型估計期間內解釋變數的解釋力，最後選出在樣本外24個月中預測力最高的模型作為下一期的預測模型。實證結果顯示，風格輪動策略在台灣股票市場具有相當顯著的潛在報酬。在大型/小型輪動策略中，預測模型表現明顯比消極策略優秀，但在價值/成長輪動策略中，預測模型並沒有辦法顯著超越消極策略。而多重風格投資策略可以帶來更高的報酬，同時也涉及更高的風險。因為規模風格消極策略在樣本期間表現不佳，使得大型/小型輪動策略可以藉由預測模型打敗消極策略。然而，雖然價值/成長輪動策略的潛在利益頗大，但價值風格消極策略在樣本期間表現不俗，使得本篇論文的預測模型不易勝過消極策略。 / The disappointing performance of style consistency strategies during 1990s told us that value and small-cap stocks may not bring us the same returns as literature showed. Recently, researchers of style timing strategies have found a great potential benefit. The aim of this paper attempts to examine the execution of the style rotation strategies in Taiwan stock market and contribute to more extensive application of international style investment empirical results. First, this paper explores the potential benefits of the style rotation strategies. Then, the paper tries to predict the style premiums and compares the style rotation results to the passive strategies. Adjusting the dynamic modeling approach applied by Bauer et al. (2004), this paper adds the selection criteria of the likelihood score statistic to assure the in-sample explanatory power of 17 financial and economic variables, and chooses the forecast models with the highest out-of-sample forecasting power in the training period. The results show that the potential benefits of style rotation strategies were significant and worth researching in Taiwan stock market. The forecast models performed well in the small/large rotation strategies, but worse in the value/growth rotation strategies. The multi-style rotation strategy could provide higher return as well as involved higher risk. Because the small/large passive strategy performed poorly during the investment period, the size rotation strategy could beat the passive strategy through the forecast model. However, although the potential benefit of the value/growth rotation strategy was still large in the sample period, it was challenging to beat the passive value/growth strategy when the value/growth passive strategy performed well.

從Fama and French(1993)提出三因子模型以降，便持續有質疑的論點認為三因子仍然不夠完整且不夠有說服力，除非能夠找到更多潛藏的風險因子。這樣的批評是因為SMB、HML是依照「規模(size)」、「淨價市值比(bm)」等『特徵』分類之投資組合的報酬，而這卻和系統風險沒有足夠經濟意涵上的關聯。Hahn and Lee(2006)研究認為違約利差(default spread)、期間利差(term spread)兩種利差可以替代規模 (size)、淨價市值比 (bm)的角色去解釋風險與報酬抵換的關係，因為違約利差(default spread)、期間利差(term spread)已為眾人所知可以預測總體股票市場報酬，甚至這兩個變數已經長期被視為信用市場狀況以及貨幣政策效果的指標，也就是說這兩個變數可以捕捉市場報酬對於信用市場狀況、利率的變化。 故本文採用Hahn and Lee(2006)的研究方法，探討違約利差與期間利差是否能夠解釋SMB、HML兩個變數與景氣循環的關聯，如果可以解釋，代表利差可以進而代替SMB、HML成為資產報酬的解釋變數，甚至更富總體經濟意涵。 實證結果發現，違約利差的變動Δdefault spread對SMB顯著，兩者為正向關係；期間利差的變動Δterm spread對HML顯著，兩者為正向關係。小規模公司由於資訊不對稱的緣故，對貨幣政策以及景氣循環都有不對稱的反應，面對經濟不景氣的時候，小規模公司的信用條件惡化、暴露在違約風險之中，迴歸實證結果亦發現此時小規模公司的表現較差，故SMB下降。同樣地，在景氣不好的時候，期間利差縮小，高淨價市值比公司的股票報酬較低，故HML縮小。

此篇論文利用馬可夫狀態轉換模型實證出在歐元區的股票市場中，以規模溢酬、價值溢酬以及市場溢酬建構的投資組合存在兩個不同的情境狀態。以歐元區市場溢酬和規模溢酬建構的投資組合(SMB portfolios)在牛市存在較高的平均報酬，另一方面以價值溢酬建構的投資組合(HML portfolios)則在熊市有較高的平均報酬。而以規模溢酬、價值溢酬以及歐元區市場溢酬建構的投資組合，其報酬率變異數在熊市皆比牛市來得高。由於此篇論文實證出不論在樣本內或樣本外的測試中，以規模溢酬以及價值溢酬建構的投資組合，其特色反轉投資策略皆優於買入並持有的投資策略，因此本篇論文建議，在歐元區以規模因素（size factor）及帳面價值與市價比因素（book-to-market factor）為考量建構投資組合時，考慮規模溢酬以及價值溢酬在不同情境狀態下的反轉異常現象是重要且不可忽視的課題。 / This paper documents the presence of two regimes in the joint distribution of stock returns on European market premium portfolio and portfolios tracking size- and value effects in the Euro area. The mean returns of the EMU market portfolio and SMB portfolios are higher in the bull state while the mean return of the HML portfolio is larger in the bear state. Volatilities of the EMU market portfolio, SMB portfolio and the HML portfolio are all larger in the bear state compared to the bull state. This paper uses the Markov regime-switching model to generate the switching signal of market, size and value portfolios in the stock market and reallocates the market, size and value portfolios in the stock market by the mean-variance approach. Since both in the in-sample and out-sample test, the performance of the style rotation strategy outperforms style consistent strategy of the SMB portfolio and HML portfolio, this paper proposes that when analyzing investments in returns of size and value portfolios in the European market, it is important for us to account for anomalies for size and value effects in European market under different regimes. In the regime-switching VAR(1) model to account for the net capital flow predictability on the stock returns of EMU market, SMB and HML portfolios and the interrelationships among these variables. The result shows that adding the European Union net capital flow in relation to the economy's size as the predictor variable to the regime switching VAR(1) model, it improves the asset allocation outcomes both in the in-sample and out-sample test. Furthermore, this paper has found that both in the bull and bear states, the impulse response function shows that a shock of one standard deviation of net capital inflows last month will reduce the EMU market return up to near three months. Besides, the net capital inflow shock in European stock market will generates appreciation of companies with low book-to-market ratios (growth stocks) and large-sized firms in the bull state, while it generates appreciation of companies with high book-to-market ratios (value stocks) in the bear state.