本論文研究美國股票市場風險與超額報酬間之關係，主要的風險衡量變數為傳統風險之已實現波動(Realized volatility, RV)與下方風險之風險值(Value-at-Risk, VaR)，並且比較何者風險變數較能適當捕捉與超額報酬間之抵換關係。其中風險值之估計異於一般採用Normal分配，本文所使用skew generalized t (SGT) 分配能捕捉一般金融性資產具有偏態、厚尾及高峽峰之特徵並且配合移動視窗法(rolling window)來估出風險值。另外，也考量金融海嘯期間之影響，探討風險與報酬間之關係有何變化。而樣本資料為美國股票市場2004年至2010年期間之日資料。由實證結果發現風險值之估計以SGT分配優於拔靴法，在愈嚴格的信賴水準下SGT分配愈能合理計算出風險值，而風險值與超額報酬間存在正向抵換關係，但已實現波動依天期有不同的見解，過短或太長天期之已實現波動對報酬不具解釋能力，唯有30、60及90天之已實現波動與超額報酬間呈正向相關。最後在考量金融海嘯期間之影響後，結果發現無論使用何種風險衡量變數，都難以解釋超額報酬之變化。 This paper examines the relationship between risk and excess returns in the U.S. stock market. The main risk measure variables are realized volatility (RV) of the traditional risk and Value-at-Risk (VaR) of downside risk. Moreover, comparing RV with VaR for the sake of finding a best explaining power of evaluating the risk-return tradeoff. In order to forecast VaR, we employ skewed generalized t (SGT) distribution, to capture skewness, fat-tails and leptokurtosis of financial assets, and rolling window method. Furthermore, we also investigate that whether the relationship between risk and returns changes during the period of global financial tsunami. The data period is from 2004 to 2010. Empirical results indicate that VaR of SGT distribution is superior to bootstrapping even at the strict level of confidence. Value-at-Risk has a positive and significant relationship between risk and excess returns. However, realized volatility only has a positive relationship with excess returns in 30, 60, and 90 days. Finally, we find that any risk measure variables is difficult to define the risk-return tradeoff during the period of global financial tsunami.
