如何有效掌握資產報酬的不確定性，是風險管理領域中重要的議題。我們是第一個嘗試融合總體經濟基要與投資人情緒等不確定要素融合在波動度模型中，並且應用在風險管理。本文參考Engle and Rangle (2008) 年提出的 spline-GARCH 模型，同時採用總體經濟基要與投資者情緒指標作為刻劃長期波動度的因子，以總體經濟基要掌握經濟環境概況，佐以投資者情緒等投資行為面的指標，可以更精確捕捉資產價格的長期趨勢。在金融危機期間，情緒指標可以將投資者因恐慌產生的不理性資訊帶入模型中，改善單獨使用總體基要資訊低估財務風險的狀況，更能有效地監控風險。結果顯示，同時考慮兩種要素的波動度模型可以更精準刻劃長、短期波動度的關係。我們計算總體與情緒要素個別波動度佔總波動的比例，結果顯示總體基要波動度的影響力較大，但情緒因子隨著時間變動有越顯重要的趨勢。最後，我們進一步使用此架構預測 2002 年與 2003 年的 99% 信心水準係下的風險值，結果顯示同時加入總體資訊與情緒因子的模型，長期風險管理績效更佳。 How to precisely forecast the uncertainty of asset return is one of the most important issues in finance. This thesis aims to construct a new innovative volatility model by reconciling both the macroeconomic fundamentals and investor sentimental variables from both the rational and behavioral literature, with an application to modern risk management. Fundamental and sentimental factors are employed to capture the long-run volatility of asset return via the spline-GARCH proposed by Engle and Rangel (2008). Our model is successful in allowing for behavioral biases from the emotional and sentimental behaviors among investors during the period of financial crisis. Specifically, we characterize how much both parts contributes to the total variation of price changes and found that the fundamental contributed variations dominated. However, the proportion from the behavioral factor explained variation tends to grow in the recent years. Our empirical results suggest that the information disclosed from both parts is shown to improve the performance of volatility modeling, particularly in the longer horizon, and in tail risk management.
