此篇研究主要利用Engle(2001)提出之DCC-多變量GARCH模型探討亞洲金融危機時期，亞洲國家，包括印尼、日本、馬來西亞、菲律賓、韓國、台灣與泰國等匯率市場和股票市場彼此間的動態條件相關性。 亞洲金融危機於1997年7月泰銖大貶而掀開序幕，其影響快速的蔓延至東南亞各國。雖然傳遞至各國的速度與各國受影響的程度皆不一樣，但是最終影響了全球的經濟以及造成一些新興國家的經濟嚴重衰退。因此，亞洲金融危機使得經濟學者對各金融市場間的關係的評估不在只是著重在國際面，而也開始著重在一個國家裡各個市場的相關性。然而，針對過去一些金融危機（financial crisis）的研究多數著重在報酬率(一階動差)的探討，如匯市報酬率和股市報酬率之間的相關性，較少研究波動性(二階動差)的變化與影響。然而，在財務理論上，不論在資產評價或動態避險等，波動性扮演著很重要的角色。隨著時間的不同，各市場間的波動變化皆不一樣，進而影響市場間波動之相關性也隨著時間之不同而不同。因此，採用固定相關係數來衡量不同市場彼此間的影響似乎易造成高估或低估彼此間的相關性，尤其在波動性發生劇烈變化之時，進而影響避險效果或風險值(Value at Risk，VaR)之計算。 本論文之研究結果主要發現市場之間有資訊或波動外溢之現象。波動性之蔓延，使市場彼此間變異與波動相關性皆受到影響，尤其在亞洲金融風暴期間，市場間之波動相關性皆發生明顯正相關之變化。此外，隨著投資種類與投資市場的多樣化及規避風險重要性之提高，採用固定相關係數來衡量市場間之相關性與投資組合之風險值計算，會造成高估或低估的情況，尤其在市場波動劇烈變動時期。 Financial Crisis and Risk Spillover: An Application of the DCC Model In this paper, we apply the Dynamic Conditional Correlation Multivariate GARCH (DCC MV-GARCH) model, proposed by Engle (2001), to investigate the effects of risk-spillover between currency and equity markets during the Asian crisis. We consider seven Asian countries including ndonesia, Japan, Malaysia, Philippines, South Korea, Taiwan, and Thailand. The Asian crisis began in July 1997 with the devaluation of the Thai baht and it spread out quickly through East Asia. Although each country experienced the crisis with differing intensity and duration, eventually the global economy is affected by this crisis and causing several emerging countries experience deep recessions. Many economists have evaluted the relationships among international financial markets and also the intermarket dependencies within each country since the Asian crisis. Studies of Asian crisis mostly focus on the first moment (return), however, volatility (the second moment) plays a key role in many areas of finance, especially in asset pricing and dynamic hedging strategies. For example, volatility and the dynamic correlation among markets play a central role as the selection of investment assets and markets and the importance of dynamic hedging are increasing. The hypothesis of a constant correlation of volatility among markets is likely to be incorrect because the correlation will likely to be more volatile as market volatility fluctuates. Thus, there would be bias if we simply use constant correlation to measure the correlation between markets. Especially when the market volatility is unstable, it would also affect the dynamic hedging effect and the calculation of VaR (Value at Risk). Our results show that there are volatility spillovers among markets. The dynamic correlations among markets are positively related during the Asian crisis. We find that the assumption of a constant correlation would introduce biases in the calculation of correlation among markets and in the estimation of VaR.
