台灣股、匯市與美國股市關聯性探討

DISCUSSING THE LINKAGE BETWEEN THE STOCK AND EXCHANGE MARKET OF TAIWAN AND THE STOCK MARKET OF US

本研究的目的在於檢驗台灣股、匯市與美國股市之間，是否存在外溢效果及蔓延效果。我們先利用Inclán and Tiao（1994）所建議的疊代累積平方加總運算法，先檢定出台灣股票加權指數、台幣對美元匯率及美國二大股票指數：那斯達克及 S&P500 等4個金融變數報酬的變異多重結構改變點的位置。之後，將各市場報酬變異結構改變點以虛擬變數的方式，納入GARCH 模型內以 Engle（2000）所建議之均數回復動態條件相關模型來估計任二國間動態條件相關係數，並以各市場結構改變點所劃分的區間，來建立聯合信賴區間以檢驗任一結構改變點前後之間的相關係數是否異同。實証結果顯示，台灣股市及匯市與美國股市之間存在區段性的條件相關係數顯著增加或減少的蔓延效果。另外，我們也以設定報酬間向量自我迴歸模型的方式來進行Granger 因果關係檢定及衝擊反應分析，用以探討報酬間的溢出效果。實證結果顯示，存在美國股市單向影響台灣的股市及匯市的因果關係，而台灣的股市與匯市間，則存在相互影響的雙向因果關係。

In this paper, we test whether there exist crisis contagion and spillover effects between the stock and exchange markets of Taiwan and two stock market indices of the US (the NASDAQ and S&P500). The time points of structural changes in the volatility of the return are detected first, based on the iterated cumulative sums of squares algorithm developed by Inclán and Tiao (1994). Second, this paper estimates the dynamic conditional correlation-multivariate GARCH models supported by Engle (2000). We obtain the dynamic conditional correlation coefficients using a standard deviation estimated by the GARCH model, including dummy variables instead of the breakpoints of the individual markets. Moreover, this system also constructs a simultaneous confidence interval to test for a contagion effect. Finally, this paper shows that the VAR model can be used to test the spillover effect using the Granger-causality test and impulse response function. Therefore, we find that the stock and exchange markets of Taiwan do demonstrate the effect of crisis contagion from the US to Taiwan, and that there exists bi-directional Granger causality ( feedback) between the stock and exchange markets of Taiwan.