美國與國際股市之極值相依行為：極值連鎖理論的應用

Extreme Dependence Behavior between US and International Equity Markets: Application of Extreme Linkage Theory

本文應用極值連鎖理論提出一個可同時捕捉報酬分配之極值相依與異質波動特性並降低估計偏誤的兩階段無母數估計方法，實證衡量1991年至2013年間美國股市與國際股市間極端報酬的相依行為。首先實證研究發現，在負面衝擊部分，美國與亞洲、拉丁美洲及歐洲股市間存在高度共移或者相互依存的現象，但不存在尾部相依；在正面衝擊部分，美國與歐洲大規模股市如英國、德國及法國等股市則具有尾部相依現象。其次，美國與各國股市間左、右尾部相依函數並無差異，但若考慮滾動樣本分析，則左、右尾部相依效應會隨時間變化，而且互有消長。再者從全球地理位置分析，無論是聯合報酬分配的左尾或右尾，越靠近美國的股市其尾部相依效應越大。最後，本文發現美國與各國股市之相依效應與傳染路徑會因不同的金融危機而有差異，但2007 年美國次級房貸危機所引發的金融大海嘯及接踵而至的歐洲債信危機，這一連串嚴峻的危機已經成為涵蓋市場風險、信用風險與流動性風險的綜合風險，其影響範圍不再侷限於某特定區域，而是蔓延到全球金融市場。

The study applies extreme linkage theory to measure the tail dependence behavior of returns distribution between US and international stock markets from 1991 to 2013. The proposed method is a two stage non-parametric estimation method which can efficiently measure the extreme dependence of the conditional distribution of the heteroscedastic returns series and minimize estimation bias. Firstly the empirical analysis demonstrates that when negative shocks occur there are high level market co-movements or interdependence between US and Asian, Latin American and European stock markets, but there are no tail dependence. When positive shocks occur there are tail dependence between US and European larger stock markets such as Britain, Germany and France. Secondly the left and right tail dependence function of return distribution between the US and international stock markets are not different, but rolling sample estimation finds the left and right tail dependence time-varying. Thirdly countries geographically near US have high left and right tail dependence with US. Finally we find that the impacts of financial crisis on dependence and contagion pattern vary across different financial crises. In particular, the US subprime mortgage crisis in 2007 triggered the financial tsunami and the European debt crisis, and spillover its effects on market, credit, and liquidity risks, from local financial market to the global financial markets.