本研究選取2012年1月2日至2014年3月31日期間內之境外人民幣NDF及香港離岸人民幣(CNH)即期匯率每日報價資料,運用Enders and Granger(1998)門檻自我迴歸模型(TAR)及動差門檻自我迴歸模型(M-TAR)進行門檻共整合檢定,並進一步利用Enders and Granger(1998)及Enders and Siklos(2001)門檻誤差修正模型(TECM),分析解釋人民幣NDF與香港離岸人民幣(CNH)之間的長短期非對稱因果關係。 實證結果發現,CNH和一個月期人民幣NDF以及CNH和三個月期人民幣之間彼此不具短期因果關係,不存在「領先-落後」之因果互動關係,唯有十二個月期人民幣NDF對CNH有正向的短期因果關係,亦即十二個月期人民幣NDF顯著領先CNH;而在CNH和人民幣NDF之長期因果關係方面,在偏離度較小而低於門檻值時, CNH無論與一個月期、三個月期或十二個月期人民幣NDF,變數之間均有正向的因果關係,也就是CNH與NDF之間具有雙向的長期因果關係,兩者互為因果,而在偏離度較大而高於門檻值時,僅十二個月期NDF對CNH具有正向的長期因果關係,一個月期及三個月期人民幣NDF則否。 This study begins by using Enders and Granger (1998) threshold autoregressive model (TAR) and the momentum-threshold autoregressive model(M-TAR), then further using Enders and Granger (1998) and Enders and Siklos(2001) threshold error-correction model(TECM) to explore the the short-term and long-term asymmetric causal relationship between RMB NDF and CNH. We obtained daily closing rates from January 2, 2012 to March 31, 2014 as the data sets. Empirical results show that, between CNH and 1-month RMB NDF, and CNH and 3-month RMB NDF there are no short-term causal relationship, and there are no "lead-lag" causal relationship. Only 12-month RMB NDF on CNH has positive short-term causal relationship, and it denote that 12-month RMB NDF leader CNH. In addition, the long-term causal relationship between CNH and RMB NDF, when the degree of deviation is less than the threshold value, there are positive causal relationship between CNH and various period RMB NDF variables. That means there is a two-way long-term causal relationship between CNH and RMB NDF. The two variables are cause-and-effect each other. While the degree of deviation is higher than the threshold value, only 12-month RMB NDF for CNH has a positive long-term causal relationship, 1-month and 3-month RMB NDF does not.