本文目的在於以各種方法實證台灣與美國之間的產出關係。全文分析主要根據新近發展的 多變數時間數列計量方法－－向量自我迥歸(VAR)，希望能夠藉此方法來採討台灣與美國之 間實質國民生產毛額的因果與動態關係。除VAR方法外，亦利用交叉光譜分析來測度台灣與 美國之間產出波動的相關性，更進一步利用移轉函數模型與移轉函數干擾模型，及實質商 業循環理論來檢定台灣與美國之間實質國民生產毛額的因果關係。 VAR模型顯示美國的產出干擾與台灣的產出干擾之間有密切的相關，美國的產出干擾在台灣 產出的波動中扮演重要的角色。衝擊反應分析顯示美國產出干擾對台灣的產出具有持續、 正面的效果與很強的彈性乘數效果，交叉光譜分析顯示台灣與美國之間的產出波動具有顯 著的一致性，這表示台灣與美國之間的產出波動存在密切的相關。 雖然由VAR型無法發現美國與台灣的實質產出之間存在任何的因果關係，但ARIMA模型顯示 美國與台灣的實質產出之間存在同時性的因果關係，即美國實質國民生產毛額的變動導致 台灣實質國民生產毛額立即的波動。根據實質商業循環模型，吾人無法發現台灣與美國之 產出的循環成分之間有任何顯著的相關，這隱含台灣的經濟波動乃因與美國的貿易連結而 非與美國受到共同的外在衝擊所致。在世界經濟緊相結合的今日，貿易連結與共同外在衝 擊應同時對一國的經濟波動扮演相當的角色。因此，進一步探討對美國出口在台灣經濟波 動中所扮演的角色，將有助於吾人瞭解台灣經濟波動的原因與台灣和美國之間產出變動的因果關係。This paper presents an empirical analysis of the output relationship between Taiwan and the V. S. The analysis is based on an econometric technique recently developed for multivariate time series: vector autoregression (VAR). The purpose of this study is to examine the causal and dynamic relationship of real GNP between Taiwan and the U. S. In addition to the VAR approach, cross-spectral analysis is also used to measure the correlation of output movements between Taiwan and the U. S. The transfer function model, the transfer function-noise model, and the real business cycle theory are used to do further causality tests of the real GNP relationship between Taiwan and the U. S.The VAR model shows that innovations in the U. S. output have a close correlation with innovations in Taiwan's output and they play an important role in Taiwan's output fluctuations at all time horizons. The impulse response analysis indicates that innovations in the U. S. output have persistent positive effect on Taiwan's output, and that they have a strong elasticity multiplier effect on Taiwan's output. The cross-spectral analysis show that there are significant coherences of output movements between Taiwan and the U. S. This indicates the existence of a close correlation of output fluctuations between Taiwan and the U. S.Although the VAR model does not find the existence of any Granger causality between the U. S. real GNP and Taiwan's real GNP, the ARIMA model shows the existence of Granger instantaneous cansality between the U. S. real GNP and Taiwan's real GNP. A change in the U.S. real GNP causes Taiwan's real GNP to change immediately. According to the real business cycle model, we do not find any significant correlation of the cyclical components of output between Taiwan and the U. S. This implies that Taiwan's economic fluctuations are caused by trade links with the U. S. rather than by the common outside shocks with the U. S. In a closely integrated world economy, both trade links and common outside should play a role in a country's economic fluctuations. Further investigation of the role of Taiwan's exports to the U. S. in Taiwan's economic fluctuations will help us understand the reasons of Taiwan's economic fluctuations and the causal relationship between Taiwan's and the U. S. outputs.