為了瞭解台灣地區貨幣總計數與經濟活動（如物價、所得）間，是會受到 金融自由化、國際化之影響，故在研究體系中，納入傳統貨幣需求函數中 ，所未放入之變數－－匯率，藉以進一步了解變數間的互動。本文採用 Johansen and Juselius (1990,1992) 發展之共整合分析，以最大概似法 估計共整合向量; 並利用概似比檢定法導出共整合向量個數之檢定統計量 ，及其極限分配；再利用 Johansen 法對匯率進行「弱外生性」檢定； 最後仍利用Granger(1969,1980,1988) 所提出之Granger Causality之觀 念，來探究何種貨幣總計數對GNP 平減指數具有較佳之解釋能力。依據本 文之實證結果，在選擇作為貨幣政策中間目標之指標變數時， M1B 或M1 C 較 M2 及 M1A 為佳。 / The relationships between money aggregates, price, income, interest rate and exchange rate in Taiwan has been investigated in this paper. The Johansen procedure is adopted to estimate cointegration vectors between these variables. Further, weak ex- ogeneity of variables is also tested by way of Johansen procedu- re. Finally, the Granger causality of money and price is invest- igated fo the sake of understanding of predictibility. Main res- ults of this paper can be summarized as follows: First, the nar- row money aggregates, say M1A, M1B and M1C, are closer related to price than M2; M2 is closer related to real GNP than narrow money aggregates. Second, the inclusion of exchange rate and in- terest rate facilitates cointegration relationships between var- iables. Third, the statement "money Granger cause price" is more adequate than "price Granger cause money" in Taiwan. According to above conclusions, it is reasonable to say that M1B and M1C could be better intermediary indicators than M2 for monetary po- licies of stablizing domestic price.