本文旨在討論新臺幣實質有效匯率指數編製過程中，包括權數與貨幣籃等兩項關鍵技術選擇的搭配議題。爲此，本文透過適當的VAR模型執行樣本外預測程序（其中並使用了平穩拔靴重抽法），檢測了32種編製方式搭配下，相應新臺幣實質有效匯率指數的預測表現。根據DM檢定結果，本文建議：(1)貨幣籃範圍方面，建議使用「核心貨幣籃」（即美、日、中、港、韓五國）或與臺灣雙邊貿易比重居前的19國「完整貨幣籃」等兩種作法；(2)若使用「雙邊貿易權數」，建議搭配「核心貨幣籃」；(3)若使用加計「第三市場出口競爭權重」的「雙邊貿易權數」，則建議搭配「完整貨幣籃」。最後，我們也對新編的新臺幣實質有效匯率指數進行若干實證應用，例如檢視央行匯率「動態穩定政策」、穩定經濟基本面目標及估算其潛在容許區間寬度等。

The purpose of this paper is to analyze the combinations of the two relevant technical choices, including the weights and currency baskets, for constructing Taiwan's real effective exchange rate (REER) index. Using the VAR model and the re-sampling method of stationary bootstrap, we execute the out-of-sample forecasting procedures for all the 32 possible kinds of Taiwan's REER indices mentioned above. Referring to the DM (Diebold and Mariano, 1995) tests and statistics, our major findings and suggestions include: (1) as for the currency basket, we can use the ”primary currency basket” of 5 countries (US, Japan, China, Hong Kong, and Korea), or the ”complete currency basket” of 19 countries (the most important 19 countries in the bilateral trade of Taiwan); (2) we should incorporate bilateral trade weight into ”primary currency basket”; (3) we should incorporate the trade weights combining the third-market exports weight, into ”complete currency basket”. Moreover, we conduct several empirical analyses, like re-examining the Central Bank's ”dynamic stabilization” policy, stabilization policy for the economic fundamentals, and estimating the potential width of the allowed interval.