有效匯率預測模型與避險績效比較

本研究利用UIP、PPP、MF、TR_s及TR_a五種匯率預測模型，以新台幣兌美元即期匯率、遠期匯率進行避險準確率及避險成效的實證分析。實證結果發現不論在短、中或長期避險上，總體經濟變數之匯率預測模型所得到的避險正確率皆較過去文獻的避險準則為佳。然而具有較佳之避險正確率的模型不一定有較好的預測績效。推測原因在於避險正確率較佳的模型雖然能對匯率升貶的方向作較準確的判斷，卻對升貶的幅度預測不良。在上述五種匯率預測模型比較分析中，以貨幣模型(MF)及不對稱泰勒模型(TR_a)之六個月期之避險準確率最佳。這兩種匯率預測模型再搭配完全避險比率，其報酬率能夠由負轉正且波動度下降。實證發現MF及TR_a模型搭配適當避險所獲取的報酬率相同，但TR_a模型之標準差下降較多；因此若企業公司以降低避險後績效之風險，則可採用TR_a模型搭配指數加權避險。

關鍵字

匯率避險；貨幣模型；不對稱泰勒模型；準確率

並列摘要

This study uses five exchange rate models, UIP, PPP, MF, TR_s and TR_a , to examine the hedging accurate rate and performance on the TWD/USD exchange rate. The empirical results show that in any length of hedging terms, the hedging accurate rates of macro fundamental models are higher than those of hedging rules from the past researches. However, the high hedging accurate rate of the model does not imply it has good forecasting ability because it cannot forecast the magnitude of exchange rate movements precisely. Our results also show that monetary fundamental model (MF) and the asymmetric Taylor rule (TR_a) outperform the other competing models in six-month out-of-sample hedging. Comparing the hedging performance between MF and a TR models, we find that the hedging performance by the MF and a TR with exponential weighted moving average (EWMA) hedge ratio will get the same return. However, the hedging performance by the TR_a with EWMA hedge ratio will get the lower volatility.