| 摘要: | 遠期匯率不偏性假說(Forward Rate Unbiased Hypothesis, FRUH)在國際金融研究領域中，被視為是一個極欲解決的謎題。以往文獻顯示：傳統即期匯率報酬與遠期溢酬的迴歸實證結果皆拒絕此假說。然而由於遠期溢酬的一階自我相關係數均非常接近1，顯示遠期溢酬具有高持續性的特徵，故近年來文獻將研究重點擺在有限樣本估計的準確性。事實上傳統 統計量在有限樣本會產生嚴重的型I誤差扭曲，已被視為一個可能拒絕此假說的原因。為了更深入探討此議題，本篇研究利用適當的統計推論重新驗證遠期匯率不偏性假說。Campbell and Yogo (2006) 提出的Bonferroni Q檢定中，將解釋變數的高持續性特徵，以Local-to-unity方法模型化，並以及Bonferroni方法估計出參數的信賴區間，故本論文將其應用於重新檢驗遠期匯率不偏性假說。 本篇文章以澳幣、加拿大幣、歐元、日元、紐西蘭幣、英鎊、瑞士法郎等七種貨幣兌換美元匯率為研究對象，探討各國即期匯率與一個月期遠期匯率的關係。本文實證結果發現：遠期溢酬自我迴歸係數的信賴區間，除了加拿大介於[0.925,0.999]之間，其餘均包含1，而殘差相關係數的估計值並不大，顯示內生性問題並不嚴重，以致於傳統即期匯率報酬與遠期溢酬迴歸在10%信賴水準之下，仍然可以拒絕遠期匯率不偏性假說。 Conventional spot return and forward premium regressions have long been known to provide a strong rejection of the Forward Rate Unbiased Hypothesis which is a puzzle in the international financial field. However, due to the highly persistent in the forward premium, which shows estimated autoregressive roots close statistically indistinguishable from one, recent literature has cast doubt on the finite sample accuracy of these tests. In fact, finite sample size distortion, which could be from conventional statistics, has now come to be considered as one of several possible explanations behind the forward premium puzzle. In particular, this paper revisit the unbiasedness hypothesis using more appropriate inference method which is called Bonferroni Q-test proposed by Campbell and Yogo (2006). The paper uses exchange rate of AUD, CAD, ECU, JPY, NZD, GBD, and SWF, which both relative to USD, to discuss the relationship between spot rate and one-month forward rate. The empirical results show that confidence interval of every forward premium’s autoregressive root include one except for CAD. In addition, the estimated correlation between the innovations to return and predictor variable is too low to cause serious problem of endogeneity which can reject the Forward Rate Unbiased Hypothesis in 10% significance level. |
