購買力平價說的實證檢定：考慮不偏估計與整合偏誤

An Empirical Test of Purchasing Power Parity under Taking Account of Unbiased Estimation and Aggregation Bias

Rogoff (1996)曾提出著名的「購買力平價說謎團」，以彰顯短期實質匯率持續性的迷思。本文嘗試利用傳統最小平方法與晚近在計量上所發展的中位數不偏估計法，對實質匯率半衰期進行推估，以瞭解「購買力平價」在受衝擊後回復的速度。此外，本文亦參考Imbs et al. (2005)與Midrigan (2007)的建議，即考慮不同商品可能會存在不同程度的可貿易性。資料則以台灣與美國間相對產業所計算出的各產業別實質匯率計17種。實證結果顯示，17種產業間的實質匯率收斂速度的確存在明顯差異，且非貿易財產業的實質匯率收斂速度顯著地較貿易財慢。此亦說明，傳統上將存在不同產業特性的價格資料，整合計算出的實質匯率並以之實證購買力平價的方法可能不恰當。整體而言，本文為「購買力平價說」無法成立的理由，提供若干適當的驗證與解釋。

The famous purchasing power parity puzzle pointed out by Rogoff (1996) emphasizes the myth for real exchange rate persistence in modern international macroeconomics. The study employs the conventional OLS and newly developed unbiased estimation to examine the real exchange rates for detecting the empirical PPP deviations. Aggregation bias is taken into account for the data classifications (see, Imbs et al. 2005; Midrigan, 2007). There are 17 industrial real exchange rates between Taiwan and the U.S. used for our estimations. Thefindings support the heterogenous assumption of real exchange rates across industries. Finally, our empirical results offer some possible reasons and explanations for PPP.