The Long-Run Behavior of the Real Exchange Rate: A Reconsideration
使用一种新的统计检验，证明实际汇率并非随机游走，从而否定了购买力平价偏差会无限扩大的观点。对六国数据的检验结果支持这一结论。
Using a new statistical test, this paper provides empirical evidence that the real exchange rate is not a random walk. If the real exchange rate were a random walk, deviations from purchasing power parity could be expected to become unbounded as the forecast horizon became longer. Recently, Christopher A. Sims proposed a test based on Bayesian posterior odds ratios that is designed to discriminate between a unit root and a large but stationary autocorrelation coefficient. This paper applies the Sims test to real exchange rate data for six industrial countries. The results reject the random walk hypothesis. Copyright 1992 by Ohio State University Press.