摘要

The study examines the half-life deviations from purchasing power parity (PPP) using both linear and nonlinear models for the sector specific real exchange rates of the Pacific Rim countries. By using a linear benchmark model, the estimated half-life deviations from PPP of about/or less than 3 years seemingly solve the PPP puzzle. After re-examining them using nonlinear forms, the PPP puzzle for most sectoral real exchange rates in the study still remains. More specifically, we find that deviations from PPP by TAR are persistent and take much more time for mean reversions especially for non-tradable sectors using logistic STAR / exponential STAR (LSTAR/ESTAR). The major reasons might be that conventional studies have failed to control the possible nonlinearity of real exchange rates, and have disregarded the possible bias of any single numeraire country. Finally, we also show the speed of mean reversion for real exchange rates by sectoral prices (tradable and non-tradable).