In studying monthly real exchange rates between the US and Britain, Canada, Germany, and Japan from 1961 to 1993, we find that the deviation of the log real exchange rate from its time-varying, long-run equilibrium value contains a statistically significant predictable component at the four-year horizon over a forecast period extending from 1985 to 1993. Fixed-effects regressions employing differentials in productivity, real interest rates, and per capita income display some predictive power but fundamentals based on simple monetary models are generally more accurate and significant.

We find nonlinear mean reverting tendencies in Southeast Asian currencies by applying the newly developed nonlinear unit-root test by Park and Shintani (2005). First, with the US dollar as the numeraire currency, we find that 63% of the real exchange rates of Southeast Asian currencies turn out to be stationary. However, with the Japanese yen as the numeraire currency, we find no evidence in favour of Purchasing Power Parity (PPP) for most currencies in Southeast Asia, except for the Korean won and Taiwanese dollar. These findings imply that Southeast Asian currencies may not form a yen-dominated Asian exchange rate system. Second, when the dollar-based real exchange rates of Southeast Asian countries are nonlinear mean reverting, we find that the mean-reverting process could be well described by the Exponential Smooth Transition Autoregressive (ESTAR) model, rather than the Double Threshold Autoregressive (DTAR) or Double Logistic Smooth Transition Autoregressive (DLSTAR) model. Our results are reinforced by impulse response function and forecasting analysis.purchasing power parity, nonlinear unit root test, Asian real exchange rates, dollar and yen,

No abstract

No abstract

No abstract