摘要：In this paper we introduce a new nonlinear Markov-STAR model to capture both the markov switching and smooth transition dynamics for real exchange rates. The Markov switching part captures the e.ect of time variations of the equilibrium exchange rates, while the smooth transition part models the nonlinear adjustment to the equilibrium. We describe the model and the estimation algorithm. In an empirical application the Markov-STAR model is applied to the real exchange rates of 18 countries. In an e.ort to make sense of the switching equilibrium rates, we relate relevant macroeconomic variables, such as output gap, in.ation rate, and economic uncertainty to the smoothed probabilities through logit regressions. We find that, consistent with economic models, a deteriorating economy relative to US economy tends to significantly increase the likelihood of the real exchange rate to depreciate relative to the US Dollar for the majority of the countries under investigation. Furthermore, a higher economic uncertainty in the US tends to significantly increase the likelihood of a real exchange rate appreciation for many advanced European economies while it is exactly the opposite for some developing countries. Finally, we also find strong evidence that rising economic uncertainty tends to be associated with a higher exchange rate volatility
