Modelling East Asian exchange rates: a Markov-switching approach

This paper compares the ability of nonlinear and standard linear models to capture the dynamics of foreign exchanges rates in the presence of structural breaks. The analysis is conducted for three East Asian countries, namely Indonesia, South Korea and Thailand. It is shown that a Markov regime-switching model with shifts in the mean and variance (rather than a STAR model) is well suited to capture the nonlinearities in exchange rates. Such a model is found to outperform a random walk specification in terms of both in-sample fitting and out-of-sample forecasting. In order to evaluate competing forecasts, accuracy measures based on both the forecast errors and the variance forecast are used.