The ESTAR Model: Nonlinearities, Price Indices, Crises and Forecasts

Due to arbitrage, real exchange rates might revert to their equilibrium value in a nonlinear fashion. Modelling such behavior could enhance our understanding and forecasting of exchange rate movements. In this report, the Exponential Smooth Transition Autoregressive model is used on monthly data on four currency pairs using Nonlinear Least Squares in an attempt to model nonlinear exchange rate behavior. The report also focuses on the role of trade price indices, the in uence of the recent nancial crisis and the forecast performance of the model. Limited evidence of nonlinearity is found when comparing the test statistics with their simulated counterparts. Furthermore, no evidence of the impact of the crisis is found and using the trade price index only adds value over the traditional price index when forecasting the exchange rate.