Identifying Permanent and Transitory Components in Latin-American Real Exchange Rates

The behaviour of the long-run real exchange rate for four Latin-American countries is investigated for the period 1957-2002. The long-run real exchange rate is derived from an unobserved component model which divides the real exchange rate into a permanent and a transitory component. The permanent component is modeled as a homoskedastic randomwalk while the transitory component is modeled as having variances which switch according to a Markov-Switching process, between low and high variance states. Following Engel and Kim (1999), we estimate this model using the Gibbssampling methodology with monthly data for Argentina, Brazil, Chile and Mexico. The analysis of results allows us to estimate diferent expected duration of the low and high-variance regimes; as well as the behavior of the permanent and transitory components and the detection of important events in the four countries related to the peaks of the high-variance state.