This work examines the nonlinear behavior of exchange rates as predicted by target zone models that allow for lack of credibility in the band. Testable implications related to episodic nonlinearity are explored as the expected nonlinear pattern depends on the position of the exchange rate inside the fluctuation band. The analysis is performed for EMS countries over the period of 1987 to 1993.; In this study a new methodology to test for episodic nonlinearity applying Lagrange Multiplier linearity tests is proposed. In the Moving Ending Observation (MEO) analysis the tests are conducted recursively to wider windows of data. At the same time, the Moving Initial Observation (MIO) analysis conducts the same tests recursively by dropping observations thus considering narrower windows. The MEO-MIO approach allows multiple bipartitions of the data set therefore avoiding contamination problems of the results of the linearity tests present in similar studies. Furthermore, in contrast to previous works that required specification of fundamentals, this study tests the model considering a univariate specification of the exchange rate at a daily frequency. In addition, the testing procedure does not impose any ad-hoc nonlinear specification of the exchange rate equation.; A Monte Carlo study to inspect the size and power of the linearity tests was conducted for different sample sizes and nonlinear alternatives. Based on the results from these simulations, it was determined which of those tests would be most appropriate to check the implications of nonlinear behavior in first and second moment for different positions of the exchange rate inside the band. Overall, this study has found no support for the nonlinear behavior of exchange rates as predicted by target zone models.
展开▼