Abstract
This paper develops a test for target-zone credibility that makes use of nonlinear forecastable dependences in time series. The test procedure, based on nearest-neighbour forecasting methods, is applied to nine EMS currencies. The results suggest credibility for most of the EMS countries before the 1992 crisis, credibility losses in all countries (except Belgium and the Netherlands) after such crisis, and some credibility gains after the widening of the fluctuation bands in 1993.