Forecast performance of nonlinear error-correction models with multiple regimes

In this paper we investigate the forecast performance of nonlinear error-correction models with regime switching. In particular, we focus on threshold and Markov switching error-correction models, where adjustment towards long-run equilibrium is nonlinear and discontinuous. Our simulation study reveals that the gains from using a correctly specified nonlinear model can be considerable, especially if disequilibrium adjustment is strong and/or the magnitude of parameter changes is relatively large.