Forecasting with estimated dynamic stochastic general equilibrium models: The role of nonlinearities

In this paper we study the e®ects of nonlinearities on the forecast- ing performance of a dynamic stochastic general equilibrium model. We compute ¯rst and second-order approximations to a New Keyne- sian monetary model, and use arti¯cial data to estimate the model's structural parameters based on its linear and quadratic solution. We and that, although our model in not far from being linear, the fore- casting performance improves by capturing the second-order terms in the solution. Our ¯ndings suggest that accounting for nonlinearities will improve the predictive abilities of DSGE models in many appli- cations.