Forecasting with Real Business Cycle Models

Forecasting at business cycle frequencies is traditionally done with statistically estimated econometric models. This paper takes a different approach, using a calibrated dynamic general equilibrium model in line with the real business cycle literature. First attempts by others have not proved very successful, most probably because the structure of the models was too simple. We take a simple real business cycle model, the Kydland-Prescott (1982) model economy sufficiently simplified to accommodate for the availability of state variables in the data, augmented by government expense shocks. The forecasts are then evaluated with the traditional tools of the econometric forecaster. It is found that the model has potential for making good forecasts when compared to estimated models that are equally parsimonious.