A Comparison of Forecast Performance Between Federal Reserve Forecasts, Simple Reduced-Form Models, and a DSGE Model,’’ (unpublished

This paper considers the forecast performance of the Federal Reserve staff, five atheo-retical reduced-form models, and an estimated dynamic stochastic general equilibrium (DSGE) model, focusing on the late 1990s through 2001. Our analysis finds that the DSGE model and atheoretical models forecast real GDP growth better than the Federal Reserve staff during this period by an economically significant margin; we find smaller differences across each method in the quality of inflation forecasts over this period. These results provide some support to the notion that richly specified DSGE models such the one used in this paper belong in the forecasting toolbox of a central bank