This work investigates whether and to what extent the instability of the structural parameters affects the forecasting performance of DSGE models. In particular, it examines whether updating the parameter estimates every time new data are released improves GDP and inflation forecasts over a short and medium horizon. Moreover, the work analyses whether the forecasting performance of the models improves in the presence of financial frictions.
The results, which are based on quarterly real-time data for the United States, show that the forecasting performance of DSGE models improves if the parameter estimates are repeated whenever new data become available. The forecasting accuracy, both in terms of point and density forecasts, is greater for the models featuring financial frictions.
Published in 2019 in: Journal of Macroeconomics, v. 61.
