A comprehensive evaluation of macroeconomic forecasting methods

This paper contributes to the academic literature and the practice of macroeconomic forecasting. Our evaluation compares the performance of four classes of state-of-art forecasting models : Factor-Augmented Distributed Lag (FADL) Models, Mixed Data Sampling (MIDAS) Models, Bayesian Vector Autoregressive (BVAR) Models and a medium-sized Dynamic Stochastic General Equilibrium Model (DSGE). We look at these models to predict output growth and inflation with datasets from the US, UK, Euro Area, Germany, France, Italy and Japan. We evaluate the accuracy of point and density forecasts, and compare models with a large set of predictors with models that employ a medium-sized dataset. Our empirical results shed light on how the predictive ability of economic indicators for output growth and inflation changes with horizon, on the impact of dataset size on the calibration of density forecasts, and how the choice of the multivariate forecasting model depends on the forecasting horizon.