Forecasting German key macroeconomic variables using large dataset methods

We study the forecasting performance of three alternative large scale approaches for German key macroeconomic variables using a dataset that consists of 123 variables in quarterly frequency. These three approaches handle the dimensionality problem evoked by such a large dataset by aggregating information, yet on different levels. We consider different factor models, a large Bayesian VAR and model averaging techniques, where aggregation takes place before, during and after the estimation of the different models, respectively. We find that overall the large Bayesian VAR provides the most precise forecasts compared to the other large scale approaches and a number of small benchmark models. For some variables the large Bayesian VAR is also the only model producing unbiased forecasts at least for short horizons. While a Bayesian factor augmented VAR with a tight prior also provides quite accurate forecasts overall, the performance of the other methods depends on the variable to be forecast.