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...