Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
In this paper we explore the forecasting performances of methods based on a pre-selection of monthly indicators from large panels of time series. After a preliminary data reduction step based on different shrinkage techniques, we compare the accuracy of principal components forecasts with that of parsimonious regressions in which further shrinkage is achieved using the General-To-Specific approach. In an empirical application, we show that the two competing models produce accurate current-quarter forecasts of Italian GDP and of its main demand components, outperforming naïve forecasts and comparing favorably with factor models based on all available information. A robustness check conducted on the GDP growth of the euro area and of its major members confirms these results.