Combining expert forecasts: Can anything beat the simple average?

This paper explores the gains from combining expert forecasts from the ECB Survey of Professional Forecasters (SPF). The analysis encompasses combinations based on principal components and trimmed means, performance-based weighting, and least squares estimates of optimal weights, as well as Bayesian shrinkage. For GDP growth and the unemployment rate, only few of the individual forecast combination schemes outperform the simple equally weighted average forecast in a pseudo-out-of-sample analysis, while there is stronger evidence of improvement over this benchmark for the inflation rate. Nonetheless, when we account for the effect of multiple model comparisons through White’s reality check, the results caution against any assumption that the improvements identified would persist in the future.