Crowdsourcing of Economic Forecast: Combination of Combinations of Individual Forecasts Using Bayesian Model Averaging

Abstract

Economic forecasts are essential in our daily lives. Accordingly, we ask the following questions: (1) Can we have an improved prediction when we additionally combine combinations of forecasts made by various institutions? (2) If we can, then what method of additional combination will be preferred? We non-linearly combine multiple linear combinations of existing forecasts to form a new forecast (“combination of combinations”), and the weights are given by Bayesian model averaging. In the case of forecasting South Korea’s real GDP growth rate, this new forecast dominates any single forecast in terms of root-mean-square prediction errors. When compared with simple linear combinations of forecasts, our method works as a “hedge” against prediction risks, avoiding the worst combination while maintaining prediction errors similar to those of the best combinations.

Keywords: Combination of combinations, Combination of forecasts, Bayesian model averaging

JEL Classification: C53, E37