Title: Combining large numbers of density predictions with Bayesian predictive synthesis
Authors: Tony Chernis - Bank of Canada (Canada) [presenting]
Abstract: Bayesian Predictive Synthesis is a flexible method of combining density predictions. The flexibility comes from the ability to choose an arbitrary synthesis function to combine predictions. Being able to choose an arbitrary synthesis function is useful, but what is the correct choice? we study this issue when combining large numbers of predictions - a common occurrence in macroeconomics. Specifically, we examine a Canadian GDP nowcasting exercise with close to 100 models and a GDP forecasting exercise using the European Survey of Professional Forecasters with around 50 experts. Estimating combination weights with so many predictions is difficult, so we consider shrinkage priors and factor modelling techniques as ways to address this problem. These techniques provide an interesting contrast between the sparse weights implied by shrinkage priors and dense weights of factor modelling techniques. We find that the sparse weights of shrinkage priors perform well across exercises.