Forecasting Swiss Exports using Bayesian Forecast Reconciliation

Abstract

This paper conducts an extensive forecasting study on 13,118 time series measuring Swiss goods exports, grouped hierarchically by export destination and productcategory. We apply existing state of the art methods in forecast reconciliation andintroduce a novel Bayesian reconciliation framework. This approach allows for explicitestimation of reconciliation biases, leading to several innovations: Prior judgment canbe used to assign weights to specific forecasts and the occurrence of negative reconciledforecasts can be ruled out. Overall we find strong evidence that in addition to producing coherent forecasts, reconciliation also leads to improvements in forecast accuracy.