Bayesian forecasting with the structural damped trend model

In this paper we consider the structural damped trend model which is standard in arsenal of forecasting analysis. We both multiple sources error (MSOE) as well single source errors (SSOE). Relative to existing research, propose Bayesian analysis for estimation and based on Markov Chain Monte Carlo techniques and, especially, Gibbs sampler with data augmentation. empirical applications (from M3 competition from Bank International Settlements) show superior performance MOSE versus SSOE model. also document MSOE its sampling-theory counterpart. Additional evidence provided by a optimal pool approach determines weights combining predictive posterior distributions.