Modelos Bayesianos Univariados Aplicados à Previsão de Séries Econômicas

Abstract

This article reports the results of an exercise in which dynamic models are "fitted" for some Brazilian macroeconomics time series, useful for the current economic analysis. Its forecasting performance is compared with classical structural and transfer function models using trading days as regressor. The Bayesian forecasting methodology is briefly presented including an alternative way to estimate the parameters of an autoregressive processo