Further evidence on forecasting international GNP growth rates using unobserved components transfer function models

  1. García Ferrer, Antonio
  2. Hoyo Bernat, Juan Luis
  3. Novales Cinca, Alfonso
  4. Young, Peter C.
Documentos de Trabajo (ICAE)

ISSN: 2341-2356

Year of publication: 1993

Issue: 12

Pages: 1-21

Type: Working paper

More publications in: Documentos de Trabajo (ICAE)


Forecast of international GNP growth rates are computed using a novel, onobserved components model that allows for estimating the trend and the perturbational components in GNPdata. The model is formulated in state space terms, and estimating using recursive methods of filtering and fixed interval smoothing, The decomposition crucially hinges on the choice of the Noise-Variance Ratio parameter. As any other signal extraction method, the choice of the relevants parameters affects the statistical characteristics of the estimated components. Here, we incororate a priori beliefs on the values of the NVR parameter leading to a decomposition with reasonable business cycle properties. Throughout the paper, forecast comparisons are made with other Bayesian and non-Bayesian alternatives.