Scandinavian Working Papers in Economics
HomeAboutSeriesSubject/JEL codesAdvanced Search
The Economic Research Institute, Stockholm School of Economics SSE/EFI Working Paper Series in Economics and Finance

No 561:
Linear models, smooth transition autoregressions, and neural networks for forecasting macroeconomic time series: A re-examination

Timo Teräsvirta (), Dick van Dijk () and Marcelo Medeiros ()

Abstract: In this paper we examine the forecast accuracy of linear autoregressive, smooth transition autoregressive (STAR), and neural network (NN) time series models for 47 monthly macroeconomic variables of the G7 economies. Unlike previous studies that typically consider multiple but fixed model specifications, we use a single but dynamic specification for each model class. The point forecast results indicate that the STAR model generally outperforms linear autoregressive models. It also improves upon several fixed STAR models, demonstrating that careful specification of nonlinear time series models is of crucial importance. The results for neural network models are mixed in the sense that at long forecast horizons, an NN model obtained using Bayesian regularization produces more accurate forecasts than a corresponding model specified using the specific-to-general approach. Reasons for this outcome are discussed.

Keywords: forecast combination; forecast evaluation; neural network model; nonlinear modelling; nonlinear forecasting; (follow links to similar papers)

JEL-Codes: C22; C53; (follow links to similar papers)

36 pages, July 14, 2004, Revised November 9, 2004

The paper will appear with Discussion by Professor Alfonso Novales and a reply by the authors.

Download Statistics

This paper is published as:
Teräsvirta, Timo, Dick van Dijk and Marcelo Medeiros, (2005), 'Linear models, smooth transition autoregressions, and neural networks for forecasting macroeconomic time series: A re-examination', International Journal of Forecasting, Vol. 21, pages 755-774

Questions (including download problems) about the papers in this series should be directed to Helena Lundin ()
Report other problems with accessing this service to Sune Karlsson () or Helena Lundin ().

Programing by
Design by Joachim Ekebom

Handle: RePEc:hhs:hastef:0561 This page was generated on 2014-12-14 19:23:05