Munich Personal RePEc Archive

Forecasting Using Functional Coefficients Autoregressive Models

Bruno, Giancarlo (2008): Forecasting Using Functional Coefficients Autoregressive Models.

[img]
Preview
PDF
MPRA_paper_42335.pdf

Download (247Kb) | Preview

Abstract

The use of linear parametric models for forecasting economic time series is widespread among practitioners, in spite of the fact that there is a large evidence of the presence of non-linearities in many of such time series. However, the empirical results stemming from the use of non-linear models are not always as good as expected. This has been sometimes associated to the difficulty in correctly specifying a non-linear parametric model. I this paper I cope with this issue by using a more general non-parametric approach, which can be used both as a preliminary tool for aiding in specifying a suitable parametric model and as an autonomous modelling strategy. The results are promising, in that the non-parametric approach achieve a good forecasting record for a considerable number of series.

UB_LMU-Logo
MPRA is a RePEc service hosted by
the Munich University Library in Germany.