Department of Economics and Business Economics

Timo Teräsvirta

Forecasting with nonlinear time series models

Research output: ResearchWorking paper

Standard

Forecasting with nonlinear time series models. / Kock, Anders Bredahl; Teräsvirta, Timo.

Aarhus : Institut for Økonomi, Aarhus Universitet, 2010.

Research output: ResearchWorking paper

Harvard

Kock, AB & Teräsvirta, T 2010 'Forecasting with nonlinear time series models' Institut for Økonomi, Aarhus Universitet, Aarhus.

APA

Kock, A. B., & Teräsvirta, T. (2010). Forecasting with nonlinear time series models. Aarhus: Institut for Økonomi, Aarhus Universitet.

CBE

Kock AB, Teräsvirta T. 2010. Forecasting with nonlinear time series models. Aarhus: Institut for Økonomi, Aarhus Universitet.

MLA

Kock, Anders Bredahl and Timo Teräsvirta Forecasting with nonlinear time series models. Aarhus: Institut for Økonomi, Aarhus Universitet. 2010., 32 p.

Vancouver

Kock AB, Teräsvirta T. Forecasting with nonlinear time series models. Aarhus: Institut for Økonomi, Aarhus Universitet. 2010.

Author

Kock, Anders Bredahl ; Teräsvirta, Timo. / Forecasting with nonlinear time series models. Aarhus : Institut for Økonomi, Aarhus Universitet, 2010.

Bibtex

@techreport{1d32b830f91b11de9c17000ea68e967b,
title = "Forecasting with nonlinear time series models",
abstract = "In this paper, nonlinear models are restricted to mean nonlinearparametric models. Several such models popular in time series econo-metrics are presented and some of their properties discussed. This in-cludes two models based on universal approximators: the Kolmogorov-Gabor polynomial model and two versions of a simple artificial neuralnetwork model. Techniques for generating multi-period forecasts fromnonlinear models recursively are considered, and the direct (non-recursive)method for this purpose is mentioned as well. Forecasting with com-plex dynamic systems, albeit less frequently applied to economic fore-casting problems, is briefly highlighted. A number of large publishedstudies comparing macroeconomic forecasts obtained using differenttime series models are discussed, and the paper also contains a smallsimulation study comparing recursive and direct forecasts in a partic-ular case where the data-generating process is a simple artificial neuralnetwork model. Suggestions for further reading conclude the paper.",
keywords = "forecast accuracy, Kolmogorov-Gabor, nearest neighbour, neural network, nonlinear regression",
author = "Kock, {Anders Bredahl} and Timo Teräsvirta",
year = "2010",
publisher = "Institut for Økonomi, Aarhus Universitet",
type = "WorkingPaper",
institution = "Institut for Økonomi, Aarhus Universitet",

}

RIS

TY - UNPB

T1 - Forecasting with nonlinear time series models

AU - Kock,Anders Bredahl

AU - Teräsvirta,Timo

PY - 2010

Y1 - 2010

N2 - In this paper, nonlinear models are restricted to mean nonlinearparametric models. Several such models popular in time series econo-metrics are presented and some of their properties discussed. This in-cludes two models based on universal approximators: the Kolmogorov-Gabor polynomial model and two versions of a simple artificial neuralnetwork model. Techniques for generating multi-period forecasts fromnonlinear models recursively are considered, and the direct (non-recursive)method for this purpose is mentioned as well. Forecasting with com-plex dynamic systems, albeit less frequently applied to economic fore-casting problems, is briefly highlighted. A number of large publishedstudies comparing macroeconomic forecasts obtained using differenttime series models are discussed, and the paper also contains a smallsimulation study comparing recursive and direct forecasts in a partic-ular case where the data-generating process is a simple artificial neuralnetwork model. Suggestions for further reading conclude the paper.

AB - In this paper, nonlinear models are restricted to mean nonlinearparametric models. Several such models popular in time series econo-metrics are presented and some of their properties discussed. This in-cludes two models based on universal approximators: the Kolmogorov-Gabor polynomial model and two versions of a simple artificial neuralnetwork model. Techniques for generating multi-period forecasts fromnonlinear models recursively are considered, and the direct (non-recursive)method for this purpose is mentioned as well. Forecasting with com-plex dynamic systems, albeit less frequently applied to economic fore-casting problems, is briefly highlighted. A number of large publishedstudies comparing macroeconomic forecasts obtained using differenttime series models are discussed, and the paper also contains a smallsimulation study comparing recursive and direct forecasts in a partic-ular case where the data-generating process is a simple artificial neuralnetwork model. Suggestions for further reading conclude the paper.

KW - forecast accuracy, Kolmogorov-Gabor, nearest neighbour, neural network, nonlinear regression

M3 - Working paper

BT - Forecasting with nonlinear time series models

PB - Institut for Økonomi, Aarhus Universitet

ER -