Structural-break models under mis-specification: Implications for forecasting

Bonsoo Koo, Myung Hwan Seo

Research output: Contribution to journalArticleResearchpeer-review

Abstract

This paper revisits the least squares estimator of the linear regression with a structural break. We view the model as an approximation to the true data generating process whose exact nature is unknown but perhaps changing over time either continuously or with some jumps. This view is widely held in the forecasting literature and under this view, the time series dependence property of all the observed variables is unstable as well. We establish that the rate of convergence of the estimator to a properly defined limit is at most the cube root of TT, where TT is the sample size, which is much slower than the standard super consistent rate. We also provide an asymptotic distribution of the estimator and that of the Gaussian quasi likelihood ratio statistic for a certain class of true data generating processes. We relate our finding to current forecast combination methods and propose a new averaging scheme. Our method compares favourably with various contemporary forecasting methods in forecasting a number of macroeconomic series.
LanguageEnglish
Pages166 - 181
Number of pages16
JournalJournal of Econometrics
Volume188
Issue number1
DOIs
Publication statusPublished - 2015

Cite this

@article{ce92061016c345ba922437404399eefb,
title = "Structural-break models under mis-specification: Implications for forecasting",
abstract = "This paper revisits the least squares estimator of the linear regression with a structural break. We view the model as an approximation to the true data generating process whose exact nature is unknown but perhaps changing over time either continuously or with some jumps. This view is widely held in the forecasting literature and under this view, the time series dependence property of all the observed variables is unstable as well. We establish that the rate of convergence of the estimator to a properly defined limit is at most the cube root of TT, where TT is the sample size, which is much slower than the standard super consistent rate. We also provide an asymptotic distribution of the estimator and that of the Gaussian quasi likelihood ratio statistic for a certain class of true data generating processes. We relate our finding to current forecast combination methods and propose a new averaging scheme. Our method compares favourably with various contemporary forecasting methods in forecasting a number of macroeconomic series.",
author = "Bonsoo Koo and Seo, {Myung Hwan}",
year = "2015",
doi = "10.1016/j.jeconom.2015.03.046",
language = "English",
volume = "188",
pages = "166 -- 181",
journal = "Journal of Econometrics",
issn = "0304-4076",
publisher = "Elsevier",
number = "1",

}

Structural-break models under mis-specification: Implications for forecasting. / Koo, Bonsoo; Seo, Myung Hwan.

In: Journal of Econometrics, Vol. 188, No. 1, 2015, p. 166 - 181.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Structural-break models under mis-specification: Implications for forecasting

AU - Koo, Bonsoo

AU - Seo, Myung Hwan

PY - 2015

Y1 - 2015

N2 - This paper revisits the least squares estimator of the linear regression with a structural break. We view the model as an approximation to the true data generating process whose exact nature is unknown but perhaps changing over time either continuously or with some jumps. This view is widely held in the forecasting literature and under this view, the time series dependence property of all the observed variables is unstable as well. We establish that the rate of convergence of the estimator to a properly defined limit is at most the cube root of TT, where TT is the sample size, which is much slower than the standard super consistent rate. We also provide an asymptotic distribution of the estimator and that of the Gaussian quasi likelihood ratio statistic for a certain class of true data generating processes. We relate our finding to current forecast combination methods and propose a new averaging scheme. Our method compares favourably with various contemporary forecasting methods in forecasting a number of macroeconomic series.

AB - This paper revisits the least squares estimator of the linear regression with a structural break. We view the model as an approximation to the true data generating process whose exact nature is unknown but perhaps changing over time either continuously or with some jumps. This view is widely held in the forecasting literature and under this view, the time series dependence property of all the observed variables is unstable as well. We establish that the rate of convergence of the estimator to a properly defined limit is at most the cube root of TT, where TT is the sample size, which is much slower than the standard super consistent rate. We also provide an asymptotic distribution of the estimator and that of the Gaussian quasi likelihood ratio statistic for a certain class of true data generating processes. We relate our finding to current forecast combination methods and propose a new averaging scheme. Our method compares favourably with various contemporary forecasting methods in forecasting a number of macroeconomic series.

U2 - 10.1016/j.jeconom.2015.03.046

DO - 10.1016/j.jeconom.2015.03.046

M3 - Article

VL - 188

SP - 166

EP - 181

JO - Journal of Econometrics

T2 - Journal of Econometrics

JF - Journal of Econometrics

SN - 0304-4076

IS - 1

ER -