Department of Economics and Business Economics

Factor-based forecasting in the presence of outliers: Are factors better selected and estimated by the median than by the mean?

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

Standard

Factor-based forecasting in the presence of outliers : Are factors better selected and estimated by the median than by the mean? / Kristensen, Johannes Tang.

In: Studies in Nonlinear Dynamics and Econometrics (Online), Vol. 18, No. 3, 05.2014, p. 309-338.

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

Harvard

APA

CBE

MLA

Vancouver

Author

Kristensen, Johannes Tang. / Factor-based forecasting in the presence of outliers : Are factors better selected and estimated by the median than by the mean?. In: Studies in Nonlinear Dynamics and Econometrics (Online). 2014 ; Vol. 18, No. 3. pp. 309-338.

Bibtex

@article{73afe2c742514ba393d04e872cd074c6,
title = "Factor-based forecasting in the presence of outliers: Are factors better selected and estimated by the median than by the mean?",
abstract = "Macroeconomic forecasting using factor models estimated by principal components has become apopular research topic with many both theoretical and applied contributions in the literature. In this paper weattempt to address an often neglected issue in these models: The problem of outliers in the data. Most paperstake an ad-hoc approach to this problem and simply screen datasets prior to estimation and remove anomalousobservations. We investigate whether forecasting performance can be improved by using the originalunscreened dataset and replacing principal components with a robust alternative. We propose to use an estimator based on least absolute deviations (LAD) as this alternative and establish a tractable method for computing the estimator. In addition to this we demonstrate the robustness features of the estimator through a number of Monte Carlo simulation studies. Finally, we apply the estimator in a simulated real-time forecasting exercise to test its merits. We use a newly compiled dataset of US macroeconomic series spanning the period 1971:2–2012:10.Our findings suggest that the chosen treatment of outliers does affect forecasting performance and that in many cases improvements can be made using a robust estimator such as the proposed LAD estimator.",
keywords = "Factors models, Forecasting, Least absolute deviations, Principal components analysis, Robust esitmation",
author = "Kristensen, {Johannes Tang}",
note = "Campus adgang til artiklen / Campus access to the article",
year = "2014",
month = may,
doi = "10.1515/snde-2012-0049",
language = "English",
volume = "18",
pages = "309--338",
journal = "Studies in Nonlinear Dynamics and Econometrics (Online)",
issn = "1081-1826",
publisher = "de Gruyter",
number = "3",

}

RIS

TY - JOUR

T1 - Factor-based forecasting in the presence of outliers

T2 - Are factors better selected and estimated by the median than by the mean?

AU - Kristensen, Johannes Tang

N1 - Campus adgang til artiklen / Campus access to the article

PY - 2014/5

Y1 - 2014/5

N2 - Macroeconomic forecasting using factor models estimated by principal components has become apopular research topic with many both theoretical and applied contributions in the literature. In this paper weattempt to address an often neglected issue in these models: The problem of outliers in the data. Most paperstake an ad-hoc approach to this problem and simply screen datasets prior to estimation and remove anomalousobservations. We investigate whether forecasting performance can be improved by using the originalunscreened dataset and replacing principal components with a robust alternative. We propose to use an estimator based on least absolute deviations (LAD) as this alternative and establish a tractable method for computing the estimator. In addition to this we demonstrate the robustness features of the estimator through a number of Monte Carlo simulation studies. Finally, we apply the estimator in a simulated real-time forecasting exercise to test its merits. We use a newly compiled dataset of US macroeconomic series spanning the period 1971:2–2012:10.Our findings suggest that the chosen treatment of outliers does affect forecasting performance and that in many cases improvements can be made using a robust estimator such as the proposed LAD estimator.

AB - Macroeconomic forecasting using factor models estimated by principal components has become apopular research topic with many both theoretical and applied contributions in the literature. In this paper weattempt to address an often neglected issue in these models: The problem of outliers in the data. Most paperstake an ad-hoc approach to this problem and simply screen datasets prior to estimation and remove anomalousobservations. We investigate whether forecasting performance can be improved by using the originalunscreened dataset and replacing principal components with a robust alternative. We propose to use an estimator based on least absolute deviations (LAD) as this alternative and establish a tractable method for computing the estimator. In addition to this we demonstrate the robustness features of the estimator through a number of Monte Carlo simulation studies. Finally, we apply the estimator in a simulated real-time forecasting exercise to test its merits. We use a newly compiled dataset of US macroeconomic series spanning the period 1971:2–2012:10.Our findings suggest that the chosen treatment of outliers does affect forecasting performance and that in many cases improvements can be made using a robust estimator such as the proposed LAD estimator.

KW - Factors models

KW - Forecasting

KW - Least absolute deviations

KW - Principal components analysis

KW - Robust esitmation

U2 - 10.1515/snde-2012-0049

DO - 10.1515/snde-2012-0049

M3 - Journal article

VL - 18

SP - 309

EP - 338

JO - Studies in Nonlinear Dynamics and Econometrics (Online)

JF - Studies in Nonlinear Dynamics and Econometrics (Online)

SN - 1081-1826

IS - 3

ER -