Weighted least absolute deviations estimation for ARMA models with infinite variance

Jiazhu Pan, Hui Wang, Qiwei Yao

Research output: Contribution to journalArticle

32 Citations (Scopus)

Abstract

For autoregressive moving average (ARMA) models with infinite variance innovations, quasi-likelihood-based estimators (such as Whittle estimators) suffer from complex asymptotic distributions depending on unknown tail indices. This makes statistical inference for such models difficult. In contrast, the least absolute deviations estimators (LADE) are more appealing in dealing with heavy tailed processes. In this paper, we propose a weighted least absolute deviations estimator (WLADE) for ARMA models. We show that the proposed WLADE is asymptotically normal, is unbiased, and has the standard root-n convergence rate even when the variance of innovations is infinity. This paves the way for statistical inference based on asymptotic normality for heavy-tailed ARMA processes. For relatively small samples numerical results illustrate that the WLADE with appropriate weight is more accurate than the Whittle estimator, the quasi-maximum-likelihood estimator (QMLE), and the Gauss-Newton estimator when the innovation variance is infinite and that the efficiency loss due to the use of weights in estimation is not substantial.
LanguageEnglish
Pages852-879
Number of pages27
JournalEconometric Theory
Volume23
Issue number5
DOIs
Publication statusPublished - Oct 2007

Fingerprint

innovation
normality
efficiency
Autoregressive moving average model
Deviation
Estimator
Innovation
Statistical inference

Keywords

  • autoregressive moving average (ARMA) models
  • statistical inference
  • variance
  • weights

Cite this

@article{26da22d9d4bc400080ae74e931c0bf62,
title = "Weighted least absolute deviations estimation for ARMA models with infinite variance",
abstract = "For autoregressive moving average (ARMA) models with infinite variance innovations, quasi-likelihood-based estimators (such as Whittle estimators) suffer from complex asymptotic distributions depending on unknown tail indices. This makes statistical inference for such models difficult. In contrast, the least absolute deviations estimators (LADE) are more appealing in dealing with heavy tailed processes. In this paper, we propose a weighted least absolute deviations estimator (WLADE) for ARMA models. We show that the proposed WLADE is asymptotically normal, is unbiased, and has the standard root-n convergence rate even when the variance of innovations is infinity. This paves the way for statistical inference based on asymptotic normality for heavy-tailed ARMA processes. For relatively small samples numerical results illustrate that the WLADE with appropriate weight is more accurate than the Whittle estimator, the quasi-maximum-likelihood estimator (QMLE), and the Gauss-Newton estimator when the innovation variance is infinite and that the efficiency loss due to the use of weights in estimation is not substantial.",
keywords = "autoregressive moving average (ARMA) models, statistical inference, variance, weights",
author = "Jiazhu Pan and Hui Wang and Qiwei Yao",
year = "2007",
month = "10",
doi = "10.1017/S0266466607070363",
language = "English",
volume = "23",
pages = "852--879",
journal = "Econometric Theory",
issn = "0266-4666",
number = "5",

}

Weighted least absolute deviations estimation for ARMA models with infinite variance. / Pan, Jiazhu; Wang, Hui; Yao, Qiwei.

In: Econometric Theory, Vol. 23, No. 5, 10.2007, p. 852-879.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Weighted least absolute deviations estimation for ARMA models with infinite variance

AU - Pan, Jiazhu

AU - Wang, Hui

AU - Yao, Qiwei

PY - 2007/10

Y1 - 2007/10

N2 - For autoregressive moving average (ARMA) models with infinite variance innovations, quasi-likelihood-based estimators (such as Whittle estimators) suffer from complex asymptotic distributions depending on unknown tail indices. This makes statistical inference for such models difficult. In contrast, the least absolute deviations estimators (LADE) are more appealing in dealing with heavy tailed processes. In this paper, we propose a weighted least absolute deviations estimator (WLADE) for ARMA models. We show that the proposed WLADE is asymptotically normal, is unbiased, and has the standard root-n convergence rate even when the variance of innovations is infinity. This paves the way for statistical inference based on asymptotic normality for heavy-tailed ARMA processes. For relatively small samples numerical results illustrate that the WLADE with appropriate weight is more accurate than the Whittle estimator, the quasi-maximum-likelihood estimator (QMLE), and the Gauss-Newton estimator when the innovation variance is infinite and that the efficiency loss due to the use of weights in estimation is not substantial.

AB - For autoregressive moving average (ARMA) models with infinite variance innovations, quasi-likelihood-based estimators (such as Whittle estimators) suffer from complex asymptotic distributions depending on unknown tail indices. This makes statistical inference for such models difficult. In contrast, the least absolute deviations estimators (LADE) are more appealing in dealing with heavy tailed processes. In this paper, we propose a weighted least absolute deviations estimator (WLADE) for ARMA models. We show that the proposed WLADE is asymptotically normal, is unbiased, and has the standard root-n convergence rate even when the variance of innovations is infinity. This paves the way for statistical inference based on asymptotic normality for heavy-tailed ARMA processes. For relatively small samples numerical results illustrate that the WLADE with appropriate weight is more accurate than the Whittle estimator, the quasi-maximum-likelihood estimator (QMLE), and the Gauss-Newton estimator when the innovation variance is infinite and that the efficiency loss due to the use of weights in estimation is not substantial.

KW - autoregressive moving average (ARMA) models

KW - statistical inference

KW - variance

KW - weights

U2 - 10.1017/S0266466607070363

DO - 10.1017/S0266466607070363

M3 - Article

VL - 23

SP - 852

EP - 879

JO - Econometric Theory

T2 - Econometric Theory

JF - Econometric Theory

SN - 0266-4666

IS - 5

ER -