On the power of portmanteau serial correlation tests

Yue Fang, Michael A. Martin, Terence J. O'neill, Steven Roberts

Research output: Contribution to journalArticleResearchpeer-review

Abstract

This paper studies properties of the portmanteau statistic proposed by Box and Pierce [1] and its modification of Ljung and Box [2].We show that these portmanteau statistics are feasible analogs to optimal tests for the class of statistics which are linear combinations of consistent estimates of serial correlations. We find, however, that for sample sizes commonly encountered in practice, the efficiency loss in power of portmanteau statistics relative to optimal tests can be substantial, although their size properties are broadly comparable. Our results indicate that tests based on some other non-optimal weighting schemes, including tests with optimal weights constructed from moderately misspecified alternatives, deliver tests with better power than the Box-Pierce or Ljung-Box statistics.

Original languageEnglish
Pages (from-to)593-604
Number of pages12
JournalJournal of Statistical Computation and Simulation
Volume76
Issue number7
DOIs
Publication statusPublished - 1 Jul 2006
Externally publishedYes

Fingerprint

Serial Correlation
Statistics
Optimal Test
Consistent Estimates
Weighting
Statistic
Linear Combination
Sample Size
Analogue
Serial correlation
Alternatives

Cite this

Fang, Yue ; Martin, Michael A. ; O'neill, Terence J. ; Roberts, Steven. / On the power of portmanteau serial correlation tests. In: Journal of Statistical Computation and Simulation. 2006 ; Vol. 76, No. 7. pp. 593-604.
@article{9e9dd219b3f34927b75a9fbce99c6f35,
title = "On the power of portmanteau serial correlation tests",
abstract = "This paper studies properties of the portmanteau statistic proposed by Box and Pierce [1] and its modification of Ljung and Box [2].We show that these portmanteau statistics are feasible analogs to optimal tests for the class of statistics which are linear combinations of consistent estimates of serial correlations. We find, however, that for sample sizes commonly encountered in practice, the efficiency loss in power of portmanteau statistics relative to optimal tests can be substantial, although their size properties are broadly comparable. Our results indicate that tests based on some other non-optimal weighting schemes, including tests with optimal weights constructed from moderately misspecified alternatives, deliver tests with better power than the Box-Pierce or Ljung-Box statistics.",
author = "Yue Fang and Martin, {Michael A.} and O'neill, {Terence J.} and Steven Roberts",
year = "2006",
month = "7",
day = "1",
doi = "10.1080/10629360500108921",
language = "English",
volume = "76",
pages = "593--604",
journal = "Journal of Statistical Computation and Simulation",
issn = "0094-9655",
publisher = "Taylor & Francis",
number = "7",

}

On the power of portmanteau serial correlation tests. / Fang, Yue; Martin, Michael A.; O'neill, Terence J.; Roberts, Steven.

In: Journal of Statistical Computation and Simulation, Vol. 76, No. 7, 01.07.2006, p. 593-604.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - On the power of portmanteau serial correlation tests

AU - Fang, Yue

AU - Martin, Michael A.

AU - O'neill, Terence J.

AU - Roberts, Steven

PY - 2006/7/1

Y1 - 2006/7/1

N2 - This paper studies properties of the portmanteau statistic proposed by Box and Pierce [1] and its modification of Ljung and Box [2].We show that these portmanteau statistics are feasible analogs to optimal tests for the class of statistics which are linear combinations of consistent estimates of serial correlations. We find, however, that for sample sizes commonly encountered in practice, the efficiency loss in power of portmanteau statistics relative to optimal tests can be substantial, although their size properties are broadly comparable. Our results indicate that tests based on some other non-optimal weighting schemes, including tests with optimal weights constructed from moderately misspecified alternatives, deliver tests with better power than the Box-Pierce or Ljung-Box statistics.

AB - This paper studies properties of the portmanteau statistic proposed by Box and Pierce [1] and its modification of Ljung and Box [2].We show that these portmanteau statistics are feasible analogs to optimal tests for the class of statistics which are linear combinations of consistent estimates of serial correlations. We find, however, that for sample sizes commonly encountered in practice, the efficiency loss in power of portmanteau statistics relative to optimal tests can be substantial, although their size properties are broadly comparable. Our results indicate that tests based on some other non-optimal weighting schemes, including tests with optimal weights constructed from moderately misspecified alternatives, deliver tests with better power than the Box-Pierce or Ljung-Box statistics.

UR - http://www.scopus.com/inward/record.url?scp=84865531270&partnerID=8YFLogxK

U2 - 10.1080/10629360500108921

DO - 10.1080/10629360500108921

M3 - Article

VL - 76

SP - 593

EP - 604

JO - Journal of Statistical Computation and Simulation

JF - Journal of Statistical Computation and Simulation

SN - 0094-9655

IS - 7

ER -