Detection of level change (LC) outlier in GARCH (1, 1) processes

Azami Zaharim, Siti Meriam Zahid, Mohammad Said Zainol, Kamaruzzaman Sopian

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

An outlier is an 'extreme' observation that may have a severe effect on data analysis. Their occurrences might cause problems such as bias or distortion of parameter estimation. When a time series with outliers is modelled and forecasted without taking into account their presence or without removing their effects, the test statistics of estimated parameters in the model will be distorted. A GARCH (1,1) process is not exceptional from being affected by outliers. In this study we construct test statistics for identifying outliers in GARCH (1,1) processes with special focus on the temporary change (LC) type. The statistic was developed by the least squares method, and consequently a simulation was carried out for the purpose of finding the critical region. This study is an extension of Chen and Liu's (1993) work on outlier detection. . The simulation was done for eight sample sizes; 50, 100, 150, 200, 250, 500, 1000 and 2000. The 90th, 95th and 99th percentiles were computed for estimating the distribution of the test statistic. The critical value C = 5.0 was selected based on the simulation study and was used in detecting the presence of level change (LC) in the return series of Industrial Product Index (IPI). For the period of analysis, the results indicate that LC outlier occurred in 1998.

Original languageEnglish
Title of host publicationProceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09
Pages316-321
Number of pages6
Publication statusPublished - 2009
Event11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, 8th WSEAS Int. Conf. NOLASC '09, 5th WSEAS Int. Conf. CONTROL '09 - Canary Islands
Duration: 1 Jul 20093 Jul 2009

Other

Other11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, 8th WSEAS Int. Conf. NOLASC '09, 5th WSEAS Int. Conf. CONTROL '09
CityCanary Islands
Period1/7/093/7/09

Fingerprint

Generalized Autoregressive Conditional Heteroscedasticity
Outlier
Statistics
Test Statistic
Critical region
Parameter estimation
Outlier Detection
Percentile
Time series
Least Square Method
Statistic
Critical value
Parameter Estimation
Data analysis
Sample Size
Simulation
Extremes
Simulation Study
Series

Keywords

  • GARCH
  • LC outlier
  • Least squares method
  • Simulation

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Control and Systems Engineering
  • Computational Mathematics

Cite this

Zaharim, A., Zahid, S. M., Zainol, M. S., & Sopian, K. (2009). Detection of level change (LC) outlier in GARCH (1, 1) processes. In Proceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09 (pp. 316-321)

Detection of level change (LC) outlier in GARCH (1, 1) processes. / Zaharim, Azami; Zahid, Siti Meriam; Zainol, Mohammad Said; Sopian, Kamaruzzaman.

Proceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09. 2009. p. 316-321.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Zaharim, A, Zahid, SM, Zainol, MS & Sopian, K 2009, Detection of level change (LC) outlier in GARCH (1, 1) processes. in Proceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09. pp. 316-321, 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, 8th WSEAS Int. Conf. NOLASC '09, 5th WSEAS Int. Conf. CONTROL '09, Canary Islands, 1/7/09.
Zaharim A, Zahid SM, Zainol MS, Sopian K. Detection of level change (LC) outlier in GARCH (1, 1) processes. In Proceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09. 2009. p. 316-321
Zaharim, Azami ; Zahid, Siti Meriam ; Zainol, Mohammad Said ; Sopian, Kamaruzzaman. / Detection of level change (LC) outlier in GARCH (1, 1) processes. Proceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09. 2009. pp. 316-321
@inproceedings{c510375b20ca4e919bbeb21472128358,
title = "Detection of level change (LC) outlier in GARCH (1, 1) processes",
abstract = "An outlier is an 'extreme' observation that may have a severe effect on data analysis. Their occurrences might cause problems such as bias or distortion of parameter estimation. When a time series with outliers is modelled and forecasted without taking into account their presence or without removing their effects, the test statistics of estimated parameters in the model will be distorted. A GARCH (1,1) process is not exceptional from being affected by outliers. In this study we construct test statistics for identifying outliers in GARCH (1,1) processes with special focus on the temporary change (LC) type. The statistic was developed by the least squares method, and consequently a simulation was carried out for the purpose of finding the critical region. This study is an extension of Chen and Liu's (1993) work on outlier detection. . The simulation was done for eight sample sizes; 50, 100, 150, 200, 250, 500, 1000 and 2000. The 90th, 95th and 99th percentiles were computed for estimating the distribution of the test statistic. The critical value C = 5.0 was selected based on the simulation study and was used in detecting the presence of level change (LC) in the return series of Industrial Product Index (IPI). For the period of analysis, the results indicate that LC outlier occurred in 1998.",
keywords = "GARCH, LC outlier, Least squares method, Simulation",
author = "Azami Zaharim and Zahid, {Siti Meriam} and Zainol, {Mohammad Said} and Kamaruzzaman Sopian",
year = "2009",
language = "English",
isbn = "9789604740949",
pages = "316--321",
booktitle = "Proceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09",

}

TY - GEN

T1 - Detection of level change (LC) outlier in GARCH (1, 1) processes

AU - Zaharim, Azami

AU - Zahid, Siti Meriam

AU - Zainol, Mohammad Said

AU - Sopian, Kamaruzzaman

PY - 2009

Y1 - 2009

N2 - An outlier is an 'extreme' observation that may have a severe effect on data analysis. Their occurrences might cause problems such as bias or distortion of parameter estimation. When a time series with outliers is modelled and forecasted without taking into account their presence or without removing their effects, the test statistics of estimated parameters in the model will be distorted. A GARCH (1,1) process is not exceptional from being affected by outliers. In this study we construct test statistics for identifying outliers in GARCH (1,1) processes with special focus on the temporary change (LC) type. The statistic was developed by the least squares method, and consequently a simulation was carried out for the purpose of finding the critical region. This study is an extension of Chen and Liu's (1993) work on outlier detection. . The simulation was done for eight sample sizes; 50, 100, 150, 200, 250, 500, 1000 and 2000. The 90th, 95th and 99th percentiles were computed for estimating the distribution of the test statistic. The critical value C = 5.0 was selected based on the simulation study and was used in detecting the presence of level change (LC) in the return series of Industrial Product Index (IPI). For the period of analysis, the results indicate that LC outlier occurred in 1998.

AB - An outlier is an 'extreme' observation that may have a severe effect on data analysis. Their occurrences might cause problems such as bias or distortion of parameter estimation. When a time series with outliers is modelled and forecasted without taking into account their presence or without removing their effects, the test statistics of estimated parameters in the model will be distorted. A GARCH (1,1) process is not exceptional from being affected by outliers. In this study we construct test statistics for identifying outliers in GARCH (1,1) processes with special focus on the temporary change (LC) type. The statistic was developed by the least squares method, and consequently a simulation was carried out for the purpose of finding the critical region. This study is an extension of Chen and Liu's (1993) work on outlier detection. . The simulation was done for eight sample sizes; 50, 100, 150, 200, 250, 500, 1000 and 2000. The 90th, 95th and 99th percentiles were computed for estimating the distribution of the test statistic. The critical value C = 5.0 was selected based on the simulation study and was used in detecting the presence of level change (LC) in the return series of Industrial Product Index (IPI). For the period of analysis, the results indicate that LC outlier occurred in 1998.

KW - GARCH

KW - LC outlier

KW - Least squares method

KW - Simulation

UR - http://www.scopus.com/inward/record.url?scp=78149306115&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78149306115&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9789604740949

SP - 316

EP - 321

BT - Proceedings of the 11th WSEAS International Conference on Mathematical Methods, Computational Techniques and Intelligent Systems, MAMECTIS '09, Proc. 8th WSEAS NOLASC '09, Proc. 5th WSEAS CONTROL '09

ER -