Decoding the Australian Electricity Market: New Evidence from Three-Regime Hidden Semi-Markov Model

Nicholas Apergis, Giray Gozgor, Chi Keung Lau, Shixuan Wang

Research output: Contribution to journalArticle

Abstract

The hidden semi-Markov model (HSMM) is more flexible than the hidden Markov model (HMM). As an extension of the HMM, the sojourn time distribution in the HSMM can be explicitly specified by any distribution, either nonparametric or parametric, facilitating the modelling for the stylised features of electricity prices, such as the short-lived spike and the time-varying mean. By using a three-regime HSMM, this paper investigates the hidden regimes in five Australian States (Queensland, New South Wales, Victoria, South Australia, and Tasmania), spanning the period from June 8, 2008 to July 3, 2016. Based on the estimation results, we find evidence that the three hidden regimes correspond to a low-price regime, a high-price regime, and a spike regime. Running the decoding algorithm, the analysis systemically finds the timing of the three regimes, and thus, we link the empirical results to the policy changes in the Australian National Electricity Market. We further discuss the contributing factors for the different characteristics of the Australian electricity markets at the state-level.
LanguageEnglish
Pages129-142
Number of pages14
JournalEnergy Economics
Volume78
Early online date1 Nov 2018
DOIs
Publication statusPublished - Feb 2019

Fingerprint

Decoding
Hidden Markov models
Electricity
Power markets
Markov model
Electricity market

Cite this

Apergis, Nicholas ; Gozgor, Giray ; Lau, Chi Keung ; Wang, Shixuan. / Decoding the Australian Electricity Market : New Evidence from Three-Regime Hidden Semi-Markov Model. In: Energy Economics. 2019 ; Vol. 78. pp. 129-142.
@article{61a897b91c09403db7cb68a0f53892a7,
title = "Decoding the Australian Electricity Market: New Evidence from Three-Regime Hidden Semi-Markov Model",
abstract = "The hidden semi-Markov model (HSMM) is more flexible than the hidden Markov model (HMM). As an extension of the HMM, the sojourn time distribution in the HSMM can be explicitly specified by any distribution, either nonparametric or parametric, facilitating the modelling for the stylised features of electricity prices, such as the short-lived spike and the time-varying mean. By using a three-regime HSMM, this paper investigates the hidden regimes in five Australian States (Queensland, New South Wales, Victoria, South Australia, and Tasmania), spanning the period from June 8, 2008 to July 3, 2016. Based on the estimation results, we find evidence that the three hidden regimes correspond to a low-price regime, a high-price regime, and a spike regime. Running the decoding algorithm, the analysis systemically finds the timing of the three regimes, and thus, we link the empirical results to the policy changes in the Australian National Electricity Market. We further discuss the contributing factors for the different characteristics of the Australian electricity markets at the state-level.",
keywords = "Australian electricity markets, Hidden semi-Markov model, Time-varying mean, Price spikes, Regime-switching",
author = "Nicholas Apergis and Giray Gozgor and Lau, {Chi Keung} and Shixuan Wang",
year = "2019",
month = "2",
doi = "10.1016/j.eneco.2018.10.038",
language = "English",
volume = "78",
pages = "129--142",
journal = "Energy Economics",
issn = "0140-9883",
publisher = "Elsevier",

}

Decoding the Australian Electricity Market : New Evidence from Three-Regime Hidden Semi-Markov Model. / Apergis, Nicholas; Gozgor, Giray; Lau, Chi Keung; Wang, Shixuan.

In: Energy Economics, Vol. 78, 02.2019, p. 129-142.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Decoding the Australian Electricity Market

T2 - Energy Economics

AU - Apergis, Nicholas

AU - Gozgor, Giray

AU - Lau, Chi Keung

AU - Wang, Shixuan

PY - 2019/2

Y1 - 2019/2

N2 - The hidden semi-Markov model (HSMM) is more flexible than the hidden Markov model (HMM). As an extension of the HMM, the sojourn time distribution in the HSMM can be explicitly specified by any distribution, either nonparametric or parametric, facilitating the modelling for the stylised features of electricity prices, such as the short-lived spike and the time-varying mean. By using a three-regime HSMM, this paper investigates the hidden regimes in five Australian States (Queensland, New South Wales, Victoria, South Australia, and Tasmania), spanning the period from June 8, 2008 to July 3, 2016. Based on the estimation results, we find evidence that the three hidden regimes correspond to a low-price regime, a high-price regime, and a spike regime. Running the decoding algorithm, the analysis systemically finds the timing of the three regimes, and thus, we link the empirical results to the policy changes in the Australian National Electricity Market. We further discuss the contributing factors for the different characteristics of the Australian electricity markets at the state-level.

AB - The hidden semi-Markov model (HSMM) is more flexible than the hidden Markov model (HMM). As an extension of the HMM, the sojourn time distribution in the HSMM can be explicitly specified by any distribution, either nonparametric or parametric, facilitating the modelling for the stylised features of electricity prices, such as the short-lived spike and the time-varying mean. By using a three-regime HSMM, this paper investigates the hidden regimes in five Australian States (Queensland, New South Wales, Victoria, South Australia, and Tasmania), spanning the period from June 8, 2008 to July 3, 2016. Based on the estimation results, we find evidence that the three hidden regimes correspond to a low-price regime, a high-price regime, and a spike regime. Running the decoding algorithm, the analysis systemically finds the timing of the three regimes, and thus, we link the empirical results to the policy changes in the Australian National Electricity Market. We further discuss the contributing factors for the different characteristics of the Australian electricity markets at the state-level.

KW - Australian electricity markets

KW - Hidden semi-Markov model

KW - Time-varying mean

KW - Price spikes

KW - Regime-switching

U2 - 10.1016/j.eneco.2018.10.038

DO - 10.1016/j.eneco.2018.10.038

M3 - Article

VL - 78

SP - 129

EP - 142

JO - Energy Economics

JF - Energy Economics

SN - 0140-9883

ER -