Access Restriction

Author Diongue, Abdou Kâ ♦ Guegan, Dominique
Source Hyper Articles en Ligne (HAL)
Content type Text
File Format PDF
Language English
Subject Keyword A-PARCH ♦ Asymmetric distribution function ♦ electricity spot prices ♦ Leptokurtosis ♦ persistence ♦ seasonality ♦ GARMA ♦ A-PARCH. ♦ Distribution asymétrique ♦ prix d'électricité ♦ persistance ♦ longue mémoire ♦ A-P-ARCH ♦ Gegenbauer. ♦ shs ♦ math ♦ stat ♦ Humanities and Social Sciences/Economies and finances ♦ Humanities and Social Sciences/Methods and statistics ♦ Mathematics [math]/Probability [math.PR] ♦ Mathematics [math]/Statistics [math.ST] ♦ Statistics [stat]/Statistics Theory [stat.TH]
Abstract Electricity spot prices exhibit a number of typical features that are not found in most financial time series, such as complex seasonality patterns, persistence (hyperbolic decay of the autocorrelation function), mean reversion, spikes, asymmetric behavior and leptokurtosis. Efforts have been made worldwide to model the behaviour of the electricity's market price. In this paper, we propose a new approach dealing with the stationary k-factor Gegenbauer process with asymmetric Power GARCH noise under conditional Student-t distribution, which can take into account the previous features. We derive the stationary and invertible conditions as well as the δth-order moment of this model that we called GGk-APARCH model. Then we focus on the estimation parameters and provide the analytical from of the likelihood which permits to obtain consitent estimates. In order to characterize the properties of these estimates we perform a Monte Carlo experiment. Finally the previous approach is used to the model electricity spot prices coming from the Leipzig Power Exchange (LPX) in Germany, Powernext in France, Operadora del Mercado Espagñol de Electricidad (OMEL) in Spain and the Pennsylvania-New Jersey-Maryland (PJM) interconnection in United States. In terms of forecasting criteria we obtain very good results comparing with models using hederoscedastic asymmetric errors.
ISSN 1955611X
Educational Use Research
Learning Resource Type Article
Publisher Date 2008-02-01