计及分布参数时变性的电力市场风险测度研究  

Value-at-risk Estimation of Electricity Market Considering the Time-varying Features of Distribution's Parameters

作　　者：王瑞庆[1] 肖自乾[1] 马杰[1] 

机构地区：[1]海南软件职业技术学院,海南琼海571400

出　　处：《电力学报》2012年第6期578-582,共5页Journal of Electric Power

基　　金：海南省自然科学基金(611126);海南省高等院校科研基金(Hjkj2011-48)

摘　　要：准确刻画电价的波动特征是有效度量电力市场价格风险的基础。以负荷作为外生解释变量,建立了考虑电价多季节、异方差、波动集聚、尖峰厚尾等特征的GARCH-VaR计算模型,以正态分布、t分布、有偏t分布为例研究了残差的分布形式设定和参数的时变性对VaR估计精度的影响。对PJM历史数据的实证表明:计及参数时变性的有偏t分布模型的VaR估计结果准确有效,正态分布模型在置信水平大于95%时低估了电力市场的VaR,t分布模型在置信水平小于97.5%时高估了电力市场VaR。该结果对于电力市场参与者准确地评估价格风险、制定有效的风险规避策略具有重要的指导意义。How to accurately characterize the volatilities of electricity price series is the foundation to effective evaluation of the price risk in electricity market.With system load as an exogenous explanatory variable,a GARCH-VaR model to estimate VaR is proposed,in which the seasonalities,heteroscedasticities,kurtosises,heavy-tails and volatility-clustering can be jointly addressed.The impacts of probability distribution assumption and the time-varying features of parameters for three innovation's distributions,namely normal,student-t and skewed student-t,on the accuracy of VaR estimatation are analyzed.The numerical example based on the historical data of the PJM market shows that the skewed student-t GARCH-VaR model with time-varying parameters performs better in predicting one-period-ahead VaR,but the one with normal distribution underestimates the higher quantiles and the one with student-t distribution overestimates the lower quantiles.These results present several potential implications for risk quantifications and hedging strategies of electricity markets.

关 键 词：电力市场 风险价值 GARCH模型 概率分布设定 时变参数 

分 类 号：TM73[电气工程—电力系统及自动化] F123.9[经济管理—世界经济]