基于条件有偏t分布ARMAX-GARCH模型和极值理论的电力市场风险测度研究

Risk Measure of Electricity Market Based on ARMAX-GARCH Model with Conditional Skewed-t Distribution and Extreme Value Theory

在对电价的基本特征综合分析的基础上,采用残差服从时变偏度和自由度的有偏t分布ARMAX-GARCH(ARMAX-GARCH-VST)模型对电价序列的相关性、有偏性、多季节性、异方差性和波动集聚性进行过滤,以获得统计特性更好的独立同分布残差序列,然后运用极值理论描述ARMAX-GARCH-VST模型归一化残差的尾部分布,以获得更加精确的VaR估计。基于PJM电力市场历史数据的分析表明,ARMAX-GARCH-VST模型和ARMAX-GARCH-EVT模型都能对现货电价的变化做出比较迅速的反应,VaR的估计结果在各个置信水平下均准确有效,但ARMAX-GARCH-EVT模型的VaR估计结果更加准确,表现出了更好的动态特征。该结果对于电力市场参与者准确地测度价格波动风险和制定有效的风险规避策略具有重要的指导意义。

On the basis of comprehensive analysis of the electricity price basic features, an ARMAX-GARCH model using a skewed-t distribution with residuals subject to time-varying skewness and degree of freedom （ARMAX- GARCH-VST） is proposed to filter spot electricity price series in terms of its dependencies, skewnesses, seasonali- ties, heteroskedasticity and volatility clustering, whereby an approximately independently and identically distributed normalized residual series with better statistical properties is acquired. Then extreme value theory（EVT） is adopted to describe the tail distribution of the normalized residuals of ARMAX-GARCH-VST model for the more accurate value- at-risk （VaR） estimate. The empirical analysis based on the historical data of the PJM electricity market shows that both the ARMAX-GARCH-VST and ARMAX-GARCH-EVT model can rapidly reflect the changes of spot electricity prices and can produce accurate VaR estimates at all confidence levels, but the ARMAX-GARCH-EVT model is su- perior, showing better dynamic characteristics. These results can provide guidance for the electricity market partici- pants in accurately forecasting price fluctuation risk and formulating effective risk-hedging strategies.