Abstract
In the literature on forecasting electricity spot prices, it is noted that the empirical distribution of the growth rates of equilibrium prices for electricity is characterized by the presence of heavy ‘tails’, so they can be described as a jump-diffusion process. However, electricity prices are associated with a variety of observable factors that can be included in the model. Within the framework of this article, a flexible model, that allows taking into account statistical features of electricity prices (multilevel seasonality, stochastic volatility), as well as fundamental price factors that directly or indirectly affect equilibrium price indices (weather factors, resource prices, industrial production index, or IPI), was developed. Using methods of Bayesian inference, it was shown that the developed two-level specification of the stochastic volatility model, which separates the factors influencing the deterministic and stochastic components of the series, fits the data best, among the considered alternatives. As a result, differences in the price dynamics between the European and Siberian price zones were revealed: the influence of the weather factor in the price zones is not the same; there are also differences in the weekly price dynamics and the effect of holidays. The effect of low-frequency economic factors (resource prices, IPI) on prices was not revealed. This model is a useful tool for analyzing short-term and long-term electricity price dynamics, building scenario forecasts, and it also can potentially be used in risk-management and electricity derivatives pricing.