Fat Tails in Power Prices

Spot power prices exhibit extreme price jumps and the tendency to oscillate around a long-term mean. Despite these well-known characteristics, electricity price models used for Monte Carlo simulations, VaR related measures, or derivatives valuation, often assume normally distributed residuals. In this paper, we examine the distributional characteristics of model residuals and show that the hypothesis of normality is rejected due to significant tail fatness and skewness. We then examine the Student-t distribution as a candidate fit for residuals and as an alternative distribution for random innovations in Monte Carlo simulations. The resulting price patterns clearly show that simulations based on the Student-t distribution resemble more closely actual power price patters. We then discuss the implications of our results for risk management.