Some statistical investigations on the nature and dynamics of electricity prices

This work analyzes the log-returns of daily electricity prices from the NordPool day-ahead market. We study both the unconditional growth rates distribution and the distribution of residual shocks obtained with a non-parametric filtering procedure based on the Cholesky factor algorithm. We show that, even if the Subbotin family of distributions is able to describe the empirical observations in both cases, the Subbotin fit obtained for the unconditional growth rates and for the residual shocks reveal significant differences. Indeed, the sequence of log-returns can be described as the outcome of an aggregation of Laplace-distributed shocks with time-dependent volatility. We find that the standard deviation of shocks scales as a power law of the initial price level, with scaling exponent around -1. Moreover, the analysis of the empirical density of shocks, conditional on the price level, shows a strong relationship of the Subbotin fit with the latter. We conclude that the unconditional growth rates distribution is the superposition of shocks distributions characterized by decreasing volatility and fat-tailedness with respect to the price level.