Empirical evidences about hourly electricity prices in some European markets

The study originates from a work about the economic risk analysis of a windpower plant to be built on an Italian site. The annual gain distribution is drawn by means of simulations of a business year. A crucial point is the simulation of the hourly sale prices of the electricity on the basis of a suitable time-series model. The model should take into account the typical features of the hourly electricity prices: (i) seasonality, particularly with daily and weekly periodicity; (ii) mean-reversion that can be formalised by including an AR(1) relation in the model; (iii) jumps and spikes that can be due to the difficulty of the electricity supply to match the demand; (iv) volatility clustering; (v) leptokurtic distribution. Analysing the 2008-2009 Italian PUN (national common price) of the electricity an easy seasonal AR-GARCH model with t-Student standardised innovation has been detected. The same type model has been tested with electricity prices of Spain and Norway markets. The model fits well both of the data. At last a singular simplification of the model is analysed: the model can be reduced to an AR(1)-GARCH model by means of an average of seasonal differences.