After completing a doctorate in mathematics from the University of Grenoble, Alain Galli specialised in stochastic modelling in the earth sciences, especially in he oil industry. Published in:
Publication date:
Co-author 1: Margaret Armstrong
Powernext started operating on 27 November 2001. Although the German exchange, EEX, has been functioning for much longer, the two have many common points. Both use the same system for fixing the day-ahead spot price, the one developed by NordPool. In contrast to Omel in Spain, power producers in France and Germany are not obliged to sell through the exchange. In addition, the crossborder transmission lines that physically link the French and German grids, not only make the electricity supply more reliable they also allow cross-border commercial transactions which should homogenize prices in both countries. So after nearly a year of operation it is interesting to compare the spot prices on the two exchanges in order to have a better understanding of the statistical properties of the prices in the two markets and the relationship between them. This information will be used when modelling the structure of the day-ahead spot prices.
The data used to carry out the study consists of the (hourly) spot prices for electricity from Powernext and EEX, for the period from 1 January 2002 to 2 December 2002. Data from the first five weeks of trading were not included because traded volumes were relatively low initially and so these data are not necessarily representative.
This report is divided into four sections. The first one presents the basic statistics, starting with the histograms of all the 8064 spot prices in the 336 days, for both exchanges. In time series data, it is usual to find three types of seasonality: daily, weekly and annual. As the available data cover less than one calendar year, it is too early to attempt to study annual trends. So we limit ourselves to studying daily and weekly fluctuations. Plotting the hourly average prices for each day of the week shows some interesting differences between Powernext and EEX, as well as a high degree of similarity. The second section studies the time correlations between the two time series. Rather than using auto and cross correlations which are sensitive to the estimate of the mean, we have chosen to use variograms. The third section focuses on the buy and sell curves that effectively determine the spot prices. Some preliminary conclusions are given in the fourth section.
Powernext started operating on 27 November 2001. Although the German exchange, EEX, has been functioning for much longer, the two have many common points. Both use the same system for fixing the day-ahead spot price, the one developed by NordPool. In contrast to Omel in Spain, power producers in France and Germany are not obliged to sell through the exchange. In addition, the crossborder transmission lines that physically link the French and German grids, not only make the electricity supply more reliable they also allow cross-border commercial transactions which should homogenize prices in both countries. So after nearly a year of operation it is interesting to compare the spot prices on the two exchanges in order to have a better understanding of the statistical properties of the prices in the two markets and the relationship between them. This information will be used when modelling the structure of the day-ahead spot prices.
The data used to carry out the study consists of the (hourly) spot prices for electricity from Powernext and EEX, for the period from 1 January 2002 to 2 December 2002. Data from the first five weeks of trading were not included because traded volumes were relatively low initially and so these data are not necessarily representative.
This report is divided into four sections. The first one presents the basic statistics, starting with the histograms of all the 8064 spot prices in the 336 days, for both exchanges. In time series data, it is usual to find three types of seasonality: daily, weekly and annual. As the available data cover less than one calendar year, it is too early to attempt to study annual trends. So we limit ourselves to studying daily and weekly fluctuations. Plotting the hourly average prices for each day of the week shows some interesting differences between Powernext and EEX, as well as a high degree of similarity. The second section studies the time correlations between the two time series. Rather than using auto and cross correlations which are sensitive to the estimate of the mean, we have chosen to use variograms. The third section focuses on the buy and sell curves that effectively determine the spot prices. Some preliminary conclusions are given in the fourth section.