Futures risk premium characterization and spot price modeling on the German electricity market

Abstract: In this thesis it was investigated how accurate the futures predict the spot prices and characterizing the futures risk premium on the German electricity market, which currently undergoes an energy transition. This was conducted twice, with realized data and with data from an own developed model. The realized spot price data and futures were obtained from the European Energy Exchange. The methodology to develop the model was according to conventional financial time series modeling. First, the data was converted from nonstationary to stationary. Then, Box & Jenkins modeling and Akaike information and Schwartz Bayesian criteria were used to determine the lags in the subsequent ARMA model. The model was chosen to be an ARMA(1,1) model. Its parameters were estimated with the maximum-likelihood method and then a 1-step forecasting method was applied to generate data points. Both the realized data and the data from the model, together with the relevant futures, suggest that there is a positive risk premium. This is consistent with financial research. Regarding the accuracy measures, the uncertainty tended to increase gradually for the realized values, meanwhile the model based ones generated a non-consistent pattern.