We present a new model for the electricity spot price dynamics, which is able to capture seasonality, low-frequency dynamics and the extreme spikes in the market. Instead of the usual purely deterministic trend we introduce a non-stationary independent increments process for the low-frequency dynamics, and model the large uctuations by a non-Gaussian stable CARMA process. The model allows for analytic forward prices, and we apply these to model and estimate the whole market consistently. An estimation procedure is suggested, where we fit the non-stationary trend using forward data with long time until delivery, and a robust L1-Filter to and the states of the CARMA process. The estimation is based on minimizing the distance between the empirical and theoretical risk premiums, and applied to data from the German electricity exchange EEX. Our empirical analysis is split into base load and peak load prices. We find an overall negative risk premium for the base load forward contracts, except for contracts close to delivery, where a small positive risk premium is detected. The peak load contracts, on the other hand, show a clear positive risk premium, when they are close to delivery, while the contracts in the longer end also have a negative premium.      «

We present a new model for the electricity spot price dynamics, which is able to capture seasonality, low-frequency dynamics and the extreme spikes in the market. Instead of the usual purely deterministic trend we introduce a non-stationary independent increments process for the low-frequency dynamics, and model the large uctuations by a non-Gaussian stable CARMA process. The model allows for analytic forward prices, and we apply these to model and estimate the whole market consistently....     »