Abstract
In the recent literature, reduced-form models and equilibrium models have been proposed for pricing electricity futures. We present the first empirical comparison between a one-factor reduced-form model by Lucia and Schwartz (2002) and an equilibrium model which is a dynamic extension of Bessembinder and Lemmon (2002). The latter one models three groups of agents - producers, retailers, and end-users - and explicitly considers the production and cost structure. This allows to better meet the distribution of spot prices, generate price spikes without modelling exogenous jumps, and endogenously derive a term structure of the risk premium. Our analysis is based on 50 months of futures prices from the Scandinavian electricity exchange Nord Pool. We estimate both models out-of-sample for risk-neutral and risk-averse participants. We find that considering risk premia in futures prices improves the forecasting quality for both models. The equilibrium model better explains future prices in volatile markets, i. e. when price jumps occur. We also find empirical evidence for the derived structure of the risk premia. The much larger effort for estimating the equilibrium model pays off by a better handling of price spikes when evaluating electricity futures.