Summary
We consider the valuation of contracts of electrical energy supply with optionalities. After discussing appropriate stochastic programming models and presenting especially suited solution algorithms, a set of price scenarios is simulated based on a probabilistic model of the electricity spot market price at the EEX. We determine empirically upper and lower bounds for the stochastic optimization over any scenario tree obtained by reduction techniques. Furthermore, we introduce constraints restricting all scenarios to have identical contract exercise amounts cumulated over various fixed subperiods. Calculation of the losses of the optimal value of the objective function caused by these constraints shows that, for subperiods of 1 month, no substantial loss is encountered. This suggests a temporal decoupling heuristic where the depth of scenario trees is reduced to a suitable subperiod, yielding a good approximation to the valuation problem with substantially reduced complexity.