A contingent claims approach to hydroelectric scheduling

We consider the problem of hydro scheduling in a deregulated market. Finding an optimal strategy for the release of water from reservoirs is done under the uncertainty of future electricity prices and runoffs. The aim is to maximize the market value of the electricity generation resulting from releases. Estimating the stochastic process for runoffs is done using historical data and hydrological forecasting models, while information about the power price process should be taken from the local electricity market. A common approach in building the price process model is to base the estimation on historical spot prices or, in the case of a short history due to recent deregulation, on spot price forecasts. We propose a different approach, based on modern asset pricing theory, where the price process is estimated based on current and/or historical derivative prices. Electricity derivatives, e.g. futures contracts, give readily the market value of future delivery of electricity. The approach has at least two advantages. First, observed market prices can be used instead of subjective, bias-prone forecasts. Second, risk free interest rates can be used for discounting, instead of using a subjective risk-adjusted rate, which is hard to estimate. Based on a stochastic programming model, we quantify the benefits of such an approach for a small hydro producer.