Structural Estimation Analysis in Hydropower Scheduling

Abstract

When planning production, a hydro power reservoir managers need to form expectations for electricity prices in the future. When forming expectations, the Nordic electricity forward market is a useful tool for predicting how the underlying spot price will move. In this paper we develop a structural estimation model for a single agent hydropower producer in Norway. With this model, we analyse how primitives in the price process, related to the forward price, can be inferred from empirical data from actual production time series. By analyzing trends and patterns in observed time series we have approximately parametrized the state space transition. Central here is the connection we model between inflow and price, to capture dry- and wet year dynamics in the two. To demonstrate the model it has been applied to a specific hydro power plant in Norway. From the results we make a preliminary analysis of to what extent the producer uses forward information when planning production. The results indicate that this specific producer is inclined to take the forward price into account when planning, and that a forward price with $6$ months to maturity is favored. An important byproduct of our model is the ability to calculate water values from the outputs.