BFS 2002

Contributed Talk

The goal of the paper is to propose a family of processes to model electricity prices in deregulated markets. Our class is meant to be broad enough to encompass different regions worldwide. Besides mean reversion, a property they share with other commodities, electricity prices exhibit the unique feature of big spikes, generated by extreme weather conditions or other sources of supply disruption in the absence of storability. To address this key characteristic, we introduce a set of discontinuous semimartingales displaying a feature of jump-reversion while preserving the Markov property. We propose a calibration procedure based on approximate maximum likelihood. Simulation tests investigate the qualitative properties of sample paths and exhibit the quality of the statistical estimation on a database of US power markets.