Experience with electricity market test suite: students versus computational agents

Abstract

This paper applies two experimental economics methods (i.e., agent-based modeling and laboratory experiment) to a market test suite that is based on a fictional European wholesale electricity market. Quantitative results of generators' strategic behavior in this market context are separated between generators played by human subjects (i.e., master students) in a laboratory experiment and generators represented by computational agents in an agent-based model. The behavior is measured through offers that students or agents make when participating in the electricity trading auction and the market outcomes under both methods are discussed in order to illustrate the difference between the behavior of human and computational agents. The paper also identifies the improvements that would need to be made to the market test suite to allow for a more conclusive comparison in future experiments.