Competition, risk and learning in electricity markets: an agent-based simulation study

This is the latest version of this item.

Official URL: http://dx.doi.org/10.1016/j.apenergy.2017.03.121

This paper studies the effects of learning and risk aversion on generation company (GenCo) bidding behavior in an oligopolistic electricity market. To this end, a flexible agent-based simulation model is developed in which GenCo agents bid prices in each period. Taking transmission grid constraints into account, the ISO solves a DC-OPF problem to determine locational prices and dispatch quantities. Our simulations show how, due to competition and learning, the change in the risk aversion level of even one GenCo can have a significant impact on all GenCo bids and profits. In particular, some level of risk aversion is observed to be beneficial to GenCos, whereas excessive risk aversion degrades profits by causing intense price competition. Our comprehensive study on the effects of Q-learning parameters finds the level of exploration to have a large impact on the outcome. The results of this paper can help GenCos develop bidding strategies that consider their rivals' as well as their own learning behavior and risk aversion levels. Likewise, the results can help regulators in designing market rules that take realistic GenCo behavior into account.