A computationally efficient approach to optimizing offers in centrally committed electricity markets

We study the incentive properties of the two primary approaches to incorporating unit-commitment decisions in wholesale electricity markets. One approach is centralized unit commitment, wherein generating firms provide complex multi-part offers that specify their non-convex fixed and variable operating costs. The market operator uses these offers to co-optimize unit-commitment and economic-dispatch decisions. The second approach is self-commitment, whereby firms determine unit-commitment decisions for their generating units individually and submit simple offers for the provision of energy. Operators of self-committed markets determine generator dispatch based on the merit order of the simple offers. Comparing the incentive properties of the two market designs is challenging because the offer-optimization problem for a firm that participates in a centrally committed market is a bi-level model with binary variables in the lower-level problem. To address this challenge, we develop a computationally efficient approach to solve such a problem and illustrate the method with examples. We use the examples to compare the incentive properties of the two market designs. Our examples show that the profit of the profit-maximizing firm does not differ significantly between the two market designs but that system costs can be higher under a self-committed design. These cost differences are because the complex offers and discriminatory payment schemes that are used under centrally committed designs can mitigate incentives for the profit-maximizing firm to exercise market power.