Co-Learning Patterns As Emergent MarketPhenomena: An Electricity Market Illustration

The denition of emergence remains problematic, particularly for systemswith purposeful human interactions. This study explores the practical importof this concept within a specic market context: namely, a double-auctionmarket for wholesale electric power that operates over a transmission gridwith spatially located buyers and sellers. Each prot-seeking seller is a learningagent that attempts to adjust its daily supply oers to its best advantage.The sellers are co-learners in the sense that their supply oer adjustmentsare in response to past market outcomes that reect the past supply oerchoices of all sellers. Attention is focused on the emergence of co-learningpatterns, that is, global market patterns that arise and persist over time asa result of seller co-learning. Examples of co-learning patterns include correlatedseller supply oer behaviors and correlated seller net earnings outcomes.Heat maps are used to display and interpret co-learning pattern ndings.One key nding is that co-learning strongly matters in this auction marketenvironment. Sellers that behave as Gode-Sunder budget-constrained zerointelligenceagents, randomly selecting their supply oers subject only to abreak-even constraint, tend to realize substantially lower net earnings thansellers that tacitly co-learn to correlate their supply oers for market poweradvantages.[...]