Wholesale electricity market failure and the new market design

This paper studies the California energy crisis from the angle of power market design. We introduce an infinitely repeated multi-unit auctions model to analyze the wholesale electricity auctions under Poolco rule. We then use a multi-agent system to simulate the model and look for its adaptive learning solutions. We find that bidders´ learning during their repeated interactions will converge to equilibria with withholding behavior. Our results not only generate mild withholding reported by other researchers, but also lead to a type of equilibria with extreme withholding where market price jumps to the cap. This type of equilibria better resemble the realistic situation in the power market hence give us important clue for a new market design. We propose that instead of imposing a price cap, the auctioneer should allow the utility companies to submit price-elastic demand schedule that better reflects the consumers´ price-substitution preferences. Simulation results show that the new design greatly improves the consumer welfare.