Evolutionary-stable strategies with increasing and decreasing marginal utilities in the Ausubel auction

A genetic algorithm has been developed to solve bidding strategies in a dynamic multi-unit auction: the Ausubel auction, with independent private values and without dropout information. The genetic algorithm aims to maximize each bidder's payoff. To this end two experimental environments have been tested with decreasing and increasing marginal utilities. The bidding strategies are analyzed, along with their effects on revenue and efficiency. With decreasing marginal utilities the computational experiments yield to sincere bidding as the evolutionary-stable strategy, which is also the weakly dominant strategy and the ex post perfect equilibrium. Nevertheless, with increasing marginal utilities there is no theory model developed in order to find the equilibrium. Therefore, the challenge of this work is to study the auction outcome where theoretical predictions are unknown. The genetic algorithm finds bidding sincerely as the evolutionary-stable strategy with increasing marginal utilities.