Algorithmic collusion in a price oligopoly

This thesis demonstrates algorithmic collusion in a price oligopoly with multinomial logit demand. Of main interest is the scenario where competing firms independently adopt copies of the same price algorithm. One of the key takeaways is that in this scenario, major challenges to achieving and sustaining supra-competitive outcomes can be overcome by the firms. We show this by constructing competitive price algorithms that have collusive capabilities when at least one of the opponents uses a copy of the same algorithm. Importantly, the firms do not have to activate the algorithm at the same time, do not need to use the same input parameters, and do not need to know that they are using the same algorithm. Moreover, the proposed algorithms do not engage in illicit forms of communication or signaling, and therefore exemplify tacit (lawful) algorithmic collusion. This thesis thus contributes to the ongoing debate on the extent to which collusion by algorithms is possible within the confines of existing antitrust law and jurisprudence, and further contributes to the fields of axiomatic bargaining, dynamic pricing, and demand learning in the presence of competitors.