Increasingly, algorithms are supplanting human decision-makers in pricing goods and services. To analyze the possible consequences, we study experimentally the behavior of algorithms powered by Artificial Intelligence ( Q-learning) in a workhorse oligopoly model of repeated price competition. We find that the algorithms consistently learn to charge supracompetitive prices, without communicating with one another. The high prices are sustained by collusive strategies with a finite phase of punishment followed by a gradual return to cooperation. This finding is robust to asymmetries in cost or demand, changes in the number of players, and various forms of uncertainty. (JEL D21, D43, D83, L12, L13) Software programs are increasingly being adopted by firms to price their goods and services, and this tendency is likely to continue. 1 In this paper, we ask whether pricing algorithms may "autonomously" learn to collude. The possibility arises because of the recent evolution of the software, from rule-based to reinforcement learning programs. The new programs, powered by Artificial Intelligence (AI), are indeed much more autonomous than their precursors. They can develop their pricing strategies from scratch, engaging in active experimentation and adapting to changing environments. In this learning process, they require little or no external guidance.In the light of these developments, concerns have been voiced, by scholars and policymakers alike, that AI pricing algorithms may raise their prices above the competitive level in a coordinated fashion, even if they have not been specifically

This paper studies the exchange of information between two principals who contract sequentially with the same agent, as in the case of a buyer who purchases from multiple sellers. We show that when (a) the upstream principal is not personally interested in the downstream level of trade, (b) the agent's valuations are positively correlated, and (c) preferences in the downstream relationship are separable, then it is optimal for the upstream principal to offer the agent full privacy. On the contrary, when any of these conditions is violated, there exist preferences for which disclosure is strictly optimal, even if the downstream principal does not pay for the information. We also examine the effects of disclosure on welfare and show that it does not necessarily reduce the agent's surplus in the two relationships and in some cases may even yield a Pareto improvement.

No abstract