Autonomous Algorithmic Collusion: Q-Learning Under Sequential Pricing

37 Pages Posted: 14 Jun 2018 Last revised: 12 Jul 2019

See all articles by Timo Klein

Timo Klein

University of Amsterdam - Amsterdam School of Economics (ASE)

Date Written: July 2019

Abstract

A recent and prominent concern within competition policy and regulation is whether autonomous machine learning algorithms may learn to collude on prices. We show in a simulated environment that when algorithmic competitors update prices sequentially, Q-learning (a simple but well-established self-learning algorithm) coordinates on high fixed-price equilibria or profitable asymmetric price cycles. This occurs even though the algorithm does not communicate and does not receive any instructions to collude. We show that results are robust to changes to the learning parameters and timing and discuss how more advanced algorithms could deal with practical limitations.

Keywords: artificial intelligence, machine learning, reinforcement learning, Q-learning, pricing algorithms, algorithmic collusion, sequential pricing

JEL Classification: L13, L41, D43, D83

Suggested Citation

Klein, Timo, Autonomous Algorithmic Collusion: Q-Learning Under Sequential Pricing (July 2019). Amsterdam Law School Research Paper No. 2018-15; Amsterdam Center for Law & Economics Working Paper No. 2018-05. Available at SSRN: https://ssrn.com/abstract=3195812 or http://dx.doi.org/10.2139/ssrn.3195812

Timo Klein (Contact Author)

University of Amsterdam - Amsterdam School of Economics (ASE) ( email )

Roetersstraat 11
Amsterdam, North Holland 1018 WB
Netherlands

Register to save articles to
your library

Register

Paper statistics

Downloads
324
Abstract Views
1,175
rank
93,544
PlumX Metrics