Can Price Algorithms Learn to Form a Cartel?
Speakers
Arnoud den Boer
Associate Professor, Korteweg-de Vries Institute for Mathematics, University of Amsterdam
Arnoud is Associate Professor at the Korteweg-de Vries Institute for Mathematics of the University of Amsterdam. He studied Mathematics at Utrecht University (2006), Mathematics for Industry at Eindhoven University of Technology (2008) and wrote his PhD thesis `Dynamic Pricing and Learning’ (2013) about data-driven price algorithms at the CWI Centrum for Wiskunde and Computer Science in Amsterdam. Arnoud’s research focuses on the interface of learning and optimization, with applications in dynamic pricing and revenue management. He is the recipient of several awards and grants, including the 2015 Gijs de Leve prize for best PhD Thesis in operations research defended in the Netherlands in the period 2012-2014, personal grants from the Dutch Science Foundation, and the INFORMS Revenue Management & Pricing Section Prize. Arnoud serves as editor for Management Science, M&SOM, and POMS, is board member of the Euro Working Group on Pricing and Revenue Management and board member of the INFORMS Revenue Management and Pricing Section.
| Date: |
| Monday, 11 March 2024 |
| Time: |
| 10:00 am - 11:30 am |
| Venue: |
|
NUS Business School Mochtar Riady Building BIZ1 0304 15 Kent Ridge Drive Singapore 119245 (Map) |
Abstract
Can price algorithms learn to form a cartel instead of compete against each other, potentially leading to higher consumer prices and lower social welfare? The question is controversial among economists and competition policy regulators. One the one hand, concerns have been expressed that self-learning price algorithms do not only make it easier to form price cartels, but also that this can be achieved within the boundaries of current antitrust legislation – raising the question whether the existing competition law needs to be adjusted to mitigate undesired algorithmic collusion. On the other hand, a number of economists believe that algorithms learning to collude is science fiction, except by using forms of signaling or communication that are already illegal, and argue that there is no need to change antitrust laws. Motivated by this discussion, I will present work that shows that under some market conditions, price algorithms can learn to collude.
Based on joint work with Janusz Meylahn, Thomas Loots, Maarten Pieter Schinkel, Ali Aouad.