Algorithmic pricing in Economics (and especially in Industrial Organization) is not new. Nonetheless, there is a renewed interest in this discipline because of the extensive use of this tool by web-based markets, such as Amazon, eBay, or Uber.
This section only reflects some of the most interesting articles of this growing literature. Given the recent interest, I may miss some works. Thus, any suggestion is more than welcome.
This is one of the first appearances of Q-learning in Industrial Organization. It is a pioneering work that showed a potential area to explore, the integration of artificial intelligence with Economics, but it was quite an early work from economists' perspectives..
A little bit different in scope, this work addresses the impact of price setting algorithms but under a different approach than the previous ones. In this case, instead of using a Reinforcement Learning approach (e.g. Q-learning), they use an Evolutionary Algorithm, specifically, a Particle Swarm Optimization algorithm (PSO), which may be more flexible to be implemented in reality, but it has not been well explored yet. This work provides a good introduction and the pseudo-code to implement it in other works.
Although it is not the first work in highlighting the possibility of collusion with algorithmic pricing, it is a good introduction to understand the renewed interest in this area. I recommend this work to spark curiosity on the topic.
An updated description of possible ways to address collusive practices derived from algorithms, as well as a detailed discussion of the possibility of such an event. Highly recommended taxonomy of possible adverse effects of pricing algorithms.
A detailed account identifying the potential harms of algorithmic pricing with a special focus on collusion. This is among the first policy papers coming from a competition authority.
A discussion around law, economics, and computer science literature that highlights the potential limitations of traditional competition policies
This paper contrasts with others highlighted in this section as it shows that algorithms can lead to lower prices and higher consumer surplus.
An empirical article showing that platforms may be incentivized to bias recommendations (which can be detrimental to users). A key observation is that: While concepts related to algorithmic recommendations (e.g., data mining, machine learning) may sound neutral, the key parameters and objective functions of algorithms are chosen by human managers or developers.
Clearly, this is the work that is leading the development of the use of artificial intelligence and industrial organization nowadays. The authors make an outstanding work by proving the hypothesis of a collusive outcome with a simple Q-learning algorithm. If you are an economist working in this area, not knowing about this work is almost a crime.
Based on the previous one, this work is a working paper now, but it also deserves mention because it passes on the baton to a new arena. Algorithmic pricing is not just collusion. It is a new kind of behavior that we should take into account when designing market rules.
Following Calvano, et al. Klein similarly addresses the same issue. It is another outstanding work that sheds more light on the current debate about the role of algorithms in collusion.
A thoughtful essay where the author discusses how unregulated AI can produce political and economic harm (e.g. damaging competition, consumer privacy and consumer choice; excessively automating work, fueling inequality, inefficiently pushing down wages, and failing to improve worker productivity; and damaging political discourse).
What if we find no collusion but algorithms set supra-competitive prices? It may be possible that algorithms are neither playing Nash equilibria nor collusive ones? This is the starting point of Asker, et al., who observe that algorithmic pricing is compatible with a different solution concept, experience-based equilibria. This work is the first one in providing a satisfactory answer to the supra-competitive prices of Q-learning far beyond collusion.
A survey about the growing literature on the economic effects of the recent technological advances in AI. The focus is mainly on labor markets, consumer behavior and competition. A good place to start if you are new to the literature.
A paper that combines theory and data to find that high-frequency algorithmic pricing may lead to higher prices than non-algorithmic players.
Since this literature is growing fast and in multiple directions, there are many notable works that I did not include. With time, I will focus the literature on more specific aspects.