Autonomous algorithmic collusion: economic research and policy implications

Assad, Stephanie; Calvano, Emilio; Calzolari, Giacomo; Clark, Robert; Denicolò, Vincenzo; Ershov, Daniel; Johnson, Justin; Pastorello, Sergio; Rhodes, Andrew; Xu, Lei; Matthijs, Wildenbeest

doi:10.1093/oxrep/grab011

Cited by 18 publications

(7 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tacit collusion is a well-known issue in oligopolistic markets Ivaldi et al (2003) and Tirole (1988), and may emerge in repeated auctions without explicit information sharing (Han, 2021;Skrzypacz & Hopenhayn, 2004). The possibility that tacit collusion may emerge from the interaction of autonomous algorithms learning from market data in a decentralized fashion, a situation sometimes referred to as "algorithmic collusion" (Assad et al, 2021;Han, 2021), has attracted the concerns of market regulators (Competition & Markets Authority, 2021). It is thus of interest to investigate the present work we investigate whether the tacit collusion exhibited by Calvano et al may also arise in the setting of financial markets with market makers competing for two-sided (buy and sell) order flow.…”

Section: Introductionmentioning

confidence: 99%

Dynamics of market making algorithms in dealer markets: Learning and tacit collusion

Cont

Xiong

2023

Mathematical Finance

View full text Add to dashboard Cite

The widespread use of market‐making algorithms in electronic over‐the‐counter markets may give rise to unexpected effects resulting from the autonomous learning dynamics of these algorithms. In particular the possibility of “tacit collusion” among market makers has increasingly received regulatory scrutiny. We model the interaction of market makers in a dealer market as a stochastic differential game of intensity control with partial information and study the resulting dynamics of bid‐ask spreads. Competition among dealers is modeled as a Nash equilibrium, while collusion is described in terms of Pareto optima. Using a decentralized multi‐agent deep reinforcement learning algorithm to model how competing market makers learn to adjust their quotes, we show that the interaction of market making algorithms via market prices, without any sharing of information, may give rise to tacit collusion, with spread levels strictly above the competitive equilibrium level.

show abstract

Section: Introductionmentioning

confidence: 99%

Dynamics of market making algorithms in dealer markets: Learning and tacit collusion

Cont

Xiong

2023

Mathematical Finance

View full text Add to dashboard Cite

show abstract

“…Margrethe Vestager, the EU Commissioner for Competition, commented in 2018 that "The challenges that automated systems create are very real … If they help companies to fix prices, they really could make our economy work less well for everyone else" (quoted inHirst, 2018). For commentary seeHarrington (2018),Schwalbe (2018),Assad et al (2021), and Veljanovski (2022).7 This is often referred to as "Q-learning." However, in the machine learning (or AI) literature, Q-Learning tends to have a broader meaning that subsumes both asynchronous and synchronous leaning.…”

mentioning

confidence: 99%

“…Similarly,Normann and Sternberg (2021) provide related experimental evidence.15 In the Bertrand game case, this would converge (under some standard conditions) to the Nash equilibrium of the game.16 Another point of distinction in that, in our setting, players do not "learn" about the probability distribution of strategies chosen by other players. Rather, what they have in memory is only the value associated with every action.17 Examples of contributions directed at this policy debate includeMehra (2015),Ezrachi and Stucke (2017),Kühn and Tadelis (2017),Schwalbe (2018), de Corniere and Taylor (2020),Assad et al (2021), andVeljanovski (2022) Goldfarb et al (2019). provide a broader overview of the likely impact of AI on the economy at large.18 We assume such a structure purely for computational convenience.…”

mentioning

confidence: 99%

“…Margrethe Vestager, the EU Commissioner for Competition, commented in 2018 that “The challenges that automated systems create are very real … If they help companies to fix prices, they really could make our economy work less well for everyone else” (quoted in Hirst, 2018). For commentary see Harrington (2018), Schwalbe (2018), Assad et al (2021), and Veljanovski (2022).…”

mentioning

confidence: 99%

“…1717Examples of contributions directed at this policy debate include Mehra (2015), Ezrachi and Stucke (2017), Kühn and Tadelis (2017), Schwalbe (2018), de Corniere and Taylor (2020), Assad et al (2021), and Veljanovski (2022). Goldfarb et al (2019) provide a broader overview of the likely impact of AI on the economy at large.…”

mentioning

confidence: 99%

See 2 more Smart Citations

The impact of artificial intelligence design on pricing

Asker

Fershtman

Pakes

2023

Economics Manag Strategy

View full text Add to dashboard Cite

The behavior of artificial intelligence (AI) algorithms is shaped by how they learn about their environment. We compare the prices generated by AIs that use different learning protocols when there is market interaction. Asynchronous learning occurs when the AI only learns about the return from the action it took. Synchronous learning occurs when the AI conducts counterfactuals to learn about the returns it would have earned had it taken an alternative action. The two lead to markedly different market prices. When future profits are not given positive weight by the AI, (perfect) synchronous updating leads to competitive pricing, while asynchronous can lead to pricing close to monopoly levels. We investigate how this result varies when either counterfactuals can only be calculated imperfectly and/or when the AI places a weight on future profits. Lastly, we investigate performance differences between offline and online play.

show abstract

AI pricing algorithms under platform competition

Sanchez-Cartas,

Katsamakas

2024

Electron Commer Res

View full text Add to dashboard Cite

Platforms play an essential role in the modern economy. At the same time, due to advances in artificial intelligence (AI), algorithms are becoming more widely used for pricing and other business functions. Previous literature examined algorithmic pricing, but not in the context of network effects and platforms. Moreover, platform competition literature has not considered how algorithms may affect competition. We study the performance of AI pricing algorithms (Q-learning and Particle Swarm Optimization) and naïve algorithms (price-matching) under platform competition. We find that algorithms set an optimal price structure that internalizes network effects. However, no algorithm is always the best because profitability depends on the type of competing algorithms and market characteristics, such as differentiation and network effects. Additionally, algorithms learn autonomously when an equilibrium is unstable and avoid it. When algorithm adoption is an endogenous strategic decision, several algorithms can be adopted in equilibrium; we characterize the conditions for the various outcomes and show that the equilibrium and platform profits are sensitive to algorithm design changes. Overall, our research suggests that AI algorithms can be effective in the presence of network effects, and platforms are likely to adopt a variety of algorithms. Lastly, we reflect on the business value of AI and identify opportunities for future research at the intersection of AI algorithms and platforms.

show abstract

Autonomous algorithmic collusion: economic research and policy implications

Cited by 18 publications

References 19 publications

Dynamics of market making algorithms in dealer markets: Learning and tacit collusion

Dynamics of market making algorithms in dealer markets: Learning and tacit collusion

The impact of artificial intelligence design on pricing

AI pricing algorithms under platform competition

Contact Info

Product

Resources

About