Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Tacit algorithmic collusion in deep reinforcement learning guided price competition: A study using EV charge pricing game

Diwas Paudel, Tapas K. Das

TL;DR

The study addresses tacit algorithmic collusion in AI-guided pricing for EV charging hubs operating under day-ahead and real-time electricity markets. It introduces a two-step data-driven approach combining a stochastic DA commitment optimization and a MADRL-based pricing model, plus a collusion index to quantify the outcome. In a two-hub Tampa case, results show a collusion index ranging from $0.14$ to $0.45$, indicating low-to-moderate tacit collusion, influenced by the DRL algorithm and network architecture; balking behavior can further modulate this effect. The work provides a practical framework for assessing AI-driven pricing collusion with implications for regulators and market designers, while acknowledging limitations such as coarse time granularity and simplified hub symmetry.

Abstract

Players in pricing games with complex structures are increasingly adopting artificial intelligence (AI) aided learning algorithms to make pricing decisions for maximizing profits. This is raising concern for the antitrust agencies as the practice of using AI may promote tacit algorithmic collusion among otherwise independent players. Recent studies of games in canonical forms have shown contrasting claims ranging from none to a high level of tacit collusion among AI-guided players. In this paper, we examine the concern for tacit collusion by considering a practical game where EV charging hubs compete by dynamically varying their prices. Such a game is likely to be commonplace in the near future as EV adoption grows in all sectors of transportation. The hubs source power from the day-ahead (DA) and real-time (RT) electricity markets as well as from in-house battery storage systems. Their goal is to maximize profits via pricing and efficiently managing the cost of power usage. To aid our examination, we develop a two-step data-driven methodology. The first step obtains the DA commitment by solving a stochastic model. The second step generates the pricing strategies by solving a competitive Markov decision process model using a multi-agent deep reinforcement learning (MADRL) framework. We evaluate the resulting pricing strategies using an index for the level of tacit algorithmic collusion. An index value of zero indicates no collusion (perfect competition) and one indicates full collusion (monopolistic behavior). Results from our numerical case study yield collusion index values between 0.14 and 0.45, suggesting a low to moderate level of collusion.

Tacit algorithmic collusion in deep reinforcement learning guided price competition: A study using EV charge pricing game

TL;DR

The study addresses tacit algorithmic collusion in AI-guided pricing for EV charging hubs operating under day-ahead and real-time electricity markets. It introduces a two-step data-driven approach combining a stochastic DA commitment optimization and a MADRL-based pricing model, plus a collusion index to quantify the outcome. In a two-hub Tampa case, results show a collusion index ranging from to , indicating low-to-moderate tacit collusion, influenced by the DRL algorithm and network architecture; balking behavior can further modulate this effect. The work provides a practical framework for assessing AI-driven pricing collusion with implications for regulators and market designers, while acknowledging limitations such as coarse time granularity and simplified hub symmetry.

Abstract

Players in pricing games with complex structures are increasingly adopting artificial intelligence (AI) aided learning algorithms to make pricing decisions for maximizing profits. This is raising concern for the antitrust agencies as the practice of using AI may promote tacit algorithmic collusion among otherwise independent players. Recent studies of games in canonical forms have shown contrasting claims ranging from none to a high level of tacit collusion among AI-guided players. In this paper, we examine the concern for tacit collusion by considering a practical game where EV charging hubs compete by dynamically varying their prices. Such a game is likely to be commonplace in the near future as EV adoption grows in all sectors of transportation. The hubs source power from the day-ahead (DA) and real-time (RT) electricity markets as well as from in-house battery storage systems. Their goal is to maximize profits via pricing and efficiently managing the cost of power usage. To aid our examination, we develop a two-step data-driven methodology. The first step obtains the DA commitment by solving a stochastic model. The second step generates the pricing strategies by solving a competitive Markov decision process model using a multi-agent deep reinforcement learning (MADRL) framework. We evaluate the resulting pricing strategies using an index for the level of tacit algorithmic collusion. An index value of zero indicates no collusion (perfect competition) and one indicates full collusion (monopolistic behavior). Results from our numerical case study yield collusion index values between 0.14 and 0.45, suggesting a low to moderate level of collusion.
Paper Structure (24 sections, 29 equations, 13 figures, 2 tables, 2 algorithms)

This paper contains 24 sections, 29 equations, 13 figures, 2 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. EV charge pricing game
  3. System model
  4. Transportation network
  5. Electric power network
  6. EV charging hubs
  7. Day-ahead commitment
  8. Competitive dynamic pricing
  9. Hub power management
  10. Solution methodology for the system model
  11. Numerical case study
  12. DRL algorithms and Neural network architectures
  13. Train and test data sets
  14. Implementation of methodology
  15. Results
  16. ...and 9 more sections

Figures (13)

  • Figure 1: Schematic of EV charge pricing game
  • Figure 2: A hub's interaction with transportation and power networks
  • Figure 3: Road intersection in Tampa, Florida, USA (between Hillsborough Avenue and Veterans expressway) considered to generate the EV traffic data. The competing hubs are assumed to be located at this intersection.
  • Figure 4: Probability of an EV on the road to seek charge over the hours of a day deng2018demand.
  • Figure 5: (a) probability distribution for EV state of charge(SOC) in percentage at the start of charging Idaho, and (b) probability of balking by an EV for various price ratios of the hub that is available for charging to the cheapest hub.
  • ...and 8 more figures