Presentation Information
[2Yin-A-19]An Analysis of the CO2 Emissions Trading Market Using an Artificial Market with LLM Agents
〇Shin Futamata1,2 (1. Tokyo Metropolitan University, 2. Nikko Research Center, Inc.)
Keywords:
Artificial Market,Agent-Based Simulation,CO2 Emissions Trading,Genetic Algorithm (GA),Large Language Models (LLM)
In Japan, a carbon dioxide (CO2) emissions trading scheme was introduced on a pilot basis in fiscal year 2023 and is scheduled to begin full-scale operation in fiscal year 2026. While emissions trading in Europe has a long history and high trading volumes, expanding CO2 emissions trading remains a major challenge in Japan.
This study analyzes the CO2 emissions trading market using a multi-agent framework. Emissions firms, carbon absorption firms, market makers, and the exchange are modeled as agents, and a genetic algorithm (GA) is used to estimate agent- and market-level parameters and to construct an artificial market.
In addition, generative artificial intelligence based on large language models (LLMs) is incorporated into emissions and absorption firm agents to generate order proposals reflecting exogenous news.
The analysis yields three main findings: (1) market maker entry increases trading volume; (2) the magnitude of this increase depends on market maker trading strategies; and (3) incorporating news affects both firm behavior and market prices.
This study analyzes the CO2 emissions trading market using a multi-agent framework. Emissions firms, carbon absorption firms, market makers, and the exchange are modeled as agents, and a genetic algorithm (GA) is used to estimate agent- and market-level parameters and to construct an artificial market.
In addition, generative artificial intelligence based on large language models (LLMs) is incorporated into emissions and absorption firm agents to generate order proposals reflecting exogenous news.
The analysis yields three main findings: (1) market maker entry increases trading volume; (2) the magnitude of this increase depends on market maker trading strategies; and (3) incorporating news affects both firm behavior and market prices.