Agent-based modeling of firm's heterogeneous preferences: Implications for trading and technology adoption in electricity-carbon markets

Firm preference and behavior significantly influence carbon market performance and emission reduction efficiency. This paper develops an electricity‑carbon coupled market model, integrating Agent-Based Modeling (ABM) and Multi-Agent Reinforcement Learning (MARL), to analyze the trading and investment behavior of heterogeneous firms in carbon markets. Using calibrated data from China's energy markets, the study uncovers four key findings. First, “compliance-based trading” behavior drives carbon price and volume surges near compliance deadlines. Second, large firms, with financial and technical advantages, act as first movers in adopting low-carbon technologies. Third, longer payback period reduces carbon prices but promote technological innovation, while stronger technology-oriented expectations boost trading activity, prices, and innovation. Fourth, trading preferences shape market outcomes: arbitrage firms increase short-term price volatility, risk-hedge firms stabilize markets and lead in innovation, and speculative firms strike a balance between price impacts and moderate innovation. Policy recommendations include extending payback period to ease financial pressures and encourage technology diffusion, leveraging large firms' resources while supporting smaller firms with fiscal incentives, and regulating arbitrage behavior during compliance periods to stabilize carbon markets.