An agent-based approach to modeling power firms' emission reduction strategies and market dynamics

This paper develops an agent-based model to study the impact of China's national carbon market on the decision-making behaviors of power generation firms by accounting for their multidimensional heterogeneity. The multi-agent deep deterministic policy gradient algorithm is employed to optimize firm strategies in production, low-carbon technology adoption, and allowance trading. An application study is conducted by using the real data from over 3000 Chinese power firms and the real price data from the China national carbon market. The modeling results show that tightening emission reduction targets leads to higher carbon and power prices, greater renewable energy generation, and increased adoption of low-carbon technologies. Additionally, our study highlights the critical role of allowance allocation methods, with auction-based rule providing more stable and higher carbon price signals that incentivize earlier emission reductions. The research also identifies a disparity between large-scale and small-medium-scale firms in terms of participation in allowance trading and low-carbon technology adoption, with larger firms leading in both areas. The findings offer valuable insights for enhancing the cost-effectiveness and incentive mechanisms of carbon market.