工业企业低碳转型是我国落实“双碳”目标的迫切要求，基于碳市场等政策的碳定价机制是激励企业自主减排的主要手段，但低碳转型中企业的复杂决策机制及其对激励型政策的非线性响应使得低碳转型过程模拟面临新的挑战。识别工业系统低碳转型过程中的关键影响因素与作用机制，将能够为碳市场政策设计和实施优化提供参考，具有重要的理论意义与实践价值。本研究运用多主体建模方法（Agent-Based Modeling，ABM），综合技术-组织-环境框架及创新决策理论的知晓、说服、决策三阶段，整合前景理论、小世界网络模型与技术传播理论，建立了以工业企业低碳技术采纳行为为核心的工业系统低碳转型模型（Agent-Based Modeling of low-carbon Technology Adoption Behavior Simulation，ABM-TABS模型），深入刻画了企业的异质性特征、有限理性、不完全信息和企业互动，量化了企业对碳定价和市场交易机制的响应，并以连云港电力、钢铁、水泥三大行业作为案例，模拟并验证了企业在技术决策中的异质性参数设置，分析评估了企业行为特征及企业主观认知、企业网络结构影响，进而将ABM-TABS模型扩展到全国钢铁行业，对多种低碳政策情景进行了模拟分析。研究发现，低碳转型过程中企业行为存在有限理性和不完全信息的特征，考虑这些特征将使得相对误差分别降低4.8%和5.3%；企业的有限理性决策特征将使得其在2035年时的碳排放平均提高7.2%，企业对碳价波动所可能带来损失的顾虑是阻碍低碳转型的主要因素；反映企业时间偏好的折现率对企业行为的影响最为明显，且整体上大规模、高碳排放水平的企业更倾向于开展低碳行动；企业的不完全信息模式将使得低碳转型过程更为滞后，但长期来看最终的碳减排效果不会有显著变化；企业与其他规模、产量等生产特征相似及相差较大的企业均有互动时更可能打破不完全信息的阻碍，网络结构中邻居数量的提升也将促进技术传播，分别能使得碳排放降幅增加1%和4%；技术的外部传播与内部传播都能促进低碳转型过程，分别使得碳排放降幅增加10%和3%，且技术传播的早期外部传播影响明显，后期则内部传播影响增大；碳价波动会通过损失规避效应阻碍企业减排、通过碳价峰值效应促进企业减排，且波动较低时前者更为明显，反之则为后者；收紧碳配额和增强市场活跃度分别使得碳排放降幅增加了19%和32%，后者影响更为明显，且过紧的碳配额将使得市场作用无法充分发挥。
Low-carbon transition of industrial firms is an urgent requirement for the achievement of carbon peak and neutrality goals in China. Carbon pricing mechanism based on carbon market is the main policy to incentivize industrial firms to cut their carbon emissions, but firms’ complex decision-making processes and non-linear response to incentive-based policy make the simulation of low-carbon transition process face new challenges. Identifying key factors and decision-making mechanisms during industrial low-carbon transition, thus simulating and analyzing firms’ low-carbon behaviors and carbon reduction under different policy scenarios will be valuable for policy optimization, and of great theoretical significance and practical value.In this context, with the approach of Agent-Based Modeling (ABM), this study integrates Technology-Organization-Environment Framework (TOE) and the three stages of knowledge, persuasion, and decision in Innovation Decision Process (IDP), combined with prospect theory, small-world network, and technology diffusion theory, to establish an Agent-Based Modeling of low-carbon Technology Adoption Behavior Simulation (ABM-TABS Model). This model characterizes heterogeneous characteristics of industrial firms, as well as firms’ limited rationality, incomplete information and interactions, and quantifies firms’ response to carbon pricing mechanism and carbon market. Taking power, iron and steel, and cement sectors in Lianyungang municipality as a case, the study simulates and validates the heterogeneity parameter settings of industrial firms in technology adoption decision-making, analyze firms’ behavioral change as well as the impacts of firms‘ cognition and interaction. Then the study extends the ABM-TABS model to iron and steel firms across China to conduct scenario analysis of different carbon policies, so as to provide suggestions for the design of carbon policy.It is found that the firms’ behavior in the process of low-carbon transition is characterized by limited rationality and incomplete information, while taking it into consideration will reduce relative errors of the simulation by 4.8% and 5.3%, respectively. Decision-making with limited rationality of industrial firms will impede the process of low-carbon transition, and the firms’ concern about the possible losses caused by fluctuation of carbon price is the main factor hindering the low-carbon transition. Discount rate that reflects time preference of firms has the most obvious influence on firms’ behavior, and large-scale firms as well as firms with higher carbon emission intensity would carry out more low-carbon actions. Incomplete information of industrial firms will delay low-carbon transition process, but in the long run, the effect will be minor. Firms will break the obstacle of incomplete information most when they interact with both similar and different firms in the aspect of production and carbon emission, and increasing of neighbors in network will also promote technology diffusion. Both external and internal diffusion of technology can promote low-carbon transition, while the external diffusion has greater impact, but with diminishing marginal effect. Fluctuation of carbon price will impede emission reduction through loss-avoidance effect, while promote emission reduction through peak price effect, and influence of the former is more obvious when the extent of fluctuation is small. Tightening carbon quotas and enhancing market activity can both promote low-carbon transition, but the latter has more significant effect, and a policy design with very tight carbon quotas will overshadow the effect of market mechanism. Government should set moderate carbon quotas and prioritize the promotion of market activity.
