Measurement, Typology, and Multi-Scenario Forecasting of Urban Marginal Abatement Costs

Under China’s “dual-carbon” goals, identifying inter-city differences in carbon marginal abatement costs and assessing their risks are crucial for region-specific policy design and optimal resource allocation. This study develops an integrated framework covering measurement, typology, driver analysis, and scenario forecasting. First, a DEA model is used to measure urban carbon marginal abatement costs. Second, within a high-quality development framework, structural indicators such as industrial structure, energy-use scale, urbanization carrying capacity, and innovation input are selected, and an endogenous city typology is identified through SOM pre-clustering and K-means partitioning, yielding six interpretable city types. Third, a random forest model combined with SHAP is employed to characterize the directional and nonlinear effects of key drivers. Finally, under the United Nations climate governance framework, three scenarios consistent with 1.5°C, 2°C, and 2.5°C warming targets are constructed, and a stacking model is used to forecast carbon marginal abatement costs from 2025 to 2060 and compare their trajectories across city types. The results show that carbon marginal abatement costs exhibit significant typological differences and fat-tail characteristics. An accelerated mitigation pathway consistent with the 1.5°C target is generally more conducive to compressing right-tail risks and promoting faster convergence across city types, whereas higher-warming scenarios are more likely to delay cost pressures and amplify uncertainty. Based on these findings, this study proposes a type-specific cost-reduction policy path centered on “controlling the right tail and promoting convergence,” providing quantitative evidence and policy implications for differentiated urban emission reduction and long-term cost governance.