Synergistic emission reductions of CO₂ and air pollutants in Chinese cities: Evidence from SHAP-enhanced machine learning and fsQCA

Synergistic reduction of CO₂ and air pollutant emissions (CAPE) is essential for achieving China's carbon and air-quality goals. Given regional heterogeneity, reducing CAPE necessitates the consideration of multidimensional determinants. Although prior studies have examined the effects of various determinants, few have systematically addressed the configurational complexity that arises from their non-linear interplay. To bridge this gap, this study integrates SHAP-enhanced machine learning and fuzzy-set qualitative comparative analysis (fsQCA) to uncover the complex configurations driving synergistic reductions of CAPE. Drawing on the technology-organization-environment (TOE) framework, we identify key determinants for CO₂, PM₂.₅, SO₂, PM₁₀, NO₂, O₃, and CO across Chinese cities from 2003 to 2023. Results show: 1) Technological factors, particularly green innovation and industrial upgrading, most effectively reduce CO₂ and SO₂ and correlate with synergistic CO₂–PM₂.₅ and CO₂–SO₂ reductions; meanwhile, organizational factors, especially population density, dominate PM₂.₅ reduction and synergistic PM₂.₅–SO₂ and CO₂–PM₂.₅–SO₂ reductions. This shows that determinants that are more influential for a certain emission also tend to yield stronger influence for other emissions. 2) Interaction effects among the most influential factors generate the strongest synergistic reductions. The most substantial interactions include green innovation with digitalization for CO₂ reduction, population density with air temperature for PM₂.₅ reduction, and green innovation with digitalization for SO₂ reduction. 3) FsQCA identifies several distinct reduction configurations for CO₂, PM₂.₅, SO₂, and PM₁₀, but reveals no comparable configuration diversity for NO₂, O₃, or CO. Further, carbon and air pollution reduction configurations exhibit asymmetric synergy: a subset of carbon-reduction configurations also lowers PM₂.₅ or SO₂, whereas the configurations that reduce PM₂.₅ or SO₂ exhibit limited spillover to CO₂. However, the reduction configurations for PM₂.₅ and SO₂ display mutual synergy. This multi-dimensional framework offers a methodological reference for environmental governance in heterogeneous cities.