Infrastructure, institutions, and the resilience dividend: A hybrid Econometric–Machine learning analysis of CO2 emissions in Central and Eastern Europe

CO 2 emissions in Central and Eastern Europe (CEE) highlight an institutional deficit, as rapid urbanization and fossil-dependent infrastructure intensify environmental pressures under uneven governance, impeding EU climate targets. This study examines the direct marginal effects of urbanization pressure and institutional capacity on CO 2 emissions; investigates infrastructure intensity versus economic capacity; and quantifies impacts of economic openness, civic engagement, and environmental policy enforcement. Utilizing annual panel data for 18 CEE countries from 2000–2024 sourced from World Bank, we apply ARDL bounds testing for cointegration and dynamics, fixed effects for robustness, and double machine learning (DML) with SHAP for causal attribution and counterfactual simulations, addressing non-linearity and endogeneity. Results indicate infrastructure intensity decisively elevates emissions (β = 1.430, p<0.001 long-run), while economic capacity strongly reduces them (β = -0.695, p<0.001), validating institutionally mediated decoupling; government effectiveness curbs emissions (β = -0.216, p<0.001), but civic engagement exhibits a positive association, potentially endogenous (β = 0.325, p<0.001) simulations reveal a 45.8% emissions drop from 1-SD infrastructure reduction. Findings extend institutional theory by operationalizing a reactive 'resilience dividend' and inform scalable policies: prioritize grid decarbonization and adaptive governance to offset urbanization in transitional economies.