Nonlinearities in the road freight–transport CO2 relationship in Europe

This study examines whether the relationship between road freight activity and transport CO 2 emissions changes as road freight activity expands and across the conditional emissions distribution in 18 European countries from 1999 to 2024. Method of Moments Quantile Regression (MMQR) is complemented by Fully Modified Ordinary Least Squares, Driscoll–Kraay fixed-effects estimation, and XGBoost–SHAP analysis. MMQR results indicate a U-shaped association: the linear freight term is negative and statistically significant across all quantiles, whereas the squared term is positive and significant. The estimated turning points lie within the observed freight range and shift downward toward higher emission quantiles, suggesting that the marginal association turns positive at lower freight levels when conditional emissions are relatively high. Motor vehicle production and economic development are positively associated with transport emissions, whereas urbanization is negatively associated across the distribution. Trade openness shows a more conditional relationship, becoming increasingly negative toward the upper quantiles. Alternative estimators preserve the nonlinear freight pattern, while XGBoost–SHAP identifies road freight as the dominant predictor and reveals corresponding nonlinear predictive structure. Overall, the findings suggest that freight decarbonization policies should account for both the scale of road freight activity and prevailing emissions conditions.