Revisiting heterogeneity in the relationship between CO2 per capita emissions and income per capita

This study applies various methods to deal with sources of potential misspecification in the drivers of CO₂ per capita emissions within an environmental Kuznets curve (EKC) framework. The proposed methodologies are as follows: (i) Bayesian model averaging analysis as a remedy for omitted variable bias due to model uncertainty, (ii) convergence club analysis to endogenously classify a large size of panel of countries as a remedy for unobserved heterogeneity, (iii) inclusion of lagged regressors as a remedy for simultaneity. The empirical findings show that the EKC is holding for all except for the first group with the highest (or fastest) long-term income trajectories. Moreover, one observes a clear negative impact of renewable energy consumption, a general positive impact of financial development, and no impact of institutions on emissions. Finally, the findings highlight the importance of constructing homogeneous country clusters.