Drivers of carbon dioxide emissions: an empirical investigation using hierarchical and non-hierarchical clustering methods

The mitigation of CO₂ emissions requires a global effort with common but differentiated responsibilities. In this paper, we identify clusters of CO₂ emissions across 72 countries. First, using the stochastic version of the IPAT and employing the dynamic common correlated effects technique, we identify three key determinants affecting CO₂ emissions (non-renewables, population, and real GDP). In the second step, both hierarchical and non-hierarchical clustering methods are considered to identify the optimal number of clusters. We identify two to four clusters with different member countries, and in particular establish that in most cases, a 2-cluster solution appears to be optimal. The contents of clusters vary slightly according to the clustering methods for each period. The clustering results from using only the overall CO₂ emissions indicate that the countries we consider form three clusters, with China and the USA each within a single member cluster. The remaining 70 countries form the third cluster. Our findings reflect the prominent roles of China and the USA in overall CO₂ emissions. Analyses with sub-period and largest emitters reflect a different clustering structure. Some policy recommendations in setting emission reductions are made, considering different clusters across countries.