Convergence analysis as spatial dynamic panel regression and distribution dynamics of \(\hbox {CO}_{2}\) emissions in Asian countries

This paper examines the conditional convergence of carbon dioxide \((\hbox {CO}_{2})\) emissions in 11 Asian countries over the period of 1972–2009. We allow for spatial dependence in emissions across countries to consider the possibility that country’s emissions are affected by emissions in neighboring countries. We utilize the spatial Durbin model (SDM) and spatial lag model (SAM) estimation for the Green Solow model and also employ distribution dynamics using the nonparametric method with three different \(\hbox {CO}_{2}\) emission samples: per capita \(\hbox {CO}_{2}\), relative \(\hbox {CO}_{2}\), and \(\hbox {CO}_{2}\) intensities. We found that \(\hbox {CO}_{2}\) emissions are significantly influenced by the emissions of neighboring countries. These test results on SDM for per capita \(\hbox {CO}_{2}\) and SAM for \(\hbox {CO}_{2}\) intensities do not support the convergence hypothesis. However, when we control for both the individual and time-period fixed effects, for relative \(\hbox {CO}_{2}\) and \(\hbox {CO}_{2}\) intensities, SAM supports the convergence hypothesis. With the distributional dynamics approach adapting spatial-filtering and HP-filtering techniques before proceeding with the estimate of the stochastic kernel, we find that per capita \(\hbox {CO}_{2}\) and \(\hbox {CO}_{2}\) intensities indicate a strong tendency toward convergence in Asian countries over the entire period, yet relative \(\hbox {CO}_{2}\) shows dynamics with a switch toward divergence. In conclusion, the \(\hbox {CO}_{2}\) emission intensities among Asian countries strongly support the convergence to the long-run steady state of \(\hbox {CO}_{2}\) emissions as Asian economies develop. However, relative \(\hbox {CO}_{2}\) emissions in Asian countries do not converge with the long-run steady-state level of developed countries as sustainable growth paths are predicted from the theoretical Solow model.