Re-examining CO2 emissions. Is the assessment of convergence meaningless?

This paper re-examines CO2 emissions in 22 OECD countries over the period 1870–2006. It contributes to the field of environmental economics trying to clarify the possible sources of the mixed evidence on CO2 emissions convergence. To this end we employ a detailed methodological strategy. First we start with standard linear tests as the ones proposed by Ng and Perron (2001). Then, using the Lee and Strazicich (2003) tests, we take into account the possible existence of structural breaks in the series. Finally, we apply a non-linear test within a smooth transition autoregressive (STAR) framework proposed by Kapetanios et al. (2003). The empirical evidence provided by our methodological strategy suggests that the original per capita CO2 emissions for the largest span, from 1870 to 2006, are stationary, so that to continue with the assessment of convergence in this context would not be adequate. However if we consider instead the period 1950-2006, per capita CO2 emissions are in a non-stationary local regime. Thus, in this case we proceed with the study of convergence. Bearing in mind plausible nonlinearities, CO2 emissions convergence is assessed using two versions of the Kapetanios et al. (2003) test, and conclude that there is no robust convergence among these 22 OECD countries.