SummaryWe study the structural differences among climate change leading 'factors' -Northern EU members -, and lagging actors -southern EU countries and the 'Umbrella group' -with regard to long run carbon-income relationships. Homogeneous and heterogeneous panel models show that the groups of countries less in favour of stringent climate policy have yet to experience a Kuznets curve, though they show relative delinking. Northern EU instead robustly shows bell shapes. Exogenous policy events such as the 1992 climate change convention appear to be relevant in shaping the EKC of Northern EU. In addition, other events such as the second oil price shock appear to have also impacted in shaping the long run emission/GDP dynamics. (OECD, 2002; EEA, 2003). Stylised facts have been proposed on the relationship between pollution and economic growth, which became know as the Environmental Kuznets Curve (EKC) hypothesis, that has gained an increasing research attention over time since the pioneering works of Grossman and Krueger (1995), Shafik (1994) and Holtz-Eakin and Selden (1992).
KeywordsApplied EKC investigations mainly focus on emissions into the air, although evidence for other types of emissions and pollutants, such as waste, has been emerging. In this paper we focus on CO 2 emissions which have been recognised as a major source of environmental pollution (Schmalensee et al., 1998), and offer the most robust data for applying advanced econometric techniques. The relevance on carbon is also depending on the fact that if one the one hand absolute delinking have been experienced and verified in the literature for local and regional air and water emissions, CO 2 (and waste generation)are environmental impacts that have not shown clear and robust EKC shapes, if not for specific countries and sectors in advanced economies i . Even in advanced economies nevertheless the evidence is far from assessing a neat absolute delinking overall (Cole, 2003, Stern, 2004 Dinda, 2004;Musolesi et al., 2009). Decoupling between income growth and CO 2 emissions is not (yet) apparent for many important world economies, and where it is observed, it is relative rather than absolute as usually assumed by the EKC hypothesis.This paper aims to contribute to the development of EKC research in two main directions. First, we use modern econometric panel approaches capable of providing new evidence on EKC long run dynamics. We employ recent homogeneous estimators -as those derived from panel cointegration analysis or those that explicitly take into account cross section correlation -as well as heterogeneous estimators which allow individual slopes to be derived from sampling or Bayesian approaches. It is difficult a priori to decide between homogeneous and heterogeneous panel estimators. On the one hand, the increasing time dimension means that the slope homogeneity implicit in the use of a pooled estimator is questionable. On the other hand, most researchers agree about the use of homogeneous estimators since the efficiency gains from pooling ofte...