Evidence from panel unit root and cointegration tests that the Environmental Kuznets Curve does not exist

The Environmental Kuznets Curve (EKC) hypothesis - an inverted U-shape relation between various indicators of environmental degradation and income per capita - has become one of the 'stylised facts' of environmental and resource economics. This is despite considerable criticism on both theoretical and empirical grounds. Cointegration analysis can be used to test the validity of such stylised facts when the data involved contain stochastic trends. In the present paper, we use cointegration analysis to test the EKC hypothesis using a panel dataset of sulfur emissions and GDP data for 74 countries over a span of 31 years. We find that the data is stochastically trending in the time-series dimension. Given this, and interpreting the EKC as a long run equilibrium relationship, support for the hypothesis requires that an appropriate model cointegrates and that sulfur emissions are a concave function of income. Individual and panel cointegration tests cast doubt on the general applicability of the hypothesised relationship. Even when we find cointegration, many of the relationships for individual countries are not concave. The results show that the EKC is a problematic concept, at least in the case of sulfur emissions.