Do emissions and income have a common trend? A country-specific, time-series, global analysis, 1970–2008

This paper uses Vector Autoregressions that allow for nonstationarity and cointegrationto investigate the dynamic relation between income and emissions in the period1970-2008, for all world countries. We consider three emission variables over the years 1970-2008 taken from the EDGARv4.2 database, namely CO2, SO2 and a composite global warming index (denoted GWP100) in which all Kyoto-protocol greenhouse chemical compounds are converted to units of tonnes CO2-equivalent with the standard UNFCCC 100-year weighting factors. EDGAR-v4.2 is the greenhouse gas and air pollutant emissions database that provides consistent global estimates and covers the full IPCC emissions category set. These emissions include energy-relatedactivities with a share varying from 60% (GWP100) to almost 90% (SO2). For all chemicalcompounds, it is found that for over two thirds of cases income and emissions are driven byunrelated random walks with drift, at 5% significance level. For one quarter of the cases thevariables are found to be driven by a common random walk with drift. Finally, for the remaining4.5% of cases the variables are trend-stationary. Tests of Granger-causality show evidenceof both directions of causality. For the case of unrelated stochastic trends, one finds a predominanceof emissions causing income (in growth rates), which accords with a production-functionrather than with a consumption-function interpretation of the emissions-income relation. Theevidence challenges the main implications of the Environmental Kuznets Curve hypothesis,namely that the dominant direction of causality should be from income to emissions, and thatfor increasing levels of income, emissions should tend to decrease.JRC.H.2-Air and Climat