Long-run determinants of atmospheric CO2: Granger-causality and cointegration analysis

Abstract

Atmospheric concentrations of CO2 grew annually 1.12 +- 0.48 parts per million (ppm) in 1958-1984, and 1.72 +- 0.54 ppm (mean +- standard deviation) in 1985-2009, so that the rate growth is growing itself. Natural phenomena that influence short-run changes in CO2 atmospheric levels (through their influence on CO2 emissions and sinks) are stationary processes that cannot explain the growth of CO2 levels at an increasing rate. Cointegration tests show at a high level of statistical significance that the annual increase of CO2 concentrations is roughly proportional to “human activities” as measured by the money value of the world economy and the size of the world population. We find that population and world GDP help to predict CO2 concentrations, but CO2 concentrations do not help to predict the othervariables; that is, there is Ganger causality from population and world economic output to CO2. Though the smallness of the time series involved and the theoretical and practical issues posed by cointegration allow only for a limited confidence in these results, they have obvious major implications. For business-as-usual conditions and a world economy growing annually 3.5%—the mean annual growth of the world economy since 1960—the required world population to maintain or reduce CO2 levels would be 1.3 billion or less. For a world population of 7 billion as the present one, CO2 atmospheric levels would decrease if the global economy contracted annually 24.5% or more.