Eric Hanley
University of KansasEconomic Growth and Climate Change:Exploring the Relationship between GDP and CO 2 EmissionsWhat do we know: What does Sociology bring to the table for studying the human dimensions of global
climate change?
What do we know about the relationships between economic growth and carbon dioxide emissions? Most
economists tell us that this relationship takes the form of an inverted-U, with per capita emissions increasing up
to a certain level of income and diminishing thereafter. Having uncovered this type of inverted-U relationship
between economic growth (measured in terms of per capita GDP) and a number of pollutants, economists have
a attached a specific label to this curve, referring to it as the environmental Kuznets curve (or EKC for short).
Most of the empirical studies on CO 2 emissions that have been published in economic journals indicate that the
relationship between per capita gross national produce (GDP) and CO 2 emissions takes the form of an EKC,
with emissions increasing up to a certain level of income and diminishing thereafter. What this suggests is that
there is a delinking of carbon dioxide emissions and economic growth at relatively high levels of income. Before
discussing some of the factors which researchers have highlighted in an attempt to explain why per capita CO 2
emissions may diminish at higher income levels, I would like to challenge the basic finding that the relationship
between economic growth and carbon dioxide assumes the form of an EKC in the first place.
The provocative idea that I am putting forward is that the relationship between economic growth and
carbon dioxide emissions does not assume a quadratic form with emissions diminishing once a certain income
level has been reached but rather takes a cubic form with emissions rising at low-income levels, stabilizing or
dropping at middle levels, and rising again at high-income levels. I base this conclusion on findings generated
from a pooled cross-sectional analysis of 92 countries for the period 1980 to 2004. These data were used to
estimate the following fixed effects model:
Yit = β 0 + β 1 Xit + yt + ci + εitwhere t and i are year and country indexes respectively, Yit represents the natural log of per capita carbon
emissions in metric tons, Xit is a vector of covariates, yt is a vector of fixed year effects, ci a vector of fixed country
effects, and εit an error term (the model corrects for first-degree autocorrelation through inclusion of the lagged
dependent variable as an independent variable). Results from this model, which are presented in Table 1, show
that a cubic (i.e. N-shaped) functional form between GDP and carbon dioxide emissions fits the data well. Not
only are the coefficients for the GDP-terms are highly significant, the size and sign of the coefficients indicate
increasing carbon dioxide emissions at lower income levels (up to $18,000 per capita in constant US dollars),
stabilization or reduction in these emissions at middle levels (between $18,000 and $32,000), and an increase in
emissions at higher levels (above $32,000). These findings call into question the optimistic assumption voiced
by some researchers that economic growth can in and of itself mitigate some of the effects associated with global
climate change.
Although a handful of studies have uncovered a cubic relationship between GDP and carbon dioxide
emissions, the authors of these studies have discounted this finding on one or more of the following grounds:
(1) the upward turning point in question lies beyond the range of observed values for income; (2) when certain