674 Panel Methods to Test for Unit Roots and Cointegration
capita GDP) and pollution (often proxied by emissions) have been conducted.
Most of the papers focus on a specific conjecture, the so-called environmental
Kuznets curve (EKC) hypothesis, which postulates an inverted U-shaped relation-
ship between the level of economic development and pollution. This hypothesis
therefore states that pollution first rises with increasing income up to a certain
point to then fall for income increasing further. The term EKC refers by analogy
to the inverted U-shaped relationship between the level of economic development
and the degree of income inequality, postulated by Kuznets (1955) in his 1954
presidential address to the American Economic Association.
The theoretical underpinnings for EKCs are discussed, for example, in Andreoni
and Levinson (2001), Brock and Taylor (2004, 2005), Jones and Manuelli (2001)
and Stokey (1998). Brock and Taylor (2005) identify three different mechanisms
that link economic activity with pollution (respectively emissions). These are the
scale, composition and technique effects. For unchanging composition of out-
put and unchanging technology, emissions rise alongside the scale of economic
activity. For given scale and technique, emissions can rise or fall when the compo-
sition of output changes towards a more or less emissions-intensive composition.
Finally, emissions (per unit of output, that is, emissions intensity), can decrease
with improvements in technology, that is, via improved abatement technology.
Depending upon the relative magnitudes of these three effects, a monotonic, a U-
shaped, or an inverted U-shaped relation (among the different patterns possible)
between per capita GDP and per capita emissions may emerge.
Disentangling the relative importance of the three effects requires detailed struc-
tural modeling. The empirical EKC literature is typically less ambitious and focuses
on reduced form modeling to address the issue of whether the three distinct
mechanisms are jointly sufficiently strong to allow for an inverted U-shaped
relationship.
The basic parametric panel formulation of a homogeneous EKC is given by:
ei,t=αi+γit+β 1 yi,t+β 2 y^2 i,t+ui,t, (13.17)whereei,there denotes the logarithm of per capita emissions,yi,tthe logarithm of
per capita GDP,ui,tdenotes the stochastic error term,i=1,...,Nis the country
index andt=1,...,Tis the time index as before. The formulation also includes
country specific fixed effects and country specific linear trends, to allow for income
independent level and slope effects that couldinter aliareflect country specific
preferences with respect to emissions.
An EKC, that is, an inverted U-shaped relationship, occurs ifβ 1 >0 andβ 2 <0, in
which case the turning point with respect to income (in the homogeneous formu-
lation identical across countries) is given byyTP=exp
(
−β 1 / 2 β 2). The primary task
in an empirical EKC analysis is to estimate the relationship (13.17) by appropriate
methods and to test the relevant hypothesis concerning the coefficients.
As discussed in the introduction and also in the convergence example at the
beginning of this chapter,yi,tis often considered to be a unit root process. Often,
due to the relatively short time series available for many countries, for both per