8
J. Swart and L. Brinkmann
performing specification tests. In addition, studies using more representative data
find sulphur dioxide emissions and concentrations—traditionally a showcase for the
EKC hypothesis—to be positively correlated with income, a finding that is in line
with the estimation results for other environmental degradation indicators.
Stern and Common (2001) reduce misspecification problems by using firstdifferenced data when estimating the EKC. The authors find that the relationship
between sulphur dioxide emissions and income is positive in both high- and lowincome countries. Stern (2002) applies an emission-decomposition model and finds
that sulphur emissions rise with increasing income at all levels of income. Many
other test the EKC hypothesis and find mixed results (e.g. Suri and Chapman 1998;
Perman and Stern 2003; Tao et al. 2008; Managi and Jena 2008; Mills and Waite
2009; Jalil and Mahmud 2009; Iwata et al. 2010; Fodha and Zaghdoud 2010; Fosten
et al. 2012; Saboori and Sulaiman 2013; Al-Mulali et al. 2015).
The empirical evidence is not more compelling in Brazil-specific studies. The
empirical literature investigating the possible existence of an EKC for Brazil focus
on the link between deforestation and economic development in Brazil. Santos et al.
(2008), for instance, analyse 782 municipalities of the Amazon basin from 2000 to
2005, and find support for the EKC hypothesis. Oliveira et al. (2011) analyse these
same municipalities for 2001–2006, but do not find evidence for a deforestation
EKC. The authors include a rich set of additional explanatory variables, such as
cattle, agricultural activity, vegetable extraction, forestry, population density, rural
credit, annual dummies, and deforestation rate in the previous year. Gomes and Braga
(2008) also analyse deforestation rates in the Amazon region and find fairly mixed
evidence for the EKC. Depending on the functional form applied, the authors find
either an inverted U-shaped or an N-shaped relationship between the deforestation
rate and income.
Teixeira et al. (2012) investigate the existence of a deforestation EKC in 139
municipalities of the state of Mato Grosso in 2006, one of nine states constituting
the Amazon River basin region. The authors control for spatial effects and add a rich
set of additional explanatory variables, including per capita wood extraction and a
ratio of cattle units over area destined for cattle. The findings depend strongly on the
functional form chosen, thus suggesting a weak relationship between deforestation
and income. Also controlling for spatial effects, Colusso et al. (2015) analyse data
for the year 2008 related to 1306 municipalities constituting the Cerrado biome.
The empirical evidence of this study is similarly mixed, ranging from an inverted
U-shaped to a U-shaped or N-shaped relationship.
To the best of our knowledge, besides the abovementioned papers focusing on
deforestation there is only one paper which examines alternative indicators of environmental degradation to test for an EKC for Brazil. Sousa et al. (2008) proxy
environmental degradation by defining the deficit in drinking water, in basic sanitation, and in waste collection as dependent variables. Using a dataset containing
5507 municipalities in the years 1991 and 2000, the empirical evidence suggests that
the relationship between each one of these three indicators and income per capita is
U-shaped.
J. Swart and L. Brinkmann
performing specification tests. In addition, studies using more representative data
find sulphur dioxide emissions and concentrations—traditionally a showcase for the
EKC hypothesis—to be positively correlated with income, a finding that is in line
with the estimation results for other environmental degradation indicators.
Stern and Common (2001) reduce misspecification problems by using firstdifferenced data when estimating the EKC. The authors find that the relationship
between sulphur dioxide emissions and income is positive in both high- and lowincome countries. Stern (2002) applies an emission-decomposition model and finds
that sulphur emissions rise with increasing income at all levels of income. Many
other test the EKC hypothesis and find mixed results (e.g. Suri and Chapman 1998;
Perman and Stern 2003; Tao et al. 2008; Managi and Jena 2008; Mills and Waite
2009; Jalil and Mahmud 2009; Iwata et al. 2010; Fodha and Zaghdoud 2010; Fosten
et al. 2012; Saboori and Sulaiman 2013; Al-Mulali et al. 2015).
The empirical evidence is not more compelling in Brazil-specific studies. The
empirical literature investigating the possible existence of an EKC for Brazil focus
on the link between deforestation and economic development in Brazil. Santos et al.
(2008), for instance, analyse 782 municipalities of the Amazon basin from 2000 to
2005, and find support for the EKC hypothesis. Oliveira et al. (2011) analyse these
same municipalities for 2001–2006, but do not find evidence for a deforestation
EKC. The authors include a rich set of additional explanatory variables, such as
cattle, agricultural activity, vegetable extraction, forestry, population density, rural
credit, annual dummies, and deforestation rate in the previous year. Gomes and Braga
(2008) also analyse deforestation rates in the Amazon region and find fairly mixed
evidence for the EKC. Depending on the functional form applied, the authors find
either an inverted U-shaped or an N-shaped relationship between the deforestation
rate and income.
Teixeira et al. (2012) investigate the existence of a deforestation EKC in 139
municipalities of the state of Mato Grosso in 2006, one of nine states constituting
the Amazon River basin region. The authors control for spatial effects and add a rich
set of additional explanatory variables, including per capita wood extraction and a
ratio of cattle units over area destined for cattle. The findings depend strongly on the
functional form chosen, thus suggesting a weak relationship between deforestation
and income. Also controlling for spatial effects, Colusso et al. (2015) analyse data
for the year 2008 related to 1306 municipalities constituting the Cerrado biome.
The empirical evidence of this study is similarly mixed, ranging from an inverted
U-shaped to a U-shaped or N-shaped relationship.
To the best of our knowledge, besides the abovementioned papers focusing on
deforestation there is only one paper which examines alternative indicators of environmental degradation to test for an EKC for Brazil. Sousa et al. (2008) proxy
environmental degradation by defining the deficit in drinking water, in basic sanitation, and in waste collection as dependent variables. Using a dataset containing
5507 municipalities in the years 1991 and 2000, the empirical evidence suggests that
the relationship between each one of these three indicators and income per capita is
U-shaped.