Economic Complexity and the Environment: Evidence from Brazil
25
turning points of the full sample regressions are taken into account. The turning
points are higher than the maximum income per capita recorded in the panel of
ca. 55,000 constant 2010 Brazilian Reais (cf. Table 11). The highest turning point
amounts to 62,231 Brazilian Reais and lies well outside the range. Reaching an
income that allows to reverse the detrimental effect of growth on the environment is
hence unrealistic, leaving the states in the upward sloping part of the EKC.
4.4 Air Pollution
Our last set of regressions examines the relationship between air pollution and economic complexity in Brazilian metropolitan regions from 2002 to 2009. In total, we
analyse ten different indicators of air pollution: maximum and average concentrations of sulphur dioxide, nitrogen dioxide, TSP and ultrafine particles, as well as
maximum concentrations of carbon monoxide and ozone. For most of these indicators the models were poorly specified and over-parameterised because of a low
number of observations. Therefore, we give more emphasis to two air pollution indicators: maximum concentration of carbon monoxide, and maximum concentration of
ozone. Tables 4 and 5 show that, neither economic complexity nor income per capita
have a statistically significant impact on these air pollutants. In the case of ozone,
population density has a robust and statistically significant impact on maximum concentrations. These concentrations decrease in more densely populated metropolitan
regions, probably due to higher industrialisation standards providing cleaner technologies and enhanced environmental awareness on the part of their inhabitants. The
other control variables have no robust impact on carbon monoxide or ozone concentrations. Finally, Tables 6 and 7 show some indication of a negative impact from ECI
to air pollution and a positive impact from GDP per capita to air pollution. These
results should be read with cautions given the low number of observations.
5 Conclusion
This paper investigates the extent to which economic complexity affects the environment in Brazil. We hypothesize that environmental degradation rises as economies
diversify their production and become more complex, but an eventual structural
change towards knowledge-intensive industries creates the technology necessary to
limit degradation. Using panel data regression techniques and a rich set of control
variables, we analyse the relationship between economic complexity and solid waste
generation, deforestation, forest fires and air pollution. We find that waste generation
decreases, but forest fires increase linearly with rising complexity. Economic complexity has no robust, if any, impact on deforestation or air pollution. In line with
the traditional EKC, whereby degradation first increases and subsequently decreases
25
turning points of the full sample regressions are taken into account. The turning
points are higher than the maximum income per capita recorded in the panel of
ca. 55,000 constant 2010 Brazilian Reais (cf. Table 11). The highest turning point
amounts to 62,231 Brazilian Reais and lies well outside the range. Reaching an
income that allows to reverse the detrimental effect of growth on the environment is
hence unrealistic, leaving the states in the upward sloping part of the EKC.
4.4 Air Pollution
Our last set of regressions examines the relationship between air pollution and economic complexity in Brazilian metropolitan regions from 2002 to 2009. In total, we
analyse ten different indicators of air pollution: maximum and average concentrations of sulphur dioxide, nitrogen dioxide, TSP and ultrafine particles, as well as
maximum concentrations of carbon monoxide and ozone. For most of these indicators the models were poorly specified and over-parameterised because of a low
number of observations. Therefore, we give more emphasis to two air pollution indicators: maximum concentration of carbon monoxide, and maximum concentration of
ozone. Tables 4 and 5 show that, neither economic complexity nor income per capita
have a statistically significant impact on these air pollutants. In the case of ozone,
population density has a robust and statistically significant impact on maximum concentrations. These concentrations decrease in more densely populated metropolitan
regions, probably due to higher industrialisation standards providing cleaner technologies and enhanced environmental awareness on the part of their inhabitants. The
other control variables have no robust impact on carbon monoxide or ozone concentrations. Finally, Tables 6 and 7 show some indication of a negative impact from ECI
to air pollution and a positive impact from GDP per capita to air pollution. These
results should be read with cautions given the low number of observations.
5 Conclusion
This paper investigates the extent to which economic complexity affects the environment in Brazil. We hypothesize that environmental degradation rises as economies
diversify their production and become more complex, but an eventual structural
change towards knowledge-intensive industries creates the technology necessary to
limit degradation. Using panel data regression techniques and a rich set of control
variables, we analyse the relationship between economic complexity and solid waste
generation, deforestation, forest fires and air pollution. We find that waste generation
decreases, but forest fires increase linearly with rising complexity. Economic complexity has no robust, if any, impact on deforestation or air pollution. In line with
the traditional EKC, whereby degradation first increases and subsequently decreases