12
J. Swart and L. Brinkmann
of exports and imports over GDP. The ratio is denoted in constant 2010 Brazilian
Reais at the state-level, and in constant 2000 Brazilian Reais at the metropolitan
region- and municipality-level. According to Antweiler et al. (2001) as well as Cole
and Elliott (2003), the effect of trade on the environment can be decomposed into
three effects: scale, technique, and composition effect. The scale effect relates to
the increase in overall economic activity ensuing from intensified trade, being hence
detrimental to the environment. In contrast, the technique effect improves the quality
of the environment due to more environmentally-friendly production brought about
by international trade. Production changes either because of increased (domestic)
competition, forcing the least energy-efficient firms to leave the industry, or the
import of cleaner technologies. The composition effect can be positive or negative
and refers to the changes in the industrial structure of an economy arising from
international trade. A country or region will specialise on the production of goods
for which it has a comparative advantage, which can be in cleaner or more polluting
industries. Overall, the net effect of trade openness on the environment is theoretically
ambiguous.
As Selden and Song (1994), Grossman and Krueger (1995), Antweiler et al.
(2001), and Fosten et al. (2012), among others, we include a time trend to capture
technological advances and changes in environmental awareness that are not related
to either income or economic complexity.
Finally, we include an additional set of agricultural variables withdrawn from
IPEA for our analyses of deforestation rates and forest fires in Brazilian municipalities
and states. Various authors (e.g. Margulis 2003; Brown et al. 2005; Aguiar et al. 2007;
Miragaya 2008; Vera-Diaz et al. 2008) investigate the drivers of deforestation in the
Amazon rainforest, identifying agriculture—most notably in terms of cocoa, coffee,
sugarcane and soy production—and cattle as the main culprits.
3 We follow Colusso
et al. (2015) and include total harvested and total planted area, both in hectares, as
explanatory variables.
4 We expect the agricultural variables to be positively correlated
with deforestation and forest fires. In other words, we expect larger areas in the
Amazon rainforest to be destroyed, and more forest fires to be observed, when there
is an expansion of agricultural activity in the municipalities of the Amazon basin
region and in Brazilian states.
Table 8 in the Appendix provides a detailed list of all the variables used in our
analysis (in regressions or, whenever necessary, for the construction of variables) and
their sources. Tables 9, 10, 11 and 12 contain descriptive statistics for the variables
in the four panels.
3 Aguiar et al. (2007) state that 70% of the total deforested area in the Amazon Basin is destined for
cattle production. 13 and three per cent were transformed into temporary and permanent harvests,
respectively.
4 Oliveira et al. (2011) and Teixeira et al. (2012) add harvested area for the aforementioned agricultural products (in hectares) and cattle units per square kilometre as control variables. IPEA provides
data on harvested area by agricultural products, but the panels are rather incomplete and lead to a
drop in the number of observations when employed.
J. Swart and L. Brinkmann
of exports and imports over GDP. The ratio is denoted in constant 2010 Brazilian
Reais at the state-level, and in constant 2000 Brazilian Reais at the metropolitan
region- and municipality-level. According to Antweiler et al. (2001) as well as Cole
and Elliott (2003), the effect of trade on the environment can be decomposed into
three effects: scale, technique, and composition effect. The scale effect relates to
the increase in overall economic activity ensuing from intensified trade, being hence
detrimental to the environment. In contrast, the technique effect improves the quality
of the environment due to more environmentally-friendly production brought about
by international trade. Production changes either because of increased (domestic)
competition, forcing the least energy-efficient firms to leave the industry, or the
import of cleaner technologies. The composition effect can be positive or negative
and refers to the changes in the industrial structure of an economy arising from
international trade. A country or region will specialise on the production of goods
for which it has a comparative advantage, which can be in cleaner or more polluting
industries. Overall, the net effect of trade openness on the environment is theoretically
ambiguous.
As Selden and Song (1994), Grossman and Krueger (1995), Antweiler et al.
(2001), and Fosten et al. (2012), among others, we include a time trend to capture
technological advances and changes in environmental awareness that are not related
to either income or economic complexity.
Finally, we include an additional set of agricultural variables withdrawn from
IPEA for our analyses of deforestation rates and forest fires in Brazilian municipalities
and states. Various authors (e.g. Margulis 2003; Brown et al. 2005; Aguiar et al. 2007;
Miragaya 2008; Vera-Diaz et al. 2008) investigate the drivers of deforestation in the
Amazon rainforest, identifying agriculture—most notably in terms of cocoa, coffee,
sugarcane and soy production—and cattle as the main culprits.
3 We follow Colusso
et al. (2015) and include total harvested and total planted area, both in hectares, as
explanatory variables.
4 We expect the agricultural variables to be positively correlated
with deforestation and forest fires. In other words, we expect larger areas in the
Amazon rainforest to be destroyed, and more forest fires to be observed, when there
is an expansion of agricultural activity in the municipalities of the Amazon basin
region and in Brazilian states.
Table 8 in the Appendix provides a detailed list of all the variables used in our
analysis (in regressions or, whenever necessary, for the construction of variables) and
their sources. Tables 9, 10, 11 and 12 contain descriptive statistics for the variables
in the four panels.
3 Aguiar et al. (2007) state that 70% of the total deforested area in the Amazon Basin is destined for
cattle production. 13 and three per cent were transformed into temporary and permanent harvests,
respectively.
4 Oliveira et al. (2011) and Teixeira et al. (2012) add harvested area for the aforementioned agricultural products (in hectares) and cattle units per square kilometre as control variables. IPEA provides
data on harvested area by agricultural products, but the panels are rather incomplete and lead to a
drop in the number of observations when employed.