Economic Complexity and the Environment: Evidence from Brazil
9
3 Model, Data and Methodology
3.1 Model and Data Description
The aim of this paper is to examine the extent to which the economic complexity
of municipalities, states and metropolitan regions in Brazil affects the quality of
their environment. We hypothesize that initially environmental degradation rises with
economic complexity, but subsequently falls as structural change towards knowledgeintensive industries takes place, allowing for environmental-friendly technologies
and products. We expect the relationship between environmental degradation—in
terms of pollution or the extraction of natural resources such as forest resources—
and economic complexity to be quadratic. In order to test this hypothesis, we estimate
the following baseline specification:
ENV it = α i + β 1 ECI it + β 2 (ECI it )
2
+ ε it
(1)
Our dependent variable, ENV, denotes four indicators of environmental degradation in Brazil, available at various disaggregation levels. We selected the regional
level of the data based on its availability. Data on domestic solid waste generation is
available for the municipalities in the state of São Paulo, the most economic complex
Brazilian state. Deforestation rates are available for municipalities in nine states constituting the Amazon basin region. Data on forest fires is available at the state-level.
Lastly, air pollution data is available for Brazilian metropolitan regions.
We employ panel regression techniques to estimate Eq. (1). Accordingly, i =
1, …, N indexes each municipality, state or metropolitan region, and t = 1, …, T
refers to the time periods covered in the different panels. Parameter α i allows for the
possibility of time-invariant effects specific to the respective municipalities, states
or metropolitan regions, and ε it is the stochastic error term.
Solid Waste Generation
First, we analyse the relationship between solid domestic waste generation and economic complexity in the municipalities constituting the Brazilian state of São Paulo.
Yearly, the Companhia Ambiental do Estado de São Paulo (CETESB), the state’s
agency for the control and monitoring of polluting activities, publishes estimates on
the average solid waste generated domestically in each municipality in total tons per
day. For our analysis, we converted these estimates into solid waste generated per
capita in kilograms per day. Our panel contains data on up to 455 municipalities and
covers the period from 2003 to 2011.
Deforestation
The second environmental degradation indicator we examine is the deforestation
of the Amazon rainforest, based on the PRODES database. PRODES is a project
by the Instituto Nacional de Pesquisas Espaciais (INPE), Brazil’s National Institute
9
3 Model, Data and Methodology
3.1 Model and Data Description
The aim of this paper is to examine the extent to which the economic complexity
of municipalities, states and metropolitan regions in Brazil affects the quality of
their environment. We hypothesize that initially environmental degradation rises with
economic complexity, but subsequently falls as structural change towards knowledgeintensive industries takes place, allowing for environmental-friendly technologies
and products. We expect the relationship between environmental degradation—in
terms of pollution or the extraction of natural resources such as forest resources—
and economic complexity to be quadratic. In order to test this hypothesis, we estimate
the following baseline specification:
ENV it = α i + β 1 ECI it + β 2 (ECI it )
2
+ ε it
(1)
Our dependent variable, ENV, denotes four indicators of environmental degradation in Brazil, available at various disaggregation levels. We selected the regional
level of the data based on its availability. Data on domestic solid waste generation is
available for the municipalities in the state of São Paulo, the most economic complex
Brazilian state. Deforestation rates are available for municipalities in nine states constituting the Amazon basin region. Data on forest fires is available at the state-level.
Lastly, air pollution data is available for Brazilian metropolitan regions.
We employ panel regression techniques to estimate Eq. (1). Accordingly, i =
1, …, N indexes each municipality, state or metropolitan region, and t = 1, …, T
refers to the time periods covered in the different panels. Parameter α i allows for the
possibility of time-invariant effects specific to the respective municipalities, states
or metropolitan regions, and ε it is the stochastic error term.
Solid Waste Generation
First, we analyse the relationship between solid domestic waste generation and economic complexity in the municipalities constituting the Brazilian state of São Paulo.
Yearly, the Companhia Ambiental do Estado de São Paulo (CETESB), the state’s
agency for the control and monitoring of polluting activities, publishes estimates on
the average solid waste generated domestically in each municipality in total tons per
day. For our analysis, we converted these estimates into solid waste generated per
capita in kilograms per day. Our panel contains data on up to 455 municipalities and
covers the period from 2003 to 2011.
Deforestation
The second environmental degradation indicator we examine is the deforestation
of the Amazon rainforest, based on the PRODES database. PRODES is a project
by the Instituto Nacional de Pesquisas Espaciais (INPE), Brazil’s National Institute