Brazilian states panel — Greenhouse gas emissions, GDP, human capital, urbanization, and population
Description
This dataset contains a strongly balanced panel for Brazil's 27 federative units covering 1995–2021 (729 state-year observations). It supports the empirical analysis reported in the accompanying article on the long-run elasticity of greenhouse gas emissions to economic growth across Brazilian states. The dataset includes total and sectoral greenhouse gas emissions (energy, agriculture, industrial processes, waste, and land-use change) from the Brazilian Greenhouse Gas Emissions and Removals Estimation System (SEEG); GDP per capita in constant 2010 Brazilian reais from IBGE Sistema de Contas Regionais; mean years of schooling from harmonized PNAD/PNAD Contínua state-level series; urbanization rates from IBGE demographic census benchmarks (linearly interpolated between census years); and total state population from IBGE annual estimates. Variables are provided in levels and natural logarithms. Pre-computed territorial classifications (macro-region, ecological profile, Legal Amazon indicator, and structural emissions regimes) are included to enable regime-based analyses. Auxiliary variables for cointegration tests, NARDL partial sums, and the income–human capital interaction are also provided. The dataset is delivered in Stata (.dta) and CSV formats, accompanied by a README file describing all variables and reproduction steps. All source data are publicly accessible.
Files
Steps to reproduce
1. EMISSIONS: Download the SEEG state-level inventory (1970–2021 release) from https://seeg.eco.br/. Filter to 1995–2021 and aggregate by sector (energy, agriculture, industrial processes, waste, land-use change) and state. 2. GDP: Download state GDP from IBGE Sistema de Contas Regionais (https://www.ibge.gov.br/estatisticas/economicas/contas-nacionais/9054-contas-regionais-do-brasil.html) for 2002–2021. Combine with the earlier IBGE state-level GDP series for 1995–2001. Deflate to constant 2010 Brazilian reais using the IPCA index, and divide by total state population to obtain GDP per capita. 3. HUMAN CAPITAL: Download mean years of schooling (population aged 25 and over) from harmonized PNAD and PNAD Contínua state-level series provided by IBGE. 4. URBANIZATION: Compute the urbanization rate (urban population as share of total) from IBGE demographic census benchmarks (1991, 2000, 2010, 2022). Linearly interpolate between census years to obtain annual values. 5. POPULATION: Download annual state population estimates from IBGE Estimativas da população (https://www.ibge.gov.br/estatisticas/sociais/populacao/9103-estimativas-de-populacao.html). 6. PANEL ASSEMBLY: Merge all sources by state code (UF) and year, producing a balanced panel of 27 states by 27 years. Generate natural logarithms for emissions, GDP per capita, and population. Construct sectoral emission shares and the structural regime classifications described in the manuscript. 7. ESTIMATION: In Stata, run the second-generation panel diagnostics: Pesaran CD test for cross-sectional dependence (xtcd2), Pesaran–Yamagata Delta test for slope homogeneity, CIPS unit root test (multipurt or xtcips), and Kao, Pedroni, and Westerlund cointegration tests (xtcointtest). Estimate long-run elasticities using the Pooled Mean Group estimator (xtpmg), with Mean Group and Common Correlated Effects Mean Group (xtmg) estimators as robustness checks. Apply the Dumitrescu–Hurlin panel Granger non-causality test (xtgcause) for temporal ordering between variables.