Curva de Kuznets Ambiental para a Amazônia Legal

Abstract

The tropical deforestation is a main issue on the global climate change discussion. In Brazil, the deforestation caused by hot spots is responsible for large emissions of carbon dioxide, a greenhouse gas. There is always a concern about the increasing pressure over the forest as development advances. Within this context, this work aims to investigate the Environmental Kuznets Curve (EKC) hypothesis applied to Legal Amazon region: whether there is an inverted “U” relationship between an environmental degradation index (annually deforested area) and economic growth (GDP per capita) at municipal level from 2001 to 2006, using a panel data model with spatial dependence. Some other variables mentioned in the literature were considered to explain deforestation: cattle size, soybean and sugar cane crops, vegetal extraction and forestry products, population density, rural credit, and previous forest area. Additionally, there can be important intra-regional differences due to occupation history, large area and geographical aspects. The spatial heterogeneity of coefficients was considered by studying spatial regimes into two sets (macro-regions and states). The extreme coefficients spatial heterogeneity was considered by estimating the model using geographically weighted regressions (GWR). The Exploratory Spatial Data Analysis (ESDA) shows the presence of clusters in a High-High pattern (municipalities with high deforestation values near municipalities with high values for the deforested area also) in the Densely Populated Arch and Central Amazon. The econometric results reveal the presence of unobserved effects, being the fixed effects estimate the most appropriate one. The global model (considering the whole region) that fits the data well is the spatial error model with spillovers; for this model, the EKC relationship appears to have an inverted “N” shape, so the annual deforested area is higher for lower incomes, it decreases when GDP per capita increases, then it goes back to increase, and for higher income levels it decreases again. The variables related to cattle size, rural credit and previous forest area seems to affect the deforested area. When the spatial heterogeneity is considered by means of different spatial regimes, the following results are found: a) for the three macro-regions set of models (Occidental Amazon, Central Amazon and Densely Populated Arch), none of the shapes for the EKC relationship is verified; b) for the nine states set of models, the EKC hypothesis is found in an inverted “U” shape for Mato Grosso state, a “U” shape for Pará state and increasingly monotonic shape for Maranhão state, whereas for the other states the EKC is not found. For the extreme level of spatial heterogeneity, the geographically weighted regressions show local coefficients: it was possible to map the different shapes for EKC found mainly in central and northeastern portions of Amazon. The results show different local relationships between deforestation and GDP per capita at municipal level, and also different local relationships between deforestation and the additional variables.