Efeitos do transbordamento da produtividade agrícola brasileira

Abstract

Agriculture is essential for the country's economic development. Brazil stands out in world agricultural production. All surplus produced is destined for exports. In this way, this research aimed to determine the agricultural spillovers effects of the main grains produced and exported in the country, analyzing the spatial dynamics of Brazilian agriculture. It is used, for this purpose, the database corresponding to the quantity produced and planted area of the main Brazilian grain crops: soybean, corn, wheat, rice, coffee and cocoa, from the 558 Brazilian microregions, corresponding to the years 1992, 1997, 2002, 2007, 2012 and 2017, totaling 20,088 observations, with annual collection. The methodology of Spatial Econometrics was used, initially an Exploratory Analysis of Spatial Data and later Spatial Econometric Modeling, to fit a representative model of the series under study. This analysis of the spatial dynamics of Brazilian agriculture made it possible to identify the agricultural pattern of the country, verifying clusters of productivity and spillovers among crops. Their results indicated the presence of positive spatial autocorrelation between the variables, which means that microregions with high or low agricultural productivity are grouped in specific areas of the map, surrounded by microregions with similar characteristics for this variable, making it possible to identify grains agricultural productivity spillover effects between neighboring microregions. As results, after confirming the presence of spatial autocorrelation in the data, by the Moran I statistic, the adjustment of the spatial econometrics models occurred. The presence of spatial autocorrelation, confirming that space is relevant to grains agricultural productivity analysis, is a decisive factor in the models adjustment by spatial econometrics. The application and adjustment of three spatial models were performed: the Spatial Auto Regressive (SAC) model was the best fit for the variables corn, rice, coffee and cocoa, the Spatial Error Model (SEM) model was the best model adjusted for the wheat variable and the Spatial Durbin Error Model (SDEM) model was the best representative model of the soybean series generating process. It can be concluded that grains agricultural productivity variable has a heterogeneous distribution among country microregions, in other words, agricultural productivity is increasingly autocorrelated spatially over time.