Bruno Reinoso Hybner

The aim of this work was to estimate the total factor productivity (TFP) of agriculture in the microregions of southern Brazil in the period 2006-2017 to verify whether there is a convergence process involving their respective growth rates. The calculation of the performance indicator, in this case, was based on the Malmquist procedure, which made it possible to obtain the TFP index with the Data Envelopment Analysis (DEA) approach. In addition, the method made it possible to break down TFP into two other components, the Technical Efficiency Change Index (EFF), which represents the way in which production units combine inputs, and the Technological Frontier Change Index (TEC), which conceives the displacement of the production function. Concerning convergence, the equations involved were estimated by the technique of spatial econometrics, whose diagnosis of spatial dependence was favorable to the spatial error model (SEM). In addition, using the geographically weighted regression (RPG) method, the necessary velocities for the microregions to reach the absolute convergence position were estimated. With the results it could be seen that the performance of the agricultural TFP was mainly driven by TEC, than by EFF. The analysis of the spatial distribution of agricultural TFP showed evidence in favor of the convergence trend, as it was found that many regions with high TFP growth rates in the period 2006-2017 had low productivity levels in 2006. On the opposite side , locations with lower growth rates had higher levels of productivity. The TFP absolute and conditional spatial econometric models corroborated the evidences indicated by the analysis of the spatial distribution of this variable, demonstrating that the productivity trajectory moves towards an equalization of growth rates. It is noteworthy that the inclusion of spatial dependence variables (spatially lagged error term) and the microregion structural characteristics vector served to accelerate the convergence process. Finally, it was found, through the coefficients estimated by the RPG method, that the microregions with higher TFP growth rates need to maintain a lower velocity to reach the convergence state, while those with low growth need a higher velocity to reach the steady state position.