Avaliação da eficiência na produção de arroz no Brazil: uma aplicação da Análise Envoltória de Dados

Abstract

Rice cultivation is an important activity, both from an economic and social point of view and from a food supply and safety perspective. Based on the theories of production, production costs and the firm, this research had the objective of investigating the efficiency of the main rice producing regions in Brazil from the perspective of production costs for the 2011/12 to 2016/17 harvests. It is a quantitative, descriptive and applied research that uses secondary data made available by CONAB. The sample is formed by the regions of Balsas, Sorriso, Cachoeira do Sul, Pelotas and Uruguaiana, which represent the systems of cultivation (dry farming and irrigated) and planting (conventional, minimum and direct / semi - direct cultivation) predominant in Brazil. We adopted the DEA, non-parametric approach, output-oriented BCC and CCR models. Pearson correlation estimation and references from previous studies were used to select the cost variables to compose the analytical model. Six inputs were defined (operations with machines, labor, seeds, fertilizers, agrochemicals and technical assistance) and an output (rice yield). Two analysis structures were developed: general evaluation - performed based on all 30 DMUs observed by the research; and the evaluation by category, in which the DMUs of the research were segregated by cultivation system and by planting system. The cost structures of the producing regions were evaluated, observing that, except for Sorriso, all the others had a negative average variation in relation to the planted and harvested areas and the level of rice production, movement also observed at the data consolidated in Brazil. Within the sample, three regions (Cachoeira do Sul, Pelotas and Uruguaiana) were found with high productivity (over 7,000 kg / ha) and two regions (Balsas and Sorriso) with productivity below the national average. By the evaluation of the costs, it was observed that the cost of farming was the one that obtained the largest participation in the total costs of the producing regions in the period evaluated, about 62%. Considering the cost variables selected to compose the DEA analysis model, it was verified the greater expressiveness of the costs with operations with machines, fertilizers and agrochemicals. By the DEA results, the averages of the measures of technical efficiencies, scale, allocative and cost, obtained in the general evaluation, were 0.972, 0.932, 0.782 and 0.764, respectively. It was observed that the constant return of the scale predominate among the inefficient DMUs the DMUs, as a rule, present both the problem of the productive scale and the efficiency. In the evaluation by categories, considering the cultivation system, there was little change in the technical efficiency scores of the dryland DMUs, and small changes in the DMUs of the irrigated system, these stemming to the change in the benchmarks. However, significant changes were observed in the averages of the allocative efficiency and cost efficiencies, especially with the increase of efficient DMUs. Finally, considering the planting systems, we obtained measures of generally more distinct efficiencies, probably due to the increased specialization of the model, since the number of DMUs per analysis group was small given the number of variables used.