Evaluation of methods parametric and non parametric in the analysis of the efficiency of the milk production. / Avaliação de métodos paramétricos e não paramétricos na análise da eficiência da produção de leite.

AUTOR(ES)
DATA DE PUBLICAÇÃO

2003

RESUMO

The objective of the study is test the most applied methods in the analysis of the relative efficiency, purposing the comparation between a producer and a group of resembled producers, or a producer with the entire sample, as the stochastic frontier’s (random border) case. The tested methods were the DEA (Data Envelopment Analysis), the stochastic frontier and the Varian procedure, analyzing mainly milk producers. Looking for testing the hypothesis that the milk producers are efficient, relating choice of isoquant and the cost minimization point, giving the assumption of the producers achieve correct choices, and it was selected two group methods. The non parametric methods (distribution free) is the best approach to compare a producer with its group. The technology basis is supported by the group, without going through the production function. The parametric procedures are more exigent because they assume a production function that has been estimated, however is more consistent to the hypothesis test. The DEA is classified of non parametric approach, because it does not consider a function with estimated parameters, but, implicitly, it generates a production function, using linear programming. Its less structuralized than the stochastic frontier, more exigent than the Varian procedure, considering the sense that this method does not estimate any frontier, explicitly, as the stochastic frontier method (implicit border), or the DEA. The DEA and the Varian procedure are deterministic, not associated with models that hold back some structure of probability. The stochastic frontier is an associate to the probability structure model, since it defines the term of the regression error making richer the hypothesis tests. Two samples had been used to test the hypotheses. The first one takes a group of 143 commercial milk producers, whose farm enterprises were distributed in the six biggest producer states of Brazil. The second is established by a group of 114 milk producers located in the state of Minas Gerais. The Varian method produced little disturb related to the inputs or outputs. The net income obtained was bigger than the stochastic frontier and smaller than the DEA. This solution is near to the decision of producer based on feasible to make. The method requires the concept of a frontier, rearranging the costs in related to the producers that had produced more and spent less, even so, the reordering procedure requires quadratic programming. Therefore, it is a very convenient method of management. The stochastic frontier doesn’t pursue increase in the net income reduce the costs. In this study the effect on the net income was negative and produced larger disturbs in the inputs and smaller in the products. Facts that will always occur when the production function is adjust to the data. The Varian method was more appropriate in testing hypothesis of the producers, using the experiment and error test, locating them in the efficient frontier, where the point minimizes the costs is located. Therefore it demands that the producer rationalize the costs following the rationalization rule, where the producers drive themselves minimum cost. This procedure can demand much time, in function of constraints of the other sectors of the economy. Considering data of one or more years the DEA and the maximization of the net income have been inadequate procedures to test the hypothesis of Schultz, when it cannot collect convergence movement.

ASSUNTO(S)

custo econômico inferência não paramétrica milk – production econometry hipothesis test. inference non parametric indicadores econômicos produtor rural economic indicator farmers testes de hipóteses. insumos agrícolas economic cost econometria leite – produção agricultural inputs

ACESSO AO ARTIGO

Documentos Relacionados