Measuring the productive efficiency of Brazilian railways transport: an application of DEA and Tobit models

Abstract

This study measures the productive efficiency of Brazilian freight railways by the application of DEA (Data Envelopment Analysis) models considering constant returns to scale and variable returns to scale returns. Data from the National Transportation Agency was used because it has a monthly actualization. The study period is among the months from January 2006 to December 2011 and the 12 dealerships of railway transportation were considered for this study. First, proxy variables representing the factors of production used by railway concessionaires in the production of transport were selected. These variables were used as inputs to production and transport as output of DEA models. After application of DEA models and obtaining the productive efficiencies, Tobit censored regression models were prepared to assess the level of correlation between inputs of the DEA models and productive efficiencies. The results of DEA models show that the railroads that specialize in the transportation of minerals and agricultural commodities have higher levels of productive efficiency than the railroads that operate in the transport of a more diversified portfolio of products. On the other hand, the Tobit regression models show that all variables used as inputs in modeling DEA, have significant correlation with productive efficiencies, and analyzing the meanings of correlations can be inferred that the personnel and operational management are significant issues in performance of Brazilian freight railways.