Estudo da produtividade de ferrovias brasileiras com modelos de fronteiras estocásticas em painel.

Abstract

Data Envelopment Analysis (DEA) models, from their simplest to the most complex form, have become popular in the analysis of productivity and efficiency in transport. Although with virtues there are deficiencies or limitations in the use of these models, as the advances in testing hypotheses regarding the estimated parameters have recently developed. Alternatives, such as indexes and parametric methods, add elements to understand the phenomenon and to test hypotheses statistically. This article explores the use of the stochastic frontier model in a panel of Brazilian railroad concessions (2006-2016) to expand the explanation of the factors that generate efficiency and productivity in this transport sector. The results indicate, given the sample size limitation, that the infrastructure configuration and the type of cargo transported affect productivity and efficiency, without reporting evidence for the influence of accidents.