Efficiency in rail systems through three different approaches and contributions to push the Brazilian rail system toward high performance

Abstract

This research investigates the efficiency of the Brazilian rail cargo system (SFBC, acronym in Portuguese). The importance of the performance of a rail system to the logistics of goods in countries with large territorial dimensions is recognized. However, its role is secondary in Brazil compared to road transport, representing high economic and environmental costs. SFBC is dedicated to the export of mineral and agricultural bulk where it has a significant capacity for insertion. Meanwhile, the SFBC has an insignificant participation in the internal distribution of goods, which is mostly done in highways, including on longer routes. Consequently, the transport sector in Brazil emits twice as much CO2 from buming fuels than the world average. SFBC presents characteristics of a medium performance scenario. The performance of the operators is heterogeneous, benchmarking is an outlier, the average efficiency is low, and the average idleness of the sections is high. To discuss the conditions to achieve a high performance scenario is the main contribution of this thesis. Three researches were done to address this issue. The first one is dedicated to estimate the efficiency of the railway concessionaires in Brazil between 2010-2014 by using Data Envelopment Analysis. In a second stage, the significance of selected variables was assessed through Bootstrap Truncated Regression, including the type of the use of the railway track (shared or monopoly), a gap in the literature. The operators were grouped according to the efficiency scores and the type of returns to scale, and measures to increase the efficiency of the clusters were discussed. The second research has the purpose to explain the heterogeneity found in the average efficiency of the different railway systems in the world, the object of researches done between 2000-2016, through a meta-analysis carried out for the first time onj railways. For greater robustness, Leave-One-Out-Cross-Validation and VVeighted Regression were applied. Public policies induding diversification of services and models for assessing efficiency by regulatory agencies were discussed. The third research is devoted to the efficiency of the SFBC raílway sections and identifies the significant conditions for high and low performance scenarios by combining the Technique for Order of Preference by Similarity to Ideal Solution and a differential evolution algorithm for estimating the weights of variables in optimized scenarios, another gap in the literature. In a second stage, the significance of the variables selected was evaluated. The research shows that the public manager should push the rail companies toward the transportation of any type of cargo, the diversification of services, the centralized control of the operation, and the sharing of the railway track. The competition and diversifìcation are significant for high performance.