Prof. S. S. Xavier
S. S. Xavier is a Professor at Federal Institute of Paraná, Brazil. She has her expertise in electricity distribution companies benchmarking, especially with Data Envelopment Analysis models. Besides, she has her experience in economic regulation. Xavier received Diploma and Doctorate in Electrical Engineering, at University of Itajubá (Brazil). She has different affiliations at Federal University of Ouro Preto and Goiano Federal Institute. Xavier received recognition and award by conferences and scientific organizations, internationally.
How Efficient are the Brazilian Electricity Distribution Companies?
During the last years, the electricity sector has experienced great changes, especially within the economic regulation. After receiving several criticisms, the rate of return regulation has been replaced by incentive regulation, whose main objective is to encourage business efficiency. Many regulators use the entire distribution company as a Decision Making Unit for price regulation when benchmarking is applied. However, in Brazil, quality is measured using small parts of the company. Given that efficiency cannot be assessed without considering quality and characteristics of the underlying environment, we try to find the trade-off among management, quality, environment and costs. The objective of this paper is to proposes an alternative application of Data Envelopment Analysis (DEA) to the Brazilian case, characterized by a large territory: the use of Unit Networks (UNs) in the distribution segment to regionalize the concession area and then to analyse the efficiencies separately. We compare the performance of 10 largest distribution utilities in Brazil in the period from 2006 to 2007. The data can be found on the Brazilian Regulator website, where it was considered the latest consistent sample available for this period. The alternative application involves a three-stage analysis. First, we aggregate the sets of consuming units forming regions within the distribution concession area using UNs technique. Second, we use three different models to determine the technical efficiency performances of the UNs using DEA. Third, treating these calculated efficiency scores as dependent variables, we used a regression technique to determine the environmental variables that may explain the efficiency scores. The results indicate that the UNs are, on average, technically efficient by approximately 0.75 under Model 1, 0.79 under Model 2 and 0.79 under Model 3. The 15 UNs in Model 1 are efficient; note that nine UNs belong to an area with a high customer density. In this model, it was not considered quality nor environment aspect. Under Model 2, to which quality of supply was added to the analysis, 17 UNs are efficient, and 11 UNs are located in low lightning incidence areas. The average efficiency shows that some UNs rank high in Model 2 while they rank low in Model 1. Under Model 3, to which environment was considered to the analysis, there are only seven efficient UNs that contrast with the results of Model 2. Some UNs have decreased/increased their performance because they are located in a more/less favorable area. For a better view of the UN influence on the company performance, each UN is mapped according to its effect (positive or negative) and its intensity (high and low) on the company efficiency score. It is important that with this alternative application, the companies’ administration can do an internal benchmarking: comparing performance among their regions, extracting lessons from efficient UNs and applying them to inefficient ones. Therefore, the main contribution of this paper is twofold: the solution for Brazilian distribution companies’ heterogeneity and the choice of variables that are better measures for an efficiency analysis.
Electricity Power Distribution, Incentive Regulation, Data Envelopment Analysis.