Estimação paramétrica de escores de eficiência em 2 estágios: impacto das variáveis ambientais no ajuste das eficiências regulatórias das empresas brasileiras de distribuição de energia elétrica para 4o Ciclo de Revisão Tarifária Periódica

• Ano de defesa: 2015
• Autor(a) principal: Aline Veronese da Silva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Estimação de parâmetros; Segundo estágio; Data envelopment analysis; Engenharia de produção
• Link de acesso: http://hdl.handle.net/1843/BUBD-A5AMH7

The Electrical Energy Distribution regulation has been facing big challenges in recent years. In addition to price control, regulators must incentive other important issues, such as productivity gains and environmental responsibility. Therefore, regulation methods that are able to induce the companies into competition, even indirectly, have been largely used in many countries. In this context, Benchmarking techniques have been used by different Energy regulators. These models aim to compare same technology firms, and then to define a relative efficiency score to each one, pointing those which are at the efficiency frontier and are, hence, the sectors benchmark. A frontier analysis technique that has been largely used in Energy Regulation is the Data Envelopment Analysis (DEA), which is used by the Brazilian National Electrical Energy Agency (ANEEL) as part of its regulatory model for energy distribution since 2011. DEA is a non-parametric method that uses linear programming techniques to define relative efficiency scores of comparable firms. These scores are used by the regulator as a reference of efficient operational costs to each Distribution company. These costs are, then, included as part of the consumers energy tariff. DEA method assumes that the compared firms have equal inputs and outputs, besides face equal environmental conditions and requirements. In practical applications, however, its common to observe that firms that face different ambient conditions must be compared. Thus, efficiency score correction in a two-stage approach is a good alternative to deal with these differences, by estimating the environment influence in the efficiency scores and minimizing them. Nonetheless, the estimation technique used in the second stage analysis may change significantly the final scores. Keeping this issue in mind, the objective of this study is to evaluate the effect of the different estimation techniques applied on the efficiency scores generated from DEA. In this study, the DEA model presented by the Brazilian Energy Regulator for the Distribution Companies in the 4th Tariff Review Periodic Cycle was analyzed. It was concluded that the most used estimation techniques in two-stage analysis, Ordinary Least Squares (OLS) and Tobit Regression, cause an effect of reversion to the mean of the sample, independently of the environmental variables combination. Other estimation techniques were applied, as the conditional approach proposed by Banker e Natarajan (2008), where the inefficiency is conditioned to the observed estimation noise. This approach has been applied in two versions: the first kept the noise framework proposed by the authors, and the second substituted it by an adaptation from SFA. The second framework presented a better adjustment than the first. In addition, the results from the conditional approach were better than OLS and TOBIT, since it doesnt make a simple mean reversion. Even though, there is a large number of viable models, which could impact the Regulators decision. Another technique was tested, the three-stage adjustment, proposed by Estelle, Johnson e Ruggiero (2010). This last one presented overestimated scores, comparing to the remaining techniques.