Weight Restrictions on Geography Variables in the DEA Benchmarking Model for Norwegian Electricity Distribution Companies

• Type: Technical Report
• Date: 2008

Authors

Ana Maria Cunha Ribeiro dos Santos Ponces Camanho (Author) FEUP; Endre Bjørndal (Author) Other; Mette Bjørndal (Author) Other

Scientific classification

FOS: Social sciences > Economics and Business

CORDIS: Physical sciences > Mathematics > Applied mathematics > Operations research

Other information

Resumo (PT): We examine weight restrictions in the DEA model for distribution networks, taking as the starting point the NVE model with one input, total cost, and several outputs. In the unrestricted DEA models, we notice large differences in absolute and relative shadow prices, and for some companies, extreme weight on “geography” variables in the cost norms. There seems to be a tendency that companies with a large weight on geographic variables and / or a low weight on transported energy and customers become super efficient. This seems unreasonable, and one remedy may be to restrict prices / weights for individual outputs, or combinations of outputs. We consider absolute, relative and virtual weight restrictions, and show how to formulate the LP problems and how to interpret the restrictions. We discuss the relative price restrictions suggested for geography and high voltage variables by NVE (2008), and consider an alternative approach, using virtual weight restrictions on the combination of the three geography variables; forest, snow, and coast. Comparing the effects of the virtual approach to the relative, we notice that with relative weight restrictions, more companies are affected, but to a lesser extent. An important task when introducing weight restrictions in the DEA analyses is to determine the specific limits on the weights. Finding reasonable limits, depends on which type of weight restrictions that are considered, and should be based on knowledge of cost and technology in the industry. An advantage of the virtual weight restrictions is that they are on a more aggregated level than the relative ones, and it may be easier to establish limits on the overall effects on the total cost norm from a subset of outputs, rather than reasonable pair-wise comparisons of outputs weights. Finally, the report discusses implementation of DEA models with weight restrictions, and gives a short overview of available software.

Language: Portuguese

Type (Professor's evaluation): Scientific

No. of pages: 80