Design of experiments applied to environmental variables analysis in electricity utilities efficiency: The Brazilian case

Benchmarking plays a central role in the regulatory scene. Regulators set tariffs according to a performance standard and, if the companies can outperform such a standard, they can retain the gains observed by such outperformance. Efficiency performance is usually assessed by comparison (or a benchmark) against either other companies or the company's own historical performance. This paper discusses the impact of environmental variables on the efficiency performance of electricity distribution companies. Indeed, such variables, which are argued to be unmanageable, may affect the electricity utilities' performance. Thus, this paper proposes a simulation methodology based on design of experiment philosophy for statistically testing environmental variables and the interactions among them, enabling regulators to build the best suited semi-parametric two-stage model of electricity utility benchmarking analysis. To demonstrate the power of the proposed approach, experimental simulations are carried out using real data published by Brazil's regulator. The results show that environmental variables may impact efficiency performance linearly and nonlinearly.