Statistical evaluation of Data Envelopment Analysis versus COLS Cobb–Douglas benchmarking models for the 2011 Brazilian tariff revision

• We evaluate the benchmarking model used by the Brazilian regulator in 2011.
• Major differences are reported between DEA and Cobb–Douglas COLS efficiency scores.
• We provide a detailed statistical analysis of the problem.
• We point out major statistical issues of the COLS Cobb–Douglas.
• We present statistical tests to compare DEA and COLS.

In 2011, the Brazilian Electricity Regulator (ANEEL) implemented a benchmarking model to evaluate the operational efficiency of power distribution utilities. The model is based on two benchmarking methods: Data Envelopment Analysis (DEA) and Corrected Ordinary Least Squares (COLS) with a Cobb Douglas production function. Although the estimated scores are highly correlated, differences between the scores are as high as 41%. For some companies differences between the efficiency scores result in substantial reduction in regulatory operational costs. We provide a detailed statistical comparison which indicates that the COLS Cobb Douglas model has major deficiencies in terms of estimating efficiency scores.