Endogenous Generalized Weights under DEA Control

Non-parametric efficiency analysis, such as Data Envelopment Analysis (DEA) relies so far on endogenous local or exogenous general weights, based on revealed preferences or market prices. However, as DEA is gaining popularity in regulation and normative budgeting, the strategic interest of the evaluated industry calls for attention. We offer endogenous general prices based on a reformulation of DEA where the units collectively propose the set of weights that maximize their efficiency. Thus, the sector-wide efficiency is then a result of compromising the scores of more specialized smaller units, which also gives a more stable set of weights. The potential application could be to precipitate collective bargaining on cost efficiency for non-marketed resources and products. The models are applied to paneldata from 285 Danish district heating plants, where the open evaluation of multiple non-priced out- puts is relevant. The results show that sector wide weighting schemes favor input/output combinations that are less variable than would individual unit