Data envelopment analysis (DEA),first introduced by Farrell (1957) and successively developed by Charnes, Cooper e Rhodes (1978), is a linear programming technique; it defines the best practice frontier that serves as a benchmark and computes the relative distance between each unit and the frontier.
This distance can be interpreted as the economic performance of the units in the sample. Within the context of composite indicators this interpretation has been used to reassess indicators, see for example Mahlberg and Obersteiner (2001); Despotis (2005), and their reassessment of the Human Development Index; also see Somarriba and Pena (2009), and Sharpe and Andrews (2010) for applications within the context of quality of life and economic well being respectively.
In this paper, however, we make use of the distance from the best practice frontier as an efficiency measure to correct a composite indicator of endowment. In fact whenever it is reasonable to assume non-substitutability among the sub-indicators, their weighted average should also take into account the combination (or relative proportion) between the sub indicators used in the aggregation function.
You can access the paper here.