Abstract
A method is presented for classifying strongly efficient units in DEA as interior or exterior, and as self-evaluators or active peers. The exterior strongly efficient units are found by running the enveloping procedure “from below”. There is no firm production-function evidence of the efficiency of exterior self-evaluators. Interior self-evaluators are more likely to have active peers as neighbours in more directions and may therefore represent technology. When performing a second stage regression analysis of efficiency scores, exterior self-evaluators should be removed. The proportion of exterior active peers also provides information on whether the variable specification is supported by the data.