The presented work provides a framework for estimating output-specific efficiencies, first simulation evidence about the performance of the corresponding estimators for a variety of settings and a first application highlighting the potential of this approach to reveal differences which otherwise would remain undetected. The presented approach should be considered as an extension of Data Envelopment Analysis (DEA), a widely used and thoroughly investigated efficiency frontier estimation technique. The proposed approach allows to address the question of systematic efficiency differences in a statistical sense across output categories, which can not be tackled by DEA itself. On the other hand it relies in fundamental ways on DEA capabilities of generating valuable information about technological possibilities (best practice) and about posterior feasible values of the unknown target output-ratios from the sample observations of input and output quantities.
KeywordsCovariance Autocorrelation Estima
Unable to display preview. Download preview PDF.