Extensions of Nonparametric Approach

Abstract

Nonparametric efficiency analysis is one of the most active fields of current research on measuring productivity of public sector enterprises. Since this involves several disciplines such as economics, management science, statistics and operations research, it is quite natural that future generalizations would be forthcoming in all these disciplines. Although Farrell developed his efficiency concept in the context of a production frontier, his interest was mainly statistical. The DEA model on the other hand generalized the input-output ratio measure to multiple inputs and outputs, which is basically a technical measure for evaluating engineering components and plants, although this tool was widely applied in school systems, hospitals and other quasi-market organizations. The DEA model was basically designed to compare managerial efficiency of a set of relatively homogeneous DMUs. The economic theory of efficiency has sought to characterize efficiency in terms of certain implicit prices associated with certain subsets of the feasible production set. In case of single output and multiple inputs the inefficiency of a set of micro units may simply be measured as a deviation from the full capacity level of utilization. Such deviations or distortions are not so simple to measure in case of multiple outputs, since the detailed data on allocation of every input to all possible outputs are not at all available. Problems also remain in the dynamic framework, when efficiency variations over time have to be related to changes in inputs, in capacity and in the technology. Finally, there is the econometric problem of estimating the production frontier but in a nonparametric way.