Measuring the technical efficiency of production

Abstract

Production technology is commonly modelled by means of a production function, which in the scalar output case specifies the maximum output obtainable from an input vector. The degree to which the actual output of a production unit approaches its maximum is called the technical efficiency of production. A technically efficient unit must operate on its production function, although this condition is not sufficient; a technically inefficient unit may operate beneath its production function, although this condition is not necessary. If the notion of technical efficiency is to have empirical content, it must be based on a proper measure, or index, of the technical efficiency of a production unit. It is the purpose of this paper to specify a set of properties such an efficiency measure should satisfy, to show that the widely used measure proposed by Farrell [4] does not satisfy these properties, and to introduce a new measure that does satisfy these properties. The pioneering work on technical efficiency is that of Farrell. Inspired by the work of Debreu [2] and Koopmans [7], Farrell suggested a measure of technical efficiency that can be interpreted, somewhat loosely for the moment, in either of two ways: as the ratio of technically minimal to actual inputs, given output and the input mix, or as the ratio of actual to technically