Linear programming approaches to the measurement and analysis of productive efficiency (Q1893517)

This paper is a survey of the operations research/management science/economics literature which uses linear programming techniques to calculate indexes of producer performance. Among the indexes are measures of technical efficiency (distance, or gauge, functions), indexes of scale efficiency, indexes of cost efficiency and indexes or revenue efficiency. Linear programs are constructed which (i) characterize the structure of best-practice production technology, and (ii) provide indexes of producer performance in terms of distance from best practice. Information from both primal and dual programs is exploited in the analysis. Among the topics covered are (i) basic models, (ii) complementary slackness, (iii) relaxing constant returns to scale, (iv) modifying strong disposability, (v) constraining shadow prices, (vi) relaxing convexity, (vii) allowing nonpositive variables, (viii) incorporating nondiscretionary and categorical variables, and (ix) introducing a stochastic element by means of chance-constrained programming techniques.