A critical review of data envelopmem analysis is attempted in respect of some of the basic postulates of economic theory, such as rationality, risk aversion and the dynamics of the learninq process.
I. IntroductionThe non-parametric techniques of efficiency measurement known as data envelopment analysis (DEA) have frequently been applied in economic studies of productivity. The DEA models compare the performance of a decision-making unit (DMU) or a firm with a standard measure. The standard measure, constructed in various ways through suitable linear programming (LP) models, yields potential or frontier output. The difference between the firm's actual output and its potential output enables one to arrive at a measure of the firm's technical efficiency. Recent applications of DEA models include the comparisons of the British prison system by Ganley and Cubbin (1992), information theory applications by Sengupta (1993) and the various public and private sector applications of Fried et al. (1993).The DEA method of measuring the relative efficiency of a set of DMUs or firms has a number of useful and flexible features. The pioneering work of Farrell (1957), who initiated this type of non-parametric method, emphasized its non-parametric feature and the fact that it is entirely data-based. Farrell developed three major efficiency concepts, two at the firm level and one at the industry level. The firm level measures are: technical efficiency and price (allocative) efficiency. The first defines the production frontier which measures the firm's success in producing maximum output from a given set of inputs; and the second defines the allocative efficiency which measures a firm's success in choosing an optimal set of inputs under a given set of market prices. Then there is the industry-level concept of 'structural efficiency' which broadly measures the degree to which an industry keeps up with the performance of its own best firms.