Scale Efficiency and Scale Elasticity in Dea-Models - a Bootstrapping Approach

This paper presents and applies different approaches to estimate returns to scale in multiple-input multiple-output technologies. Scale efficiency gives quantitative information of scale characteristics. A primal based approach to estimate the scale elasticity is proposed as an alternative to the earlier established dual based hyperplane approach. The different scale measures are defined in terms of technical efficiency measures, that can be estimated by Data Envelopment Analysis (DEA). The DEA-method is inherently deterministic and hence offers no measures of uncertainty in the returns to scale estimates. Conditional on the sample at hand, the bootstrap offers a possibility to perform formal statistical hypotheses testing of the scale efficiencies and elasticities. Empirical evidence from (public) eye-care departments in Sweden from the years 1992 and 1993 is presented