Bootstrapping DEA-based Efficiency Measures and Malmquist Productivity Indices. A Study of Swedish Eye-Care Service Provision

The purpose of this paper is to use the bootstrap resampling technique to calculate confidence intervals for efficiency measures and Malmquist productivity indices. The efficiency and productivity measures are obtained from non-parametric linear programming models using primal production data. The bootstrap procedure is applied on a sample of Swedish eye-care departments, using data from the years 1992 and 1993. Significant positive changes in productivity are found mainly for departments located at medium sized hospitals. On average, the positive productivity change is due to a progressive development of the technology.