Analyzing firm performance in a glass industry: a non-parametric frontier approach

Abstract: A non-parametric data envelopment analysis (DEA) method is applied in this study in order for calculating the technical and scale efficiency of a glass factory by means of a production frontier, despite the common trend among contemporary literatures to use stochastic parametric methods and ‘estimate’ the frontier production function. The available aggregated structure of the firm data paved the way for the authors to rethink and adopt a mathematical (linear programming) approach which would make it possible to conduct the study with a limited number of observations and without the detailed form of the production function. Encouraged by a significant number of comparative studies of parametric and non-parametric methods along with the increasing strength of the updated DEA models; an informed choice was made to determine an input-oriented, variable returns to scale (VRS), multi-stage slack calculation DEA model. The results of the efficiency analysis found four decision making units (DMUs) inefficient in the constant returns to scale (CRS) frontier with the existence of scale efficiencies (SE), the rest of the sixteen DMUs were efficient in both the CRS and the VRS frontiers. In general, the results turned out to be quite in line with prior observation, coherent with the pattern of fluctuating production levels given certain unbalanced amount of labor employed for different decision-making units (DMUs) and the respective level of wastes in production.