Comparison of different methods for estimation of returns to scale in nonradial data envelopment analysis models

The data envelopment analysis (DEA) approach has been actively developed in recent years and is used to analyze the activities of complex production units (regions, financial institutions, industrial enterprises, etc.). An important role in such an analysis is played by the calculation of various indicators of the activity of units: returns to scale, efficiency scores, marginal rates of substitution, transformations, etc. Dependences between variables in the DEA models are not explicitly specified; therefore, special optimization models are used to calculate these indicators. Much attention is given in the scientific literature to the estimation of returns to scale. This paper describes and compares some of the best known methods for calculating returns to scale. Computational experiments show that under certain conditions, the approach proposed by the authors has advantages over other methods.