An investigation of returns to scale in data envelopment analysis

This paper discusses the determination of returns to scale (RTS) in data envelopment analysis (DEA). Three basic RTS methods and their modifications are reviewed and the equivalence between these different RTS methods is presented. The effect of multiple optimal DEA solutions on the RTS estimation is studied. It is shown that possible alternate optimal solutions only affect the estimation of RTS on DMUs which should be classified as constant returns to scale (CRS). Modifications to the original RTS methods are developed to avoid the effects of multiple optimal DEA solutions on the RTS estimation. The advantages and disadvantages of these alternative RTS methods are presented so that a proper RTS method can be selected within the context of different applications. (C) 1998 Elsevier Science Ltd. All rights reserved.