Abstract: This paper discusses the estimation of returns to scale using data envelopment analysis.The previous researches mainly focus on calculating the scale elasticity with the same proportion changes of entire output and input combination.By introducing a more general model,it is possible to calculate the scale elasticity of any input-output combination.Based on this model,the key input components,which are in the dominant position to determine scale elasticity,can be determined.The investment of expanding input resources can be used in a more efficient way if only the key input components are expanded.With the best investment combination,we may receive maximum output utility with minimum input.

Key words: DEA, scale elasticity, returns to scale, best investment combination