The range adjusted measure (RAM) in DEA: Comment

Cooper, Park, and Pastor (1999) propose the range adjusted measure (RAM) as a measure of inefficiency in Data Envelopment Analysis (DEA), This comment purports to show that some maintained properties of RAM hold under specifi c assumptions only. Moreover, RAM is a misleading measure: large and ineffi cient decision making units (DMU) seem to be less efficient than small and inefficient DMUs, according to RAM. An important advantage of conventional radial inefficiency measures is that the solutions of CCR and BCC variants of DEA are unit invariant. On the ot her hand, the radial inefficiency fails to take slacks into account. Thus, radial inefficiency is not strongly monotone in the slacks, which is a desi rable property of an inefficiency measure. As a consequence, radial ineffic iency does not increase if any input or output that has positive slack in t he solution to the envelopement problem (or a. multiplier equal to zero in the multiplier problem) increases in the CCR or BCC variants. Thus, a valid ranking of decision making units (DMUs) with respect to their radial ineff iciency is not possible. To overcome this shortcoming, slack-based measures such as additive DEA mod els have been introduced. These additive models are strongly monotone, but fail to be unit invariant. Moreover, they do not result in a meaningful ine fficiency measure without additional information about the value of inputs and outputs. The results of an additive model is nothing but a sum of (unwe ighted) slacks. If the slacks were weighted by their market prices, ineffic iency would then be simply the sum of excess expenses and foregone revenues (with the technically most successful DMUs, which would serve as the bench marks, being characterized by no excess and foregone revenues).