Efficiency Measurement of Organizations in Multi-Stage Systems

Traditional Data Envelopment Analysis (DEA) characterizes decision making units by a vector of external inputs and outputs. By the use of a scalarizing function the inputs and outputs are aggregated to an efficiency measure for each unit. DEA models are based on the assumption that the production process is a "black box", i.e. inputs are transformed in this box into outputs. In many cases more information about the production process is available. This is especially the case in multi-stage production systems. Decision making units of the underlying network employ intermediate and external inputs simultaneously. Unlike external inputs, which are assumed in classical models, intermediate inputs are provided directly by decision making units of the network. This means that intermediate goods affect the performance measure of at least two decision making units, the unit providing services and the unit applying these services. For this very reason a general DEA-model is introduced, which takes the special features of units in a multi-stage System into consideration.