Congestion measurement and elimination under the framework of data envelopment analysis

Congestion is a wasteful stage of the production process where outputs are reduced due to excessive amount of inputs. This paper investigates three types of models prevailing in the data envelopment analysis literature, i.e., the Färe-family model, the Cooper-family model, and the Wei–Tone model, for measuring the congestion effect. The Färe-family model is able to decompose technical efficiency into the product of pure technical efficiency and congestion effect, yet fails to detect congestion in some situations. The Cooper-family model successfully detects congestion in all situations, yet is unable to differentiate pure technical efficiency with the congestion effect. The Wei–Tone model does not have the aforementioned deficiencies, yet is unable to identify the excessive amounts of each input. To illustrate how to eliminate congestion in reality, the Wei–Tone model is utilized to measure the congestion effect of Taiwan forests. There are three forest districts found to have different extents of congestion. The theoretical method of deleting excessive inputs to improve efficiency is actually impractical in this case. By means of reorganization, this paper finds that the congestion effects of two districts are eliminated and the other is alleviated.