Using DEA to identify and manage congestion

This paper deals with identifying and managing congestion. For this purpose, data envelopment analysis (DEA) is used to identify congestion when the data show it to be present, estimate its amounts, and separate it from other forms of inefficiency. DEA is also used to identify where improvements may be made in the management of congestion and to estimate input decreases and output increases that may be made after managerial inefficiencies in managing congestion are eliminated. The treatment here differs from the usual approaches that are restricted to identifying sources and amounts of technical inefficiency and congestion to be eliminated. The focus is directed rather to efficiency of performances in the presence of inefficiencies imposed by, say, labor contracts or government regulations and policies. Other developments include a use of rates of substitution formulated in terms of slack variables that help to avoid instabilities associated with the very small values that are often encountered in the use of dual variables to determine the rates of substitution. These rates of substitution are intended for use in guiding allocations ( or reallocations) of inputs between different plants ( or other entities) in ways that can further improve performance without reducing the congesting inputs that are to be employed. Hence modi. cations are needed to extend the usual restrictions to movements on the efficiency frontier so that frontiers associated with congestion and other inefficiencies can be dealt with.