Weight-restricted DEA in action: from expert opinions to mathematical models

Abstract

A common problem in real-world DEA applications is that all inputs and outputs may not be equally relevant to the organizations analysed and their stakeholders. In many cases, one is also faced with a data set where the decisionmaking units do not clearly outnumber the quantity of inputs and outputs. This study reports an application where DEA embellished with weight restrictions is used to analyse the efficiency of public organizations to overcome the above-mentioned problems. Whereas there are numerous documented applications of weight-restricted DEA in the literature, the process of defining the actual weight restrictions is seldom described. However, that part -defining the actual weights restrictions based on price, preference or value information -is the most difficult step involved in using the weight-restricted DEA. Comparing various weight restriction schemes with real data suggests that the ability to consider and include preference information in DEA adds important insights into the analysis.

Making forestry decisions with multiple criteria: A review and an assessment

“…In this case study, six inputs and three outputs were defined, within an input-oriented model, in which, finally, a Tobit composite efficiency model was included. Joro and Viitala (2004) used this example to explain a weight-restricted DEA model. Another exercise with a DEA approach oriented towards public forest services can be seen in Zhang (2002), where one input and three outputs were considered for a Chinese case study.…”

Weight restrictions in DEA: misplaced emphasis?

“…In the few papers that have tried to find weight restrictions by working with sector experts it is difficult to see any generalising principles appearing from the often considerable effort spent (Joro and Viitala 2004).…”

Assessing the performance of biogas plants with multi-criteria and data envelopment analysis

“…Moreover, the inputs and outputs can be weighted in a manner that contradicts the views and/or preferences of the organizations and their stakeholders, or even in a quite counterintuitive manner by valuing secondary inputs or outputs more than priority ones (Joro and Viitala, 2004). In fact, the inputs and outputs are not generally equally relevant and some preference information must be included in the analysis.…”