In this paper an overview of DEA (Data Envelopment Analysis) is given. The assumptions underlying its use for project selection is discussed. A proposal is made to form all combinations of projects that satisfy the budget restriction, and perform the analysis on these combinations. One and only one combination will then have to be chosen, and this will be either the do nothing alternative or one of the combinations found efficient by DEA. In the final stage of the selection process, a comparison is made between the shadow prices of the efficient combinations and the preferences underlying each scenario. The authors apply this method with some real data from the next Norwegian National Road Plan, and relate the experiences and how the method is received by administrators. The method is simpler and has more intuitive appeal than the "cone-ratio" model that was used in previous applications of DEA to project selection.
Abstract