Title: The trichotomic approach for dealing with uncertainty in project portfolio selection: combining MCDA, mathematical programming and Monte Carlo simulation

Authors: George Mavrotas; Olena Pechak

Addresses: Laboratory of Industrial and Energy Economics, School of Chemical Engineering, National Technical University of Athens, Zographou Campus, Athens 15780, Greece ' Laboratory of Industrial and Energy Economics, School of Chemical Engineering, National Technical University of Athens, Zographou Campus, Athens 15780, Greece

Abstract: Project portfolio selection refers to the problem of selecting a certain number from a wide set of proposed projects. In this work, we also address inherent uncertainty, present either in project characteristics (e.g., costs, performances) or in decision environment (e.g., criteria weights, total budget). We model uncertainty with probability distributions and we use Monte Carlo simulation to produce MCDA parameters as well as parameters for mathematical programming model. In this way, set of projects is divided into three subsets: 'green' projects that are always present in portfolio, 'red' projects that are never selected and 'grey' projects that are selected in some of Monte Carlo runs. Subsequently, the focus is moved on grey projects for further elaboration. The so called trichotomic approach offers much more fruitful information to the decision maker as it quantifies the degree of certainty with which each project is selected or not in the final portfolio.

Keywords: project portfolio selection; capital budgeting; multicriteria decision analysis; integer programming; uncertainty; MCDA; mathematical programming; Monte Carlo simulation; modelling; probability distributions; decision making.

DOI: 10.1504/IJMCDM.2013.052474

International Journal of Multicriteria Decision Making, 2013 Vol.3 No.1, pp.79 - 96

Received: 04 Oct 2012
Accepted: 28 Nov 2012

Published online: 05 Mar 2013 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article