Multiobjective Bayesian network model for public-private partnership decision support

To improve the chance of success of a public-private partnership (PPP) scheme, it is essential to consider the feasibility of the scheme both from the economical and noneconomical perspectives according to the interests of all three key stakeholders, namely the government, the private investor, and end-users. Acknowledging the diverse and sometimes conflicting interests of the stakeholders, decision makers must identify a viable scheme that could satisfy public accountability, commercial interests, and social consideration of the government, investor, and community, respectively. However, because each decision item could have several possible values or states, it is difficult for decision makers to come up with different PPP schemes by adopting the conventional analytical methods. This paper proposes the use of Bayesian network (BN) techniques to imitate human reasoning and conduct multiobjective decision making. By establishing a decision network that connects the decision items, evaluating criteria, and the ultimate objectives (i.e., the satisfaction of the three main stakeholders), evaluation can be conducted through the BN and the noisy-OR gate concepts. A weighted score approach is applied to combine the objectives of the three stakeholders into a single value. This enables decision makers to evaluate and compare different PPP alternatives and identify a suitable strategy that could minimize the conflict, thereby ultimately increasing the chance of success of a PPP scheme.