A Robust Risk/Reward Distribution Model for Integrated Project Delivery


Student thesis: Doctoral Thesis

View graph of relations



Awarding Institution
Award date9 Jun 2020


Integrated Project Delivery (IPD) is an innovative procurement method in which the client, designer (architect and/or engineer), and the contractor will work in an integrated team and under one contract. Therefore, in an IPD context, resources and capabilities of team members are integrated, individual interests are bonded with project interest, and risks and rewards are aligned with project outcomes. The integration under IPD aims to enhance project outcomes and fulfill the team members’ expectations.

Despite the benefits IPD can offer, it is not widely adopted. Among all the reasons, two of the key concerns on the use of IPD are the project risks and risk/reward sharing mechanism. First, the integration advocated may lead to new risks if not all the team members are committed. New methods are needed to deal with the new risks, in particular how they affect the project outcomes. Second, it is stated that all participants who work together contribute to enhancing project value as well as their own interests. However, without understanding what and how to distribute the risk/reward makes the participants reluctant to use the IPD procurement method. Therefore, it is imperative to establish an IPD risks-outcomes evaluation oriented risk/reward distribution mechanism for effective implementation of IPD.

In this regard, two issues should be addressed. The first one is how to estimate the project outcomes through the analysis of the effect of risks on project outcomes in the IPD context. Because new risks arise in an IPD context, the way to align project objectives and risk/reward sharing mechanism should be captured to reflect an IPD environment. Specifically, on one hand, the components of project objectives should be defined in terms of the pertinent monetary value for the client. On the other hand, the critical project risks in IPD projects need to be identified to evaluate IPD project outcomes. The second issue is how to distribute the risk/reward, or pain/gain among project participants. In other words, the risk/reward distribution ratios should be determined. To address these issues, two questions should be answered. First, how to evaluate the effect of project risks on outcomes in an IPD context? Second, how to simulate and optimize the risk/reward distribution mechanism among the project participants?

Though some previous studies have been done in regards to risk/reward distribution mechanism, none of them have addressed the aforementioned two issues, due to the following reasons. First, the integration anticipated in IPD will bring new risks. Thus, enough flexibility should be allowed to deal with these risks. The effect of both new risks and general risks on project outcomes have to be investigated. Secondly, outcome measures were not to be confined to cost performance as conventionally used. Quality and schedule should also be taken into account. Third, the methods used in previous researches are not capable to identify the risk/reward distribution solution, because the researchers assumed the project outcomes are determined before observation, instead of considering the uncertain project outcomes. Fourth, the participants’ risk preferences, i.e. risk averse, risk neutral and risk seeking, have to be considered when setting risk/reward distribution mechanism.

To address the above problems, this research aims to establish a project outcomes evaluation that features risk/reward distribution for IPD type of procurement. Two main objectives of the study are: i) to develop a project risks-outcomes evaluation model by establishing the IPD project risks-outcomes conceptual model incorporating both general risks as well as new risks arising from inappropriate integration. This could include identifying significant project risks, developing the project risks-outcomes evaluation model using Bayesian belief networks, and validating the project risks-outcomes evaluation model. And ii) to develop a risk/reward distribution model by establishing an IPD risk/reward distribution conceptual model considering uncertain project outcomes, expected marginal contribution, and risk preference of each participant. Besides, a validated risk/reward distribution among participants would be proposed. To accomplish these objectives, the prospect theory and two distribution concepts, i.e. uncertain Shapley value and core solution in stochastic cooperative game theory are the theoretical base.

Regarding the research methods, the quantitative method was used because this study aims to develop an IPD project risks-outcomes evaluation oriented risk/reward distribution model. A questionnaire survey was used to collect the data for analysis by statistical tools. T-test was carried out to identify the critical risks that influence the IPD project outcomes. The Bayesian belief network approach was used to analyze the joint impact of IPD pertinent risks on project outcomes. An IPD project risks-outcomes evaluation model was developed using the data collected from the questionnaire survey. Based on the evaluated IPD project outcomes, an optimal distribution model of risk/reward in IPD context has been developed using Karush–Kuhn–Tucker conditions and zero utility principle calculation method. A case study was used to validate the proposed model.

The findings indicate that the most critical project risks in an IPD context are associated with an inefficient multidisciplinary team. In particular, design changes during the construction phase, poor leadership of the multidisciplinary team, poor coordination and iterative design changes throughout the design phase, poor communication, unfair share of risk/reward, project participants’ resistance to taking extra/new responsibilities in the integrated project, and insufficient client involvement throughout the design process is the most critical project risks. The structure of Bayesian belief networks also reveals the criticality of project risks arising from inappropriate integration practices. Furthermore, the form of contract for IPD in Hong Kong remains those used for traditional delivery methods, notwithstanding some integration practices are implemented. The risk/reward distribution model shows that the project participants’ optimal distribution ratio increases with the increase of his loss aversion coefficient λ, whereas decreases with the increases of his gain concavity and loss convexity coefficient α. The IPD project outcomes evaluation oriented risk/reward distribution model was consistent with one IPD Hong Kong case.

This study contributes to a growing body of knowledge on IPD in general and in risk/reward sharing mechanism in particular. First, the insight has gained concerning the effect of new risks as well as general risks on project outcomes in an IPD context. Second, the research not only can evaluate the effect of project risks on project outcomes but also can capture the influence of the uncertain project outcomes, marginal contribution, and risk preferences of project participants on risk/reward distribution solution. Thus, the proposed tool is able to evaluate the project outcomes and suggest an optimal risk/reward distribution among project participants. In practice, the model proposed in this research provides insight into the real motivating shared risk/reward scheme in an IPD context. A good understanding of the joint effect of IPD relevant risks on project outcomes helps project participants to fashion appropriate approaches to manage risks and improve project outcomes. In addition, the risk/reward distribution model can help the client to craft a suitable risk/reward sharing scheme to facilitate IPD adoption. Therefore, the proposed risk/reward distribution model in this research can ease the participants’ resistance to join the risk/reward sharing scheme and promote the use of IPD.

    Research areas

  • Integrated project delivery, project risk, project outcome, risk/reward sharing mechanism, distribution concepts, fair distribution, optimal distribution, Bayesian belief networks