Dynamic Contracting in Infrastructures

(duration: 10/2010 - 10/2014)

Project summary

Tendering is the main method in the public sector for attracting contractors to execute infrastructural projects, such as road construction. The problem is that such tenders result in long-term contracts, while circumstances may change during the building process, leading to new risks and opportunities for both parties. Also, contracts may be based on lacking or false information. Both the government agency and the contractor may have reason to over- or underestimate the actual costs or time frame. This may lead to the building project being rushed or executed sloppily with inferior materials, seriously harming social welfare.

A method for creating more dynamic contracts is thus called for, allowing for renegotiation of details during execution of the project. This PhD-project aims to find a better way for setting up rules for tendering. Among others, the project looks for a way to give contractors incentives to provide more information beforehand; and to support contractors and managers in dealing with changes. Central concerns are:

* Does more exchange of information lead to better cooperation?

* How to build theoretical model for contracting? How to define aspects like social benefit and rationality?

* How can we provide incentives for contractors to ensure project success?

* How to determine which contractor wins the tender? 

* What relation is there between winner determination process and the behavior of contractors?

* How to deal with changing circumstances? 

* How to design an improved tendering procedure?

Almende activities

Dynamic contracting in infrastructures” is a joint project, combining the efforts of a Technology, Policy and Management group and a Computer Science group. Within the Computer Science group, Almende will focus on multi-agent modeling. Both mechanism design and game theory will be used to attempt optimization in problems where agents are self-interested and (partial) information is only privately known to each agent.

Almende will be concerned with the theoretical investigation (modeling and design) of tendering procedures in multi-agent simulations as well as modeling current practice. As a result of this research, Almende is expected to develop a user tool which can help train project managers and contractors and provide more insight into this relatively new field of innovative contracting.

Supervisors

Mathijs de Weerdt (TU Delft)

Andries Stam (Almende)

Links

Dynamic Contracting project page at NGI

Partners