Introducing an Integrated Agent-Based and Reinforcement Learning Model of Contracting and Subcontracting in Construction Sector

The construction sector in many countries, including New Zealand, faces resourcing challenges due to the complexity of construction planning. Contractors play vital roles in providing the capacity and capability to deliver the national pipeline of construction works. Time and expenditure extension and schedule conflicts are barriers to the efficient management of multiple projects in contractors’ portfolios. Understanding the construction contracting and subcontracting mechanism is crucial to comprehending the sector’s complex structure and the origins of the gap between resources and projected demands. This research hypothesised that the contractors and projects are modular entities of a complex system. Accordingly, a model of the construction sector has been developed, and its feasibility has been proved using a sample from the New Zealand Transport Agency’s pipeline of construction projects and prequalified contractors. It investigates the impact of overestimating resources during pipeline development and the tendering process. The study introduces an integrated agent-based modelling and reinforcement learning technique to analyse the mechanics and dynamics of the construction sector. The results inform stakeholders about specific contractors’ capacity and capabilities, providing insights into resource allocation and the risk of delays and overruns. While the model shows promise in improving construction project planning and management, challenges related to the availability and assessment of contractors’ resources need to be addressed for real-world validation. Further development and validation in the next phase of research are necessary to enable the model application on a national-scaled construction projects management and measure the gap in the sector for delivering the projects.