Evolutionary game analysis on decision-making behaviors of participants in mega projects

Mega-projects, characterized by large investments, long durations, high complexity, and significant socioeconomic impacts, are prone to collusion among participants seeking additional benefits. The commissioned-agent construction system prevalent in China, where the government manages projects through owners, creates a principal-agent relationship. Collusion between the owner and constructor may lead the supervisor to join, forming a collusion-body. This poses significant risks. Existing research addresses specific aspects of mega-project governance, such as lifecycle approaches, digital technologies, and institutional mechanisms. However, a comprehensive model integrating the interactions of the collusion-body, government, and public is lacking. This study aims to address this gap by utilizing evolutionary game theory to analyze the strategic choices of these participants.

Prior research on mega-project governance focuses on various aspects including lifecycle management, the implementation of AI and digital technologies, establishing robust organizational procedures, and improving governance modernization. Studies have explored the “enterprise-government-society” framework for social responsibility and the “government-market” duality for organizational models. Existing evolutionary game models in this context have analyzed the interactions between subsets of participants, such as local governments and the public or government, contractor, and public, but not a comprehensive model including the collusion-body, government, and the public. This study fills this gap by incorporating all three.

This paper develops a tripartite evolutionary game model involving a collusion-body (owner, constructor, supervisor), the government, and the public. The model assumes limited rationality for all participants, who dynamically adjust strategies based on cost-benefit analysis. Each participant has two strategic options: the collusion-body chooses to collude or not; the government chooses active or formal intervention; and the public chooses to participate in supervision or not. Probabilities are assigned to each strategy choice (x, y, z). The payoff matrix outlines the costs (C) and benefits (R), fines (P), losses (L), and compensations (R) associated with each strategic combination. Replicator dynamic equations are derived for each participant to model the evolution of their strategy choices over time. These equations are analyzed to determine evolutionary stable strategies (ESS) and their stability conditions. Simulation analysis is used to illustrate the evolutionary paths of strategies under different parameter settings, focusing on cost differences, additional benefits of collusion, cost differences of government intervention, government intervention rewards, costs of public participation and rewards for public participation.

The analysis reveals several key findings: 1. **Collusion-body and Public:** The probability of public participation increases with the additional benefits of collusion and the cost difference between collusion and non-collusion. There is a critical probability of public participation; below this, the collusion-body always colludes; above it, they never collude. 2. **Collusion-body and Government:** Increased government rewards for active intervention reduce the likelihood of collusion. Surprisingly, a larger cost difference between active and formal intervention increases the likelihood of collusion. 3. **Government and Public:** Government rewards for public supervision and the cost of public supervision influence government intervention strategy. A critical probability for active government intervention exists, influencing public participation. 4. **System Stability:** Eight pure-strategy Nash equilibrium points exist. Three are unstable. Five (E3, E4, E6, E7, E8) may be ESS, each under specific cost-benefit conditions. E4 (no collusion, active intervention, public participation) represents the ideal state. 5. **Simulation Analysis:** Simulation confirms theoretical findings and illustrates the dynamic evolution of strategies over time under different parameter changes. Thresholds exist for several key parameters, influencing the overall system's evolution towards different ESS. For example, increased costs of collusion or increased rewards for government action shift the system towards the ideal strategy (E4).

The findings highlight the interconnectedness of the decision-making behaviors of the collusion-body, government, and public. The model demonstrates that active government intervention and public participation are crucial for effective governance. The existence of multiple ESS reveals the complexity of the problem, emphasizing that the optimal outcome depends on the specific cost-benefit structure. The counterintuitive finding that a greater difference between active and formal government intervention costs increases the likelihood of collusion suggests the importance of preventing governmental inaction or formalism.

This paper provides a novel framework for analyzing collusion in mega-projects using evolutionary game theory. The model's findings underscore the critical role of active government intervention and public participation in deterring collusion. Future research could extend the model to incorporate more nuanced factors, such as asymmetric information, risk aversion, and the impact of specific policy instruments. Further empirical investigation to validate the model’s predictions across various mega-project contexts would enhance its practical applicability.

The model relies on several simplifying assumptions, including limited rationality and the quantification of costs and benefits. The accuracy of the model's predictions depends on the reliability of parameter estimation. The model's generalizability may be affected by variations in the specific institutional contexts of different mega-projects.