A novel single-principal-multiple-agent repeated game model and solution method for performance assurance of emergency supplies

Lin Zhang,Chuangyin Dang,Richard Y.K. Fung,Jun Tian

DOI: https://doi.org/10.1016/j.cie.2021.107466

2021-09-01

Abstract:<p>Guaranteeing the quality or performance of emergency supplies is crucial for the prompt execution of disaster relief or meeting emergency needs of the affected people. However, given the incentives of gaining subsidies or reducing costs, agents who operate reserve rotation commonly take hidden actions for their own benefits. Such repeated moral hazard problems become serious for the capacity-constrained principal who should simultaneously monitor multiple agents. Considering the physical state transition determined by the agents' hidden actions, this study proposes the definition, conditions, and generating algorithms of (Restricted) Experience Based Equilibria in multi-agent systems with the action selection constraint of the principal to tackle this type of problem. Grain quality assurance problem in a two-enterprise system is then adopted to test the validity of the game model and solution methods, where the rationality and generality of the proposed strategies are further verified in three-enterprise systems. An interesting result from comparing the on-equilibrium strategies with and without the monitoring capacity constraint shows that the lower the monitoring capacity, the higher monitoring effort the principal must exert. This finding reveals that the monitoring capacity is not for actual use but for guiding and threatening the action choices of the agents. Therefore, when the principal fails to improve the monitoring capacity in a short period, making strategies that could show "virtual capabilities" is better. However, from the perspective of long-term stability, the necessity for the principal to improve this capacity is absolute.</p>

computer science, interdisciplinary applications,engineering, industrial