Decentralized beneficiary behavior in humanitarian supply chains: models, performance bounds, and coordination mechanisms

Effectiveness in humanitarian supply chain operations depends on the critical last mile between beneficiaries and needed supplies or services. Often, the last mile is traveled by the beneficiaries themselves. This paper’s focus is on systems in which beneficiaries make autonomous decisions about where to seek supplies or services using a utility function that captures distance, congestion, and the relative importance of the two factors. We model beneficiary behavior as a network congestion game where the resources are a set of facilities from which individuals choose. Importantly, our models capture the fact that the relative importance of distance and congestion may be specific to both the individual and the facility; we represent this using a factor called the congestion weight. We prove new bounds on the system performance that results from decentralized beneficiary decisions in comparison to centralized optimal assignments, and we introduce mechanisms for achieving centrally optimal outcomes even in the presence of decentralization. We demonstrate the methods with data from the international public health response to the Haiti cholera epidemic.