Site Reassignment for Mobile Outreach Teams: Investigating the Effectiveness of Decentralized Decision Making

Problem definition: : United Nations Sustainable Development Goal (SDG) 3.8 states that health coverage should be universal by 2030. This is challenging in rural and poor areas. To address this challenge, mobile outreach teams of healthcare workers visit a fixed set of remote sites to provide healthcare services. Because of dynamics in demand and supply, once-rational site-to-team assignment decisions can become far from optimal over time. This paper considers the problem of reassigning sites to teams to maximize effectiveness. Solving this problem through a central planner does not fit the context: outreach teams commonly have a high degree of decision-making autonomy. We study a decentralized approach where subsets of teams collaborate in a series of team meetings to reassign sites. The key question for decision makers is whether and when such an approach is effective. Methodology/results: : We propose a mixed-integer programming model for centralized site reassignment. We extend this model to represent the decentralized approach and develop a set of simple decision rules for this approach. We use empirical data from six country outreach programs of the nongovernmental organization MSI Reproductive Choices. Our results suggest that, when properly designed, decentralized decision making performs close to centralized decision making, and even outperforms it in the case of inaccurate information at the central level. The finding remains valid when demand and supply fluctuate, and is insensitive to the chosen objective. Managerial implications: : Humanitarian organizations currently deploy mobile outreach teams to provide a wide variety of health services. We present several useful insights for decision makers in humanitarian organizations when making design choices, taking account of context. In particular, we show that decentralized site reassignment provides a feasible and effective alternative to centralized approaches in many contexts.