Supply, Utilization, and the Long Run Demand for Residential Care: Results of An Agent Based Simulation

Background and Purpose: What is perhaps most striking about the foster care system in the U.S. is how little we use what little we know to pursue better services for what is arguably the nation's most vulnerable group of children. For example, despite a growing body of evidence, the practice of placing children into group and residential care persists. The question is why and what can be done to align practice with what is known about the benefits of group care? In this paper, we explore the interaction between bed supply and placement in residential care using agent-based simulation. The model is organized around the assumption that placing children into group settings lingers to some extent because the bed supply dictates demand. The supply lingers because public officials are unable to describe the long run demand beds other than to assert that the true demand is not zero. Providers, on the other hand, are unwilling to manage their supply of services up or down without a clear indication of where the “market” is headed. We address these complex interactions in this paper.

Methods: Our principal method is agent-based simulation. Taking utilization and supply data from a single state, we first model entries and exits into out-of-home care, including transfers between regular foster care and group care and reentry. These transition matrices are then used to parameterize an agent-based model. With regard to the overall structure of the model, we use the linear partial adjustment model (LPA) described in Social Dynamics (Tuma and Hannan, 1983). The LPA connects the future size of a population to an adjustment parameter that describes how quickly a system responds to the gap between the actual population and a target (preferred) size. The target is the number of children in residential care relative to beds (utilization). We adjust the response time and the tolerance for under/over utilization to observe how the system responds to changes in the entry, exit, and transfer rates.

Results: First, the results demonstrate the utility of agent-based models for understanding systems that exhibit complex micro and macro level behaviors. The results also show that the supply of beds is most sensitive to the interaction between changes in ‘intake' and the time needed to respond to the supply/utilization gap. Slow response times produce large waiting lists whereas fast response times lead to surplus capacity, which drives placements up through feedback mechanisms that operate at the micro-level.

Conclusions and Implications: Viewed from afar, the residential care system exhibits behavior that is consistent with the complex cross-level interactions. For planning purposes, public and private child welfare agencies, not to mention children, benefit when the supply of residential care beds stands at a level that is responsive to the level of need at any given time. Agent-based simulation provides a way to understand how one might better manage the supply/demand equilibrium.