Panel Paper: Supply-Induced Demand and Black/White Differences in the Use of Congregate Care

Friday, November 8, 2019
Plaza Building: Concourse Level, Plaza Court 8 (Sheraton Denver Downtown)

*Names in bold indicate Presenter

Fred Wulczyn, Chapin Hall, University of Chicago

Racial disparities in child welfare systems are well documented. Compared to white children, Black children are more likely to be reported as maltreated, more likely to have a maltreatment report substantiated, more likely to be placed in out-of-home care, and leave foster care at slower rates. Although these particular disparities are widely acknowledged and important, other forms of disparity have attracted less attention. For example, although Black children are 35% more likely than whites to be placed in congregate care, little is known about why the gap is so large. Because the clinical benefits of congregate care are uncertain, the Black/white difference in the use of congregate care warrants more focused scholarship.

To understand Black/white differences in the use of congregate care, we adopt a two-stage model organized around the idea that the supply of congregate care beds influences congregate care utilization. To measure whether demand is supply induced, we adopt a technique developed by population biologists who study population size in the midst of resource constraints. Known as convergent cross-mapping (CCM), the CCM coefficient measures the relationship between admissions and discharges. Theory suggests that a relationship between admissions and discharges is indicative of processes organized around a resource constraint. In our adaptation of the theory, bed supply represents the resource constraint. In keeping with the model, we computed a CCM coefficient for each of several hundred counties from weekly time series data. We regard the CCM coefficient as a measure of the supply effect. We hypothesize that children in counties where the supply effect is strong will be more likely to go into congregate care.

In the second stage, we built a multilevel model that predicts placement into congregate care. Placement into congregate care refers to whether a teenager entered congregate care. Individual-level covariates include race, age, history of placement, and gender. At the county-level, we use a measure of urbanicity, population-level measures of social disadvantage, and the county supply effect. We also included state fixed-effects to account for between-state differences in the use of congregate care.

Our analysis was organized around supply effects and their influence on differences in congregate care utilization. Simple regression models, without other covariates, show that Blacks are 35% more likely to enter congregate care. Adding age and prior placement history reduces the gap from 35% to 20%. In a full model with a supply effect included, the Black/white difference is reduced to 4 percent because young people are about 80% more likely to enter congregate care if they live in a county where supply effects are strongest.

Supply effects imply that the supply of congregate care beds induces demand. Our results suggest that where admissions and discharges are strongly coupled (i.e., the supply effect), entries to congregate care are more likely. Results also show that Blacks live in counties where supply effects tend to be strongest. Efforts to reduce disparities often focus on decision-making processes. Our research suggests, however, that efforts to reduce disparities should consider how the supply of services affects utilization.