The child welfare system in the United States operates as a parallel state — one with the power to enter homes without warrants, separate parents from children, and terminate the legal relationship between a parent and child forever. It does not exercise this power evenly. Black children are roughly twice as likely as white children to enter foster care. Native children are placed at even higher rates relative to their share of the population. In some cities and counties, more than half of all Black children will be the subject of a child protective services investigation before they turn eighteen. The numerical pattern is so persistent across jurisdictions, decades, and methodologies that the discussion has shifted from whether disproportionality exists to what causes it and what to do about it.
The standard explanation offered by agency administrators is that Black families experience more poverty, more housing instability, more exposure to violence, and therefore more of the conditions that produce neglect — and that the system is responding to genuine need. The counter-explanation, developed most forcefully by Dorothy Roberts, is that what the system calls "neglect" is largely poverty itself, that surveillance is concentrated on Black families through schools, hospitals, and public benefits offices, and that the cumulative effect is not protection but a regime of family policing. Both accounts agree on the pattern. They disagree on whether the system is correctly identifying risk or generating disparate outcomes through its own design.
At the collective scale, the dispute is not academic. It determines what reforms make sense. If the problem is underlying social conditions, the response is more services, better screening, and trained caseworkers. If the problem is the system itself, the response is to shrink the system's footprint — narrow the grounds for investigation, raise the evidentiary bar for removal, redirect funds from foster care to direct cash assistance, and end mandatory reporting laws that channel poor families into the agency's purview. The two reform directions are not complementary. Expanding services delivered through the child welfare apparatus deepens surveillance. Cutting surveillance reduces the agency's ability to deliver services through its existing infrastructure. Communities that have organized around this question — Movement for Family Power, upEND, Rise — have largely chosen the abolitionist direction. State legislatures have largely chosen the expansion direction.
The plan that disproportionality demands is therefore not a single program but a choice of model. The expansion model treats foster care as a regrettable but sometimes necessary intervention, and seeks to reduce its racial disparities through bias training, blind referral systems, and structured decision-making tools. Evidence that these tools reduce disparity is thin; some predictive risk models, like the Allegheny Family Screening Tool, have been shown to amplify disparity by treating prior agency contact as a risk factor, which is itself a function of surveillance. The shrinkage model treats the foster care system as the disparity-producing mechanism, and seeks to remove the entry ramps — by repealing universal mandatory reporting, raising the threshold for investigation, providing housing and cash assistance outside the agency, and funding kinship arrangements that do not require parental rights termination.
A second collective question is the relationship between the child welfare system and the criminal legal system. Reiko Boyd and others have documented that the two systems share informants, share data, and share a population. A parent arrested for a low-level drug offense often loses custody before they have been convicted of anything; a parent who fails a drug test in a hospital can lose a newborn before leaving the maternity ward. The 1997 Adoption and Safe Families Act tied federal funding to adoption targets and imposed a fifteen-month clock on reunification, which means a parent serving even a short prison sentence can lose their child permanently regardless of the underlying conduct. The disproportionality in child welfare is not a separate phenomenon from mass incarceration; it is the same population processed by an adjacent agency with a different vocabulary.
The deepest collective implication is that the metric matters. If success is measured by adoptions completed, the system is performing — adoptions hit record highs through the 2000s. If success is measured by families intact, the system is failing the populations it most often touches. If success is measured by long-term outcomes for children — educational attainment, mental health, criminal legal involvement in adulthood — foster care performs worse on average than leaving children in marginal homes, controlling for case characteristics, as studies by Joseph Doyle and others have shown. A planning framework that takes Law 4 seriously cannot accept the agency's own success metrics. It has to ask what the children, the parents, and the communities at the receiving end of the system actually experience over decades, and design accordingly. The current system was not designed that way. It was designed to remove children from circumstances officials found morally objectionable, and the racial pattern of who gets removed reflects whose circumstances officials have historically found objectionable.