Organizing Logic of the Child Welfare System

This post is even more half-baked than most in the sense that I’m unsure about its conclusions. I’d love to hear feedback.

The design and operation of Child Protective Services (CPS) is a topic often shrouded in pain and emotion, with big risks at play. This essay aims to take a dispassionate step back and survey the incentives that shape CPS, as part of the justice system, and how that system interacts with child and parent wellbeing.


As a starting point, I assert that society chooses some minimum standard of care for children. The standard originates from voters’ preferred tradeoffs between child safety, family autonomy, and everything else. Those preferences get translated into policy, subject to budget constraints. Naturally, what behaviors are called child neglect and abuse can drift over time.

Having established a minimum standard of care, the state monitors for violations. When a violation is found, several mechanisms are available for remediation, but they are all backed by legal power to remove children from their parent’s custody. Child removal is used to (1) directly stop individual cases of abuse or neglect and (2) deter parents who may never be observed from violating the minimum standard of care.

Removing children from their parents is usually traumatic for both parties. Since this pain has a deterrent effect for parents, there is no reason to improve parents’ experience. There is not a deterrence goal for children, however, so CPS involvement should be made as beneficial for children as possible.

Incidence of CPS Involvement

Parents have a budget. They spend some on themselves and some on their children. Even with wide variation in how parents make this split, parents with smaller budgets must spend less on their children on average. If parents’ ability to meet the CPS’ minimum standard of care is associated with money (or other correlates of income like education), most violators will be low-income parents.

Getting poor violators to change their parenting requires more coercion than for rich violators. Since we typically think each next dollar of consumption is worth a little bit less happiness (i.e., decreasing marginal utility of consumption), it’s easier for a rich parent to re-allocate consumption from herself to her child. A poor parent must dig deeper and, at the limit, may not be able to meet CPS’ standard for care at all.

With an imperfect ability to detect child neglect and abuse, the state focuses most heavily on communities with the highest incidence. Reliance on stereotyping creates over-monitoring of some communities beyond what is justified by the policy. This higher level of monitoring also incentivizes those families to devote more effort to performative signals of care, perhaps skewing the family’s budget away from what is optimal.

Impacts on Children

Imagine that CPS cases fall along a continuum between gross abuse and healthy homes. For some children, removal is a clear benefit, and, for others, it would be a clear harm. Choosing the cutoff for which children get removed is not simple.

If CPS’ sole motivation were to prevent child harm by direct intervention, they would only separate children from parents when it was expected to be a net benefit to the child. If there were a societal bias toward keeping children in their home, perhaps only extreme cases would be removed, leaving some children at home who could benefit from entering CPS custody.

However, as mentioned, given that CPS cannot observe every parent’s behavior, their enforcement actions have a deterrence effect. Fear of CPS involvement scares parents into taking better care of their kids. If CPS detects only a small share of violations, their interventions must be draconian to maintain an effective deterrent.

When deterrence is a motivation for CPS removal, there’s a tradeoff between the wellbeing of (potentially) removed children and the rest of the child population. By removing some children who would be better off at home, they can scare more parents into careful compliance – a net improvement for the whole child population.

Even though an explicit deterrence policy would upset many members of the public, deterrence can still by pursued implicitly since it is hard to predict which children benefit from removal.

Which policy is used in practice to set the child removal cutoff? A straightforward policy of removing children when it’s beneficial should result in zero impact on the marginal child; a pro-family bias would produce a positive impact at the margin; and a deterrence-based policy would produce a negative impact on the marginal child. Empirical studies of the net impact of removal on marginal children (based on randomized assignment to case workers) show mixed results so there is no clear answer.

Importantly, the impact of removal in marginal cases will be more negative than for the average case. And, hard as it is to swallow, even if CPS removals hurt some children, that may be a feature not a bug. A perfectly executed CPS policy might provide immense societal benefit even while its marginal decisions hurt individual children.

Carrots and Sticks

Many critiques of child protective services focus on the large share of its budgets spent on child removals and foster care, leaving little money for supporting families to prevent child removal. Especially in cases of neglect, its reasonable to think that cash transfers to parents would do better than the current approach.

The biggest problem with resolving child neglect at scale via supportive policies is targeting. If parents knew they could receive cash or valuable services if they were suspected of maltreating their child, it would create a perverse incentive toward maltreatment. That is, CPS would achieve the exact opposite of the deterrence effect discussed above.

In the status quo, family supports that CPS provides are either limited or not something most parents would pick for themselves, like counseling. CPS involvement remains something that parents do not want. More substantial family support is provided by the government, but it is broadly targeted to avoid creating perverse incentives. The Child Tax Credit, Earned Income Tax Credit, WIC, TANF, Medicaid, and free public education are all ways the government funnels money and services to parents.

A second reason why larger subsidies for at-risk families would not prevent child removals is that CPS’ bar for neglect and abuse is not fixed. I asserted in the introduction that the minimum standard of care is determined by voters’ preferred tradeoff between child safety and family autonomy. If parents start taking better care of their children, voters can “buy” more child safety without further compromising family autonomy. Standards will always rise in line with parents’ budgets.

Brazil’s Bolsa Familia provides one clear example of these shifting standards. Since 2001, Brazil’s federal government has provided conditional cash transfers to parents in poverty if their children attend school and receive vaccines. It’s a longstanding successful policy that has dramatically reduced child poverty in Brazil. Nevertheless, registered cases of child neglect and abuse in Brazil have risen over the same time period.

As a counterexample, somebody might point to Nordic countries with generous subsidies for families and low rates of child removals. But all that contrast demonstrates is that Nordic voters have a lower taste for child removals than American voters. If standards for child maltreatment were fixed and providing resources to parents would prevent maltreatment, then we should see a long-term downward trend in CPS cases as society gets richer. Instead, CPS case counts have stayed roughly steady, in both Norway and the United States.

Parallels to Criminal Justice

Every point in my description of Child Protective Services has a parallel to the criminal justice system. Like police, CPS caseworkers do a thorny, important job – usually motivated by a sense of public service. Like the police, CPS wields an incredible capability for violence whose use should be scrutinized. There is plenty of room to improve. But, like the police, the organizing logic of CPS is sound.

Finding Your Role

Anybody reading this far probably has a personal interest in improving the lives of children. But figuring out what role to play or which role you already play can be tough.

The figure below illustrates how I think about it. One must choose between programs that interdict and mitigate bad parenting (i.e., CPS, foster agencies) or support good parenting (e.g., family subsidies, community programs). Interdiction is always run by the government because of its monopoly on violence; programs for mitigation (foster care and adoption) and family support can be a mix of public and private programs. Because of incentive problems, interdiction and mitigation target families that need help most closely; family support programs must target a more general audience. Organizations specialize in one of these roles both because of the different levels of targeting and because mixing them creates problems with institutional trust: parents are wary of accepting help from an organization that can take their children away.

One must also choose between changing the quantity or quality of these programs. Quantity is set by voters in the long run so changing it is a matter of political advocacy.1 Changing the quality of programs normally requires working in them, innovating, and doing the job better than somebody else would do it.


  1. You could try to increase quantity of family support programs by raising private funds. But voters could compensate by lowering public support in response. 

Written on March 1, 2023