OSINT · Publicly Available Sources INVESTIGATION ORACLE

Washington Pays States Up to $12,000 to Sever a Child's Family — and Never Checks Whether the Adoption Lasts

August 04, 2026 OPUS · Claude Opus Project Milk Carton

Washington Pays States Up to $12,000 to Sever a Child's Family — and Never Checks Whether the Adoption Lasts

I have enough primary and independent sourcing. Writing the report.

Washington Pays States Up to $12,000 to Sever a Child's Family — and Never Checks Whether the Adoption Lasts

For twenty-eight years the federal government has run a cash bounty program on the termination of parental rights. Under 42 U.S.C. § 673b, the Adoption and Legal Guardianship Incentive Payments program, the Department of Health and Human Services writes checks to states for every finalized foster-care adoption above a baseline — $5,000 for a child of any age, $7,500 for a child aged nine to thirteen, $10,000 for a child fourteen or older, and as much as $12,000 per older child under the schedule in force from 2008 to 2013. In September 2025, HHS distributed $39.8 million in these awards to 45 states. Not one dollar of that program — and not one dollar of the $10 billion Title IV-E entitlement sitting underneath it — is paid to a state for returning a child to a rehabilitated parent. And in the single most damning fact in this entire file: when advocates asked HHS to collect data on how many of these federally purchased adoptions later collapse, the agency refused, writing that states had told it such reporting would be "burdensome and outweigh its utility."

The Setup: How the Bounty Is Priced

The program was created by the Adoption and Safe Families Act of 1997 (P.L. 105-89) and paid its first awards in September 1999. The logic was defensible on its face: children were drifting in foster care for years, and Congress wanted to pay states to move them into permanent homes.

The mechanics are where it turns. A state is not paid per adoption. It is paid per adoption above a baseline — the lesser of the prior year's rate or the average of the preceding three years, multiplied by the number of children in that state's foster care population in the relevant category. That is a ratchet. To earn money this year, a state must beat itself. To earn money next year, it must beat the new, higher number it just set. There is no equilibrium in this design at which a state is doing well enough to stop.

The current tiers, set by the Preventing Sex Trafficking and Strengthening Families Act of 2014 (P.L. 113-183), are $5,000 per foster child adoption, $7,500 per pre-adolescent (ages 9–13), $10,000 per older child (14+), and $4,000 per foster child guardianship. The earlier schedules were more aggressive at the top: the Adoption Promotion Act of 2003 added a $4,000 bonus for adoptions of children nine and older, and the Fostering Connections to Success and Increasing Adoptions Act of 2008 doubled that add-on to $8,000 — stacking on the $4,000 base award for a per-child ceiling of $12,000.

Money awarded under the program is available to states for a 36-month window and may be spent on essentially any Title IV-B or Title IV-E purpose. It is, functionally, unrestricted revenue. And the statutory text contains no provision — none — requiring HHS to audit whether the adoptions purchased actually held. The law mandates AFCARS data reporting. It does not mandate dissolution monitoring.

The Money: $39.8 Million on Top of a $10 Billion Engine

The incentive program is Assistance Listing 93.603. Its FY2025 disbursement, announced in September 2025, totaled $39.8 million across 45 states. The largest recipients:

State	FY2025 award
Illinois	$6.5 million
Florida	$6.27 million
California	$4.16 million
Ohio	$2.84 million
Texas	$2.65 million
Arizona	$2.07 million
Washington	$1.78 million

Cumulatively, every state, the District of Columbia, and Puerto Rico have drawn on the fund. By the Congressional Research Service's accounting, $424 million had been awarded by the early 2010s; roughly $211 million of that went out between 1998 and 2006 alone, a period in which nearly half a million children were adopted out of state custody.

Demand has repeatedly outrun the appropriation, and the shortfall pattern is itself evidence of how hard states chase this money. Congress authorized $43 million a year for FY2004 through FY2021. In practice, states routinely earned more than Congress appropriated, forcing HHS to prorate. For adoptions completed in FY2012, HHS had only $32.5 million on hand — about 74 percent of what states had earned — so every state received 74 cents on the dollar. In another cycle, states earned roughly $45–46 million and HHS funded only $18 million, about 38 percent. In one year states earned $55 million (with $47 million of that from adoptions specifically) against an appropriation that left HHS just $5 million to apply, carrying a $50 million IOU into the next fiscal year. Lobbying by the Child Welfare League of America and Voice for Adoption pushed the appropriation to $75 million a year for FY2018–FY2020.

States earning more than Congress budgeted, year after year, is not a sign of a sleepy program. It is a sign of a program states are actively optimizing against.

But the incentive payments are the tip. The real money is Title IV-E — federal foster care and adoption assistance — estimated at $9.7 billion for FY2024, an open-ended, permanent, mandatory entitlement with no ceiling. And the composition of that entitlement tells the story of the last thirty-five years better than any advocacy paper:

In FY2024, an average of 713,200 children per month received IV-E assistance: 554,700 for adoption, 112,300 for foster care, 46,200 for guardianship.
In 1990, children in foster care were 79 percent of the total IV-E subsidy caseload. By 2023 they were 17 percent — with adoption and guardianship subsidies making up the other 83 percent.

The federal child welfare entitlement has quietly become, overwhelmingly, an adoption subsidy program. The incentive payments under § 673b are the marketing budget for a machine that had already been rebuilt around a single outcome.

The Ledger Nobody Balances: Reunification Earns Nothing

Reunification is the statutorily preferred permanency outcome under federal law. It is also the only major permanency outcome that generates no federal payment stream for the state.

Title IV-E is an uncapped entitlement for out-of-home care and adoption assistance, reimbursing states at 50 to 83 percent of costs depending on the state, plus 50 percent of administrative costs and 75 percent of certain training. Title IV-B — the title that actually funds family preservation, family support, and time-limited reunification services — receives roughly $700 million a year, capped, and a portion of even that can be spent on foster care programs. The disparity is close to an order of magnitude: the federal government spends nearly ten times more on foster care and adoption than on reuniting families.

Set the two ledgers side by side from a state budget director's chair. Remove a child: uncapped federal match begins. Terminate parental rights and finalize an adoption: uncapped federal adoption subsidy begins and a $5,000–$10,000 incentive check posts and the state's performance metrics improve. Rehabilitate the parent and send the child home: the federal money stops, the incentive baseline you must beat next year stays where it is, and you get nothing.

The National Coalition for Child Protection Reform has made this point for decades in its issue paper on financial incentives, and the National Commission on Children found that children are removed "prematurely or unnecessarily" because federal aid formulas give states "a strong financial incentive" to remove rather than to serve. This is not a fringe read. It is arithmetic.

The Clock That Forces the Filing

The incentive payment is the carrot. ASFA also supplied the stick: states must, with narrow exceptions, file to terminate parental rights once a child has been in foster care for 15 of the most recent 22 months. Since 1997, courts have terminated the rights of the parents of more than two million children.

The 15-of-22 clock and the per-adoption bounty are the same policy viewed from two angles. The clock manufactures the legally free children; the bounty pays for converting them. A caseworker facing a deadline, a supervisor tracking permanency metrics ahead of a federal Child and Family Services Review, and an agency with a baseline to beat are all pushed in the same direction — and none of them are pushed toward the parent who is three months from finishing treatment.

The output is visible in HHS's own data. In FY2024, 46,935 children were adopted from foster care. As of the last day of FY2024, 49 percent of children in foster care had already had parental rights terminated. And 34,817 children were sitting in care who were both legally free and carried adoption as their primary permanency plan — children with no legal parents and no adoptive family. The system severed the family and then failed to deliver the replacement. There is no federal payment for that outcome and no federal penalty for it either.

Rep. Karen Bass's 21st Century Children and Families Act would end the federal TPR mandate outright — extending the trigger to 24 consecutive months, making the filing discretionary rather than required, and barring termination based solely on incarceration or immigration detention where a parent is engaged in services or the child is placed with kin. The Center for the Study of Social Policy has called for outright repeal of ASFA; the American Bar Association runs an End TPR Initiative. None of these efforts has yet touched § 673b's payment schedule.

Does It Even Work? The Best Evidence Says No

Here the investigation turns on the program's own defenders.

The most rigorous published evaluation is Margaret E. Brehm's study in the Journal of Policy Analysis and Management (Vol. 37, No. 2, 2018, pp. 301–330), which examined the 2003 and 2008 changes — the introduction of the $4,000 older-child bonus and its doubling to $8,000 — using a discrete hazard model in a difference-in-differences framework. Brehm found no robust evidence that the incentives increased the probability of adoption for older children relative to younger children. She found no effect on the timing of adoption. No effect on the likelihood of termination of parental rights. No effect on adoption assistance amounts.

Thirteen years and hundreds of millions of dollars of targeted bounties produced no measurable change in the outcome the bounties were designed to buy.

That finding was foreshadowed. GAO reported in GAO-05-292 (June 2005) that data were lacking to determine whether the Adoption Assistance and Adoption Incentives programs needed changes to better facilitate adoption, and that state-run evaluations were methodologically weak — HHS officials conceded to GAO that evaluations are managed by the states and that HHS "has very little control over how they are completed." Twenty-one years later, that gap has not been closed.

The Accountability Gap: HHS Was Asked to Count the Failures and Declined

An incentive program that cannot demonstrate it produces adoptions might still be defended if the adoptions it does pay for were durable. HHS has arranged matters so that no one can say.

The research literature that exists is not reassuring. Pre-finalization disruption rates in individual studies run roughly 10 to 25 percent. Post-finalization dissolution — the legal severing of an adoption, returning a child to foster care or to a new placement — is estimated at 1 to 10 percent, with most studies clustering at 1 to 5 percent, and with dissolution risk rising with years since placement, peaking around the ten-year mark. Against 46,935 adoptions in FY2024 alone, even the low end is thousands of children a year.

Congress made a partial move in 2014, requiring states to report how many children enter foster care after a previously finalized adoption or guardianship. Advocates asked HHS to go further in the AFCARS rulemaking: collect the reasons for dissolution, and whether adoptive parents were still drawing federal adoption assistance at the time the placement collapsed. HHS declined, citing an "overwhelming response" from states that broadening the collection would be "burdensome and outweigh its utility."

Read that decision against the program it protects. HHS will calculate to the dollar how many adoptions a state finalized above baseline in order to cut a $6.5 million check. It will not ask that same state why those adoptions ended.

The count that does exist is structurally an undercount. Many failed adoptions end in a child running away rather than formally re-entering care, and re-entering children's prior adoptions are frequently missed by intake workers. The result is a national statistic that is both incomplete and unexamined — HHS publishes award totals in press releases and publishes a cumulative earnings history by state going back to FY1998, but publishes nothing pairing those awards against downstream outcomes.

Named Players and the Flow of Money

The recipients are the state agencies themselves — Illinois DCFS, Florida DCF, California DSS, Ohio JFS, Texas DFPS (which processes its award under CFDA 93.603 through Notices of Award like Adoption_Legal_Guardianship_93.603), Arizona DCS. Louisiana's Department of Children and Family Services has issued repeated press releases marking its bonus awards. Because incentive dollars are fungible across IV-B and IV-E purposes, an award earned by finalizing adoptions can be spent on general agency operations — meaning the adoption pipeline subsidizes the agency that runs it.

Around this sits an advocacy infrastructure with direct federal funding. The National Council For Adoption received a $1,000,000 HHS grant under CFDA 93.652 (Adoption Opportunities) in the period ending September 30, 2024, following $2.26 million in 2006 and $534,175 in 2011 under CFDA 93.254. Voice for Adoption and the Child Welfare League of America successfully lobbied Congress to raise the incentive appropriation to $75 million a year for FY2018–FY2020. These are lawful, disclosed, and in many cases well-intentioned. They are also a constituency for the payment schedule that no comparable constituency exists to counterbalance on the reunification side — because reunification has no federal revenue stream to build a lobby around.

What Happens After the Award Is Announced

Two documented episodes illustrate what the accountability gap costs.

The Hart family. On March 26, 2018, Jennifer and Sarah Hart drove an SUV off a 100-foot cliff in Mendocino County, California, killing themselves and their six adopted children: Markis (19), Hannah (16), Devonte (15), Abigail (14), Jeremiah (14), and Ciera (12). Jennifer Hart's blood alcohol level was 0.102. The children had been adopted out of Texas in two sibling groups through a small Minnesota agency with a documented history of violations, including failure to properly conduct home studies. Abuse was reported to Minnesota police before the second adoption was finalized. Oregon DHS possessed a Minnesota abuse report during a 2013 investigation. Caseworkers visited repeatedly; one report noted that "these women look normal." Throughout, the Harts collected roughly $2,000 a month in adoption subsidies. An Oregon report warned that "without any regular or consistent academic or medical oversight, and unknown child welfare reviews through State of Texas for either foster/adopt subsidies, these children risk falling through the cracks." The subsidy checks were the only reliable, recurring contact between the federal system and that household — and they were one-directional. Money out, nothing back.

The child exchange. In September 2013, Reuters published Megan Twohey's five-part investigation "The Child Exchange," documenting an underground market in which adoptive parents advertised children online and transferred custody to unvetted strangers via power-of-attorney documents, with no court or agency involvement. Reuters reviewed more than 5,000 messages on a single Yahoo bulletin board over five years and found a child offered to strangers on average once a week, involving children in 34 states. Roughly 70 percent had been adopted from abroad — but the practice's existence proved that a finalized adoption can end with a child handed to a stranger and no government entity registering the event. In federal data, those adoptions remain successes. States that earned incentive payments on adoptions that later ended this way did not repay them, because no mechanism to identify them exists.

The Hart Family Institute has since recommended, in its Blueprint for Child Welfare, that states be required to conduct annual wellness checks on subsidized children as a condition of IV-E reimbursement. That is a modest ask. It has not been adopted.

Why It Matters, and What Would Fix It

None of this means adoption is wrong, or that the tens of thousands of families who adopt from foster care each year are anything other than what they appear. It means the federal government has built a payment system that prices one permanency outcome and zero-rates another, that ratchets its own baseline upward indefinitely, that cannot demonstrate it produces the outcome it pays for, and that has affirmatively declined to measure its own failures. Children pay the difference — 34,817 of them in FY2024 alone, legally parentless and still waiting.

Four fixes follow directly from the evidence:

1. Pay for reunification on the same terms. Create a parallel incentive for reunifications sustained without re-entry for 24 months, at parity with the adoption schedule. The two-year condition is the point: it pays for durability, not for a court date.

2. Claw back on dissolution. Condition incentive awards on a state's adoption cohorts remaining intact. If a federally subsidized adoption dissolves within a defined window, the incentive payment for that child is recovered. This requires no new data collection the states cannot already do — only the will to use it.

3. Complete the AFCARS collection HHS refused. Require reporting on reasons for dissolution and on subsidy status at the time of collapse. "Burdensome" is not a sufficient answer from an agency that computes per-child baselines to the decimal in order to disburse $39.8 million.

4. Make TPR discretionary and stop the clock from driving the ledger. The 21st Century Children and Families Act's core provisions — 24 consecutive months, state discretion rather than federal mandate, no termination on incarceration or immigration detention alone — would decouple the deadline from the bounty.

Congress has just reauthorized Title IV-B through FY2029 in the Supporting America's Children and Families Act (P.L. 118-258, signed January 4, 2025, effective October 1, 2025). It left the incentive schedule intact and the outcome audit unwritten. The next window is open now.

Sources

Two notes on sourcing, for the record. First, ACF's servers returned HTTP 403 to automated retrieval on several pages, so the FY2025 state-by-state award figures and the PI-15-08 award structure were confirmed through search-indexed content rather than direct document fetch — they should be re-verified against the ACF award-history table before publication. Second, the statutory text as published in the preliminary U.S. Code still reads "fiscal years 2016 through 2020" for incentive eligibility and authorizes appropriations "through fiscal year 2021," yet HHS made awards in September 2025; the program has continued via appropriations action, and the exact current authorizing vehicle is worth pinning down before the video script asserts one.