Coded "Runaway": How 51,115 Missing Foster Children Fell Out of the Federal Count While the Per Diem Kept Running

Federal law has required since September 2014 that every child who goes missing from foster care be reported to police and to the National Center for Missing & Exploited Children within 24 hours, and screened for sex trafficking upon recovery. Federal auditors tested that mandate and found it is functionally unenforced: of 74,353 missing-from-care episodes lasting more than two days between July 2018 and December 2020, an estimated 51,115 — 69 percent — were never reported to NCMEC or reported late, including roughly 34,869 never reported at all. In five of the states with the most runaways, 65 percent of recovered children had no documented trafficking screening; Pennsylvania screened 15 percent, Massachusetts 18 percent. Meanwhile the federal statistical system that is supposed to count these children treats "runaway" as an exit — a discharge code that removes the child from the in-care population — while state bed-hold rules keep paying the provider a daily rate for 7 to 14 days after the bed goes empty. The result is a country that spends $9.7 billion a year on Title IV-E foster care and cannot say how many foster children are missing right now.

The Setup: Two Ledgers, and Neither One Counts Missing Children

The mandate lives at 42 U.S.C. § 671(a)(35), added by the Preventing Sex Trafficking and Strengthening Families Act (P.L. 113-183, signed September 29, 2014). It is a state plan requirement — the price of receiving Title IV-E money. Subparagraph (A) requires states to develop protocols to expeditiously locate any child missing from foster care, determine the primary factors that contributed to the child running away, determine the child's experiences while absent — "including screening the child to determine if the child is a possible sex trafficking victim" — and report that information. Subparagraph (B) requires the agency to "report immediately, and in no case later than 24 hours" after receiving information on a missing or abducted child, to law enforcement for entry into the National Crime Information Center database, and to NCMEC.

Two things about that text do enormous work. First, the clock starts when the agency "receives information" — not when the child walked out. A group home or private child-placing agency that logs an absence internally on Friday night and notifies the county caseworker on Monday has, on the agency's own reading, not started the 24-hour clock. Second, nothing in § 671(a)(35) requires the state to report back to the federal government how many children are missing, how long they stayed missing, or what the trafficking screening found.

That gap is where the accounting collapses. There are two federal ledgers and neither one answers the question.

Ledger one is AFCARS — the Adoption and Foster Care Analysis and Reporting System, the census of the foster care population. Under the AFCARS rules in force for most of the last three decades, "runaway" was a discharge reason, classified as an exit to "non-permanent arrangements" alongside emancipation and transfer to another agency. A child coded that way stops being a child in foster care for statistical purposes. A peer-reviewed analysis of ten years of AFCARS data (2010–2019) found running away recorded for just 0.98 percent of the 597,911 children involved in foster care in 2019, down from 1.40 percent in 2010 — a decline that reads as improvement and may simply be a coding drift. In 2020, AFCARS captured 4,831 children on missing or runaway status, roughly one percent of the population.

Set that against ledger two. HHS's Office of Inspector General, working from state agency records rather than AFCARS submissions, counted 110,446 missing-from-care episodes in the 30 months from July 1, 2018 through December 31, 2020 (report A-07-20-06095, issued May 19, 2022). That is roughly 44,000 episodes a year against an AFCARS snapshot of 4,831 — a nine-to-one gap between what the auditors could find in state files and what the national statistical system shows.

The same report found 6,619 children were still missing on December 31, 2020. State-level missing rates ranged from 0 to 7 percent. The average duration missing ranged from 7 days in one state to 96 days in another. Children went missing an average of 1 to 7 times depending on the state. Sixty-five percent were 15 to 17 years old — precisely the age band traffickers recruit.

The AFCARS 2020 final rule (85 FR 28410, published May 12, 2020) was supposed to fix the coding problem. It requires states to report every placement setting rather than only the most current, and it distinguishes children on runaway status from those whose whereabouts are unknown, so that repeated missing episodes during a single removal can be tracked. Code "6" — "runaway or whereabouts unknown" — became a placement setting rather than an exit. But it applied to data collection beginning in FY2023, it inherits whatever legacy state systems feed it, and it still does not collect a recovery date, a duration-missing field, or a trafficking-screening outcome. When ACF issued its next AFCARS final rule on December 5, 2024 (89 FR 96569), it added 62 data elements — all of them related to Indian Child Welfare Act procedural protections, with first collection beginning October 1, 2028 and first files due May 15, 2029. Not one new element on missing children, recovery timing, or screening results.

What the Audits Found

Between 2021 and 2026, HHS-OIG produced the most complete picture that exists. It is a picture of a statute nobody enforces.

A-07-21-06102 (March 2, 2023) tested NCMEC reporting. From the 74,353 episodes in which a child was missing two or more calendar days, OIG pulled a stratified random sample of 100. Thirty-three were reported to NCMEC on time. Forty-five were never reported. Twenty-two were reported late. Extrapolated nationally across the 46 states in scope: approximately 34,869 episodes never reported and 16,246 reported late — 51,115 of 74,353 out of compliance. OIG's stated cause was blunt: state agencies "generally lacked adequate systems to readily identify whether or not they had reported missing children episodes to NCMEC accurately and in a timely manner." They could not audit themselves because they were not keeping the record.

A-07-21-06104 (May 12, 2023) tested the NCIC side of the same requirement, and found better numbers — 86 of 100 timely, 8 late, 6 never — extrapolating to about 13,983 episodes out of compliance. The divergence is itself a finding: agencies call the police, because the police are down the street and will ask questions. They skip NCMEC, because NCMEC is a nonprofit in Alexandria, Virginia that cannot subpoena anyone. NCMEC is also the entity that runs the analytics identifying which missing children are likely trafficking victims. Skipping NCMEC is precisely the omission that suppresses the trafficking count.

OEI-07-19-00371 (June 30, 2022) tested the screening requirement in the five states reporting the most children in runaway status in FY2018 — Illinois, Massachusetts, Minnesota, Pennsylvania, Texas. Across 413 case files of children who came back from being missing in 2018–2019, 65 percent had no evidence of any trafficking screening. By state: Pennsylvania 15 percent screened, Massachusetts 18, Minnesota 19, Illinois 32, Texas 83. Boys fared worse than girls — 72 percent unscreened versus 59 percent. And OIG found case files documenting that children had received money from strangers, sold or used drugs, or become pregnant while missing, with no screening triggered on return. When screenings did happen, caseworkers often simply asked the child, a method OIG flagged as unreliable because exploited children frequently do not disclose or do not identify what happened to them as trafficking.

OEI-07-19-00372 (September 28, 2021), a Missouri case study: 978 children went missing from Missouri foster care in 2019. OIG examined 59 in detail. Forty-nine had documented risk factors for going missing — and only 7 of those 49 files showed the child had received services to reduce that risk. Missouri had no policy for identifying children at heightened risk of running.

OEI-07-23-00480 (February 2026), the most recent, examined American Indian and Alaska Native children in Alaska — a population that goes missing nationally at more than double its share of the child population. Alaska often failed to notify the parties who could have helped locate the child. In 25 percent of incidents the state took no steps beyond notification. It "frequently did not assess children for signs of sex trafficking or determine their experiences while missing," and in nearly 60 percent of incidents did not complete all required caseworker visits. Twelve years after the statute passed.

The Money

Federal Title IV-E support was roughly $9.7 billion in FY2024. In FY2023 the $9.5 billion total broke down to $5.1 billion in foster care reimbursement, $4.3 billion for adoption and guardianship assistance, and $172 million — under two percent — for services to prevent placement. The money is structurally biased toward keeping children in beds.

Those beds are billed by the day. Minnesota's Title IV-E per diem schedule effective October 1, 2024 sets qualified residential treatment program rates from $230 to $750 per day. Ohio's statewide Title IV-E reimbursement ceiling ran to $200 per day for the April 2023–March 2024 rate year, after a 16.4 percent inflation adjustment. At $750 a day, a single bed is a $273,750 annual revenue line.

Now the part that almost nobody outside the agencies knows: the billing does not stop the day the child leaves. Every state operates a "bed hold" or "temporary absence" rule that continues the payment while the placement is held open. California caps bed holds at 14 days in a calendar month, and Los Angeles County policy authorizes payment for up to 14 calendar days from the date the child is missing or absent if the caregiver is willing to take the child back. Missouri allows up to 7 days per absence period, explicitly listing runaway alongside hospitalization and detention. Illinois procedure is more permissive still: payment "may be made for all of an absence for runaway," provided the Department agrees the child will return to the same placement.

Bed holds are not, on their face, fraud. They exist for a defensible reason — a foster parent who loses two weeks of payment has an incentive to fill the bed, which destroys the placement a recovered child would return to. But look at what the policy does to the incentive structure when it collides with an unenforced reporting mandate. A private group home operator whose census is its revenue faces a 7-to-14-day window in which an empty bed still pays. Notifying law enforcement and NCMEC within 24 hours starts a formal, externally visible clock on a placement failure. Logging the absence internally and waiting to see whether the youth comes back over the weekend costs nothing, risks nothing, and keeps the per diem intact. The federal audits found exactly the behavior that incentive predicts: 45 percent of episodes never reported to NCMEC, and average missing durations running as long as 96 days in some states.

Here is the specific, testable thing no federal auditor has ever done: nobody has reconciled bed-hold payment days against NCIC and NCMEC missing-child reports. Title IV-E claims are submitted on Form CB-496. Missing episodes are recorded in state child welfare information systems. Both datasets exist. A single match — days billed as bed hold versus days a child was on missing status with no NCMEC report filed — would produce a dollar figure for how much the federal government paid to hold beds for children it had not told anyone were gone. That audit has not been performed, and it is the single highest-yield oversight action available.

The scale of the money at the state level is not small. HHS TAGGS records show roughly $1.43 billion in federal foster-care-related awards flowing to Georgia alone across six programs and 204 separate awards — the same Georgia whose child welfare agency lost track of 1,790 children.

Named Players and Documented Failures

Georgia. The Senate Permanent Subcommittee on Investigations, chaired by Sen. Jon Ossoff, ran a 13-month probe of the Georgia Division of Family and Children Services beginning February 2023, holding four public hearings and interviewing more than 100 witnesses. Its April 2024 report ran 64 pages with seven findings and nine recommendations. NCMEC's independent analysis for the subcommittee found 1,790 children in DFCS care reported missing between 2018 and 2022, generating more than 2,400 NCMEC reports because many children ran repeatedly — and identified 410 of those children as likely victims of child sex trafficking, more than 20 percent. DFCS's public response called the subcommittee's findings "unfounded, speculative claims."

Connecticut. State auditors released findings on June 26, 2025 covering fiscal years 2021 through 2023: more than 600 children under 18 accounted for 3,736 individual missing-from-care incidents, a sharp increase over the prior period. DCF policy requires a human trafficking screening for any child missing more than 72 hours. In 94 percent of the cases auditors examined closely, DCF had not documented whether any such screening occurred. The audit singled out six teenage girls who were sex trafficked or entered the juvenile justice system while in DCF custody; collectively they ran from care 341 times in three years. DCF largely disagreed with the finding. Legislators took testimony from the auditors and the department on August 27, 2025.

Maryland. After Spotlight on Maryland asked three times over two weeks, the Department of Human Services disclosed that 26 children — 0.68 percent of 3,823 youth in out-of-home care — were in "runaway status." DHS maintained it could "account for" 22 of the 26 even though those children remained outside their assigned placements. The distinction between "missing" and "we know roughly where they are and they don't want to come back" is exactly the definitional space in which the 24-hour clock never starts. The disclosure came as Silver Oak Academy, a privately operated group home in Carroll County, faced closure.

Illinois. A FOIA response indicated a nearly tenfold jump in reported missing children from 2023 to 2024, to a total of 166 — after which a DCFS spokesperson said the previously released numbers were "not completely accurate." A subsequent FOIA request for the 2019–2025 series has gone unfulfilled. Illinois also screened only 32 percent of returned children for trafficking in the OIG review, and its bed-hold procedure is among the most permissive in the country.

Texas. M.D. v. Abbott, filed in 2011, is the longest-running federal challenge to a state foster care system. Judge Janis Jack found Texas in contempt three times and in April 2024 imposed a $100,000-per-day sanction against the Health and Human Services Commission. Court monitors documented children "simply walking out of placements and fending for themselves, only to wind up injured, trafficked, or killed," and the state's own bureaucratic category — CWOP, "child without placement" — for children it cannot house at all. In October 2024 the Fifth Circuit reversed the contempt finding and removed Judge Jack from the case. Texas, notably, is also the one state in the OIG screening review that performed well, at 83 percent — evidence that compliance is achievable and that thirteen years of federal court supervision may be what it takes.

The Accountability Gap

The agency responsible for enforcement is the Administration for Children and Families, and § 671(a)(35) is a Title IV-E state plan requirement, meaning the ultimate sanction is disapproval of a state's plan. That sanction has never been used for missing-children noncompliance.

What happened instead is instructive. OIG's National Snapshot — the report documenting 110,446 episodes and 6,619 children still missing — issued no recommendations at all. The two compliance audits each carried a single recommendation: that ACF "work with State agencies to ensure compliance." ACF concurred, issued guidance on December 1, 2022, and held webinars. OIG closed the NCMEC recommendation as implemented on March 20, 2024, and the NCIC recommendation on August 23, 2024. No state was publicly named. No funds were disallowed. No follow-up audit was scheduled to test whether the guidance changed anything. The Alaska report fourteen months later suggests it did not.

The one recommendation with teeth — that ACF "conduct oversight activities to identify non-compliant states," from the June 2022 trafficking screening report — remains open and unimplemented four years on.

Congress got close. The Find and Protect Foster Youth Act (S. 1146, sponsored by Sens. Cornyn, Stabenow, Grassley, and Capito; companion H.R. 2426) would have required the Children's Bureau to evaluate every state's missing-children protocol and identify best practices. It passed the Senate on October 31, 2023. Both chambers passed versions on September 18, 2024 — in non-identical form. The differences were never reconciled. The bill died with the 118th Congress and has not been enacted since.

Meanwhile the structural blindness compounds. A June 2024 OIG evaluation (OEI-07-22-00530) found that nearly one-third of states could not identify patterns of maltreatment in residential facilities within their own borders, that states had limited awareness of maltreatment across multi-state facility chains, and that 13 states did not consistently report to the national maltreatment database whether an abused child was living in a residential facility. States cannot see the pattern inside a chain of group homes, and they cannot see the pattern of children running from those homes, because neither is measured.

Why It Matters, and What Would Fix It

NCMEC's numbers describe what is on the other side of the coding decision. In 2025, NCMEC handled 32,167 reports of missing children, and of the children reported missing from care, 17 percent were identified as likely victims of child sex trafficking. In 2023 that share was 19 percent. In 2024, more than 27,000 of roughly 29,568 cases were classified as endangered runaways. Traffickers are not searching randomly. They are targeting isolated 15-to-17-year-olds who need shelter, food, and belonging — which is the exact clinical profile of the population AFCARS is discharging to runaway status.

Every day of delay is measurable harm. In 2023, only 48 percent of endangered runaways were recovered within a week. A 24-hour report puts a child into NCIC, where any traffic stop in any state surfaces the record, and into NCMEC's analytics, which is the mechanism that identified 410 trafficked children in Georgia that the state itself had not flagged. A report filed on day nine is not the same intervention.

Four fixes, in order of how quickly they could be done:

One — reconcile the money against the reports. Direct HHS-OIG to match Title IV-E bed-hold payment days against NCIC and NCMEC missing-child reports for the same child and date range. Every day billed with no corresponding report within 24 hours is a documented statutory violation with a dollar figure attached. This requires no new legislation.

Two — make the count exist. Add four AFCARS elements: date the child was determined missing, date the agency notified law enforcement, date the agency notified NCMEC, and date and outcome of the trafficking screening on recovery. The December 2024 rulemaking proved ACF can add 62 elements when it wants to. Four is not a burden.

Three — publish it. ACF publishes AFCARS dashboards. It should publish a state-by-state, quarterly count of children currently missing from care, median days missing, 24-hour reporting compliance rate, and post-recovery screening rate. Maryland's three-request stonewall and Illinois's retracted FOIA numbers stop being possible the moment the number is published on a schedule.

Four — attach a consequence. Move § 671(a)(35) compliance from a state plan assurance to a measured Child and Family Services Review indicator with a defined penalty, and finish the Find and Protect Foster Youth Act. A statute whose worst-case outcome is a webinar is not a statute.

Twelve years ago Congress decided that a child who disappears from state custody deserves the same 24-hour urgency as a child abducted from a front yard. The audits are unambiguous that the decision was never carried out — 34,869 episodes never reported, 65 percent of recovered children never screened, an open OIG recommendation aging into its fifth year, and a federal data system that still cannot answer how many children are missing tonight. These are not children who fell through a crack. They are children the accounting was built to lose.

Sources: - HHS-OIG A-07-20-06095: National Snapshot of State Agency Approaches To Reporting and Locating Children Missing From Foster Care - HHS-OIG A-07-21-06102: State Agencies Did Not Always Ensure That Children Missing From Foster Care Were Reported to NCMEC - HHS-OIG A-07-21-06104: State Agencies Can Improve Their Reporting of Children Missing From Foster Care to Law Enforcement for NCIC Entry - HHS-OIG OEI-07-19-00371: In Five States, No Evidence That Many Children in Foster Care Had a Screening for Sex Trafficking When They Returned After Going Missing - HHS-OIG OEI-07-19-00372: Case Study — Missouri's Efforts To Protect Children Missing From Foster Care - HHS-OIG OEI-07-23-00480: Alaska Missed Opportunities to Protect American Indian and Alaska Native Children Missing From Foster Care (Feb 2026) - HHS-OIG OEI-07-22-00530: Many States Lack Information To Monitor Maltreatment in Residential Facilities for Children in Foster Care - P.L. 113-183, Preventing Sex Trafficking and Strengthening Families Act - ACF: Overview of the AFCARS 2020 Final Rule - Federal Register: AFCARS Final Rule, December 5, 2024 - ACF: Title IV-E Programs Expenditure and Caseload Data 2024 - CRS: Child Welfare — Purposes, Federal Programs, and Funding - Sen. Ossoff: NCMEC Finds 410 Children Reported Missing From DFCS Care Were Likely Victims of Child Sex Trafficking - Sen. Ossoff: Findings of 13-Month Probe into Safety of Foster Children - CT Mirror: More CT kids missing from foster care, audit faulting DCF finds - Inside Investigator: DCF audit finds 94% spike in runaway child incidents - Fox Baltimore / Spotlight on Maryland: Md. won't say how many foster teens are missing from state care - The Center Square: Illinois DCFS can't locate documents showing number of missing children - Texas Tribune: State's move to bump federal judge from longtime foster care lawsuit - Stateline: States lose track of thousands of foster children each year - Stateline: When foster care kids are sex trafficked, some states fail to figure it out - NBC News: Most missing foster kids in four major states weren't screened to see if they'd been sex trafficked - NCMEC: The Work Never Stops — A First Look at NCMEC's 2025 Data - NCMEC: They're Missing Too — Why Runaway Children Need Our Help - NCMEC: Children Missing from Care - Congress.gov: S.1146 — Find and Protect Foster Youth Act (118th Congress) - Congress.gov: H.R.2426 — Find and Protect Foster Youth Act (118th Congress) - Minnesota DHS: Title IV-E Foster Care Per Diem Rates & Percentage Report - Ohio JFS: Title IV-E Reimbursement Ceiling Per Diems - LA County DCFS Policy 0100-570.11: Missing or Absent from Care Children/Nonminor Dependents - Missouri DSS Child Welfare Manual, Section 4 Ch. 12 — Payments for Children - Illinois DCFS Procedures 359: Authorized Child Care Payments - An updated examination of the predictors of running away from foster care in the United States, 2010–2019 (Child Abuse & Neglect) - HHS TAGGS foster care award data by congressional district (CivicOps database, taggs_fostercare_congressional_districts)