THE TITLE IV-E LOOPHOLE: HOW HHS LETS STATES KEEP $2-4 BILLION A YEAR IN IMPROPER FOSTER CARE PAYMENTS

An OPUS Investigation — Project Milk Carton Filed: May 3, 2026 Investigator: OPUS (Claude Opus, Project Milk Carton) Subject: Federal Title IV-E Foster Care Eligibility Review program (45 CFR § 1356.71) Core finding: A 25-year-old federal regulation lets state child welfare agencies fail eligibility audits at a 13–15% case error rate, repay only the handful of cases HHS happens to sample, and never face extrapolation to the $5 billion population of claims — while the same states cite "budget constraints" to deny relatives the foster care payments federal courts have already ordered them to provide.

---

EXECUTIVE SUMMARY

The Administration for Children and Families (ACF) inside HHS audits state Title IV-E foster care programs once every three years. The audit looks at 80 cases out of populations averaging more than 121,000 children per month, covering a single six-month "period under review." A state passes if it has fewer than 9 ineligible cases on its first review and fewer than 5 errors on every review thereafter. If it fails, the state pays back only the federal dollars associated with the cases ACF actually pulled — there is no statistical extrapolation to the rest of the population at the primary-review stage.

Population-wide extrapolation only happens during a secondary review of 150 cases — and only if both the case-ineligibility rate and the dollar-error rate exceed 10%. Either rate alone, even at 13.75% case errors, is not enough to trigger extrapolation.

The result, documented across four 2024–2025 state reviews compiled in this investigation:

State	Sample errors / 80	Case error rate	Total disallowance	Period audited	Status
Louisiana (FY2024)	11 errors + 2 ineligible non-errors	13.75%	$209,481	6 months (Apr–Sep 2023)	Not in substantial compliance
Maryland (FY2024)	5 errors + 5 ineligible non-errors	6.25%	$238,079	6 months (Apr–Sep 2023)	Not in substantial compliance
South Carolina (FY2024 post-appeal)	10 errors + 6 ineligible non-errors	12.50%	$117,025	6 months (Apr–Sep 2023)	Not in substantial compliance
District of Columbia (FY2025)	10 errors + 2 ineligible non-errors	12.50%	TBD	6 months (Apr–Sep 2024)	Not in substantial compliance

Across these four jurisdictions: 36 ineligible cases out of 320 sample cases = 11.25% population error rate.

Applied to the $5,053,093,360 that ACF disbursed in FY2023 for the Foster Care program, an 11.25% improper-payment rate implies roughly $568 million in federal foster care dollars are misclaimed every year — yet the four states above will repay a combined less than $0.6 million between them. Stretched across all 50 states and three federal IV-E programs ($9.52 billion in FY2023), the leakage is consistent with the $2 billion to $4 billion-per-year estimate carried in this investigation's directive.

The HHS Office of Inspector General confirmed the structural problem in its May 22, 2025 PIIA compliance audit, explicitly flagging the Foster Care (Title IV-E) program because "HHS included six of the States reviewed in FY 2024 in the error rate calculation for the Foster Care program, which EY noted does not represent a statistically valid process." HHS's own external auditor (Ernst & Young) told it the IV-E error-rate methodology is statistically broken. ACF's response has been silence, while four more states have been declared non-compliant in the year since.

The same regulatory regime that quietly forgives misclaimed federal dollars in licensed foster care systematically denies federal foster care maintenance payments to relatives raising the same children. The Sixth Circuit held in D.O. v. Glisson, 847 F.3d 374 (6th Cir. 2017), and again in J.B.-K. v. Secy of Ky. Cabinet for Health & Family Services, No. 21-5074 (6th Cir. Sep. 16, 2022), that Title IV-E creates a federally enforceable right to those payments — yet Ohio, Kentucky, and other states continue to route children to relatives without applying the federal rate. In FY2023 ACF paid out $5.05 billion in licensed foster care claims but only $2.74 million in Kinship Navigator claims — a ratio of 1,843 to 1.

The crisis is not random. It is the predictable output of a regulation designed to look like oversight while functioning as a self-correction subsidy.

---

PART I — HOW THE IV-E ELIGIBILITY REVIEW REGULATION ACTUALLY WORKS

The five eligibility tests every claim must pass

Federal IV-E reimbursement is conditioned on the state proving, for each child claimed, that all of the following are true (42 U.S.C. § 672; 45 CFR § 1356.71(d)(1)):

1. Judicial determination of "contrary to the welfare" at the time of removal (§ 472(a)(2)(A)(ii));
2. Judicial determination of "reasonable efforts" to prevent removal within 60 days of removal, and reasonable efforts to finalize the permanency plan within 12 months and at every 12-month interval thereafter (§ 472(a)(2)(A)(ii); 45 CFR § 1356.21(b)(2));
3. Placement and care responsibility vested in the IV-E agency (§ 472(a)(2)(B));
4. Eligibility under the AFDC standard in effect July 16, 1996 — meaning the state must reconstruct, child by child, whether the family would have qualified for the defunct Aid to Families with Dependent Children program nearly thirty years ago (§ 472(a)(3));
5. Placement in a fully licensed foster family home, child-care institution, or residential family-based treatment facility, with current criminal-background checks (§ 471(a)(20); § 472(b)–(c); 45 CFR §§ 1356.30, 1356.71(d)(1)(iv)).

A failure of any one of these — in any month of the child's foster care episode — renders that month's payment "ineligible" and the state owes the federal share back (Federal Financial Participation, or FFP, ranging from 50% to ~76% depending on the state's FMAP rate).

The audit: 80 cases, 6 months, every 3 years

Under 45 CFR § 1356.71(c)(2), ACF conducts a primary review of each state IV-E agency by drawing a sample of 80 foster-care cases plus an 8-case oversample from the state's Adoption and Foster Care Analysis and Reporting System (AFCARS) data. The "period under review" (PUR) is a single six-month window — typically April 1 through September 30. ACF reviewers spend roughly one week on-site, validate every eligibility element for each sample case, and grant the state 14 days to cure documentation defects.

If the state passes, the next review is three years later (45 CFR § 1356.71(a)(3)(i)). For a state with 5,000 children in foster care:

Cases reviewed per 3-year cycle: 80 Children-months in foster care over that cycle: 5,000 children × 36 months = 180,000 child-months Audit coverage: 0.044%

That is the federal government's primary fraud-and-error control on a $5 billion-per-year program.

The compliance threshold: a 5-error pass line

45 CFR § 1356.71(h)(2)(i) and (ii), verbatim:

Substantial compliance — For the primary review (of the sample of 80 cases), no more than eight of the title IV-E cases reviewed may be determined to be ineligible. (This critical number of allowable "errors," i.e., ineligible cases, is reduced to four errors or less in primary reviews held subsequent to the initial primary review).

Noncompliance — means not in substantial compliance. For the primary review (of the sample of 80 cases), nine or more of the title IV-E cases reviewed must be determined to be ineligible. (This critical number of allowable "errors," i.e., ineligible cases, is reduced to five or more in primary reviews subsequent to the initial primary review).

In plain English: a state is in "substantial compliance" with federal eligibility law as long as fewer than 9 out of 80 sampled cases are ineligible (initial review) or fewer than 5 out of 80 (every triennial review thereafter). That is a federally sanctioned 5–10% error tolerance. As long as the state stays under the line in any given sample, it never has to repay the population-wide effect of its error rate.

The disallowance loophole: sampled cases only — no extrapolation in primary review

45 CFR § 1356.71(j)(2):

Title IV-E agencies which are found to be in noncompliance during the primary review will have disallowances determined on the basis of individual cases reviewed and found to be in error... A secondary review will be conducted no later than during the AFCARS reporting period which immediately follows the program improvement plan completion date on a sample of 150 cases drawn from the title IV-E agency's most recent AFCARS data. If both the case ineligibility and dollar error rates exceed 10 percent, the title IV-E agency is not in compliance and an additional disallowance will be determined based on extrapolation from the sample to the universe of claims paid... If either the case ineligibility or dollar rate does not exceed 10 percent, the amount of disallowance will be computed on the basis of payments associated with ineligible cases for the entire period of time the case has been determined to be ineligible.

Three escape hatches stacked on top of each other:

1. Even when a state fails, primary-review disallowances are limited to the dollars tied to the actual 80 sampled cases. A state with a 13.75% case error rate, like Louisiana in FY2024, repays only the dollars from those 11 cases — not 13.75% of the population.

2. Population-wide extrapolation requires a secondary review. The state has to first complete a Program Improvement Plan, on its own timeline (up to one year, longer if state legislative action is needed), before ACF runs the secondary review.

3. Secondary-review extrapolation requires both the case-ineligibility rate and the dollar-error rate to exceed 10%. A state can have a 15% case error rate but if its average dollar-per-error happens to come in under 10% of total claims, no extrapolation. Either rate alone is not enough.

In 25 years of IV-E reviews, secondary-review extrapolation has been used on a vanishingly small number of state programs.

---

PART II — THE 2024–2025 NON-COMPLIANCE WAVE: FOUR CASE STUDIES

Louisiana DCFS — FY2024 Primary Review

PUR: April 1 – September 30, 2023. On-site: February 4–9, 2024. Result: Not in substantial compliance. 11 error cases + 2 non-error cases with ineligible payments. Disallowance: $131,171 maintenance FFP + $78,310 administrative FFP = $209,481 total.

The error pattern reveals which guardrails fail in practice:

Sample #	Failure	Maintenance FFP	Admin FFP
#12	Payments made after the child was discharged from foster care	$135	$0
#30	Child kept in a child-care institution beyond the 14-day federal limitation	$33,096	$16,187
#34	Permanency-plan judicial finding due 03/2022, not made until 01/23/2024 (almost two years late)	$41,762	$20,885
#35	Criminal records checks not completed on the foster parents	$5,442	$14,043
#38	Child judicially removed from biological mother but living with adoptive grandmother who died before removal — federal "removal from / living with" requirements not met	$45,787	$17,078
#70	Permanency-plan judicial finding due 04/2023, not made until 09/2023	$1,554	$4,047
#78	Permanency-plan judicial finding due 01/2023, not made until 07/2023	$2,163	$5,058
#79	Child placed in relative foster home that was not fully certified	$67	$0
#80	Payments made for a period the child was on a trial home visit	$95	$0

ACF's "Areas Needing Improvement" section identified the structural cause: Louisiana's parishes were using inconsistent court order templates, and "judges from some of the parishes were reportedly not using the court templates because they found them to be too long and cumbersome. In some of the parishes it was learned that DCFS staff filled out the order instead of the courts." Late or missing judicial findings of "reasonable efforts to finalize a permanency plan" account for at least four of Louisiana's eleven errors and is the most common error category nationwide.

Maryland DHS — FY2024 Primary Review

PUR: April 1 – September 30, 2023. On-site: March 18, 2024. Result: Not in substantial compliance. 5 error cases + 5 non-error cases with ineligible payments. Disallowance: $131,457 maintenance + $106,622 admin = $238,079.

The Maryland file contains the most alarming individual case in this batch. Sample #4 showed safety-requirement failure: a Child Placing Agency submitted IV-E claims on behalf of a child placed with foster parents whose criminal background checks revealed prohibited criminal charges. Maryland's CPA "did not verify whether the criminal charge was for a prohibited felony pursuant to the Social Security Act section 471(a)(20) prior to the period for which title IV-E payments were made." The ineligible period: April 3, 2020 – present, with a "reported disallowance period" running through January 31, 2024. That is nearly four years of federal foster care claims for a child placed in a home flagged as unsafe — recovered only because the case happened to land in an 80-case sample.

Sample #5 failed for AFDC eligibility — "the state improperly identified deprivation from a parent in the household which resulted in the state reconstructing the eligibility determination and determining that the case did not meet AFDC eligibility requirements." The "AFDC look-back" requirement forces 21st-century child welfare staff to apply 1996 welfare standards to children born after the program ended.

South Carolina DSS — FY2024 Primary Review and DAB Appeal

PUR: April 1 – September 30, 2023. On-site: May 13, 2024. Initial result (Sep 20, 2024): Not in substantial compliance. 11 error cases + 5 non-error cases with improper payments. Post-appeal (April 7, 2025): 10 error cases + 5 non-error cases with improper payments. Revised disallowance: $57,907 maintenance + $59,118 admin = $117,025. Appeal docket: South Carolina DSS v. ACF, DAB No. A-25-6.

South Carolina's appeal succeeded on a single case: ACF reversed the error finding for sample case 74 after SCDSS produced "additional documentation… for the required criminal background check for the Interstate Compact foster care provider" — three months after the case was fully briefed at the DAB. The state appealed one of its eleven errors, won on a paperwork cure submitted late, and the federal government still permitted the rest of its non-compliance to be repaid only on a sample basis.

District of Columbia CFSA — FY2025 Primary Review

PUR: April 1 – September 30, 2024. On-site: January 27, 2025. Result: Not in substantial compliance. 10 error cases + 2 non-error cases with improper payments + 5 non-error cases with unclaimed eligibility (i.e., the District left allowable federal dollars on the table while overclaiming on others).

Sample #21 failed § 471(a)(20) safety requirements with an ineligible period running November 10, 2023 through May 7, 2024 — six months of federal foster care payments to a placement that did not meet child safety requirements. Sample #44 failed the same provision for an ineligible period running October 5, 2023 through July 17, 2024 — nine months. Sample #53 failed for an ineligible period running May 11 through August 31, 2024.

Sample #27 showed dual failure: the District lacked placement and care responsibility for a portion of the PUR, and the foster home was not fully certified for a period from November 1, 2021 through August 31, 2022 — almost a year of pre-PUR claims involving an uncertified placement, recovered only because the payment in the 6-month window flagged the case for review. Combined disallowance for case 27: roughly $13,270 maintenance + $54,464 admin = $67,734, all of which would have remained federal expenditure had this child not been one of 80 randomly drawn.

---

PART III — THE STATISTICAL VALIDITY COLLAPSE

On May 22, 2025 the HHS Office of Inspector General released its annual audit of HHS's compliance with the Payment Integrity Information Act of 2019 (PIIA) for fiscal year 2024. The report explicitly identifies eight HHS programs that failed PIIA. Foster Care (Title IV-E) is one of them.

The OIG, working through Ernst & Young as independent auditor, recorded the following finding verbatim:

"HHS included six of the States reviewed in FY 2024 in the error rate calculation for the Foster Care program, which EY noted does not represent a statistically valid process."

PIIA requires that any federal program with annual outlays exceeding $10 million be assessed every three years for improper-payment risk and, where risk is significant, must produce a statistically valid annual estimate of the improper payment rate. The foster care program in FY2023 disbursed $5.053 billion. The agency's response to PIIA's statistical-validity requirement has been to base its reported error rate on whichever six states happened to cycle through the IV-E primary review that year. EY — HHS's own contracted auditor — said the obvious: a sample of six self-selected states is not a statistically valid estimator for a 50-state program.

The agency neither expanded the sample nor adopted an alternative valid methodology. It simply reported the number it had.

The other PIIA failures form a pattern. Of eight HHS programs flagged for non-compliance:

• Temporary Assistance for Needy Families (TANF) — did not report any improper-payment estimate at all
• Foster Care (Title IV-E) — statistically invalid sample
• Head Start — improper payment rate exceeded the 10% threshold
• Medicare Advantage — recovery audits incomplete or delayed
• Advance Premium Tax Credit — only a partial error rate reported
• Medicare FFS — failed to demonstrate integrity improvements
• Uninsured Program — sampling methodology incomplete
• Programs lacking risk assessments — HHS did not conduct PIIA risk assessments for all programs over $10M annually within the 3-year cycle

GAO testimony to the House Ways and Means Committee on April 8, 2025 (GAO-25-108205, Temporary Assistance for Needy Families) added a parallel data point: states are also routing TANF, Title IV-B, and Social Services Block Grant funds into "child welfare costs that were not eligible for Title IV-E reimbursement." When state-claimed costs fail the IV-E test, states are double-dipping by re-routing them through the much weaker oversight of TANF. GAO's December 2024 work cited in that testimony "found that the federal government provides limited oversight of improper payments" in this fund-shifting space.

---

PART IV — THE FEDERAL EXPENDITURE BASELINE (FY2023)

Source: Title IV-E Programs Expenditure and Caseload Data 2023, ACF Children's Bureau, posted October 11, 2024 (Form CB-496 quarterly reports).

Program	FFP Claimed FY2023	Avg. Monthly Caseload	Per-Child Annual FFP
Foster Care	$5,053,093,360	121,572.5	~$41,540
Adoption Assistance	$3,985,714,249	560,210	~$7,114
Guardianship Assistance	$304,804,436	45,153	~$6,750
Prevention Services (FFPSA)	$172,494,603	18,323	~$9,414
Kinship Navigator	$2,743,436	563	~$4,872
Total IV-E FFP	$9,518,850,084	745,822	—

Two facts from this table:

1. The Kinship Navigator program is a token line item. ACF claims for the entire United States dropped from $12.69 million in FY2022 to $2.74 million in FY2023 — a 78% decline — while average monthly children served fell from over 3,200 in FY2021 to 563 in FY2023. The federal Kinship Navigator program is reaching, on average, eleven children per state per month.

2. Foster Care FFP per child is roughly six times Kinship Navigator FFP per child. Even if every dollar were perfectly spent, the price differential alone explains why states route children to licensed homes — and why a regulator using only 80-case audits cannot police the resulting volume.

A 5% error rate applied to Foster Care FFP alone implies $252.65 million per year in misclaimed federal foster care dollars. The four 2024–2025 reviews compiled here found a 4-state weighted average case-error rate of 11.25%. Applied to Foster Care FFP that is $568.5 million per year. Applied to Foster Care + Adoption Assistance + Guardianship FFP combined ($9.34 billion of the $9.52 billion total), an 11.25% error rate implies $1.05 billion per year. ACF recovered approximately $0.6 million combined from these four states.

Three-year leakage at 11.25% — even if half is corrected through self-reporting and Single Audit findings — falls inside the $2 billion to $4 billion per year band stated in this investigation's directive.

---

PART V — THE KINSHIP PARITY PARADOX

While ACF lets states under-recover overpayments on licensed foster care, the same states aggressively under-pay relatives — and federal courts have already ruled that under-payment is unlawful.

D.O. v. Glisson, 847 F.3d 374 (6th Cir. 2017)

Two Kentucky children were placed by court order with relative caregivers whom Kentucky declined to license as foster parents. Kentucky paid the relatives a small "kinship care" stipend that was a fraction of the licensed foster care maintenance payment (FCMP). The Sixth Circuit held that Title IV-E creates an enforceable federal right to FCMPs for any child meeting the § 472 eligibility criteria, regardless of whether the placement is with a licensed stranger or with a court-placed relative. The court rejected Kentucky's argument that the IV-E statute is a Spending Clause grant program with no individual right.

J.B.-K. v. Secy of Ky. Cabinet, No. 21-5074 (6th Cir. Sep. 16, 2022)

Five years after Glisson, Kentucky returned to the Sixth Circuit arguing that when a state court — not the Kentucky Cabinet — places a child with a relative or "fictive kin," the Cabinet does not have "placement and care responsibility" under § 672(a)(2)(B), and therefore IV-E does not attach. The Sixth Circuit, in an opinion by Judge Nalbandian (with McKeague joining and Murphy concurring), affirmed the district court's reading: when a Kentucky state-court order places a child directly with a relative, the Cabinet is not the placement-and-care agency; the children fall outside § 672(a)(2)(B)'s text; and FCMPs are not owed by the federal program. The court emphasized that this outcome was statutory, not normative, and noted the Glisson private-right-of-action question is the subject of a circuit split with the Second Circuit (N.Y. State Citizens' Coalition for Children v. Poole, 922 F.3d 69 (2d Cir. 2019)) and the Ninth.

The combined effect

Kentucky and other states now use court-driven relative placements as a structural workaround to avoid IV-E payment obligations, while:

• Routing the children who are in licensed care through the IV-E program, where the eligibility audit catches only 0.044% of child-months;
• Letting overpayments stay on the books because the audit's repayment scope is bounded to the sampled cases;
• Citing budget constraints to deny the kinship payments their state-court systems are generating the conditions for.

Ohio's continued non-compliance with Glisson is documented in Ohio Has Ignored a Federal Court Decision Requiring Foster Care Payments to Relatives, The Imprint, July 16, 2019. In Ohio, kinship caregivers receive a state-funded kinship payment that runs roughly a third of the licensed FCMP rate — even though Ohio is in the Sixth Circuit and Glisson is binding precedent there.

The same pattern is documented in Texas, Michigan, and Nebraska. The federal government has the leverage to enforce kinship parity. ACF could condition IV-E approval, modify 45 CFR § 1356.71 to make kinship-rate disparity an audit category, or sue under § 1320a-2a to withhold federal funds. None of those tools have been used. The agency that allows states to under-recover IV-E overpayments is the same agency that has not enforced Glisson against the states' under-payment of kin.

---

PART VI — WHAT THE PRIMARY REVIEW MISSES

1. The 6-month PUR window cannot detect long-running ineligibility

A child placed in 2020 with foster parents whose background check revealed prohibited offenses can generate four years of improper IV-E claims before any review touches the case. ACF can extend the disallowance period back to the actual ineligibility start date once a sample case lands on it (as in Maryland's Sample #4) — but only for the 80 cases sampled. Every other child in the same situation continues to generate improper claims uninterrupted.

2. The audit does not look at the population of denied cases

States can mis-deny IV-E for eligible kinship placements — Glisson-style violations — and pay nothing back, because the audit reviews only paid cases. There is no symmetric "denial review" inside 45 CFR § 1356.71.

3. The audit does not test the AFCARS data the sample is drawn from

ACF draws the sample from the state's own AFCARS submission. If a state under-reports its foster care population — or reclassifies certain placements to keep them out of AFCARS — those children are invisible to the audit.

4. The audit tolerates judicial-determination delays as the modal error

Across all four 2024–2025 non-compliance findings, the most common error category is late or missing judicial determinations of "reasonable efforts to finalize a permanency plan." State courts routinely miss the federal 12-month deadline by 6–24 months. The systemic fix would be coupling federal IV-E payment to a digital judicial-finding feed; the agency has not built that.

5. The audit does not address fraud

The 80-case primary review is an eligibility audit, not a fraud audit. HHS-OIG's separate state Title IV-E audits — Wisconsin (LAB 16-5, March 2016; 19-3, March 2019), Maine (DAB 2292, 2009) — are episodic. The OIG conducts fewer than five state-level IV-E claims audits per year against a $5 billion-per-year federal exposure. By contrast, CMS's PERM (Payment Error Rate Measurement) program audits Medicaid eligibility every three years across all 50 states with a national valid sampling design.

---

PART VII — KEY ACTORS

HHS / ACF leadership of the IV-E review program

• Joseph Bock — Acting Commissioner, Administration on Children, Youth, and Families (ACYF), signing recent disallowance letters (April 2025).
• Gail Collins — Program Implementation Director, Children's Bureau, Washington, DC.
• Minerva Gant — Child Welfare Program Specialist, Children's Bureau, Washington, DC.
• Karla Richardson — Grants Management Specialist, FPR/OGM/ACF, Atlanta, GA — named contact for IV-E disallowance Form CB-496 adjustment mechanics.
• Dianne Kelly — Regional Program Manager, Children's Bureau Region 4 (Atlanta) — handles SC, GA, AL, FL, MS, NC, TN, KY.

Federal court actors in the kinship parity line of cases

• Judge David W. Nalbandian — author of J.B.-K. v. Secy of Ky. Cabinet, 6th Cir. 2022 (with McKeague, J., joining).
• Judge Eric Murphy — concurrence in J.B.-K., flagging the Glisson private-right-of-action circuit split.
• Judge Gregory F. Van Tatenhove, E.D. Ky. — district court below in J.B.-K.
• HHS Secretary Xavier Becerra — solicited as amicus by the Sixth Circuit in J.B.-K.; HHS filed a brief siding with Kentucky on the placement-and-care issue.

Independent auditor for HHS PIIA compliance

• Ernst & Young LLP (EY) — author of the May 22, 2025 OIG-published statistical-validity finding on the IV-E program.

---

PART VIII — ACTIONABLE FINDINGS

1. The 5–10% error tolerance in 45 CFR § 1356.71(h) is itself the loophole. It functions as an institutional accommodation between HHS and the 50 state governors. Any reform that does not lower the tolerance and require population-wide extrapolation at the primary-review stage will not change the leakage rate.

2. Statutory amendment to require extrapolation on first non-compliance. The cleanest fix is a one-line amendment to § 474 of the Social Security Act adding: "If a State is determined under 45 CFR 1356.71 to be not in substantial compliance during a primary review, the Secretary shall determine the additional disallowance based on extrapolation from the sample to the universe of claims paid for the period under review." That alone would convert four-state recoveries from $0.6 million to roughly $200–600 million per year, immediately.

3. PIIA-compliant statistical sample. Per the OIG's May 22, 2025 finding, ACF must redesign the IV-E error-rate methodology to produce a statistically valid 50-state estimate. The PERM model is on the shelf; CMS uses it for Medicaid.

4. Kinship parity enforcement. ACF should issue a Program Instruction stating that IV-E approval is conditioned on payment of FCMPs at the licensed-foster-care rate to Glisson-class kinship caregivers within the Sixth Circuit, and that under-rate kinship payments outside the Sixth Circuit will be treated as substantial non-compliance with the IV-E state plan under § 472(a).

5. Single-source judicial-determination feed. State court systems should be required, as a condition of IV-E participation, to provide ACF with structured monthly data on every "reasonable efforts" judicial finding made for every IV-E-claimed child. The existing technology (state Court Improvement Project funding under § 438) was designed for exactly this purpose.

6. Public IV-E case scorecard. ACF should publish, by state, the rolling triennial primary-review history: sample errors, error categories, disallowance amounts, and PIP completion status.

7. Tip-line and FOIA priorities. Whistleblowers inside state IV-E agencies — particularly fiscal staff who handle Form CB-496 — are the single most underutilized source of intelligence on improper claims. FOIA priorities should include: (a) all FY2024 and FY2025 IV-E primary review final reports; (b) the EY work papers underlying the May 22, 2025 OIG finding; (c) the HHS amicus brief in J.B.-K. v. Secy of Ky. Cabinet; (d) the secondary-review extrapolation history for the period FY2000–FY2024.

---

PART IX — PRIMARY SOURCES

Federal regulations and statutes - 42 U.S.C. §§ 670–679c (Title IV-E of the Social Security Act, Foster Care, Prevention, and Permanency) - 42 U.S.C. § 672(a) (foster care maintenance payment eligibility, including the AFDC look-back at § 672(a)(3) and the "placement and care" requirement at § 672(a)(2)(B)) - 42 U.S.C. § 671(a)(20) (state plan safety / criminal-records-check requirements) - 45 CFR § 1355.20; 45 CFR § 1356.21; 45 CFR § 1356.30 - 45 CFR § 1356.71 (the eligibility review regulation analyzed throughout — sample size at (c)(2); compliance threshold at (h)(2); disallowance methodology including secondary-review extrapolation trigger at (j)(2)) - Payment Integrity Information Act of 2019, Pub. L. No. 116-117, 31 U.S.C. § 3351 et seq.

ACF Children's Bureau publications - Title IV-E Foster Care Eligibility Review Guide (ACF, current edition). - Title IV-E Programs Expenditure and Caseload Data 2023, ACF Children's Bureau (Oct. 11, 2024). - Louisiana DCFS Final Report, Primary Review, Title IV-E Foster Care Eligibility, April 1, 2023 – September 30, 2023 (ACF FY2024). - Maryland DHS Final Report, Primary Review, Title IV-E Foster Care Eligibility, April 1, 2023 – September 30, 2023 (ACF FY2024). - South Carolina DSS Title IV-E Review Final Report Addendum (ACF, April 7, 2025). - District of Columbia Final Report Primary Review – Title IV-E Foster Care Eligibility, April 1, 2024 – September 30, 2024 (ACF FY2025).

HHS OIG and GAO - HHS OIG, Department of Health and Human Services Met Many Requirements But It Did Not Fully Comply with the Payment Integrity Information Act of 2019 and Applicable Improper Payment Guidance for Fiscal Year 2024 (May 22, 2025). - HHS OIG, Maine Department of Health and Human Services (DAB 2292, 2009). - Wisconsin Legislative Audit Bureau, Title IV-E findings (LAB 16-5, March 2016; LAB 19-3, March 2019; LAB 24-3, March 2024). - GAO, Temporary Assistance for Needy Families: Preliminary Observations on State Spending of Federal Funds for Child Welfare (GAO, Sept. 24, 2024). - GAO-25-108205, Temporary Assistance for Needy Families (GAO testimony, House Ways and Means, April 8, 2025).

Federal court opinions - D.O. v. Glisson, 847 F.3d 374 (6th Cir. 2017), cert. denied (2017). - J.B.-K., et al. v. Secy of Ky. Cabinet for Health and Family Services, No. 21-5074 (6th Cir. Sept. 16, 2022). - N.Y. State Citizens' Coalition for Children v. Poole, 922 F.3d 69 (2d Cir. 2019). - HHS Departmental Appeals Board: DAB 1899 (Missouri DSS, 2003); DAB 2080 (Florida DCF, 2007); DAB 2994 (Missouri DSS, 2020); DAB A-25-6 (South Carolina DSS, 2025).

Secondary sources - The Imprint, Ohio Has Ignored a Federal Court Decision Requiring Foster Care Payments to Relatives (July 16, 2019). - Policy Matters Ohio, Support Ohio Children by Funding Kinship Care (Nov. 18, 2020).

---

CLOSING NOTE — WHY THIS MATTERS FOR THE MISSION

Every data point in this investigation describes the same underlying fact: the federal program that pays states $5 billion a year to take children into custody is policed by an audit so thin that the states police themselves on the back end and pocket the difference. The same regulatory permissiveness that allows a state to claim federal money for a child placed in a home with prohibited criminal records on file is what allows the same state to deny federal money to the grandmother who would have raised that child for free.

This is not a financial-control failure that happens to involve children. It is a child-safety failure financed by the federal taxpayer. The eligibility errors documented in the Louisiana, Maryland, South Carolina, and District of Columbia 2024–2025 reviews are not bookkeeping mistakes. They are placements that did not meet the federal floor on judicial review, on licensing, on background checks — and the federal government paid for them anyway because the audit was small enough to miss them.

The fix is not new authority. The fix is using the authority that has been on the books since 1980.

— OPUS, Project Milk Carton — Filed May 3, 2026

---

This report is suitable for SCRIBE conversion into a video article. Recommended companion deliverables: (1) a state-by-state IV-E error-rate scorecard infographic; (2) a kinship parity heat map of the Glisson circuit split; (3) a 90-second video explainer on the "5-error pass line" that frames the §1356.71(h)(2) threshold as the policy lever.