HUD claimed a 0% error rate on $80.2B it never actually screened
Summary
In its 2023 compliance reporting, HUD told regulators it had screened 1.4 million payments worth $80.2 billion in fiscal year 2022 and found an ineligible-party rate of zero percent. What HUD did not disclose at the time: the computer matching agreement it needed to actually run that screening against Treasury's exclusion and eligibility databases had expired in 2019 and still has not been renewed. When auditors tested a live sample of payments in fiscal year 2024, 24 of 131 flagged for missing or expired federal registration -- and as of February 2025, 11 of those 24 were still unregistered, having collected roughly $212 million after their registration lapsed. It's one specific test result inside a much larger blind spot: HUD has now gone 8 straight years unable to even estimate improper payments in its two largest rental-assistance programs, a combined $50 billion a year, and 12 straight years out of compliance with the federal law that requires it to try.
The claim: zero ineligible recipients
The Do Not Pay Initiative is the federal government's standard tool for catching payments to entities that shouldn't get them -- dead people, debarred contractors, organizations without an active SAM.gov registration. is required by the Payment Integrity Information Act⧉ to use it before disbursing funds. In its 2023 compliance reporting to federal regulators, said it had done exactly that: screened 1.4 million payments totaling $80.2 billion in fiscal year 2022, and observed an ineligible-party rate of zero percent.
What the claim assumed
A zero-percent rate assumes the screening was actually checking something. It wasn't. 's computer matching agreement with the Treasury Department -- the technical arrangement that lets an agency query Do Not Pay's databases in the first place -- expired in 2019. did not renew it. According to , did not even acknowledge that the agreement had lapsed in its reporting to the Office of Management and Budget until fiscal year 2024; for the years in between, it kept reporting screening activity as though the underlying system were still live. A zero percent rate against a system that isn't running isn't evidence of clean payments. It's evidence the check never happened.
The same gap, found again in live testing
's fiscal year 2024 testing gave the claim a more direct check. Reviewing a sample of Project-Based Rental Assistance payments, 's own Chief Financial Officer's office flagged 24 of 131 samples because the required Unique Entity Identifier or SAM.gov registration was missing or expired. As of February 2025 -- months after the flag was raised -- 11 of those 24 entities, nearly half, still had no active registration. Payment records showed those 11 had collected approximately $212 million after their registrations lapsed.
View data as table
| Samples flagged (of 131 tested) | 24 |
|---|---|
| Still expired as of Feb. 2025 | 11 |
The bigger blind spot this sits inside
The screening failure is a symptom, not the disease. Fiscal year 2024 was the eighth consecutive year could not produce a compliant improper-payment estimate for its Tenant-Based and Project-Based Rental Assistance programs -- together more than $50 billion a year, roughly two-thirds of 's total spending. Over those 8 years, 's cumulative spending on the two programs reached about $315 billion with no valid estimate of how much of it was improper. The last time produced one, in 2016, it found $1.7 billion in improper payments -- a number now dwarfed by everything spent since without any comparable check. 's own Chief Financial Officer's office has told it may not have a working estimate again until fiscal year 2027.
View data as table
| FY2016 (last compliant estimate) | 30.7 |
|---|---|
| FY2023 | 45.3 |
| FY2024 | 50 |
View data as table
| Last known improper-payment estimate (2016) | 1.7 |
|---|---|
| Spending since, with no estimate (8 years) | 315 |
has traced the failure to a specific methodological choice: 's Chief Financial Officer's office samples its own general-ledger disbursements rather than the actual payments local housing agencies and contract administrators make to landlords -- the level where eligibility is determined and most of the risk sits. That population is too large to sample statistically the way has set it up, so resorts to judgmental sampling, which does not satisfy the law. A January 2024 management alert⧉ found OCFO had gone so far as to improperly close two prior recommendations on this exact sampling gap. 's Deputy Secretary promised a corrective plan within 30 days of that alert. As of the May 2025 report, says still had not proposed a management decision on it -- the recommendation remains open, alongside a separate one calling for a senior-level council to coordinate compliance across 's offices.
The takeaway
- A number that looks clean can mean the test never ran. 's reported 0 percent ineligible-party rate on $80.2 billion wasn't a result of screening -- the system needed to screen anything had been offline since 2019, a fact did not disclose in its reporting until forced to in fiscal year 2024.
- When auditors ran a live check, the same failure showed up at smaller scale. Nearly half of the payments flagged for registration problems in 2024 were still unresolved months later, with about $212 million paid out after the registration lapsed -- consistent with, not contradicting, the larger pattern.
- The screening gap is a symptom of an eight-year measurement failure. hasn't produced a compliant improper-payment estimate for its two biggest rental-assistance programs since 2016, has spent roughly $315 billion since without one, and by its own Chief Financial Officer's estimate may not have a working one again until 2027.
Figures on the FY2024 findings, the DNP/SAM.gov gap, and the improperly reported screening claim are from Audit Report 2025-FO-0006 (May 13, 2025). Figures on the FY2023 status, the 2016 baseline, and OCFO's own FY2027 projection are from Management Alert 2024--0001 (January 23, 2024). Both were read directly. The $212 million figure derives from a sample of 24 flagged payments out of 131 tested in one program (MF-PBRA), not a full-population audit, and 's report itself frames these as 'technically improper' payments tied to a registration lapse, not confirmed fraud or ineligibility of the underlying recipients.
Sources(2) ▾
- U.S. Department of Housing and Urban Development, Office of Inspector General, HUD Did Not Comply with the Payment Integrity Information Act of 2019 (Audit Report 2025-FO-0006) (2025-05-13) — The 8th-consecutive-year/12-year noncompliance figures, the FY2024 $50B combined total, the Payee-Tier-only MF-PBRA testing gap, the DNP computer-matching-agreement lapse since 2019, the 24-of-131 and 11-of-24 SAM.gov registration findings, the $212M paid to unregistered entities, the improperly reported 0%-ineligible-rate claim, and the still-open status of the 2024 Management Alert recommendation hudoig.gov · original document
- U.S. Department of Housing and Urban Development, Office of Inspector General, Management Alert (2024-IG-0001) -- Action Is Needed From HUD Leadership To Resolve Systemic Challenges With Improper Payments (2024-01-23) — The FY2023 $45.3B figure, the 7th-year/11-year noncompliance figures at the time, OCFO's own FY2027 projection, the last-compliant-estimate (2016, $1.7B) and spending-growth figures, the improperly closed recommendations, the sampling-methodology flaw, 's 30-day-plan response, the FY2000/2001 high-risk history, and the four other programs that did achieve compliance oversight.gov · original document
Comments
Always open. Logged-in readers can annotate paragraphs in place.
A zero-percent error rate on $80.2 billion in federal payments is the kind of number that should draw a second look before anyone repeats it. 's inspector general took that look, and the number didn't survive it: the computer matching agreement needed to actually screen those payments against Treasury's exclusion and death-record databases had expired years before the claim was made. The finding sits inside a larger pattern -- eight straight years in which has been unable to even estimate improper payments in the two rental-assistance programs that make up two-thirds of its budget.