Still Can't Count Its Own Overdue In-Bond Cargo
Summary
A Treasury Department inspector general audit found U.S. Customs and Border Protection cannot reliably track $716.76 billion in merchandise that moved through the country in-bond -- allowed to skip duty payments at the border so long as it reaches its destination or leaves the country by a deadline. When fixed an error in the software logic behind its own overdue-shipment reports, its count of overdue in-bond shipments for fiscal year 2020 fell from 1,233,554 to 736,848, a change of 496,706 records could not explain; one of its two reports alone dropped from 516,080 overdue shipments to zero. Auditors ultimately judged 's underlying data too unreliable to test, and confirmed 's official $90.55 million revenue-loss estimate for the year excludes in-bond merchandise altogether.
A shipment count that swings by half a million
tracks overdue in-bond shipments through two computer-generated reports, INBND-005 and INBND-006, run inside its Automated Commercial Environment (ACE) system. In September 2022, auditors asked to explain the criteria behind those reports; the inquiry led 's own technology staff to discover the underlying business logic was wrong -- it relied on a date that had nothing to do with whether a shipment was actually overdue. fixed the logic in December 2022. Run against the identical fiscal-year-2020 period, the fixed reports told a very different story: INBND-005 alone, which had identified 516,080 overdue shipments under the old logic, identified zero under the new one. 's Trade Transformation Office -- the office that maintains the reports -- could not explain why.
View data as table
| INBND-005 -- original logic | 516,080 | Overdue-at-destination-port report, before 's December 2022 fix |
|---|---|---|
| INBND-005 -- updated logic | 0 | Same report, same FY2020 period, after the fix -- could not explain the drop to zero |
| INBND-006 -- original logic | 717,474 | Overdue-for-export report, before the fix |
| INBND-006 -- updated logic | 736,848 | Same report, after the fix |
The check that couldn't check itself
Auditors then tried to verify whichever count was closer to correct by comparing both versions of the reports against 's own raw data extract of the same fiscal year's in-bond transactions. Neither matched. Under the original business logic, 361,551 of the reports' 1,233,554 flagged records -- 29% -- had no corresponding record in 's own extract. Under the updated logic, that share got worse, not better: 359,039 of 736,848 records, 49%, were missing. 's Trade Transformation Office told auditors an unknown filter may have caused the exclusions, but could not identify what it was or why it existed. Unable to reconcile 's reports with 's own underlying data, auditors concluded the data was not reliable enough to use for further testing -- meaning the actual number of overdue in-bond shipments in fiscal year 2020, and the revenue at risk from them, remains unknown to the agency responsible for collecting it.
View data as table
| Missing, original business logic | 29% | 361,551 of 1,233,554 records |
|---|---|---|
| Missing, updated business logic | 49% | 359,039 of 736,848 records |
Why the blind spot is a revenue problem, not just a data problem
An overdue in-bond shipment is not automatically fraud, but it is exactly the scenario 's own accounting exists to catch: merchandise that may have been diverted into the U.S. economy without anyone ever paying the duties, taxes, and fees owed on it. 's official estimate of revenue lost to trade noncompliance -- $90.55 million for fiscal year 2020, reported in the Department of Homeland Security's consolidated financial statements -- comes from a statistical sample of "entry summary" reviews, the paperwork filed only after merchandise has already, formally entered U.S. commerce. In-bond shipments that never reach that step, because they were diverted, lost, or simply never resolved in 's system, are excluded from that estimate by design. Making the gap harder to close, the in-bond application itself only records a shipment's insurance value, not the invoice value duties are actually calculated from -- so even a perfectly accurate count of overdue in-bonds would not, on its own, tell how much revenue was actually at stake.
does run a separate compliance-risk program, In-Bond Compliance Measurement (IBCM), created in December 2022 in response to Treasury 's prior audit. But IBCM only samples in-bonds that reached their destination -- it excludes overdue shipments, the category auditors say could be illegally diverted into U.S. commerce before ever reaching a port. Auditors cited 's Standards for Internal Control in the Federal Government as the baseline fell short of: agencies are expected to obtain reliable data in a timely manner, document their internal-control processes, and identify and assess risks to their own objectives. On all three counts, auditors found had not.
What Treasury's watchdog wants fixed, and by when
Auditors issued four recommendations to the Commissioner: verify the business logic behind the INBND reports and sample the results annually for three years; document how ACE identifies overdue in-bonds; continuously monitor the reports going forward; and, separately, actually assess the revenue risk that overdue in-bonds could be illegally diverted into U.S. commerce. concurred with all four. In its written response, told auditors it expects to close the first three recommendations by January 29, 2027, once a broader, already-underway overhaul of its in-bond systems and policies is further along. The fourth recommendation -- assessing how much revenue is actually at risk -- carries no target date in 's response; management tied it instead to the outcome of that same broader review, with no fixed point at which it becomes due.
- 's own recount of its overdue shipments moved by 496,706 records, and could not explain why. One of its two tracking reports fell from 516,080 overdue in-bond shipments to zero for the identical time period.
- Even after the fix, 's reports don't match 's own data. Auditors found 49% of the records in the post-fix reports missing from 's underlying data extract -- worse than the 29% mismatch before the fix -- and ultimately judged the data too unreliable to test.
- The government's official revenue-loss number for this system, by design, doesn't cover the shipments most at risk. 's $90.55 million trade-noncompliance estimate excludes in-bond merchandise -- the exact category this audit is about -- entirely.
- agreed to fix it, on two different timelines. Three recommendations carry a January 29, 2027 target date; the recommendation to actually measure the revenue at risk has no date attached at all.
This audit is a statutorily mandated follow-on to Treasury 's March 2022 audit (-22-033), which first raised unresolved concerns about 's overdue in-bond counts and revenue-loss estimates; both audits examined fiscal-year-2020 data. Fieldwork for this report ran from September 2022 through June 2024; Treasury states the report's June 18, 2026 issuance was delayed by lapses in funding that held up 's written management response. A separate recommendation from the 2022 audit, to modernize ACE's in-bond tracking and automation, remains open on its own timeline with an estimated completion date of December 2027.
Sources(1) ▾
- Department of the Treasury, Office of Inspector General, Revenue Collection: CBP Needs to Improve its Accountability of Merchandise Transported In-Bond to Assess the Impact on Revenue (OIG-26-034) (2026-06-18) — Full 32-page audit report, fetched directly and read page-by-page. A mandated-by-statute (TFTEA Section 112) follow-on to Treasury 's March 2022 audit (-22-033) of the same In-Bond Program, covering fieldwork from September 2022 through June 2024 on fiscal-year 2020 in-bond transaction data. Issuance was delayed to June 2026 by lapses in funding that held up 's written management comments. Includes Table 1 (FY2020 INBND-005/INBND-006 overdue counts, original vs. revised business logic), Table 2 (share of in-bond records missing from 's own data extract), the $716.76 billion total FY2020 in-bond transaction value, the $90.55 million FY2020 trade-noncompliance revenue-loss estimate that explicitly excludes in-bond merchandise, all four recommendations, and 's full written management response with target closure dates. oig.treasury.gov · original document
Comments
Always open. Logged-in readers can annotate paragraphs in place.
U.S. Customs and Border Protection lets some imported merchandise -- "in-bond" cargo -- move through the country to a different port, or leave it entirely, without duties being assessed at the border, as long as the shipment arrives or is exported within a set deadline. A June 2026 audit by the Treasury Department's Office of Inspector General⧉ -- Treasury's independent watchdog, required by statute to check 's revenue protection every two years -- found cannot reliably say how much of that cargo, worth an estimated $716.76 billion in fiscal year 2020 alone, actually made those deadlines. When fixed an error in the computer logic behind its own overdue-shipment reports, its count of overdue in-bond shipments for that year fell from 1,233,554 to 736,848 -- a change of 496,706 records neither nor the auditors could fully explain.