By Cynthia Wu

Abstract

Following the Supreme Court’s holding that the International Emergency Economic Powers Act (IEEPA) does not authorize the challenged tariffs, refund implementation has shifted from merits to mechanics: liquidation status, administrative capacity, and the remedial scope of the U.S. Court of International Trade (CIT). In early March 2026, the CIT (Eaton, J.) issued a refund order in Atmus Filtration, Inc. v. United States directing U.S. Customs and Border Protection (CBP) to liquidate unliquidated entries “without regard to IEEPA duties,” and to reliquidate liquidated entries “for which liquidation is not final” without IEEPA duties. The court then amended and temporarily suspended the order as proceedings moved onto an appeal- and implementation-management track. This paper (i) situates these orders within the customs “refund engine” (liquidation/reliquidation → refund), (ii) analyzes the CIT’s asserted systemwide authority and the government’s anticipated challenges, (iii) examines CBP’s operational constraints and proposed ACE modifications, and (iv) identifies the principal unresolved cohort—entries already liquidated and final—and evaluates plausible pathways for addressing that gap.

1. Background: From Merits to Mechanics

The Supreme Court’s decision in Learning Resources, Inc. v. Trump resolved the core statutory question—IEEPA does not authorize the tariff program—but did not prescribe a refund program or timeline.¹ That omission predictably pushed the dispute into the customs system’s existing architecture: liquidation and finality, protests and reliquidations, and judicial remedies calibrated to those mechanisms.

Under the Tariff Act, refunds generally do not occur merely because a measure is declared unlawful in the abstract; refunds are typically paid when liquidation or reliquidation determines that duties deposited exceed duties lawfully owed.² The system therefore makes “entry posture” decisive: whether an entry is unliquidated, liquidated but still within a correction window (e.g., voluntary reliquidation or protest posture), or liquidated and final.³

The institutional consequence is that “refund litigation” often becomes “implementation litigation.” Plaintiffs seek relief that produces corrected duty outcomes—commonly by directing liquidation or reliquidation without the unlawful duty component—because those corrected accounting events activate the statutory refund obligation and interest mechanics.⁴ That dynamic set the stage for the CIT’s March 2026 refund order and the government’s feasibility objections.

1. The CIT’s March 2026 Refund Order and Subsequent Amendments
2. The March 4 Refund Order in Atmus

On March 4, 2026, Judge Eaton issued an order in Atmus Filtration, Inc. v. United States stating that “all importers of record whose entries were subject to IEEPA duties are entitled to the benefit of the Learning Resources decision.”⁵ The order directed CBP to “liquidate all unliquidated entries … without regard to IEEPA duties,” and further provided that “[a]ny liquidated entries for which liquidation is not final shall be reliquidated without regard to IEEPA duties.”⁶

Critically, the order did not address entries that were already liquidated and final—the cohort most exposed to finality arguments and missed statutory deadlines.⁷

Doctrinally, the order tracks the structure of the customs refund engine. Liquidation is CBP’s final computation of duties; reliquidation revises that computation; and refunds flow downstream from those corrected computations.⁸ The court’s directive thus attempted to translate a merits holding (no tariff authority) into the operational predicate for repayment (liquidation/reliquidation without the unlawful duty line items).

1. Amendment and Suspension

Public reporting and practitioner alerts describe that the court issued an amended order and later temporarily suspended the operative order while proceedings moved into a managed posture, including closed conferences and anticipated appellate activity.⁹ The sequence underscores that the central battleground is no longer whether the IEEPA tariffs were unlawful, but whether the CIT can order systemwide corrective liquidation activity and how quickly CBP must (or can) implement it.

III. The Scope Question: Systemwide Relief, Uniformity, and Post-CASA Constraints

A recurring legal issue is whether the CIT may issue relief that effectively benefits non-parties. The March 4 order’s opening declaration—“all importers of record”—reads as systemwide in effect.¹⁰ Commentators have emphasized that Judge Eaton distinguished the Supreme Court’s universal-injunction discussion in Trump v. CASA by pointing to the CIT’s national jurisdiction and its specialized statutory role in trade and customs disputes.¹¹

The government, by contrast, has signaled resistance to universal relief, arguing in court that refunds should not automatically issue to every importer absent entry-by-entry preservation (including litigation filing).¹²

This scope dispute is not academic—it determines whether the refund process is (i) litigation-driven, pressuring large numbers of importers into protective suits, or (ii) administratively scalable, requiring CBP to design a uniform program that identifies eligible entries, computes principal and interest, and issues payments while maintaining compliance checks and audit integrity.

1. Feasibility and the “ACE Problem”: CBP’s Operational Objections
2. CBP’s Position: Current Systems Cannot “Flip the Switch”

In reporting on CBP’s submissions to the CIT, CBP stated it could not immediately comply with an order to begin refunding at scale using existing technology, processes, and manpower.¹³ CBP also suggested that, with time to retool the Automated Commercial Environment (ACE), it could implement functionality intended to streamline and consolidate refunds and interest payments on an importer basis, rather than issuing tens of millions of entry-level refunds.¹⁴

On the record described publicly, CBP has tied the scale problem to entry volumes and staffing burdens, framing the implementation issue as a systems-and-controls challenge, not simply a policy preference.¹⁵

1. Why “Aggregated Payment” Can Still Be Legally Compatible

Even if the legal trigger remains entry-based (liquidation/reliquidation determines an overdeposit), the payment form need not be “one entry, one refund.” In principle, CBP can aggregate refunds by importer so long as the underlying calculations remain (i) lawful, (ii) traceable to entries/lines, and (iii) auditable—i.e., accompanied by a schedule that reconciles the aggregate payment back to entry numbers, dates, deposits, offsets, and interest computations.

1. The Four Implementation Frictions CBP Must Solve

Even with improved ACE tooling, feasibility turns on at least four frictions:

1. Data integrity and mapping. The system must associate each IEEPA duty assessment with the correct entry, liquidation status, and any post-entry adjustments or payments.¹⁶
2. Netting and offsets. CBP must avoid refunding amounts that should be offset by other liabilities (e.g., AD/CVD deposits, unpaid fees, or other duty components), and must do so in a way that is reviewable and consistent.¹⁷
3. Interest computation. Interest is statutory, not discretionary; CBP must apply the applicable interest conventions consistently across millions of deposits, partial payments, and timing scenarios.¹⁸
4. Finality boundaries. The system must draw a clean line between “not final” entries eligible for reliquidation and “final” entries requiring different legal vehicles or new authorization.

Taken together, CBP’s feasibility narrative is more than tactical delay: it reflects that the customs system was not designed for instant mass unwinding of a year-long tariff program across tens of millions of entries.¹⁹

1. The Unresolved Cohort: “Liquidated and Final” Entries
2. Why “Final” Matters

The March 4 order expressly addressed unliquidated entries and liquidated-but-not-final entries, leaving unaddressed those entries that have become final through the passage of time and operation of customs finality rules.²⁰ Finality is the core barrier: once liquidation is final and conclusive, entry-level correction ordinarily requires a timely protest and, if denied, CIT review through the protest-denial pathway.²¹

In other words, the system’s default posture is that a merits unlawfulness holding does not automatically reopen final liquidations.

1. Plausible Solutions and Their Practical Limits

1. Litigation expansion and targeted equitable relief. Importers may litigate whether the CIT’s remedial authority can reach final entries in narrow circumstances (e.g., ultra vires exactions) and whether equitable relief can function as a “functional equivalent” of reliquidation where ordinary protest routes are unavailable or manifestly inadequate. This approach is fact-dependent, contested, and may yield inconsistent outcomes unless tightly consolidated.²²
2. Congressional intervention (programmatic refunds). Reported legislative proposals contemplate mandatory refund processing, timelines, and prioritization (including small businesses). A statutory program could override entry-by-entry finality constraints by creating a refund entitlement not conditioned on protest posture.²³ Feasibility depends on political timing and drafting choices (scope, appropriations, interest, offsets, fraud controls).
3. Administrative claims process. CBP could build a claims intake to avoid litigation-by-default. But absent new statutory authority, CBP’s ability to pay on final liquidations without reopening them is constrained—meaning a claims program either needs (i) a lawful reopening mechanism tied to existing statutes, or (ii) a statutory directive.²⁴
4. Hybrid model. The most operationally realistic near-term model is: automatic liquidation/reliquidation relief for unliquidated and non-final entries, while final entries remain subject to litigation strategies or legislative repair. The equity problem is obvious: similarly situated importers could experience different outcomes solely due to liquidation timing.

1. Forecast: What a Workable Refund Program Likely Looks Like

A practicable refund regime will likely converge on three design features:

1. Status-based segmentation. Unliquidated entries are handled by liquidation without IEEPA duties; non-final liquidations by reliquidation; final entries require separate legal authorization or adjudicated exceptions.²⁵
2. Importer-level aggregation. This is likely the only scalable payment method. Even if calculations remain entry-true, CBP can reduce transaction volume dramatically by consolidating payments per importer while providing downloadable reconciliation schedules.²⁶
3. Bounded submissions. Even if CBP seeks “minimal submission,” some importer validation may be required for identity, banking information, offsets, and dispute handling.²⁷

The remaining uncertainty is less about engineering than about authority: whether appellate courts will permit a systemwide program reaching beyond parties, and how finality constraints will be addressed for the liquidated-and-final cohort.

VII. Case Consolidation and Managed Proceedings: From “Should Refund” to “How to Refund”

Reporting indicates the CIT moved toward managed proceedings, including closed conferences characterized as settlement or implementation conferences.²⁸ These mechanisms can improve uniformity and reduce contradictory rulings. But they also mean importers should treat “refund timing” as dependent on: (i) judicial scope rulings, (ii) CBP implementation readiness, and (iii) appellate stays—rather than assuming an immediate, automatic credit will hit accounts simply because the merits are resolved.

VIII. CBP’s Execution Difficulty: Why “CBP Issues Refunds Every Day” ≠ “CBP Can Refund All IEEPA Tariffs Now”

A common refrain in court coverage is that CBP “does refunds every day,” so mass IEEPA refunds should be straightforward. That comparison misses how different this situation is from ordinary overpayment refunds.

Ordinary refunds usually involve a discrete set of entries, limited importer populations, and well-established workflows (including routine liquidation/reliquidation outcomes). By contrast, the IEEPA refund problem is a policy-level unwind across a massive entry universe, with three scaling multipliers:

• Volume and concurrency. Public reporting describes more than 330,000 importers and over 53 million entries involving IEEPA duties.²⁹ This is not a normal refund queue; it is a parallel system-wide workload.
• Interest complexity. Refund interest is statutory and timing-sensitive; applying consistent interest rules across millions of deposits and timelines is not merely arithmetic—it requires reliable deposit-date and liquidation-date data, plus consistent treatment of additional deposits and statement payments.³⁰
• Cross-liability checks. CBP has indicated that it must verify that no other duties, taxes, or fees are owed (e.g., AD/CVD or other trade remedies) before paying, which is a control function that becomes heavy at scale.³¹

CBP’s proposed solution—new ACE functionality that consolidates refunds and interest on an importer basis—should be understood as an attempt to preserve two competing requirements: speed and auditability. An approach that is “fast but not reconcilable” invites error, fraud exposure, and downstream litigation; an approach that is “perfect but manual” collapses under workload. The legal system can order refunds; it cannot wish away the controls and data requirements that make refunds defensible.

1. Political and Administrative Variables: Personnel Shocks, Agency Funding, and Uncertainty

Although the refund dispute is fundamentally legal and operational, contemporaneous political and administrative events can shape execution timelines. For example, reporting describes major Department of Homeland Security (DHS) leadership change and an agency funding lapse/shutdown environment, with the DHS Secretary’s departure announced amid shutdown conditions and broader congressional scrutiny.³²

These events should not be overstated as direct causes of any particular court order or judicial pause. But they are relevant as execution risk factors because CBP (as a DHS component) depends on budget stability, leadership continuity, staffing, and IT prioritization—especially when the agency is simultaneously managing enforcement surges and attempting a major systems build-out for refunds.³³

The practical takeaway is modest but important: importers should track not only judicial developments (scope, stays, appellate posture) but also whether CBP’s implementation plan is being resourced and prioritized. In a mass-refund environment, administrative bandwidth is a real constraint.

Conclusion

The March 2026 Atmus orders illustrate the customs system’s basic truth: major refund events are implemented through corrected liquidation outcomes, not by abstract declarations alone. The CIT attempted to translate the Supreme Court’s merits holding into operationally meaningful accounting events—liquidation/reliquidation without IEEPA duties—while the government pressed feasibility and scope objections. CBP’s execution posture signals that any workable program must be staged, systematized, and scalable—likely through importer-level aggregation and new ACE functionality. The most consequential unresolved issue remains liquidated-and-final entries, which fall outside the cleanest administrative pathways and may require targeted litigation or legislative intervention.

Footnotes (Authorities)

1. Learning Res., Inc. v. Trump, 607 U.S. ___ (2026) (slip op. at 26).
2. 19 U.S.C. § 1505(b)–(c); 19 C.F.R. § 24.36.
3. 19 U.S.C. §§ 1501, 1504, 1514; 19 C.F.R. pt. 174.
4. 28 U.S.C. § 2643(a)–(c); 19 U.S.C. § 1505(b).
5. Atmus Filtration, Inc. v. United States, Order (Ct. Int’l Trade Mar. 4, 2026) (as summarized in practitioner alerts).
6. Id.
7. Practitioner alerts noting the order did not address liquidated-and-final entries.
8. 19 U.S.C. § 1505(b)–(c); 19 C.F.R. § 24.36.
9. Coverage describing amendment/suspension and managed proceedings.
10. Atmus reporting/alerts describing “all importers of record” framing.
11. Discussion of CASA and the CIT’s asserted distinction in commentary/alerts.
12. Government position quoted in reporting/alerts (refunds limited to litigants).
13. Reporting on CBP telling the CIT it cannot immediately comply.
14. Reporting on CBP proposing ACE modifications and a ~45-day timeline.
15. Reporting quoting CBP on entry counts and workload hours.
16. Id.
17. Alerts discussing offsets/other duty regimes and implementation friction.
18. 19 U.S.C. § 1505(c); 19 C.F.R. § 24.36(a)(1).
19. Reporting on CBP’s feasibility constraints and system redesign.
20. Practitioner alerts noting the court did not address final liquidations.
21. 19 U.S.C. § 1514(a), (c)(3); 28 U.S.C. § 1581(a); 28 U.S.C. § 2636(a).
22. 28 U.S.C. § 2643(a)–(c); 28 U.S.C. § 1581(i).
23. Reporting and commentary describing proposed legislation prioritizing small businesses and requiring refunds.
24. Discussion in alerts/news noting administrative limits absent new authority.
25. Status-based segmentation described in alerts and reporting of the Atmus order.
26. Reporting on importer-level aggregation in ACE.
27. “Minimal submission” framing in CBP reporting and implementation commentary.
28. Reporting describing closed conferences/settlement conferences at the CIT.
29. Reporting quoting entry/importer counts.
30. 19 U.S.C. § 1505(c); 19 C.F.R. § 24.36(a)(1).
31. Practitioner commentary describing CBP compliance checks and other duties/taxes considerations.
32. Report on DHS leadership change and shutdown conditions.
33. Id.