Skip to content
un-tariff / IEEPA Refund
Methodology · Form MET-01

How un-tariff computes refunds.

Every classification, every interest computation, every validation rule — what source we consult and how we derive the answer. No black boxes.

Data sources

un-tariff consumes four primary reference datasets, all versioned:

  1. IEEPA HTS registry — derived from the Harmonized Tariff Schedule Chapter 99 subheadings published by the USITC, cross-referenced to the executive orders that imposed them (EO 14193, 14194, 14195, 14245, 14257, 14323, 14329, 14380, 14382 — see Federal Register Vol. 90, No. 226). Each row has effective_from and effective_to dates, a source_citation field, and never overwrites. Historical classifications remain reproducible.
  2. CBP quarterly interest rates — published under 19 CFR 24.3a, stored quarter by quarter with effective dates. The current rates are confirmed against CBP's IEEPA FAQ.
  3. CAPE excluded-entry types — the categories of entries CBP has explicitly excluded from CAPE Phase 1, per the CAPE Trade Information Notice.
  4. IEEPA exclusions (USMCA, CAFTA-DR, etc.) — preferential-trade programs that exempted qualifying entries from some IEEPA tranches. Flagged per rule IEEPA-006 for human review.

Classification (rule IEEPA-001 through IEEPA-004)

For every line in your ACE Entry Summary export, the classifier:

  1. Looks up the HTS subheading in the IEEPA registry with the line's entry-filing date. If no registry row matches the date, the line is NOT_IEEPA and drops out of the refund.
  2. If multiple rows match (overlapping executive orders), selects the most specific row by effective_from proximity to the filing date. Ties go to the later row.
  3. Confirms the paid-duty amount on the line is consistent with the registry row's expected rate. Material deviations (> 0.5%) flag for human review.
  4. Records the rule ID, registry row version, and deviation amount in the audit log.

Stacking (rule IEEPA-005)

Where an entry line paid IEEPA alongside Section 232 or Section 301, the classifier separates the paid duty into three buckets using the applied-rate breakdown from the ACE export:

Only the IEEPA bucket is summed into the claim.

USMCA / preferential-trade exclusions (rule IEEPA-006)

For entries with a country of origin matching a preferential program (USMCA for Canada/Mexico, CAFTA-DR for Central America, etc.), the classifier checks whether the executive order that imposed IEEPA on that entry's HTS subheading had an explicit preferential-program carve-out. If yes, the line is flagged PREFERENTIAL_EXCLUSION_CANDIDATE — refund path may be reconciliation, not CAPE. Human review decides.

Cliff check (rule CAPE-002 / CAPE-003)

For each classified entry:

Interest computation (CBP 19 CFR 24.3a)

For each CAPE-eligible entry:

  1. Identify the payment date of the entry from the ACE record (date of entry summary submission / duty payment, not liquidation).
  2. Walk daily from payment date to the refund-issue date (defaulted to "today + 60 days" for forecasting).
  3. At each day, apply the CBP quarterly overpayment rate in effect on that date. Rates change on the first day of each calendar quarter.
  4. Compound daily: balance = balance × (1 + rate/365).
  5. Record the per-quarter rate walked and the resulting principal-plus-interest number.

Validation (CBP VAL-F / VAL-E / VAL-I)

Before the CAPE CSV is downloadable, the tool runs every CBP-published validation rule locally:

Any failure blocks generation. The file you download is one that will pass CBP validation.

Audit trail

Every entry's classification, stacking separation, cliff check, and interest computation is logged with:

If CBP questions a number, the chain is reconstructable end-to-end.

What we do not do

Every source cited on this page resolves to a live URL. If you want to dig further, the SCOTUS guide consolidates the primary and secondary citations.