Why Firms Pay Duties They Could Legally Avoid

The theory of preference utilisation under trade agreements is well covered elsewhere. This is about the version you meet in practice.

The gap between eligible and claimed

Trade agreements reduce or remove duties on qualifying goods, yet a meaningful share of eligible trade still enters paying the standard rate. Utilisation rates vary widely by agreement, by product and by firm size, and low utilisation is common in newer agreements.

The reason is rarely that firms do not know the agreement exists, since duty is a visible line in any landed cost model. It is that claiming requires proving origin, and proving origin requires information the claimant may not control. The decision is a straightforward comparison of saving against cost, made product by product rather than agreement by agreement.

What proving origin actually requires

The claimant must show that the goods satisfy the specific rule for their tariff line under that specific agreement. For a value content rule, that means costing the bill of materials accurately, including the origin status of every input.

Upstream of that, for a tariff shift rule, it means knowing the classification of each non-originating input before it entered the process. Either way the data sits with suppliers, and suppliers who are not party to the claim have limited incentive to provide it. Collecting supplier declarations across a multi-tier chain is the real work, and it recurs whenever the specification changes.

When the maths says do not bother

A low duty rate on a low-value shipment can produce a saving smaller than the internal cost of assembling the evidence. Products with volatile bills of materials require re-verification often enough that the compliance cost never becomes a one-off.

On the manifest, firms shipping many small consignments to many markets face this calculation repeatedly and rationally decline much of the time. Where duty rates are already near zero on a most-favoured-nation basis, the preference margin may not justify any effort at all. That last case is why utilisation is often lowest exactly where agreements are politically most celebrated.

The risk that outlives the shipment

Customs authorities can verify an origin claim years after clearance, by which time the goods are sold and the margin recognised. A failed verification usually means repaying the duty saved across every affected entry, with interest and possibly penalties. Because the exposure is retrospective and cumulative, the downside of a wrong claim is not symmetric with the upside of a right one.

Line by line in the tariff schedule, conservative firms therefore claim only where documentation is complete, even when they believe the goods would qualify.

That asymmetry, more than any single rule, explains cautious behaviour among firms with strong compliance functions.

Making claims cheaper

Self-certification regimes, where the exporter or importer declares origin without an issued certificate, reduce transaction cost considerably. Approved exporter and registered exporter schemes let firms certify their own shipments after an initial approval process. Standing supplier declarations covering a period, rather than a shipment, cut the recurring administrative load substantially.

On the manifest, software that maintains bills of materials with origin status attached turns an annual scramble into a query. The investment makes sense above a volume threshold, which is precisely why utilisation correlates with firm size.

Trade data lags by months and is revised afterwards, so recent figures are provisional.

Reading utilisation as a signal

Low utilisation of an agreement is evidence about its rules of origin and administration, not necessarily about its economic value. Where rules are restrictive relative to how the industry actually sources, firms cannot qualify however much they want to.

Where rules are workable and utilisation is still low, the obstacle is usually procedural and can be addressed without renegotiation. Distinguishing the two requires looking at product-level data rather than headline agreement coverage. That distinction is the difference between a rules problem and an implementation problem, and they have different fixes.

The takeaway

A preference is worth exactly the duty saved minus the cost of proving it. Run that sum per product line, not per agreement.

Capacity takes a decade to build and one quarter to look like a mistake.