Tariff Arithmetic

World merchandise trade is now expected to fall in 2025, roughly three percentage points below where it would have been without recent policy shifts. This report separates the direct price effect from the redirection and uncertainty effects, and argues the second pair is doing most of the work.

Decomposing the revision

World merchandise trade volume is now expected to fall 0.2 percent in 2025 before a modest 2.5 percent recovery in 2026. The number that matters is not the level but the revision: the 2025 estimate is nearly three percentage points below where it would have been without recent policy shifts, and it marks a reversal from the start of the year, when continued expansion was expected on improving macroeconomic conditions.

A three point swing in a global aggregate inside a few months is unusual. It is worth decomposing rather than reporting.

The direct price effect

The direct effect of a tariff is straightforward: it raises the landed price of an affected good, reduces the quantity demanded, and the elasticity determines the rest. If that were all that were happening, the revision would be smaller than it is, because the tariffs in question cover a large but not universal share of one country's imports.

Two things amplify it. The first is that trade policy uncertainty spills well beyond the directly linked trade relationships. Firms making irreversible commitments in supply chain assets discount for the variance of policy, not just its expected level, and that discount applies to routes that are not themselves affected. The second is that the announced measures are not the end state. If reciprocal tariffs were enacted, world merchandise trade growth would fall by a further 0.6 percentage points on the WTO's estimate.

A tariff that might be imposed is not free. It is priced by every firm deciding whether to sink capital into a route that might become uneconomic.

The redirection effect

Trade does not simply stop. It moves. Chinese merchandise exports are projected to rise by 4 to 9 percent across all regions outside North America as trade redirects. At the same time US imports from China are expected to fall sharply in textiles, apparel and electrical equipment.

That combination creates gaps in the US market that other suppliers can fill, and the WTO explicitly notes this could open the door for some least developed countries to increase their exports. Redirection is not a neutral reshuffle: it produces winners with spare capacity and correctly positioned product mix, and losers holding fixed assets pointed at the old pattern.

For anyone with exposure, the operational consequence is that the aggregate volume forecast is close to useless. What matters is bilateral flows at product level, and the aggregate averages over exactly the variation that determines outcomes.

Evidence from services trade

Services trade is not directly subject to these tariffs and has been revised down anyway, to 4.0 percent in 2025 and 4.1 percent in 2026 against baseline projections of 5.1 and 4.8 percent. That is a downgrade of over a percentage point in each year for a category nobody taxed.

Three channels explain it. Lower goods volumes reduce derived demand for transport and logistics. Uncertainty dampens discretionary spending on travel. And slower investment reduces investment-related services. The services revision is therefore a clean read on the indirect and confidence channels, stripped of the direct price effect, and it is large.

We regard this as the most under-discussed number in the release. It implies that a substantial share of the total trade effect operates through uncertainty and derived demand rather than through relative prices, which in turn implies that removing the tariffs would not restore the baseline quickly.

Indicators to monitor

• Frontloading. If importers bring purchases forward ahead of effective dates, 2025 volumes will beat this forecast and 2026 will undershoot it. That is a timing shift, not a demand improvement, and we would expect the aggregate forecast to be revised in both directions before it settles.
• Exemptions and suspensions. The realised effect depends heavily on scope carve-outs and on whether announced measures take effect on schedule. Announced rates are a poor guide to collected duty.
• Retaliation. The current projection assumes a limited retaliatory response. That assumption is doing a lot of work.
• Investment, not trade. The uncertainty channel appears in capital expenditure and foreign direct investment data before it appears in customs data, and it persists after policy settles because a shortened value chain reduces the trade intensity of any given level of output.

Our working expectation is that the 2025 outturn comes in better than this forecast and the 2026 outturn worse, for reasons of timing rather than because the analysis is wrong. We will mark this projection against outturn when the data arrives.

Source register