Global Trade Did Not Beat The Tariffs, It Has Not Met Them Yet

The consensus reading of this summer is that the trade war turned out to be survivable. Global goods trade hit roughly 13.7 trillion US dollars in the first half of 2026, up 12.5 percent on the same period of 2025, according to UN Trade and Development. US headline inflation cooled to 3.4 percent in the year to July, with core at 2.5 percent. The S&P 500 closed above 7,800 for the first time on 13 August. The story writes itself: the tariffs came, commerce adapted, nothing broke.

That reading is wrong, and it is wrong in a specific and checkable way. The first half of 2026 did not test the tariff regime, because the two largest tariff actions of the year had not yet taken effect when the measurement window closed. Global trade has not beaten the tariffs. It has not met them yet.

The window everyone is quoting closed before the tariffs landed

On 23 July the US Trade Representative finalized Section 301 forced labor tariffs on 60 economies, at 10 percent for those with import prohibitions in place and 12.5 percent for those without. They took effect at 12:01am Eastern on 24 July. Fifty four of the sixty had neither imposed nor effectively enforced a prohibition; the remaining six, including Canada, Mexico and the European Union, had rules they were judged not to be enforcing.

Three days earlier, on 20 July, the President signed three proclamations invoking Section 338 of the Tariff Act of 1930 for the first time, adding a 50 percent duty on selected Canadian dairy, alcoholic beverages and motor vehicles, with coverage extending to wine, cement and other goods. USTR put the affected trade at nearly 20 billion dollars a year. There is no USMCA carve out. That duty takes effect at 12:01am Eastern on 19 August, four days from now.

So the celebrated first half figure covers January to June. The forced labor duties bit for none of it. The Section 338 duties bite for none of it. Citing H1 resilience as evidence that these measures are absorbable is like citing a dry January as evidence the roof does not leak.

Roughly a third of the growth was price, not goods

The composition of that 12.5 percent matters more than its size. UNCTAD is explicit that a significant share of the increase reflects higher prices rather than larger volumes. Prices of traded goods rose about 3.6 percent in the first quarter and an estimated 5 percent in the second.

Average those and you get something near 4.3 percent across the half. Strip it out of the 12.5 percent headline and real volume growth is on the order of 8 percent. That is still a good number. But it means roughly a third of what is being reported as a trade boom is simply the same goods costing more.

The direction of travel is the part to watch. Traded goods prices did not merely rise, they accelerated, from 3.6 percent to about 5 percent quarter on quarter. Energy did much of that work. Brent settled around 87.72 dollars a barrel during the week of 10 August and traded above 88 dollars on Friday, with the Strait of Hormuz still constrained: US officials put flows through the waterway at up to 9 million barrels a day, below normal levels, after the 17 June memorandum of understanding with Iran collapsed.

This is not a logistics breakdown, which is why it matters

It would be comfortable to file rising traded goods prices under shipping chaos, on the assumption that chaos passes. The freight data does not support that.

Sea-Intelligence put global container schedule reliability at 62.4 percent in April, 64.7 percent in May, the highest reading of the year, and 62.6 percent in June. April was up 4.0 percentage points year on year. Average delays for late vessel arrivals ran at 5.34, 5.52 and 5.31 days across those three months. That is a network under strain but functioning, and functioning better than a year ago.

Which removes the easy explanation. Goods crossing borders got about 5 percent more expensive in the second quarter while the ships ran roughly on time and volumes kept growing. That is not congestion. That is the cost of the goods themselves going up, through energy and through duty.

Calm consumer prices are not exoneration, they are a location problem

Here is the tension nobody is resolving. Traded goods prices accelerated to around 5 percent in the second quarter. US core inflation went the other way, easing to 2.5 percent in the year to July from 2.6 percent in June, with core CPI up 0.2 percent on the month. Both cannot be describing the same pass through.

A cost wedge that opens at the border and does not appear at the till has to be sitting somewhere in between. It is sitting in gross margin. Research cited around the Fed system finds evidence of tariff pass through to importers with only limited pass through to consumer prices, which is precisely the shape of a cost being absorbed by the firms doing the importing rather than avoided.

That absorption is not permanent. It is a function of hedges, inventory bought at pre tariff cost, and supply contracts written before the duty existed. All three roll off. The consumer price data is not telling you the tariff was free. It is telling you the invoice has not been forwarded yet.

Why the Fed disagreeing with itself supports this reading

The Fed's own researchers are split. The Federal Reserve Board estimated that tariffs implemented through November 2025 raised core goods PCE prices by 3.1 percent through February 2026, enough to account for the excess inflation in that category. Minneapolis Fed economists Neil Mehrotra and Michael Waugh have argued the opposite, showing that the goods categories with the highest tariff exposure are not reliably the ones with the largest realized price increases.

That disagreement is usually presented as evidence the tariff effect is uncertain. Read it again. If importers absorb duty into margin at wildly different rates depending on their contract cycles, hedging and pricing power, then tariff exposure would fail to predict consumer prices category by category, which is exactly what the Minneapolis work found. Their result is not a refutation of tariff pass through. It is a fingerprint of pass through being intercepted before it reaches the shelf.

The AI boom is flattering the trade data

The volume growth that did occur is not evenly spread, and its concentration is the last reason to distrust the resilience story. UNCTAD reports trade up 38 percent in critical minerals, 25 percent in semiconductors, 15 percent in batteries, 14 percent in ICT products and 11 percent in electric vehicles.

These are data center and electrification inputs. They are bought against build deadlines by buyers with capital budgets already committed, which makes them about as price insensitive as traded goods get. A 10 percent duty does not stop a semiconductor shipment when the fab schedule is the binding constraint.

So the aggregate is being held up by the one category that would keep moving at almost any price. The AI capital expenditure boom and the trade resilience number are routinely cited as two separate pieces of good news. They are substantially the same piece of news counted twice, and it says very little about how tariff exposed consumer goods will behave.

What would change my mind

This argument is falsifiable, so let me say how.

• If third quarter traded goods prices decelerate back toward 3 percent or below while volumes hold, the cost was absorbed upstream by exporters cutting prices rather than by importers cutting margin, and my mechanism is wrong.
• If fourth quarter reporting from import heavy sectors shows gross margins flat or expanding, the wedge is not where I say it is.
• If Hormuz reopens and traded goods price growth falls with it, then this was an energy shock wearing a tariff costume and the duties are a minor term.
• If the Section 338 and Section 301 duties are materially rolled back in negotiation before they accumulate, the test simply does not happen.

What I do not accept is the argument being made now: that a data set ending in June, in which a third of the growth was price and the volume growth was concentrated in deadline driven technology inputs, demonstrates that tariffs taking effect in late July and mid August are survivable.

The market is priced for that demonstration. The FTSE 100 set an intraday record of 10,989 on 31 July. The S&P 500 finished the week at 7,785.76 after its first close above 7,800, a third consecutive weekly gain, on cooler inflation prints. Those prices embed a conclusion the evidence has not yet earned. The first genuinely informative data will be the third quarter trade figures and the goods components of autumn inflation reports, and 19 August is when the clock actually starts.