Article — "The Decoupling That Wasn't: Chinese Value-Added Barely Falls as Exports Surge 24% Elsewhere": Here is the contradiction hiding inside two numbers from the same Peterson Institute brief. One says the trade war worked. The other says it barely happened.
China's share of bilateral US imports fell 7 percentage points from 2017 to 2024. China's share of value-added in those imports — the actual Chinese content, wherever it entered from — fell just 2.
That 5-point spread is the number that matters. It is the amount of decoupling that was pure paperwork.
This week the ADB multiregional input-output tables let us put a figure on the cosmetics, and add the part nobody advertises: the bill went to allies.
Start with the honest metric. Bilateral import share measures where a box was stamped before it crossed the border. Value-added traces where the resource base actually sits — the refined minerals, the precursors, the components embedded inside a "Vietnamese" or "Taiwanese" good.
When the first number falls three times faster than the second, you are not watching supply chains leave China. You are watching them acquire a second passport.
When the first number falls three times faster than the second, you are not watching supply chains leave China. You are watching them acquire a second passport.
The redirected flow had to land somewhere, and PIIE names the runways. Between 2017 and 2025, the top gainers in aggregate US goods import share were Taiwan at 4.1 points, Vietnam at 3.7, and Mexico at 2.3.
Those are the countries that learned to stamp the box. The Chinese content rode along inside their exports — which is exactly why the value-added line refused to drop.
Furniture shows the mechanics in close-up. Section 301 slapped up to 25% on Chinese wood furniture, with antidumping rates on wooden bedroom furniture that have at times exceeded 200%.
So sourcing moved to Vietnam, whose share of US wood furniture imports reportedly climbed past 30%, per a Vietnamese furniture supplier's marketing materials. Vietnamese suppliers advertise the arithmetic openly: landed costs reportedly 20–22% lower than China, "due to tariff exemption," per NAFOCO, a single supplier.
Read that phrase carefully. The savings are not from cheaper labor or better wood. They are from the tariff exemption itself — the same product, the same buyer, a different label, minus the 25%.
That is the routing algorithm in one line. The tariff did not change what gets made or where the inputs come from. It changed where the paperwork says the box was born, and charged the US importer a premium for guessing wrong.
Now the twist that undoes the whole project.
The rerouting only worked while the tariff gap was wide. In the second term, Trump extended high tariffs to Southeast Asia — and the gap the entire strategy depended on collapsed.
Consider Alliance Consumer Group, a US consumer goods firm that had moved sourcing out of China. It is now shifting a large share of orders back. Flashlights reportedly cost 15% less to make in China than in Thailand, and the tariff difference has shrunk to a single point: roughly 20% on China, about 19% on Thailand, Vietnam, and Cambodia, per Aju Press.
One percentage point of tariff no longer offsets a 15% production cost gap plus China's near-complete component supply chain. So the factories reverse.
The evidence is not anecdotal. QIMA, the supply-chain inspection firm, reportedly saw US-client demand for China inspections fall 18% in 2025 while Vietnam surged 61% and Cambodia 26%.
Then it flipped. QIMA's China inspection share, 30% last year, climbed back to 35% in Q2 2026 — a six-quarter high, with Southeast Asia's share falling.
SHEIN tells the same story physically. It reportedly built a 15-hectare logistics center in Vietnam and pushed Chinese clothing makers to follow; that center has since been cut to 6 hectares with local staffing slashed, while SHEIN commits over 10 billion yuan to a smart supply chain in Guangdong.
The rerouting was never even durable. It was an arbitrage on a tariff gap, and once the gap closed, the arbitrage closed with it — routing traffic straight back to Guangdong.
The lever that closed the gap deserves one line, because it is the whole reason the tariff differential shrank. China imposed rare earth export controls in response to US tariffs, disrupted the US supply chain, and both sides then negotiated tariffs down. Without that leverage, per the reporting, Beijing could not have bargained on equal footing. The tariff gap did not narrow because the US relented on principle; it narrowed because China held a chokepoint.
So who actually paid?
Not Beijing. PIIE's terms-of-trade data through Q2 2026 show little improvement after the second-term trade war began — and what improvement exists is more plausibly tied to wartime disruptions than to tariffs, which have swung wildly. Far larger terms-of-trade moves came after COVID. The US did not extract a better deal from China; the price of what it buys barely moved.
The people who paid were importers and allies.
Take Korea. The US imposed a new 15% Section 232 tariff on polysilicon derivatives — ingots, wafers, cells — plus a minimum import price system. Polysilicon is a core input for both semiconductors and solar panels.
Seoul convened a public-private meeting on August 11 at the Korea Chamber of Commerce, pulling in Samsung Electronics, Hanwha Qcells, OCI, and SK Siltron to assess the hit to exports and to Korean firms' US operations. That is an ally scrambling to contain damage from a tool aimed at China.
Aluminum shows the same collateral pattern. Between August 2025 and May 2026, US steel and aluminum imports fell 8.5%. Chinese imports reportedly dropped 34% — about $7.9 billion — under an effective 48% tariff, per the German Economic Institute (IW).
But the flow did not come home. It reshuffled. Thailand's US aluminum imports rose 31%, lifting its share 1.3 points to 4.4%, with Vietnam, Japan, and India also gaining. Germany, reportedly at a 21% tariff versus China's 48%, still slipped 0.2 points to 7.4%.
That is the signature of a routing algorithm, not a decoupling tool. The Chinese metal loses direct share; the dependence relocates one border over; and an ally like Germany loses ground anyway.
Step back and the pattern holds across furniture, flashlights, polysilicon, and aluminum. Tariffs reliably move where a box is stamped. They do not move where the resource base sits.
Consumer assembly is portable, so it reshored — or rather re-labeled. The chokepoint inputs — refined minerals, precursors, the components China makes to near-perfection — did not budge, which is why even the 2-point value-added drop overstates the real decoupling.
Meanwhile China reportedly posted 24% year-on-year export growth in July and a $113 billion surplus, on track for another trillion-dollar-plus surplus, per Reuters. An economy genuinely being decoupled from its largest market does not export like that.
The bilateral import share was always the wrong number to watch. It measures labels. Value-added measures dependence, and dependence fell two points, then started climbing back as the factories returned.
The lesson for anyone building a China strategy on tariff differentials: you are renting a routing advantage, not buying independence. The advantage lasts exactly as long as the gap. And the US just spent seven years, a set of alienated allies, and a polysilicon tariff on Samsung and Hanwha to learn that Beijing still sits where it always did — inside the box, under someone else's label.