Tariffs Hit China. China Stayed in the Supply Chain

 

U.S. imports moved away from China faster than Chinese production disappeared from the goods America buys

U.S. tariffs reduced direct imports from China, but Chinese components and production remain embedded in supply chains running through countries such as Vietnam and Mexico.

On a working morning in Karachi, Vietnam can appear in front of me as the destination behind a machinery payment. The payment tells me where the money goes, but it cannot identify the origin of every component inside the machine. A motor may have crossed another border before assembly. Tariffs hit China. China stayed in the supply chain.

A reader pushed me toward that problem after reading my argument about China's industrial power. He credited Donald Trump with confronting America's dependence on Chinese production more forcefully than other Western leaders. I think Trump deserves recognition for forcing the industrial cost of dependence into American presidential politics.

Tariffs changed trade. The harder question concerns what they changed underneath it, because a different country on an import document does not necessarily mean a different industrial system produced the goods.

Tariffs Hit China. China Stayed in the Supply Chain

Washington imposed major Section 301 tariffs on Chinese goods beginning in 2018 after investigating Chinese practices involving technology transfer and intellectual property. During 2018 and 2019, the United States raised tariffs on roughly two-thirds of Chinese imports. The Federal Reserve calculates that the average tariff increase reached about 12 percentage points.

Trump changed the argument. American policy had long treated commercial integration with China mainly as an economic opportunity, although disputes over Chinese trade practices were already substantial. His administration placed industrial dependence much closer to the centre of national-security politics.

The change survived his first presidency. The Biden administration reviewed the Section 301 measures instead of dismantling them, then increased tariffs in several strategic sectors. USTR concluded that the existing tariffs had reduced some American exposure to the Chinese practices identified in the original investigation.

Trump returned to office with that architecture largely intact. USTR began the second statutory four-year review of the China Section 301 actions on 6 May 2026. Trade confrontation had moved from a presidential disruption into a continuing feature of American economic policy.

Yet one comparison now makes the outcome difficult to describe as decoupling.

Measure	Change
China's share of direct U.S. imports, 2017 to 2024	Down 7 percentage points
Chinese value added embedded in U.S. imports, 2017 to 2024	Down only 2 percentage points
Mexico's share of U.S. imports by 2025 Q1	About 16%
Chinese production or processing associated with Mexico's U.S. export gains	About 14%

The first two figures come from a 2026 Peterson Institute for International Economics analysis using multiregional input-output data. China's direct share of American imports fell sharply, but Chinese value embedded in the wider American import basket declined far less.

The border moved faster than production.

Seven Points Disappeared. Only Two Really Left

The seven-point versus two-point gap changes how I read the trade war. A bilateral import statistic records the country supplying America directly. Value-added analysis reaches further upstream and asks where the economic content inside imported goods originated.

Consider a product assembled outside China. Its final exporter can sit in another country while Chinese parts remain inside the product. American customs data can therefore show declining direct dependence even while Chinese industrial participation persists further back in production.

PIIE found exactly that divergence. Between 2017 and 2024, China's direct share of U.S. imports fell by seven percentage points. Its share of value added embedded in American imports declined by only two points.

A five-point gap sits between those measurements.

I would not interpret it as proof that tariffs failed. USTR found that the measures contributed to lower imports from China while encouraging alternative sourcing. Economic analyses reviewed by USTR also found positive production effects in the ten American sectors most directly affected by the tariffs.

The evidence instead tells me to distinguish three political claims that often blur together.

Diversification means buying from a wider range of countries. Decoupling means reducing underlying economic dependence on China.

Reshoring asks a different question: did production return to the United States?

A product bought from Mexico can diversify American imports without returning manufacturing to Ohio. If Chinese inputs remain upstream, the same transaction may also achieve less decoupling than its customs origin suggests.

Tariffs can redirect trade quickly.

Industrial geography moves more slowly.

Mexico Shows What the Tariffs Actually Changed

Mexico provides a useful test because it became America's largest import supplier after the 2018 and 2019 tariff increases. By the first quarter of 2025, it accounted for about 16 percent of U.S. imports. Federal Reserve researchers examined how much of that growth came from the tariff shock.

Their answer complicates both sides of the argument.

Trade diversion associated with the China tariffs accounted for about 53 percent of Mexico's export gains to the United States. Yet Chinese production or processing in Mexico accounted for about 14 percent of Mexico's total gains.

Direct transshipment was tiny.

The Fed researchers estimate pure transshipment at below 1 percent. The data fit a more consequential explanation: some Chinese firms responded to tariffs by moving production or processing into Mexico rather than simply relabelling finished Chinese goods.

The distinction matters.

A Chinese product secretly routed through Mexico would mainly represent tariff evasion. Chinese-owned production operating inside Mexico represents something deeper because the industrial network itself has adapted to the tariff boundary.

Nor did China account for most of Mexico's gains. The Federal Reserve attributes another 38 percent to trade diversion from sources outside China, including Mexican producers and foreign multinationals.

Mexico therefore did not become a giant Chinese back door.

The evidence shows something more interesting. Tariffs altered production geography while leaving a measurable Chinese industrial footprint inside the new geography.

From Karachi, Origin Looks Different

I work in a city where financial flows and physical trade sit unusually close together. Karachi's financial district lies near the port infrastructure through which imported machinery enters Pakistan. Geography compresses the distance between payment and production.

Banking does not.

A SWIFT payment message can identify parties to a transaction and carry payment information. It does not provide a bill of materials showing where every component inside an imported machine originated. The financial message answers a different operational question.

Imagine a Pakistani company buying machinery from Vietnam. Its commercial documentation can correctly identify a Vietnamese supplier while an important component still comes from China. No contradiction exists between those facts.

Trade statistics face a related problem.

Customs origin tells policymakers something important about where goods enter the trading relationship. Industrial dependence can sit further upstream, where components require specialized suppliers and manufacturing capacity takes years to reproduce.

Finding another seller is not enough.

A replacement supplier must deliver compatible output at sufficient scale. Price matters as well, because theoretical production capacity offers little relief if substitution makes the final product commercially unviable.

Paulie's examples from his American workshop make more sense through that lens. A plumbing fitting sounds geopolitically trivial until millions are required. An electrical component attracts little attention until replacing its supplier becomes slow or expensive.

Industrial power often hides inside ordinary objects.

Diversification Is Not Decoupling

Trump deserves credit for identifying the vulnerability more forcefully than previous American presidents. His tariffs also produced measurable changes. USTR found lower imports from China alongside increased sourcing from alternative suppliers.

Costs accompanied those gains. USTR's review found that economic studies generally showed small negative effects on aggregate U.S. welfare, with Chinese retaliatory tariffs contributing significantly to the losses. The same review found positive production effects in the ten sectors most directly exposed to the tariffs.

So the record resists a simple verdict.

Tariffs changed incentives, but they could not manufacture an entire replacement ecosystem on command. Factories require capital and workers. Supplier relationships need time.

Washington increasingly treats tariffs as only part of the answer. The Biden administration retained the Section 301 structure while pairing trade restrictions with domestic industrial measures. Its 2024 tariff changes targeted sectors including semiconductors and electric vehicles.

American sourcing shifted at the same time.

PIIE calculates that Taiwan increased its share of U.S. goods imports by 4.1 percentage points between 2017 and 2025. Vietnam gained 3.7 points, while Mexico added 2.3 points.

Diversification can reduce risk.

But it should not automatically be called reshoring, and the value-added evidence warns against assuming that every non-Chinese exporter represents independence from Chinese production. Supply chains can cross borders without severing their upstream relationships.

China also pays a price when production leaves. American demand remains commercially valuable, while manufacturers outside China can capture business that Chinese factories once supplied directly. The Federal Reserve's Mexico findings show that most Mexican export gains cannot be attributed to Chinese production relocation.

Mutual dependence does not settle the argument.

The harder question is which dependence can be replaced faster.

The Real Measure of Power Is Replacement Time

Industrial dependence should not be measured only by import share. I would also measure how quickly another supplier can replace lost production at sufficient scale. That turns an abstract geopolitical argument into an operational one.

Replacement time changes the meaning of coercion.

Financial power can restrict access to payment networks or capital. Military power can threaten physical assets. Industrial concentration creates another constraint when losing a supplier produces shortages that cannot quickly be repaired.

None makes the others irrelevant.

America retains immense financial capacity and major technological advantages. China remains exposed to foreign demand and to technologies it cannot always replace easily. A serious confrontation would test vulnerabilities on both sides.

Yet tariffs have already provided an experiment.

Direct Chinese imports fell. Supply chains diversified. Chinese value added inside America's wider import basket proved harder to remove.

I can look at a payment in Karachi and identify where the money is going. An import document can tell me where a machine arrived from, but neither necessarily reveals how quickly the production behind it could be replaced.

The invoice can tell me where the container came from. It still cannot tell me how quickly America could replace what China made inside it.

---

• AI Transparency Statement: "This analysis was drafted under editorial direction with AI technical assistance, then verified and edited by Munaeem Jamal."

---