Trump actually started to decouple America from China
The "economic divorce" between the two countries is proceeding slowly, but it is proceeding.
Donald Trump is headed to China with a whole bunch of top U.S. CEOs in tow to talk about trade. There is probably a post to be written here about how Trump is creating a new kind of “America, Inc.” centered around his own person, using a combination of tariffs, export controls, federal government equity stakes, and personal bullying. But this is not that post. Instead, this is a post about decoupling. Trump was elected in 2016, and again in 2024, on promises to reduce American economic dependence on China. How well has he succeeded?
First, some background. In the mid-2010s, when Trump came to power, the U.S. and China had a pretty well-understood economic relationship. America did R&D and designed products, then shipped the designs to China where they were manufactured — often using components from Japan/Korea/Taiwan, but sometimes using Chinese components. China would then ship the products back to America, where they were marketed and sold and serviced by the American companies.
Both countries chafed at this arrangement. Americans complained that the relocation of labor-intensive assembly to China put American factory workers out of a job (which was true) and worried that outsourcing assembly would eventually lead to the outsourcing of more valuable activities (which was probably true), while Chinese leaders were annoyed at being stuck in the low-value-added middle of the production chain. So both countries implemented policies to break up this arrangement and create a new trading system.
China used industrial policy to onshore high-value component manufacturing and create its own “national champion” brands, while U.S. Presidents Trump and Biden strove to reduce U.S. trade dependence on China.1 I wrote about this breakup in 2022:
Everyone agrees that China has succeeded in its half of the decoupling — far more Chinese-made goods are now made with Chinese components. The country has climbed up the value chain, and developed top brands like BYD, Huawei, Xiaomi, DJI, CATL, and so on.
Whether the U.S. has succeeded in reducing its dependence on Chinese manufacturing, however, has been a subject of hot debate. On one hand, the percentage of America’s imports that it gets from China has plummeted:
That’s from a WSJ story in February of this year, entitled “The American and Chinese Economies Are Hurtling Toward a Messy Divorce”. A few more details:
Some businesses have moved production from China to the U.S. to avoid tariffs, but the flow is still modest. Mexico and Southeast Asian nations are more common destinations for manufacturers leaving China…About 9% of Ohio manufacturers in a recent survey said they had reshored some production to the U.S. in 2025, up from 4% in 2021. About 60% of the reshoring in 2025 relocated from China[.]
It’s clear that tariffs have had an effect on the shifting of U.S. imports away from China. Even Trump’s far weaker tariffs in his first term showed results — America started buying tariffed goods from countries other than China, even as it kept buying non-tariffed goods from China:
Despite all the hullabaloo of “Liberation Day”, Trump’s tariffs on China — which built on previous tariffs on China by the Biden and first Trump administrations — dwarfed his tariffs on friendly countries:
Where have the imports shifted to? Mostly to other Asian countries, and to Mexico:
Which kind of products is America no longer importing from China? Trump’s first-term tariffs mostly hit low-value products like furniture, shoes, and clothing (where China’s share was slowly declining anyway as its labor costs rose). But more recent tariffs have hit China’s sales of electronics — PCs, phones, etc. Two years ago, most of America’s PCs were made in China; now, most of them are made in Vietnam.
It’s not just trade, either; on the investment side, too, decoupling has been very apparent. There were a whole bunch of stories in 2025 about U.S. businesses wanting to relocate their production out of China. These anecdotes represented a trend that was highly visible in the data — the collapse of foreign direct investment into the Chinese economy:
Much of this investment was shifting to Southeast Asia, though in advanced manufacturing it shifted to Europe.
Why has investment shifted? Tariffs are one reason. Traditionally, a lot of what the U.S. imported from China was made by American companies — for example, Apple manufacturing iPhones in Shenzhen and shipping them back to the U.S. Tariffs make this a more expensive thing to do, so they provide an incentive for American companies — and any multinational companies that sell stuff to the U.S. — to stop investing in Chinese factories.
A second reason is what I call the “China Cycle”. Multinationals have learned the painful lesson that when they put their factories in China, their technology will be appropriated by Chinese indigenous companies — often with the help of the Chinese government — and then later used to outcompete them in global markets. Again and again, companies fell for the lure of the huge Chinese domestic market, only to lose their technological crown jewels to fierce Chinese competitors who rarely played fair. This has naturally chilled the desire to invest in China.
A third reason, of course, was the threat of war. As China grew more bellicose over Taiwan and the South China Sea, multinationals began to realize that having their factories in China, where they would be either blockaded or expropriated in the event of a conflict, posed a big risk.
So it’s possible to tell a pretty coherent story here. U.S. companies had plenty of reasons to move out of China, but tariffs gave them a big extra push. And with the exodus of those companies, China’s exports to America plunged.
But in fact, there are lots of people who don’t believe the decoupling is real. One group — call it the “macro camp” — has argued that because U.S. trade deficits and Chinese trade surpluses are still about the same size (or larger), there must be some sort of hidden conduit by which Chinese products are still reaching American shores, possibly by a circuitous route.
The macro camp included some strange ideological bedfellows — people like Brad Setser and Robin Brooks who were frustrated with tariffs’ inability to curb global imbalances and wanted to see sterner protectionist measures taken, and free-traders like The Economist and the Peterson Institute who seemed to think that if Trump & co. can be convinced that tariffs are futile, the free-trade consensus will reappear. I had some ferocious battles2 with some of these folks back in 2023:
My key argument was that you can’t just look at macro imbalances — China’s trade surplus with the whole world, and America’s trade deficit with the whole world — and conclude that Chinese goods must be making their way into America. It just doesn’t follow. China could be finding alternative markets for its exports, while America found alternative sources for its imports, and these could roughly be the same countries. The macro imbalances would persist, but China and America would have decoupled.
That said, it’s also possible that the macro camp was right — China might be finding some way to get around tariffs. And sure, multinational companies are divesting from China, but that doesn’t mean China’s exports to America have to fall; China’s indigenous companies, like BYD and Huawei, are perfectly capable of selling their own products to America.
So before we conclude that decoupling is definitely real, we need to actually check the data in greater detail.
How might Chinese goods be sneaking into America? Decoupling skeptics often posited transshipment — basically, the idea that Chinese companies responded to tariffs by slapping a “Made in Vietnam” label on their products and sending them through Vietnamese ports on their way to American shores. But while a little of this probably did happen, Gerard DiPippo estimates that transshipment is minor — at most 18% of China’s lost exports to America, and probably a lot less.
He got this estimate by looking at specific products — examining what China stopped selling to the U.S., and what it started selling to Vietnam, in the wake of tariffs. If products are being transshipped through Vietnam, the two numbers should line up. But they usually don’t — the things China has started selling to Vietnam since Trump’s tariffs went into effect are, by and large, not the same products Vietnam has been selling more of to America. Transshipment can’t be the big story here.
A more convincing argument is mismeasurement. There is a gap between how much the U.S. says it imports from China, and how much China says it exports to the U.S. As of 2024, the latter had fallen by much less than the former:
The biggest reason for this was probably the “de minimis” exemption, which let China ship small packages to America without paying tariffs. Chinese manufacturers took advantage of this rule by breaking down their shipments into a bunch of small packages:
But Trump closed the de minimis loophole by executive order in the summer of 2025. So that loophole can’t explain the continued collapse in China’s exports to the U.S. over the last year.
There is one far more believable way that Chinese-made products might still be flooding into America: intermediate goods. Just as a “Made in China” iPhone was mostly made out of Japanese and Korean and Taiwanese parts back in 2011, a “Made in Vietnam” iPhone today will contain a lot of Chinese parts. Since complicated components represent a lot more of the actual value of an electronics product than the actual final assembly, this means that it’s still mostly China selling stuff to America. Hsu, Peng, and Wu estimated in 2024 that this effect was substantial:
Utilizing transaction-level customs import-export data, we develop a novel measure to assess firm-product-level indirect dependence of U.S. importers on China via their suppliers in Vietnam and Mexico. Our findings indicate a substantial increase in indirect dependence on China post-Trade War…suggesting that despite efforts to reduce dependence on China, U.S. supply chains remain indirectly dependent on China via third-party nations.
Annoyingly, however, this data is only through 2022. In fact, we also have another data source on indirect trade — the OECD’s value-added trade numbers. But that’s also released very slowly; the most recent data set also only goes through 2022.
Looking at that data is still interesting, though. In fact, before the pandemic, America’s share of imports from China was falling on a value-added basis. The pandemic bumped it back up, but then it started to fall again in 2022:
The pandemic throws a wrench into the trend, making it hard to see if there’s been a recent drop that mirrors the recent drop in gross import flows. It’ll take some time to get that data. But in the meantime, it looks like Trump’s first-term tariffs really did reduce America’s import dependence on China a bit — and that decoupling might have resumed in 2022.3
Intermediate goods trade changes the basic story about decoupling. Tariffs and other factors broke the old arrangement between the U.S. and China, where American companies outsourced production to China and sold the products back to American customers. That old world is gone. In its place is a new relationship, in which Chinese companies sell parts and components to assemblers in other countries, who then sell the goods to America.
This is not a trivial change. On one hand, it shows how Chinese companies have moved up the value chain, becoming direct competitors to multinationals. On the other hand, final assembly of goods isn’t trivial or meaningless. It’s the least profitable part of the value chain, but it’s still important — after all, China industrialized in the 1990s and 2000s while doing mostly that sort of work.
So the fact that American tariffs are causing that assembly work to move out of China is significant. It doesn’t remove U.S. dependence on Chinese manufacturing, but it reduces it. China itself started out doing assembly but later moved into component manufacturing; there are some signs Vietnam may be starting to do the same. And if Vietnam can do it, so can India, Mexico, Indonesia, and so on. China doesn’t have some magic secret sauce that makes it the only country that can make physical objects; other countries can learn, just like China did.
A non-Chinese supply chain won’t be built quickly or easily, and it hasn’t been happening as fast as the headline numbers suggest. But we’ve made a promising start, and the tariffs on China were part of that. A lot of Trump’s protectionist policy has been haphazard, misdirected, stupid, and downright corrupt, but this one — which was continued by Biden and the Democrats — was actually starting to yield some results. It would be a shame if Trump throws that all away on this trip in exchange for the promise of a few soybean purchases or whatever.
The U.S. also started using export controls to limit China’s development in key strategic industries like semiconductors, and China eventually followed suit with its own export controls on rare earths.
OK, fine. I wrote some blog posts criticizing them, which they pretty much completely ignored. But in my mind, the battles were ferocious indeed.
One additional note of caution here: Even when the components are also made in Vietnam or Mexico, they may be made by Chinese-owned factories, meaning that some portion of what America pays to its Vietnamese and Mexican suppliers flows through to Chinese shareholders. Those profit flows won’t show up in any trade numbers at all.
My wife worked for a global electronics company in Trump 1. They were early decouplers. When the first trade war happened they discussed it with their factory counterparts in China and they shifted production to Vietnam. Same factory owners, same parts suppliers, different country of origin.
A couple of years later the company itself was sold to a group of Chinese investors, who didn't change this arrangement.
Soybeans? I wonder about Nvidia chips. Until proven otherwise, I presume that the most corrupt and damaging thing is what he wants to do. Tell me I'm wrong, I'd welcome that.