AIThis post was created with the assistance of artificial intelligence (AI).

By Thorsten Meyer | November 17, 2026 | ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

Work After Work · Episode 15 · Arc 2: The basics everyone needs

Turn over your phone, your trainers, your kitchen chair. Made in China? Look again. More and more often, the label says Vietnam.

In the first nine months of 2026, the United States imported goods worth $205 billion from Vietnam and $215 billion from China, according to the US Census Bureau. In 2017, America still bought more than ten times as much from China as from Vietnam.

Last time, we looked at unemployment and its three kinds. Machines are not the only thing blamed for lost jobs. The other suspect is trade. This episode closes our second arc with the oldest argument in economics for trade, the hardest evidence against its easy version, and what AI is already doing to it.

Key takeaways

  • Trade pays even when one side is better at everything. David Ricardo showed it in 1817 with cloth and wine: each country should make what costs it the least, not what it is best at. That is comparative advantage.
  • The gains are spread thin, the losses are concentrated. In the United States, the “China shock” hit towns that made furniture, toys and clothing. It peaked around 2010, and the damage was still visible in 2019.
  • The number to remember is 86%. In more exposed US regions, 86% of the factory jobs lost to Chinese imports were not replaced by other local jobs. Most people did not move away; they left work.
  • Germany came out ahead, so far. Trade with Eastern Europe and China added an estimated 442,000 jobs from 1988 to 2008. In 2025, though, imports from China rose 8.8% while exports to China fell 9.7%.
  • AI is already moving trade. In the first half of 2026, goods that AI runs on were 14.8% of world goods trade but delivered 47% of its growth, according to the WTO.
Watch the video version of this episode (about 7 minutes). Narration: AI voice clone of Thorsten Meyer.

Products Worth Considering

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

Bose QuietComfort Wireless Noise-Cancelling Headphones, Sandstone

Bose QuietComfort Wireless Noise-Cancelling Headphones, Sandstone

As an affiliate, we earn on qualifying purchases.

1817: Ricardo’s puzzle

In episode 12, we met David Ricardo asking who gets the pie. In the same book, On the Principles of Political Economy, and Taxation (London, 1817), he set a puzzle about trade in the chapter “On Foreign Trade”.

His example uses two goods and two countries. The numbers are workers needed for one year:

  • England: 100 workers for a certain amount of cloth, 120 for a certain amount of wine.
  • Portugal: 90 workers for the cloth, only 80 for the wine.

Portugal is better at both. So why would Portugal ever buy English cloth?

Ricardo’s answer: because Portugal gives up less when it makes wine. The wine of 80 Portuguese workers buys English cloth that would have cost 90 workers to make at home. England, in turn, gets wine worth 120 English workers for cloth made by 100. “Thus, England would give the produce of the labour of 100 men for the produce of the labour of 80,” Ricardo wrote. Both countries end up with more.

Economists call this comparative advantage: not what you are best at, but what costs you the least to do, measured by what you give up. It is the same logic as a lawyer who types faster than her assistant but still hires one: her hour is worth more in court. (That example is ours, not Ricardo’s.)

Terraced vineyards on a steep slope above a calm river valley in soft morning light
Ricardo's example: Portugal makes the wine, England the cloth, and both end up with more. AI-generated illustration.

Ricardo took sides in politics too. Britain’s Corn Laws of 1815 put tariffs on imported grain to protect landowners; Ricardo argued against them, as David Autor, David Dorn and Gordon Hanson note in their 2021 paper. Parliament repealed them in 1846, under Prime Minister Robert Peel, according to the UK National Archives.

1930: the world tries the opposite

A century later, the world tried the opposite. In 1930, the United States passed the Tariff Act known as Smoot-Hawley. The US State Department’s historians are careful: scholars disagree about how much protection it actually added and whether it set off the wave of retaliation that deepened the Great Depression. What is certain, they write, is that it “did nothing to foster cooperation among nations” and became a symbol of the “beggar-thy-neighbor” policies of the 1930s.

The numbers from that time are brutal. US imports from Europe fell from $1,334 million in 1929 to $390 million in 1932. Overall, world trade declined by some 66% between 1929 and 1934.

After the war, governments built the opposite system. On 30 October 1947, 23 countries signed the General Agreement on Tariffs and Trade (GATT), according to the World Trade Organization. For 47 years it remained a provisional agreement, while rounds of talks cut tariffs step by step. On 1 January 1995, the GATT was replaced by the World Trade Organization (WTO).

Trade grows, then stalls

And trade grew. The World Bank measures exports plus imports of goods and services as a share of world GDP:

  • 1970: 26%
  • 1990: 38%
  • 2008: 60%, the high point
  • 2024: 57%

Since the financial crisis of 2008, the share has stopped rising: trade still grows, but no longer faster than the world economy.

Thorsten Meyer AI
Work After Work · Ep. 15

Infographic 01 · Who America buys from

From China to Vietnam

China's share of US goods imports peaked in 2017 and has more than halved since. Vietnam is close behind in 2026.

$215 bnUS goods imports from China, Jan-Sep 2026
$205 bnfrom Vietnam, Jan-Sep 2026: almost as much
21.6%China's share of US goods imports in 2017
7.7%China's share, Jan-Sep 2026

China · share of all US goods imports

  • 19913.9%
  • 2001joins the WTO in December9.0%
  • 200716.4%
  • 2017the peak share21.6%
  • 20259.0%
  • Jan-Sep 20267.7%

Vietnam · share of all US goods imports

  • 20010.1%
  • 2007joins the WTO in January0.5%
  • 20172.0%
  • 20255.7%
  • Jan-Sep 20267.4%

Shares are our calculation from Census goods trade totals in nominal US dollars, not seasonally adjusted; 2026 = January to September. Census counts imports by their declared country of origin; how much of the shift reflects goods partly made in China is not measured here.

Sources: US Census Bureau, trade in goods with China, trade in goods with Vietnam, US trade in goods, world totals (data to September 2026).
Powered by Thorsten Meyer AI

China’s wave

The biggest wave of the last half-century came from China. According to Autor, Dorn and Hanson (2016), China’s share of the world’s manufacturing exports grew from 2.3% in 1991 to 18.8% in 2013. China joined the WTO on 11 December 2001.

In US Census Bureau data, American imports of goods from China rose from $19 billion in 1991 to $321 billion in 2007, and peaked at $539 billion in 2018. Measured as a share of all US goods imports, China went from 3.9% (1991) to 16.4% (2007) and 21.6% (2017).

As episode 6 showed, American factory jobs had started to decline long before China joined the WTO, and the national effect of Chinese imports was real but smaller than the long decline. What surprised economists was something else: where the jobs went.

The China shock

For a long time, economists expected the losses from trade to spread thinly. If a furniture factory closes, the logic went, its workers move to other industries or other regions, and the pain is shared by everyone with similar skills. Autor, Dorn and Hanson describe that prevailing view in their 2021 paper, and then the evidence against it.

In “The China Syndrome” (2012 working paper, American Economic Review 2013), they studied 722 local labour markets (“commuting zones”) covering the US mainland from 1990 to 2007. Places whose factories made what China began to make, like furniture, toys and clothing, saw imports hit hard. In their words, rising exposure “increases unemployment, lowers labor force participation, and reduces wages in local labor markets”, and payments for unemployment, disability, retirement and healthcare “rise sharply in exposed labor markets”. The research came to be known as the China shock.

An empty factory hall with idle machines and stacked unfinished wooden chair frames in soft daylight
The China shock hit places that made furniture, toys and clothing, and the damage lasted for years. AI-generated illustration.

The number to remember: 86%

And it lasted. In “On the Persistence of the China Shock” (2021), the same authors show that the shock plateaued in 2010. But adverse effects on manufacturing jobs, employment and income per person in more trade-exposed places were still present in 2019, nearly a decade later.

That’s the number to remember: 86%. Over 2000 to 2019, greater import competition lowered the share of working-age people in manufacturing jobs by 1.54 percentage points in more exposed regions compared with less exposed ones, and 86% of that loss showed up as a drop in the overall employment rate. In plain words: for every 100 factory jobs lost to Chinese competition, 86 were not replaced by any other job in the area.

Where did the people go? Mostly nowhere. Population counts fell only for foreign-born workers and native-born adults aged 25 to 39. The authors conclude that “exit from work is a primary means of adjustment”. The effects were worse in places that had fewer college-educated workers and depended on a narrow set of industries.

Thorsten Meyer AI
Work After Work · Ep. 15

Infographic 02 · The number to remember

Gains spread thin, losses concentrated

What the China shock did to American places, and how many ended up worse off even after cheaper prices.

86%of the factory jobs lost to Chinese imports in more exposed US regions were not replaced by other local jobs (2000-2019)
2010the year the China shock plateaued
2019effects on jobs and income still visible
1.25%lower US consumer prices from trade with China: a high-end estimate (Jaravel & Sager)

Share of the US population living in places with net losses from trade with China · estimates

  • Standard estimate of the gainsaggregate gains of 0.22% or less · 223 commuting zones32.8%
  • High estimate of cheaper pricesconsumer prices 1.25% lower (Jaravel & Sager) · 82 commuting zones6.3%

How to read the 86%: in more exposed regions compared with less exposed ones, the share of working-age people in manufacturing fell by 1.54 points, and the overall employment rate fell by 86% of that. Most people did not move away; the authors call “exit from work” a primary means of adjustment. All figures are research estimates, not job counts.

Source: Autor, Dorn & Hanson, On the Persistence of the China Shock (NBER WP 29401, 2021), abstract and section 6.
Powered by Thorsten Meyer AI

Germany: a different story

Germany’s experience was different. Wolfgang Dauth, Sebastian Findeisen and Jens Suedekum studied German local labour markets during the “rise of the East”, trade with China and Eastern Europe, from 1988 to 2008 (Journal of the European Economic Association, 2014).

  • Regions specialised in industries that competed with imports lost jobs.
  • Regions specialised in export industries gained even more.
  • In the aggregate, they estimate that this trade integration caused some 442,000 additional jobs and “contributed to retaining the manufacturing sector in Germany”. That, they add, “is almost exclusively driven by the rise of Eastern Europe, not by China”. (An earlier working-paper version gave 493,000.)

Today the picture is shifting. According to Germany’s Federal Statistical Office (Destatis), China was again Germany’s most important trading partner in 2025, with trade of €251.8 billion. But the balance tipped further: imports from China rose 8.8% to €170.6 billion, while exports to China fell 9.7% to €81.3 billion. The import surplus with China grew to €89.3 billion.

Myth check: “Free trade makes everyone better off”

Verdict: partly true.

  • For the country as a whole, yes. Autor, Dorn and Hanson write that, as expected, quantitative models find US aggregate gains from trade with China are positive, and that few economists would read their results as a case for more protection. Shoppers gained through lower prices.
  • For every place, no. The gains were spread thin; the losses were concentrated. With the standard model estimate of aggregate gains (0.22% or less), the authors count 223 commuting zones, home to 32.8% of the US mainland population, that suffered net losses.
  • Even with a high estimate of the cheaper prices (based on Jaravel and Sager, 2019, implying 1.25% lower consumer prices), 82 commuting zones with 6.3% of the population still ended up worse off.

Our reading: Ricardo’s arithmetic works for countries. It does not promise that it works for every town, and in the 2000s, for many towns, it didn’t.

2025 and 2026: tariffs return

In 2025, the United States raised tariffs sharply. The US Treasury’s monthly statements show what that meant in money: net customs duties of $195 billion in fiscal year 2025 (October 2024 to September 2025), against $77 billion a year earlier, two and a half times as much.

On 20 February 2026, the Supreme Court ruled 6-3 in Learning Resources, Inc. v. Trump that the International Emergency Economic Powers Act, the emergency law behind many of these tariffs, does not allow them: “IEEPA does not authorize the President to impose tariffs.” Refunds followed: Treasury data show customs refunds of about $134 billion from May to September 2026.

But tariffs under other laws stayed. In September 2026, gross customs duties were still $25.3 billion for the month, about three times the $7.9 billion of January 2025. Net of refunds, customs revenue for fiscal year 2026 came to $173 billion.

Meanwhile, trade adjusted rather than stopped:

  • China’s exports set a record. According to China’s General Administration of Customs, as reported by Xinhua on gov.cn, exports in 2025 grew 6.1% to 26.99 trillion yuan, imports 0.5% to 18.48 trillion yuan. The difference, about 8.5 trillion yuan, is a goods trade surplus of more than a trillion dollars (our conversion).
  • China’s share of US imports halved. From 21.6% in 2017 to 9.0% in 2025 and 7.7% in January to September 2026.
  • Vietnam took a big part of it. Its share of US goods imports rose from 2.0% (2017) to 7.4% (January to September 2026). Vietnam has been a WTO member since 11 January 2007.

Vietnam: a sector clock on fast-forward

What a trade boom does to a country’s workforce shows up on our sector clock. By the International Labour Organization’s modelled estimates (World Bank data), three in four Vietnamese workers farmed in 1991, and only 7% worked in industry and construction. By 2025, one in four farmed, and 35% worked in industry and construction. Services grew from 19% to 40%.

Thorsten Meyer AI
Work After Work · Ep. 15

Infographic 03 · The sector clock

Vietnam: a sector clock on fast-forward

Share of employed persons by sector. In one generation, Vietnam went from three in four workers on the land to one in four, as factories for export arrived.

Primary field: farming, fishing, forestrySecondary factory: industry, buildingTertiary services: shops, offices, care
1991
  • Field74.0%
  • Factory7.3%
  • Services18.7%

ILO modelled estimate

2007
  • Field49.3%
  • Factory20.4%
  • Services30.3%

joins the WTO (11 Jan)

2025
  • Field25.0%
  • Factory34.8%
  • Services40.1%

ILO modelled estimate

Secondary = industry and construction (ILO definition). Modelled estimates get revised; shares may not add up to exactly 100 because of rounding.

Sources: World Bank WDI, employment by sector, Viet Nam (ILO modelled estimates) (data updated 8 October 2026); WTO, Viet Nam member page.
Powered by Thorsten Meyer AI

And AI?

AI is already moving trade, mainly through the hardware it needs. The WTO’s Global Trade Outlook of October 2026 counts “AI-enabling goods” such as semiconductors, servers and network equipment:

  • They made up 14.8% of world merchandise trade in the first half of 2026, up from 7-8% in 2016-2023.
  • They accounted for 47% of the growth in world merchandise trade by value in that half-year.
  • The trade is highly concentrated: in 2025, the top 10 traders of these goods accounted for around 85% of world exports.
  • Partly because of this AI investment boom, the WTO raised its forecast for world merchandise trade volume growth in 2026 to 3.9%, up from 1.9% in March.

Looking further ahead, the WTO’s World Trade Report 2025 simulates that by 2040, AI could raise the value of world trade by 34% to 37% across its scenarios, and global GDP by 12-13%, with the largest growth in digitally deliverable services (42%), services that can be sent over the internet. These are model simulations under policy assumptions, not forecasts you can bank on.

What it means for businesses

  • Know where your inputs really come from. The China shock and the 2025 tariffs both show how fast a cost base can change. A second source is insurance.
  • Plan for policy swings. Tariffs can change within months, and courts can reverse them, as the refunds of 2026 show.
  • Watch the AI supply chain. If your products depend on chips and servers, you depend on a trade that is about twice as concentrated as goods trade overall.

What it means for you

  • Your exposure depends on your town as much as your job. The China shock hit hardest where one industry carried the local economy.
  • Ask how much of your work could be delivered over the internet. If a task can be sent over a wire, it can be traded, to someone abroad or to a machine.
  • My reading: the next trade shock may hit offices rather than factories. That is an interpretation of the WTO’s services projection, not a finding.

Thorsten’s thesis: what trade says about it

My thesis for this season is that automation could become its own sector, up to 90% of the economy within a decade or a few decades. That is my thesis, not a fact, and this season tests it with evidence for and against.

Ricardo gives the strongest argument against it. Comparative advantage does not vanish: even if machines become better at everything, it can still pay to let humans do what costs them the least to do. As I read it, that holds as long as machine time, chips and energy are not free.

But there is a counterpoint for it, and it comes from episode 3. As I see it, horses had a comparative advantage too. It did not save them once tractors got cheap enough to do their work for less than the cost of feeding them. Whether people are more like Portugal or more like the horse is one of the questions episode 47 will test.

Limits of this episode

  • Estimates, not counts. The China-shock figures are research estimates that compare more and less exposed regions; they are not a national job count. The 86% compares changes in employment rates, not individual workers.
  • Different studies, different periods. Episode 6 cited a national estimate for 1999-2011 (Acemoglu, Autor, Dorn, Hanson and Price, 2016). This episode uses the regional studies (1990-2007 and 2000-2019).
  • Census shares. Import shares are our calculation from US Census Bureau goods trade totals (nominal dollars, not seasonally adjusted). 2026 covers January to September. Census counts imports by declared country of origin; how much of Vietnam’s rise reflects goods partly made in China is not measured here.
  • Customs data. Treasury figures are net receipts by fiscal year unless marked gross; refunds are not broken down by legal basis in the monthly statement.
  • China. The surplus in dollars is our conversion of official yuan figures at the rate Xinhua used for total trade.
  • World Bank and ILO. Trade-to-GDP and Vietnam’s sector shares are modelled or compiled estimates and get revised; the 2025 world trade-to-GDP value was left out because it looked incomplete.
  • Interpretations. The lawyer example, the lessons for businesses and for you, “offices rather than factories” and the horse argument are my reading.

Series navigation

Sources

  • David Ricardo, On the Principles of Political Economy, and Taxation (London: John Murray, 1817), chapter “On Foreign Trade”. gutenberg.org
  • The National Archives (UK), “The Corn Laws: What was the impact of the 1815 Corn Laws in Britain?” (education resource). nationalarchives.gov.uk (PDF)
  • US Department of State, Office of the Historian, “Protectionism in the Interwar Period”. history.state.gov
  • World Trade Organization, “The GATT years: from Havana to Marrakesh”. wto.org · Member pages China (since 11 December 2001) and Viet Nam (since 11 January 2007). wto.org · wto.org
  • World Bank, World Development Indicators: Trade (% of GDP), world (NE.TRD.GNFS.ZS); employment in agriculture, industry and services, Viet Nam (SL.AGR.EMPL.ZS, SL.IND.EMPL.ZS, SL.SRV.EMPL.ZS; ILO modelled estimates; data updated 8 October 2026). data.worldbank.org
  • US Census Bureau, Trade in goods with China (5700), Vietnam (5520) and world totals (0004), through September 2026. census.gov · census.gov · census.gov
  • David H. Autor, David Dorn, Gordon H. Hanson, “The China Syndrome: Local Labor Market Effects of Import Competition in the United States”, NBER Working Paper 18054 (2012); American Economic Review 103(6), 2013. nber.org
  • David H. Autor, David Dorn, Gordon H. Hanson, “The China Shock: Learning from Labor Market Adjustment to Large Changes in Trade”, NBER Working Paper 21906 (2016); Annual Review of Economics 8, 2016. nber.org
  • David Autor, David Dorn, Gordon H. Hanson, “On the Persistence of the China Shock”, NBER Working Paper 29401 (2021); Brookings Papers on Economic Activity, Fall 2021. nber.org
  • Wolfgang Dauth, Sebastian Findeisen, Jens Suedekum, “The Rise of the East and the Far East: German Labor Markets and Trade Integration”, Journal of the European Economic Association 12(6), 2014, 1643-1675 (abstract). ideas.repec.org
  • Destatis, “China is Germany’s most important trading partner once again in 2025”, press release No. 056 of 20 February 2026 (German version with import and export figures). destatis.de · destatis.de
  • US Department of the Treasury, Monthly Treasury Statement, Table 4 (customs duties: gross, refunds, net), via Fiscal Data. fiscaldata.treasury.gov
  • Supreme Court of the United States, Learning Resources, Inc. v. Trump, No. 24-1287, decided 20 February 2026. supremecourt.gov (PDF)
  • Xinhua via gov.cn, “China’s foreign trade hits new high, up 3.8 pct in 2025” (14 January 2026; General Administration of Customs data). gov.cn
  • World Trade Organization, Global Trade Outlook and Statistics, update: October 2026. wto.org (PDF)
  • World Trade Organization, World Trade Report 2025: Making trade and AI work together to the benefit of all, press release of 15 September 2025 and executive summary. wto.org · wto.org (PDF)
  • John Maynard Keynes, “Economic Possibilities for our Grandchildren” (1930), quoted for the next episode. econ.yale.edu (PDF)

This article is for information only. It is not career, legal, financial or investment advice.

Researched and drafted with AI assistance, including Anthropic’s Claude.

You May Also Like

218 Arrests at a Microsoft Data Center Site in Amsterdam: Why Europe’s Backlash Against AI Data Centers Is Growing

On November 6, 2026, protesters entered the construction site of a data center leased to Microsoft in Amsterdam. Police arrested 218 people for suspected trespassing and released all of them. What is confirmed, what is only claimed, why Europe’s backlash against AI data centers is growing, and what it means for companies and users.

Reality Check: Will Every Job Really Be Automated?

Facing the automation wave, many wonder which jobs will truly disappear and which will adapt—discover the surprising realities ahead.

Is This a Cyberattack? A Week of Outages and Breaches, and What Actually Caused Them

A Microsoft outage, a nationwide FAA ground stop and five confirmed attacks in one week. What the people responsible say caused each, and why no evidence so far does not mean ruled out.

ASI-ARCH: A New Era of Autonomous AI Research

AIThis post was created with the assistance of artificial intelligence (AI).The AI…