Case 7 · Events 5, 7
Why does a chip shortage hurt more than its size, and how did controls reshape the trade?
The output loss from a constraint on energy or semiconductors is “on average 3.5 to 4.5 the size of the initial impact”; “a 10% contraction in world semiconductor production would reduce global GDP by about 0.2%”.
Quote checked against the paperRead the paper
With the 2018–19 US tariffs on China as a natural experiment, the restructuring of chip supply chains is “not led solely by the direct consequences of the tariff hikes” but also by shifts in comparative advantage between the countries in the chain.
Quote checked against the paperRead the paper
The papers’ questions measured again on UN Comtrade records. Each reading below is written from these numbers, not from the papers.
Taiwan held 25.7% of world integrated circuits exports in 2019 and 29.6% in 2024; the largest gain among the top ten: Taiwan, +3.9 pp.
| 2019 | 1 | Taiwan | 25.7% |
| 2019 | 2 | Korea | 15.4% |
| 2019 | 3 | China | 12.6% |
| 2019 | 4 | Malaysia | 12.4% |
| 2019 | 5 | United States | 5.9% |
| 2019 | 6 | Japan | 5.8% |
| 2019 | 7 | Singapore | 5.8% |
| 2019 | 8 | Vietnam | 3.4% |
| 2019 | 9 | Philippines | 3.2% |
| 2019 | 10 | Thailand | 1.8% |
| 2024 | 1 | Taiwan | 29.6% |
| 2024 | 2 | Korea | 17.2% |
| 2024 | 3 | China | 15.4% |
| 2024 | 4 | Malaysia | 7.9% |
| 2024 | 5 | Singapore | 6.1% |
| 2024 | 6 | Japan | 5.1% |
| 2024 | 7 | United States | 3.7% |
| 2024 | 8 | Vietnam | 2.9% |
| 2024 | 9 | Philippines | 2.1% |
| 2024 | 10 | Thailand | 1.6% |
Japan’s share of China’s chip-making machines (side input) imports: 23.7% in 2015, 26.4% in 2025. Highest 31.2% in 2021.
| 2015 | 23.7% | 11.5% | 31.5% | 14.2% | 8.2% | 11% | – | +0.3% | yes |
| 2016 | 26.6% | 11.8% | 29.3% | 8.3% | 9.8% | 14.2% | +2.9 pp | +20.1% | yes |
| 2017 | 28% | 11.1% | 25.8% | 9.7% | 7.6% | 17.9% | +1.4 pp | +25.5% | yes |
| 2018 | 29.4% | 16.1% | 24.5% | 8.2% | 8.5% | 13.3% | +1.5 pp | +66.4% | yes |
| 2019 | 29.6% | 10.5% | 24.7% | 13% | 8.1% | 14.1% | +0.2 pp | +1.1% | yes |
| 2020 | 29.1% | 13.2% | 24.9% | 11% | 9.9% | 12% | −0.6 pp | +30.5% | yes |
| 2021 | 31.2% | 11.4% | 22.6% | 11% | 11.2% | 12.8% | +2.2 pp | +43.5% | yes |
| 2022 | 31.1% | 10.2% | 19.7% | 9.7% | 12.2% | 17.1% | −0.1 pp | −11.5% | yes |
| 2023 | 29.1% | 22% | 12.9% | 6.3% | 13.9% | 15.9% | −2 pp | +28.6% | yes |
| 2024 | 29.6% | 24.5% | 10.9% | 7.1% | 13.7% | 14.3% | +0.5 pp | +18.3% | yes |
| 2025 | 26.4% | 24.1% | 6.9% | 6.9% | 15.6% | 20.1% | −3.2 pp | +5.1% | yes |
China’s share of United States’ integrated circuits imports: 6.8% in 2015, 3% in 2025.
| 2015 | 6.8% | 13% | 37.2% | 8.7% | 6.8% | 0.57% | 27% | – | −2.9% | yes |
| 2016 | 7.2% | 11.4% | 38.3% | 7.4% | 5.9% | 0.58% | 29.3% | +0.3 pp | +6.7% | yes |
| 2017 | 8.1% | 11.7% | 38% | 6.5% | 6.1% | 4.4% | 25.3% | +0.9 pp | +8.9% | yes |
| 2018 | 9.1% | 12.1% | 43.6% | 4% | 6.3% | 4.1% | 20.9% | +1 pp | +4% | yes |
| 2019 | 4.7% | 11.2% | 47.5% | 5.8% | 6.6% | 3.9% | 20.3% | −4.4 pp | −4.8% | yes |
| 2020 | 5.5% | 11.6% | 47.6% | 8% | 6.2% | 4.4% | 16.6% | +0.8 pp | −3.6% | yes |
| 2021 | 5.7% | 13.1% | 45.8% | 6.6% | 6.2% | 4% | 18.7% | +0.1 pp | +29.4% | yes |
| 2022 | 5.9% | 18.6% | 32.7% | 5.4% | 7.2% | 4.3% | 26% | +0.3 pp | +5.8% | yes |
| 2023 | 5.1% | 20.4% | 27% | 4.8% | 6.7% | 4.7% | 31.3% | −0.8 pp | −17.4% | yes |
| 2024 | 3.6% | 28.3% | 23.8% | 3.4% | 6% | 3.7% | 31.1% | −1.5 pp | +11.8% | yes |
| 2025 | 3% | 30.8% | 25.4% | 2.6% | 4.5% | 1.1% | 32.6% | −0.6 pp | +5.1% | yes |
World trade, against 2019: −6.1% in passenger cars and +39.8% in integrated circuits.
| Passenger cars | 2021 | 2019 | −6.1% |
| Integrated circuits | 2021 | 2019 | +39.8% |
Cut chip output in one economy by 10% (Taiwan by default; China, Malaysia and others to choose) and follow the shortfall into phones, computers and cars
| Importer | At risk, share of its imports | Its other suppliers would ship |
|---|---|---|
| 5.9% | +15% | |
| 4.9% | +10% | |
| 4.9% | +10% | |
| 4.2% | +7% | |
| 4.2% | +7% | |
| 3.8% | +6% | |
| 3.3% | +5% | |
| 2.8% | +4% | |
| 2.7% | +4% | |
| 2.5% | +3% | |
| 2.5% | +3% | |
| 2.5% | +3% |
Who could fill the gap: Japan, 5.9% of its imports
The suppliers it already buys from, in proportion to their shares: each would have to ship 15% more. Share now, then the part of the gap it would take:
The largest other exporters, by share of world exports
Smartphones
Upper bound (100% pass-through, no use share). This stage feeds only this chain, so the chain file gives it no use share and the whole shortfall passes down.
3.3% of world smartphones exports at risk in the first round, summed over the 22 large exporters with an exposure upstream. The 8 listed below carry 3.3%.
| Exporter | Its exposure upstream | World exports at risk |
|---|---|---|
| China | 4.2% | 2.8% |
| India | 2.7% | 0.2% |
| Hong Kong | 2.1% | 0.07% |
| United States | 2.8% | 0.06% |
| Vietnam | 0.7% | 0.06% |
| Korea | 3.8% | 0.06% |
| Singapore | 4.9% | 0.02% |
| Japan | 5.9% | 0.02% |
Laptops and computer units
Upper bound (100% pass-through, no use share). This stage feeds only this chain, so the chain file gives it no use share and the whole shortfall passes down.
2.7% of world laptops and computer units exports at risk in the first round, summed over the 24 large exporters with an exposure upstream. The 8 listed below carry 2.6%.
| Exporter | Its exposure upstream | World exports at risk |
|---|---|---|
| China | 4.2% | 1.8% |
| Mexico | 2% | 0.29% |
| United States | 2.8% | 0.15% |
| Thailand | 4.9% | 0.09% |
| Singapore | 4.9% | 0.07% |
| Vietnam | 0.7% | 0.06% |
| Malaysia | 2.1% | 0.05% |
| Germany | 2.3% | 0.03% |
An exporter of the next stage loses its share of world exports × its own exposure upstream (× the use share). Exporters on a closed route lose their own exports × the cut. Exposure upstream is the share of its imported integrated circuits that came from the closed supplier, not of all the integrated circuits it uses: where most are made at home (Japan, Korea, Germany for car parts), it overstates the risk.
First-round arithmetic on trade shares, not a forecast: an importer’s exposure is its import share from the shocked suppliers × the cut. No stocks, no new suppliers, no price changes.
The multiplier in the BIS bulletin comes from a global input-output table, not from trade: a network shows who buys chips, not how much output stops without them. Export cuts stand in for output cuts, a chip crosses borders several times under the same code, and controls act on specific tools and grades that HS 8486 and 8542 do not separate. Miki and Tamanyu go further with upstreamness and wage gaps; Funke and Wende (further reading) model export bans in general equilibrium.
datasteg: the flow sites
money-flow
Where money is made and where it goes: central bank and bank money, and cross-border positions by country pair.
Open money-flowtrade-flowYou are here
The world trade network and ten product chains from ore to product, with the events that moved them.
global-network
Funding and trade together: how economies earn, who funds whom against who trades with whom, and what past shocks did to both.
Open global-network