This summer at the Washington Campus program, I got a front-row seat to how policy and business really mix. We sat with agency staff, lobbyists, representatives and even met with a group of European delegates to compare notes on how the global economy is changing. As a die-hard semiconductor nerd, I channeled all of that into a team discussion paper on trade, tariffs, and chips.
Here’s the short version of what we found -
The question we chased
When the U.S. imposed Section 301 tariffs on China in 2018, did trade simply slow between those two countries? Or did the flows reroute through other places?
Our answer: the flows rerouted notably through Mexico and investment moved to make that rerouting easier.
Why old models miss the new reality
Classical trade theories (comparative advantage, Heckscher-Ohlin) assume a tidy world: two countries, simple products, low friction. That’s not chips or frankly, anything modern.
Semiconductors are a relay race: design here, wafer there, packaging somewhere else, final assembly in yet another place. Firms optimize each “leg” for cost, skill, risk, and policy. So if you tax one leg, the baton doesn’t drop it’s handed off on a different route.
We argue for a network-aware view: look at the full map of suppliers, transit points, and investors, not just two flags at the ends of a trade line.
What the data says
We stitched together public trade and investment data from 2010-2024 and followed the direction and timing of changes rather than obsessing over perfect levels. Three patterns stood out:
- Direct China → U.S. imports fell after 2018. That’s the headline you’d expect from tariffs and you do see it, including in tech-heavy categories.
- Mexico’s role grew. U.S. imports from Mexico rose at the same time that China’s exports to Mexico increased. That’s the telltale sign of trade diversion: components and subassemblies moving into Mexico, getting finished or rerouted there, then shipped to the U.S.
- Investment followed the trade. Chinese foreign direct investment into Mexico picked up especially in auto and electronics-adjacent activities consistent with firms redesigning supply chains to keep U.S. market access under USMCA rules.
Tariffs changed the path, not the destination.
What this looks like in chips
Semiconductors make the pattern obvious:
- Modular production. Chip value chains split into design (EDA, IP), fabrication (wafers), packaging & testing (OSAT), modules/boards, and final devices. Those steps already sit in different countries.
- Easy-to-reroute legs. Packaging/testing and electronics assembly are the most footloose. If direct China → U.S. is costly, firms shift those legs to Mexico (or elsewhere) and keep selling into the U.S., sometimes with essentially the same underlying content.
- Rules-of-origin gray zones. Under “substantial transformation,” a product can change its tariff “nationality” if enough is done in a third country. In practice, that can mean a China-origin component gains a Mexico label after additional processing legal, but policy-relevant if the goal was to curb China-linked inputs.
Why it matters
If we judge policy by bilateral numbers alone (“Are U.S. imports from China down?”), we’ll miss what’s happening two hops away. In a networked economy:
- Trade relabels itself as it moves.
- Capital repositions to keep the network working.
- Security and resilience goals can be undone by detours unless we can see the full map.
So what should policy actually do?
If trade behaves like a network, policy has to become network-aware. We outline four practical shifts:
- Trace components, not just finished goods. Build real traceability for critical nodes (chips, substrates, advanced packaging). Think lot-level reporting and standardized supplier attestations so content doesn’t vanish in a country-of-origin relabel.
- Tighten and enforce rules of origin. Aim at the steps that matter for value and security (e.g., advanced nodes, trusted packaging), not blanket thresholds that encourage box-checking.
- Screen enabling investment, not just imports. Coordinate trade tools with investment review so capital can’t quietly rebuild the same pathway via a third country.
- Measure the network to manage it. Publish network-style metrics (how much China-linked content reaches the U.S. via partners; which hubs are systemic). If we can’t measure the detours, we can’t design better roads.
My takeaway from the summer
Visiting policymakers and talking with European counterparts made one thing clear: everyone is grappling with the same puzzle how to keep trade open, safe, and resilient when production is spread across the map. For semiconductors, the stakes are higher because the chain is long and the nodes are specialized.
Section 301 didn’t end the flow; it redrew it. If we update our tools to see the network inputs, waypoints, and money we can make smarter choices about when tariffs help, when they just reroute, and where targeted incentives (or guardrails) will actually move the needle.
Want the full paper? DM me on LinkedIn.
