On April 2, the United States imposed tariffs ranging from 10% to over 60% on nearly every imported good, from nearly every country, regardless of sector. Not targeted duties on dumped steel. Not retaliation against a specific bad actor. A wall, raised in a single executive order under emergency economic powers, on the grounds that persistent trade deficits have hollowed out American manufacturing and compromised the defense industrial base.
The average U.S. tariff rate has hovered between 2% and 3% for years. It is now set to reach roughly 18.8% — the highest since 1933.
Within a week, global markets shed something on the order of $5 trillion in value, JPMorgan raised its odds of a global recession within the year from 40% to 60%, and the administration suspended most of the policy for ninety days while raising tariffs on China to 125%.
I want to argue that this was a serious mistake. But I want to do it properly, which means starting with the best version of the opposing case rather than the worst — because the best version is considerably stronger than most of the commentary this week has allowed.
The strongest case for the tariffs
There is a real argument here, and it does not depend on economic illiteracy.
Start with the observation that free trade’s accounting is incomplete. The conservative economist Oren Cass puts it this way: “Domestic production has value beyond what market prices reflect. Those individual decisions to save money add up to collective economic, political, and societal harms.” Each decision to offshore a plant or buy the cheaper import is privately rational. Aggregated across four decades, those rational decisions hollowed out communities, destroyed accumulated process knowledge, and left the country dependent on others for things it may one day need urgently.
This is not a fringe position. It is a claim about externalities, and it is the same structure of argument economists use to justify carbon pricing. If a transaction imposes costs on third parties that the price does not capture, the market will produce too much of it. Tariffs, in this framing, are a Pigouvian correction: they force the buyer to internalize the collective cost of the offshoring decision.
The reciprocity argument has teeth too. For decades the United States has run one of the world’s lowest tariff regimes — around 3.3% — while trading partners maintained higher barriers and thickets of non-tariff obstacles. India charges 70% on certain vehicles against America’s 2.5%. Brazil charges 18% on ethanol against 2.5%. Turkey charges 60% on American apples, which enter the U.S. duty-free. Whatever one thinks of tariffs in the abstract, it is difficult to argue the existing arrangement was symmetric. Why, the argument runs, should the United States be the only committed free trader in a world of mercantilists?
Then there is security, which is the strongest leg of the three. COVID demonstrated what supply-chain concentration means in practice when China controlled the overwhelming majority of protective equipment production. Advanced semiconductors are made almost entirely in Taiwan, South Korea, and Japan — which is to say, within artillery range of the most likely great-power conflict of the next decade. Rare earths, pharmaceutical precursors, and grid transformers are similarly concentrated. A supply chain optimized purely for cost is a supply chain optimized for peacetime.
And the record on China is genuinely bad. State-backed industrial policy, forced technology transfer, and IP appropriation helped Beijing take industry after industry. The best available estimate puts U.S. job losses from Chinese import competition at roughly 3.7 million between 2001 and 2018. Even economists who loathe Trump concede the China Shock literature is real and that the policy response to it was close to nonexistent.
Finally, the universality has a coherent rationale. Previous tariffs aimed at China simply diverted production to Vietnam and Mexico rather than bringing it home. If you tariff one door, traffic moves to another door. Tariff every door, and the only remaining option is to come inside. That is the theory, and it is not stupid.
Taken together: a high-risk bid to reboot American industrial capacity and correct real defects in the trading order. That is the most charitable reading, and it deserves to be stated before it is answered.
Why the case fails
It fails on four counts, and only one of them is about history.
The reshoring mechanism does not work as advertised. This is the central defect, and everything else follows from it. Tariffs raise the cost of importing. They do not lower the cost of domestic production, supply skilled machinists, build substations, or shorten permitting timelines. A firm facing a 25% import duty has at least four options before relocating to Ohio: raise prices, absorb margin, reroute through a third country, or automate. Relocating is the most expensive and least reversible of the five.
The evidence from the 2018–2019 round is unambiguous. Imports from China fell. Imports from Vietnam, Mexico, and Malaysia rose to meet them. The overall goods trade deficit did not close — it hit a record $1.2 trillion in 2024, after six years of tariffs. The deficit is driven by the savings-investment balance and the strength of the dollar, not by tariff schedules. You cannot fix a macroeconomic identity with a customs form.
The universality argument answers this, but only on paper. Closing every door assumes firms respond by coming home. It is at least as likely that they raise prices and stay put, because the tariff changed the cost of importing without changing any of the reasons production left.
The distributional effect is the reverse of the stated intent. The Yale Budget Lab estimates the measures will cost a typical household about $3,800 a year, roughly $2,100 of that from the April tariffs specifically. The Tax Foundation scores the plan as a $1.8 trillion tax increase over ten years, reducing GDP by about 0.5% and average after-tax incomes by about 2.1%.
Because lower-income households spend a larger share of income on tradeable goods — clothing, food, fuel, appliances — and a smaller share on services, the burden falls hardest on precisely the people the policy claims to defend. This is a regressive consumption tax with an industrial policy attached, and the tax is certain while the industrial policy is speculative.
The employment arithmetic runs the same direction. Protected sectors gain; sectors that use the protected inputs lose. The 2002 steel tariffs and the 2018 steel and aluminum duties both saved steelworker jobs and destroyed a larger number in steel-consuming industries — auto parts, appliances, construction. There are far more Americans working with steel than making it. That ratio does not improve when you extend the logic to every input in the economy.
Retaliation is not a risk; it is the plan working as designed. Smoot-Hawley is invoked so reflexively that the actual lesson gets lost. The lesson is not “tariffs bad.” It is that trade barriers are reciprocal by nature: within two years of 1930, more than twenty countries had raised their own, and global trade fell by roughly 65% between 1929 and 1934.
China retaliated within hours this time. The EU began drafting target lists. American farmers, who required multibillion-dollar bailouts when China stopped buying soybeans in 2018, are now exposed on a much larger scale. Every dollar of protection for an import-competing producer is matched by a dollar of exposure for an exporter, and the United States is very good at exporting.
Hitting allies forfeits the only strategy that could actually work on China. This is the strategic error, and it is the one I find least defensible.
If the real concern is Chinese industrial policy and supply-chain coercion — and it should be — then the countries you need are Japan, South Korea, Taiwan, Canada, and the EU. They share the analysis. They have the complementary capabilities. Coordinated pressure from that bloc is the only thing Beijing has historically had to take seriously.
Instead they were tariffed alongside everyone else. Mark Carney’s line that the U.S. has “abandoned its historic role as a champion of international economic cooperation” is diplomatic language for a durable strategic loss. Allies who are punished indiscriminately acquire domestic political cover to retaliate, to deepen ties elsewhere, and to reduce their exposure to American suppliers and American finance. That process, once started, does not reverse when the tariffs do.
There is also a lesson in the case the administration likes to cite. Germany built a formidable industrial base without high tariffs — and is now stagnating anyway, with 2024 industrial output at about 90% of its 2015 level after two consecutive years of contraction, undone by energy costs and a slow pivot to new technology. A manufacturing base is not a trophy you win once. Protection can preserve a factory; it cannot make that factory competitive. Shielded industries reliably become the thing they were shielded from.
The week the policy reversed itself
Then, on April 9, the administration suspended the reciprocal tariffs for ninety days, kept a 10% universal baseline, and raised tariffs on China to 125%.
The reversal is more informative than the policy.
On April 2 this was declared a national emergency. Peter Navarro published an op-ed insisting “this is not a negotiation.” The Treasury Secretary said flatly that talks were not on offer. On April 7, the President posted that negotiations with other countries “will begin taking place immediately” — while Navarro’s op-ed was still on newsstands. By April 9 the pause was announced, and Scott Bessent was in the White House driveway explaining that “President Trump created maximum negotiating leverage for himself. This was his strategy all along.”
Perhaps. But the President subsequently indicated that the near-panic in financial markets factored into the decision, which is difficult to reconcile with a plan executed as designed. The S&P 500 had fallen about 15% from recent peaks. Goldman put recession odds at 45%. Larry Fink and Jamie Dimon — not natural adversaries of this administration — had both objected publicly. On the announcement the S&P rose 5%, its largest single-day gain since 2008. The following session it fell 3.5%.
That two-day round trip is the market’s assessment, and I think it is correct. The relief was about the retreat. The renewed selling was about what the retreat implied: the tariffs were postponed, not resolved; China was escalated against rather than negotiated with; and a July deadline now hangs over every capital-allocation decision in the country.
There is a version of the madman strategy that works. It requires the threat to be credible, which requires the willingness to absorb pain. What this week established is that the pain threshold is roughly one bad fortnight in equities. Allies now know the hard line softens under market pressure. Adversaries have learned they can wait. Whatever leverage the shock generated, the climb-down spent it.
And the underlying grievances — the hollowed-out industrial base, the asymmetric barriers, the concentrated supply chains — remain exactly where they were on April 1, now with less allied goodwill available to address them.
What would actually work
The critique above is only worth making if there is something better on offer. There is, and most of it is already partly built. Four pillars, none of which requires a trade war.
Targeted industrial policy. The CHIPS and Science Act commits roughly $52 billion — about $39 billion in direct manufacturing incentives, $13 billion for R&D, workforce training, and regional innovation — plus a 25% investment tax credit on semiconductor manufacturing equipment. It is already producing: TSMC in Arizona, Samsung in Texas, Intel breaking ground in Ohio. The Inflation Reduction Act’s roughly $369 billion in clean manufacturing incentives catalyzed over 270 projects and something like $278 billion in private investment within a year.
The objection is that governments cannot pick winners. The record is more interesting than that. South Korea’s Heavy and Chemical Industry drive built POSCO and Hyundai into world-class firms — but the aid was conditioned on export performance and technological benchmarks, and firms that missed them lost support. Taiwan’s ITRI identified semiconductors as a strategic opening and spun out TSMC, which now handles more than half the world’s contract chip production. Germany’s Fraunhofer institutes and vocational system sustain manufacturing at nearly 20% of GDP against America’s 11%.
The pattern in every success is the same: support is conditional, performance-measured, and export-oriented. The failures are the cases where it became permanent and unconditional. That is a design constraint, not a refutation — and it is a far more tractable problem than “make forty years of offshoring reverse itself via customs.”
Crucially, this does what tariffs cannot. It lowers the cost of producing in America rather than raising the cost of buying from abroad. One of those builds capacity. The other just taxes.
Resilient supply chains through alliances. The goal is not autarky, which is unachievable and would be ruinous if achieved. It is that critical goods be produced within a network of countries that will not use them as leverage.
This is already underway: the Quad’s critical technology working group, the Minerals Security Partnership coordinating investment in lithium, cobalt, and rare earths, the Chip 4 arrangement among the U.S., Japan, South Korea, and Taiwan, and the U.S.-EU Trade and Technology Council. The logic is that an EV supply chain running from Australian lithium through Korean processing to American assembly is both resilient and efficient, because each step sits in a country that will not weaponize it.
Every one of those arrangements was made harder on April 2. You cannot simultaneously ask Japan to coordinate export controls on advanced chips and tariff Japanese cars as a national emergency.
Workforce development. The binding constraint on reshoring is not the price of imports. It is that the fabs, plants, and grid projects being funded right now cannot find the electricians, technicians, and machinists to build and run them. This is the least glamorous pillar and probably the highest-return one, and it is the part of the German model that transfers most cleanly.
Selective trade tools. Nothing here argues for unilateral disarmament. Anti-dumping duties, countervailing duties against subsidized exports, and targeted Section 232 action on genuinely strategic goods are legitimate and effective — because they are aimed at specific, documented conduct, they are defensible to allies, and they invite narrow rather than general retaliation. Trade Adjustment Assistance, properly funded rather than nominally funded, handles the displacement that remains.
The distinction throughout is precision. A tariff on Chinese solar panels dumped below cost is a policy. A tariff on everything from everyone is a mood.
The thing worth keeping
The tariff advocates are right about the diagnosis and wrong about the treatment. That is a genuinely uncomfortable position for their critics, most of whom spent two decades insisting the diagnosis was imaginary.
Manufacturing communities were hollowed out. The policy response to the China Shock was close to nonexistent. Supply chains optimized for cost really are fragile in ways that matter for national security. The trading order really was asymmetric. Anyone who wants to argue against these tariffs while dismissing all of that is not going to persuade anyone who lives in Ohio.
But the response to a real problem still has to be a mechanism that solves it. Blanket tariffs raise consumer prices with certainty, invite retaliation with certainty, and produce reshoring only speculatively, through a causal chain that the 2018 round already tested and falsified. They alienate exactly the countries whose cooperation is required to address the underlying problem. And as of this week, they are not even credible, because the administration demonstrated it will suspend them when the market objects.
America does not need a wall. It needs to be a better place to build things — which is a matter of capital, skills, energy, permitting, and allies, and can be worked on directly.
That work is slower and much less satisfying than a single executive order. It is also the only version that ends with factories.