What Economics Actually Knows About Trade Policy
I. The Trigger
I was listening to the All-In podcast interview with Commerce Secretary Howard Lutnick a few days ago. It’s a wide-ranging conversation; tariffs, GDP growth, the administration’s economic strategy. But one of Howard’s introductions of the adminitration’s high level economic philosophies bothered me.
They were discussing steel. Lutnick was making the case for why tariffs are necessary, and he laid out the math:
“How much does it cost in America? $700 a metric ton… Japan subsidizes, Korea subsidizes it. So maybe they can do it for $400 a metric ton.”
He went further: if China gives free power to steel producers, they can make steel for $250 a metric ton. American steel at $700, subsidized foreign steel at $400 or less. The implication was clear; American producers can’t compete against governments willing to subsidize their industries into profitability.
The numbers sounded precise. The logic sounded airtight. But I realized I had no idea how to evaluate whether any of this was true, or what it meant if it was. Is that $300 difference a gift to American consumers or a weapon against American workers? Is there a framework for thinking about this, or just competing tribal narratives dressed up as economics?
II. The Gift and Its Complications
The basic economics aren’t complicated. When a foreign government subsidizes its exports, there are winners and losers in the receiving country.
Winners: Consumers pay less. Downstream industries; automakers, construction companies, appliance manufacturers; get cheaper inputs. They can invest the savings elsewhere or pass them to customers.
Losers: Domestic producers in that sector can’t compete at the subsidized price. Workers lose jobs. Communities built around steel production hollow out.
There’s a seductive argument here: if Japan and Korea want their taxpayers to subsidize American consumers, why not take the gift? Classical economists would say exactly that. Use the savings productively elsewhere. Let them pay for our cheap steel.
But the complications pile up quickly. What if you need domestic steel production during a war? What happens to the workers, towns, and suppliers built around the industry? Does accepting the “gift” encourage more aggressive subsidies in other sectors? Can you actually compensate the losers, or do they just lose?
I found myself with competing frameworks that each seemed reasonable and pointed in opposite directions. Free trade maximizes aggregate welfare; but aggregate welfare doesn’t help the steelworker in Ohio. Strategic autonomy matters; but you can’t be autonomous in everything. The arguments were coherent, just irreconcilable.
III. Is Economics Just Tail-Chasing?
This is where I hit a wall. I started asking a more uncomfortable question: can we actually know anything here? Is economics just philosophical tail-chasing; smart people building elaborate frameworks to justify whatever they already believed?
The honest answer I found was worse than I expected.
Economics is worse than weather prediction. At least weather systems don’t change because you predicted rain. Economies do. If enough people believe a recession is coming, they spend less, and the recession arrives. The prediction changes the outcome. This is called reflexivity, and it makes economic forecasting fundamentally different from predicting physical systems.
We can measure flows; how much steel crossed borders, at what price, where jobs were lost. These are accounting facts. But the causal questions I actually cared about; “did that subsidy cause this job loss?”; are brutally hard.
No control group. We can’t run the American economy twice, once with tariffs and once without.
Everything moves at once. Currency fluctuations, oil prices, technology changes, demand shifts; all happening simultaneously.
Long causal chains. A steel tariff affects auto prices affects consumer spending affects… what exactly? Who knows.
I started to suspect that a lot of trade policy debate is motivated reasoning wrapped in economic language. Free traders cite consumer surplus. Protectionists cite strategic autonomy. Both can build models that “prove” their point. The math exists, but it’s not predictive the way physics is. It’s more like history with equations.
IV. What Can We Actually Know?
But here’s where the despair lifts a little. If prediction is mostly impossible, that doesn’t mean all knowledge is impossible. Policymakers have to decide something. What can they actually base decisions on?
Accounting identities. These are constraints on what CAN happen, not predictions of what WILL happen. Trade deficit equals capital surplus; if money leaves for goods, it comes back as investment. GDP equals consumption plus investment plus government plus net exports. These aren’t debatable; they’re definitions. If someone’s story violates an accounting identity, it’s wrong.
Historical base rates. Not predictions, but priors. When countries did X in the past, Y usually followed. Doesn’t mean Y will follow this time, but the pattern is informative.
This is where I had a small epiphany. We have 150 years of trade policy history across dozens of countries. This isn’t perfect experimental data, but it’s something. Countries tried protectionism and free trade, subsidies and sanctions, currency manipulation and market liberalization. We can see what happened.
The question shifts. Instead of “what will happen if we impose tariffs?”; which is probably unknowable; the question becomes “what has consistently happened when countries imposed tariffs?” That’s answerable.
V. Five Things History Actually Tells Us
I dug into the historical record. Not theory; empirical patterns that recur across different eras, different countries, different ideologies. Here’s what holds up.
Finding 1: Tariff Retaliation Is Near-Certain and Mutually Harmful
This is as close to a law as trade economics gets. When one country raises tariffs, trading partners retaliate. Both sides end up worse off.
The canonical example is Smoot-Hawley in 1930. The US raised tariffs on over 20,000 imports. Within months, 60 nations imposed retaliatory tariffs. World trade collapsed by 66% between 1929 and 1934. Smoot-Hawley didn’t cause the Great Depression, but it made everything worse.
The pattern holds in every subsequent trade war. The 2018 US-China tariffs triggered immediate retaliation. The EU responds to US steel tariffs with tariffs on bourbon and motorcycles. The other side always hits back. Always.
Finding 2: Trade Imbalances Are Macroeconomic, Not Trade Policy Problems
This one surprised me. The US-Japan trade conflict of the 1980s is the definitive case.
The US ran massive trade deficits with Japan. We imposed Voluntary Export Restraints on automobiles. We forced semiconductor agreements. We used the Plaza Accord to depreciate the dollar by 40%. Decades of sectoral protectionism.
The bilateral trade deficit persisted. Why? Because trade deficits reflect the difference between national savings and investment. If Americans save less than they invest, we run a trade deficit. Tariffs shuffle which countries we import from; they don’t change the underlying gap.
The Plaza Accord’s exchange rate adjustment actually did more than all the protectionism combined. But even that didn’t “fix” the deficit, because the deficit wasn’t the problem; it was a symptom of macroeconomic patterns.
Finding 3: Industrial Policy Works Only With Export Discipline
This finding resolves a longstanding puzzle. Why did South Korea and Taiwan industrialize successfully while Latin America stagnated with similar policies?
The Asian Tigers (Korea, Taiwan, Singapore, Hong Kong) averaged over 7% GDP growth annually for three decades. They used infant industry protection; subsidies, tariffs, preferential treatment. But they attached a condition: protected industries had to prove competitiveness by exporting. No export performance, no continued support.
Latin American countries did import substitution without the export requirement. Protected industries never had to compete globally. They stayed inefficient forever. Brazil’s computer industry is the poster child; decades of protection produced technology that fell further and further behind.
The lesson isn’t “industrial policy doesn’t work.” It’s “industrial policy without competitive discipline doesn’t work.”
Finding 4: Downstream Industries Pay Upstream Protection Costs
Every protected industry is an input to other industries. Protect steel and you raise costs for automakers. Protect semiconductors and you raise costs for computer manufacturers. The 1980s semiconductor restrictions left American computer companies with higher costs and chip shortages.
The political economy here is brutal. The protected industry’s jobs are visible and concentrated; easy to see, easy to organize. The downstream job losses are diffuse and politically invisible. No one marches on Washington for the auto workers harmed by steel tariffs.
Studies suggest every $1 spent saving steel jobs costs $5-7 in downstream industries. But the downstream costs are spread across so many companies and workers that no one can point to them.
Finding 5: Sanctions Succeed 34% of the Time; Modest Goals Only
A meta-analysis of over 200 sanctions cases from 1914 to 2000 found a 34% success rate. That sounds low, but it depends heavily on what you’re trying to achieve.
Sanctions aimed at modest policy changes have a decent track record. Sanctions aimed at regime change have roughly a 10% success rate. Cuba has been under US embargo for over 60 years with no political objectives achieved.
The successful cases share characteristics: multilateral coalitions, modest goals, targets that care about economic integration. Unilateral sanctions against isolated regimes almost never work.
VI. What This Means
So where does this leave me; and you; when listening to Howard Lutnick or anyone else make trade policy arguments?
Not nihilism. The historical patterns offer real guidance. They’re not predictions of what will happen in any specific case, but they’re base rates that any proposed policy has to overcome.
When someone proposes tariffs, I now ask:
Have they accounted for retaliation? History says the other side will hit back. Always. What’s the plan for that?
Is the goal to fix a trade deficit? If so, they’re probably using the wrong tool. Trade deficits are macroeconomic.
If this is industrial policy, is there export discipline? Protection without competition breeds inefficiency. Are beneficiaries required to compete globally?
What happens downstream? Who uses this industry’s output as an input? What happens to their costs?
If there are sanctions involved, are the goals modest? Regime change through economic pressure almost never works.
These aren’t guarantees. Any specific situation could be the exception. But the base rates are what they are. Someone claiming their policy will avoid the patterns that have held for 150 years needs to explain why this time is different.
Lutnick’s steel argument has force. The strategic dependency concern is real. But history suggests the response matters as much as the diagnosis. Tariffs invite retaliation. Protection without discipline breeds inefficiency. The downstream costs are real even if invisible.
I still don’t know what the “right” trade policy is. I’m not sure anyone does, or can. But I’ve learned to be skeptical of confident predictions; and to pay attention to historical patterns even when the confident people ignore them.