The Tariff Trap

Start with the receipt, because nobody disputes a receipt. In the first ten months of fiscal 2026 the United States collected $154.5 billion in customs duties. In the same ten months of 2024 it collected $62.7 billion. That is about $92 billion more, an increase of 146 percent.

And it shows up in the Treasury’s own books on the money-coming-in side, in the same column as the income tax and the payroll tax. Which is the first thing worth saying plainly. Whatever else a tariff is, it is a tax. The effective rate is now 11.8 percent, the highest since the early 1940s.

There is a real argument about who ends up paying it, foreign sellers or American buyers, and honest people land in different places depending on the product. But nobody argues about where it gets collected, or that the money is real. The Budget Lab’s read is that the adjustment comes through prices rather than wages, which works out to the price level rising something like 0.5 to 0.7 percent, and the average household losing somewhere between $760 and $940 a year.

That is where most analysis stops. It is where this one starts, because the average household is exactly the thing the Toilet Paper Theory says will fool you.

Counting the wrong thing

Take the published numbers exactly as they are. In today’s dollars, the tariff costs a household in the poorest tenth about $517 a year. It costs a household in the richest tenth about $2,175. Read that quickly and it sounds like the rich are paying four times more, which is the version that gets quoted when somebody wants to defend a tariff.

Now measure it as a share of what each family actually has. It flips. The poorest tenth loses 1.1 percent of its income. The richest tenth loses 0.4 percent. Three times heavier at the bottom. This is the standard regressive-tax finding, it is correct, and it is what an economist will tell you.

The Toilet Paper Theory says both of those numbers are too small, and for a reason that is built into how they are made.

A tariff taxes things you buy. The rich household pays its $2,175 and changes nothing about how it lives, because the tax lands on a family already at the ceiling for necessities with plenty of room above it. The money comes out of savings. Every dollar of it gets recorded.

The poorest tenth cannot do that. There is no room above its necessities. So part of the tariff never gets paid in dollars at all. It gets paid by not buying the thing.

Which makes a tariff regressive in a way the regressive statistic cannot capture. The published burden at the bottom is capped by what the family could afford to spend in the first place. It measures spending, and at the bottom of the income scale, spending is a measurement of constraint. The tables are not wrong. They are answering a different question, and the gap between the two questions is the entire subject of the Toilet Paper Theory.

A shortage somebody decided to have

The treatise asks one question: has the thing actually run out? Here the answer is unusually easy, because it is a matter of public record rather than inference.

Nothing has run out. The shirt that costs more is being sewn at the same rate in the same factory. The ships still sail and the ports still work. There was no drought, no strike, no closed strait, no seam that ran dry. What changed was a schedule of rates. Somebody put a wedge between a buyer who wants a thing and a seller who has it. The thing is not scarcer. The claim on it is.

That is this framework’s definition of a fake shortage, and it is worth pausing on how unusual this example is. Every other one the treatise examines is an accident. A rush for cash. A credit squeeze. A lender pulling out at the worst moment. Nobody designs those. This one was designed, debated and voted on, and the result is precisely the condition the treatise spends its length warning about.

That matters for more than the politics of it. The treatise says a fake shortage left standing long enough stops being fake, and starts wrecking the very capacity it was only impersonating a shortage of. A shortage somebody meant to create runs the same clock as an accidental one. It does not get a discount for having been intended.

The trap itself

Now put the two halves together, because it is the collision between them that earns the title.

Nobody defends a tariff by saying it raises money. They defend it by saying it moves demand to American producers, and that this calls American factories into being. Take that seriously for a moment and ask what it requires. It requires that the demand which stops going abroad turns up at a factory here instead.

The Toilet Paper Theory says a good chunk of that demand does not turn up anywhere at all. A family at the bottom does not switch to the American version at the higher price. It buys less, or it buys none. So the tariff moves some demand, destroys some demand, and the part it destroys is concentrated in exactly the cheap end of the market that a new American factory would need in order to be worth building.

The second jaw is time, and another piece here already measured it. American factories arrive on a timescale of years. A TNT plant lands in 2029. Submarines run four years behind their contracts. A master welder is thirty years of accumulated practice.

Now look at the legal instrument carrying much of this tariff. After the Supreme Court ruled the President could not impose the reciprocal tariffs under emergency powers, the replacement was Section 122 of the Trade Act of 1974. It allows duties without an investigation. It is capped at 15 percent. And it lasts 150 days.

Nobody breaks ground on a factory because of a price that expires in five months. The tariff charges its cost on a timescale of days and asks for an answer on a timescale of years, and the law that lets it exist runs out long before the concrete on the plant it was meant to summon has finished curing. The Budget Lab expects the effective rate to drop to 9.7 percent when Section 122 lapses, which tells you how much of the current wall is standing on a timer.

The jaw that closes on the government

There is one more turn, and this one catches the people who set the thing up.

Customs duties are now running at $154.5 billion over ten months, inside a budget with a $1.799 trillion hole in it over the same period. That money has become load-bearing. A tax introduced as a bargaining chip is now a line the Treasury is counting on, which is the ordinary way temporary taxes become permanent ones.

So the government ends up somewhere awkward. The tariff is hard to drop because the revenue is needed, and the revenue is needed partly because the tariff suppressed the activity that would have generated revenue through income and payroll taxes instead. A regressive tax on everyday goods that has become fiscally necessary is the most durable kind there is, and it is durable in inverse proportion to who can afford it.

Nobody is counting this either

Five times in three months this framework has hit the same wall, and at this point the wall is the finding. The gauge reads loose while subprime borrowers post a thirty-two year record. Corporate profits and small-business health pull apart with nothing able to reconcile them. The federal survey of who is going hungry is being switched off. Unemployment and average hours cannot see a person working two jobs to cover one rent. And here the standard tariff table measures dollars paid, which is a measurement of what a family could still afford, and therefore reports the smallest burden for the families priced out completely.

It has happened often enough to state as a rule. Every instrument built to sum up an economy is built out of transactions. A person who cannot transact does not appear in it. That is not a flaw in any particular index. That is what an average of dollars is.

What would prove this wrong

Two warnings. Who really bears a tariff is genuinely contested: the share carried by foreign sellers rather than American buyers varies by product, and any estimate assuming buyers pay nearly all of it will overstate the damage for some categories. This piece leans on the Budget Lab’s assumption that the adjustment comes through prices, which is a defensible reading of the evidence and is still a reading. Second, a tariff can be the right tool for reasons this framework does not price. Depending on an adversary for something critical is a real problem, and the argument here is about how the thing works and who pays, not about whether any particular protection is warranted.

What this framework does say is that the case for a tariff and the effect of a tariff are separated by the distribution. The case gets made in big numbers: a trade balance, an industry, a negotiating position. The effect lands on families who were already at their limit, and it lands as things not bought rather than as prices paid, which is the one form the big numbers cannot see. The shelves stay full. The ships keep sailing. The factories abroad keep running. And somebody buys less of something ordinary, on purpose, by design, and appears nowhere in any account as having done so.