America quadrupled its tariffs. The imports didn't stop — they moved.

Between January and October 2025, the average tariff on everything entering the United States went from 2.44% to 11.09% — more than four times higher in nine months. Customs collections went from $97 billion a year to $364 billion. Both numbers come from the same government accounts that measure GDP, and neither is a projection.

What did all that money buy? Not the outcome most people assume. America did not stop importing — 2025 set a record for goods imports, and 2026 is running higher still. What changed is where the containers come from. This is the story of a wall that redirected traffic rather than stopping it, told with the receipts.

Start here: what a tariff actually is

A tariff is a tax on a product at the moment it crosses the border. Three things about it surprise almost everyone:

• The foreign company does not pay it. The US importer — the American business bringing the goods in — writes the cheque to US Customs. That is why the money shows up in US federal receipts.
• It is a percentage of the invoice value, so a 10% tariff on a $30 pair of shoes is $3, collected before the shoes ever reach a shelf.
• Who ultimately bears it is a separate question. The importer can absorb it, push it onto the shopper, or force the foreign supplier to cut their price. In practice it is some mixture, and economists argue about the proportions. What is not in dispute is the total collected.

So the cleanest measure of “how tariffed are we?” is simple division: duties collected ÷ value of goods imported. No policy documents, no announcements, no politics — just what the Treasury took in against what came through the door. Here is that number, quarter by quarter.

One nuance worth understanding, because it explains the shape of everything that follows: the rate has been falling since late 2025 — from 11.09% to 8.76% — without any tariff being cut. It fell because the denominator grew and the mix shifted. Buyers moved toward goods and countries carrying lower rates, which mechanically drags the average down. A falling average tariff rate can mean tariffs got cheaper, or it can mean importers got better at avoiding them. Here, it is mostly the second.

China: hit first, hit hardest

China is the clearest case in the data, and the numbers are blunt. In 2018 the US bought $538 billion of Chinese goods. In 2025 it bought $309 billion — a drop of $230 billion, or 43%, from the peak. The single-year fall from 2024 to 2025 was 30%.

Share of the whole basket tells the same story more sharply. In 2017, 21.4 cents of every dollar America spent on foreign goods went to China. In 2025 it was 8.9 cents. Through the first seven months of 2026 it is 7.4 cents.

The tariffs did not bring the factories home. They moved the address on the container.

The beat-the-clock month you can see in the data

Zoom into months instead of years and something very human appears. January 2025 was the single biggest month for Chinese imports in the series — $41.8 billion, more than half again as much as a typical month today. Importers were racing to land goods before duties rose. Five months later, imports had collapsed to $18.9 billion: a 55% fall from January.

India’s chart has the same shape, and this is the part that gets reported wrongly. India had a record year in 2025 — but not a record second half. Monthly imports peaked at $11.2 billion in March 2025, slid to $7.1 billion by November (a 36% fall), and have since recovered to $9.6 billion in July 2026, the highest monthly figure in more than a year.

The wall moved the traffic. It did not stop it.

Here is the fact that undercuts almost every confident claim made about tariffs in either direction: total US goods imports set a record in 2025, at $3.45 trillion. Through July, 2026 is running at an annualised $3.63 trillion — higher again. The most heavily tariffed stretch in the ten years these series cover is also a record stretch for imports.

That combination only makes sense if three things happened at once:

• Substitution, not abstinence. Buyers switched countries rather than cancelling orders. China’s lost 12.5 points of market share went somewhere, and only 0.9 of it went to India.
• Inventory timing. The record Q1 2025 import quarter ($3.97 trillion annualised, far above trend) was goods pulled forward to beat the deadline — a cost that lands in one quarter and flatters the next.
• Price, not volume. These are dollar figures, not container counts. If tariffs raise landed costs, import values can rise while the quantity of stuff falls. The trade data cannot tell you which of those you are looking at, and anyone claiming otherwise from these series alone is guessing.

The export side is worth a glance too, because tariffs invite retaliation. US goods exports to China fell from $130.0 billion in 2017 to $106.0 billion in 2025, down 18%. Exports to India went the other way: $25.6 billion to $45.4 billion, up 77%.

What this means for your money

Somebody pays $326 billion a year. Spread across the 134.8 million US households the Census counts, the current collection rate works out to about $2,420 per household per year — up from roughly $720 at the start of 2025. Treat that as a scale check, not a bill: businesses pay the duty, and how much reaches shelf prices versus supplier margins versus corporate profits is genuinely contested. Some of it lands on business inputs you never buy directly.

It shows up unevenly. An average tariff of 8.76% is an average. Categories carrying high rates — some electronics, steel-intensive goods, certain consumer imports — moved far more than your grocery bill. When you read that “tariffs added X% to prices”, ask which basket.

It is a live variable for the Fed. Tariffs push prices up once, which looks like inflation for about a year, then drops out of the annual comparison. Telling a one-off level shift from genuine ongoing inflation is the single hardest judgement the Fed faces right now — which is the subject of our note on the October meeting.

For investors, the signal is dispersion. Companies that import components with thin margins and companies that sell domestically made substitutes are on opposite sides of the same policy. That kind of split shows up as sector-level volatility rather than index-level volatility — the same pattern documented in the bond-market note.

Data appendix

Every figure cited above with the period it belongs to. All series are public and free: the FRED codes below return exactly these numbers.

Pulled from FRED (Federal Reserve Bank of St. Louis) on September 26, 2026. The effective tariff rate is our own arithmetic — quarterly customs duties at an annual rate, divided by goods imports over the same quarter annualised — so you can reproduce or dispute it with two public series. It is an average across everything imported, not the tariff rate on any single product or country.