What Tariffs Change and What They Do Not

What a tariff is and who pays it

A tariff is a tax levied on goods as they cross national boundaries, usually by the government of the importing country. The legal question of who pays at the border is separate from the economic question of who bears the cost. The foreign exporter can absorb it through a lower price, the importer or retailer through a thinner margin, or the final buyer through a higher shelf price. We use the word pass-through for the share of the tax that ends up in a price.

Recent evidence is that foreign exporters absorbed little of the 2025 and 2026 United States tariffs. In an analysis published on 6 October 2026, New York Fed economists and a Columbia University colleague estimate that pass-through to US import prices is about 90 percent, with import prices rising almost one for one with tariffs in the first month. A July 2026 note from the same bank cites other research that nearly 90 percent of the economic burden has fallen on US firms and consumers.

What tariffs change

Tariffs change prices, and not only the prices of imports. An NBER working paper on the 2025 and 2026 tariffs finds that about 26 percent of the tariff increase passes through to consumer prices. The direct effect, on foreign varieties of a good, is 64 percent of that. The remaining 36 percent is indirect: domestic producers face higher input costs, and face less competition from imports, so they raise their own prices. The direct effect arrives quickly, while the indirect effect takes nine to twelve months.

The authors summarize the result as a rule of thumb. Consumer goods prices rise about 0.25 percent for each percentage point of average tariff after one year, so a 10 percent tariff on all imports would raise them about 2.6 percent. By February 2026 the authors estimate that tariffs had added 2.9 percentage points to goods price inflation, and that without them goods prices would have fallen slightly.

Pass-through also takes time inside firms. In a New York Fed regional survey from May 2026, nearly half of firms that had paid tariffs still planned further price increases, some because fixed-price contracts had not yet expired. As a measure of scale, the Yale Budget Lab put the average effective US tariff rate at about 11.8 percent on 8 April 2026.

A worked example

The numbers in this section are illustrative, not real data. Suppose an imported item has a customs value of $100 and sells in a shop for $200. A 10 percent tariff is introduced. If the exporter changes nothing, the duty is $10. Now suppose the exporter cuts its price by $1, to $99. The duty is 10 percent of $99, which is $9.90, so the importer's landed cost is $108.90. That is $8.90 more, or 8.9 percent. Pass-through to the importer's cost is 8.90 / 10.00, or 89 percent, close to the 90 percent estimate above.

At the shelf, the New York Fed authors report that a 10 percent rise in import and producer prices raises retail prices by 5.6 percent, because distribution costs are about half of what consumers pay. Applying that ratio, 8.9 percent x 0.56 is about 5.0 percent, or roughly $10 on the $200 item. The consumer's extra cost is close to the headline duty. A reader who sees only the 10 percent figure would expect the shelf price to rise by $20, which is the common mistake of treating the tariff rate as the price change.

What the evidence does not show

Estimates differ, and we think that is the honest state of knowledge. In an April 2026 note, Minneapolis Fed economists attribute 0.5 percentage points of core PCE inflation to tariffs, at the lower end of other estimates of 0.5 to 1 point. They also find that categories with the largest 2025 inflation contributions mostly faced low tariffs. Fortune reported in May 2026 a Dallas Fed estimate that core inflation would have been 0.80 points lower without tariffs. We have not read the Dallas Fed study itself, so we treat that figure as reported.

The New York Fed estimates also carry limits. They compare more exposed and less exposed goods, so they cannot say how much of broader price movements tariffs caused. They exclude services, about two thirds of the consumer basket, and hold exchange rates, wages, demand and monetary policy fixed. Their forecast assumes tariffs stay at end-September 2026 levels, apart from one announced increase.

A second common mistake is confusing the price level with the inflation rate. A one-time tariff raises the level of prices and lifts measured inflation only while the step works through. The same authors projected that the tariff contribution to twelve-month inflation would fall to around zero by August 2026, while the price level stayed higher, near 2 percent above where it would have been. That was a model projection from data through February 2026, so it should be checked against later releases.

Going deeper

The arithmetic above reduces to a few relationships. All are simplifications, and the coefficients are estimates for one period and one set of goods.

1. Pass-through = Percentage change in price / Percentage change in tariff-inclusive cost.
2. Consumer goods price effect after one year = Tariff increase in percentage points x 0.25 (New York Fed estimate; the reported 2.6 percent for 10 points reflects rounding of the coefficient).
3. Retail price change = Change in import and producer prices x 0.56.
4. Dollar pass-through = Extra dollars paid by consumers / Tariff dollars paid.

The fourth matters because percentages can mislead. In a study of a 25 percent tariff on certain European still wines, retail prices rose about 6.9 percent, which looks modest, yet consumers paid about $1.59 more per bottle against $1.19 in tariffs. Dollar pass-through was therefore 1.59 / 1.19, about 134 percent, because margins applied along the chain amplified the dollar cost.

To check any of this yourself, start with the data the authors used: consumer prices for 67 non-oil goods categories from the Bureau of Labor Statistics, trade data from the Census Bureau, and tariff data from the US International Trade Commission. We may have misread these papers, and we welcome scrutiny of every figure and link above.