Three Prices Between a Tariff and a Shopping Basket
The subject here is the path the 2025 U.S. tariffs took from a policy announcement to the prices and choices faced by households. The specific problem is that “the tariff rate” and “the price increase” describe different points in that path. Distinguishing those points makes the evidence useful: the reader can tell what an announced rate, a collected duty, an exporter’s price, a duty-inclusive import price, and a retail price each mean—and why none of them, by itself, measures the full household burden.
The distinction is not academic bookkeeping. A tariff may be written as a percentage of an imported good’s value, but a household usually encounters it only after the good has moved through customs, sourcing decisions, distribution, and retail pricing. At each stage, the relevant comparison changes. The 2025 evidence becomes much easier to interpret once those comparisons are kept separate.
Why the announced rate is not the exposure a household feels
A statutory tariff is the rate prescribed by policy. It is the number that appears in an official schedule or legal measure. An effective tariff is the duty actually collected relative to the value of imports. Those rates can differ. Exemptions may apply, implementation may lag, shipments may arrive under earlier arrangements, and traders may change what they import. The effective rate is therefore closer to the charge realized at the border, while the statutory rate is closer to the policy variation that officials intended to create.
That difference matters for two reasons. First, a policy can be large on paper but smaller in the customs data because not every shipment is treated identically. Second, the effective rate can itself respond to trade behavior. If import values change while collected duties do not move proportionally, the ratio of duties to import value changes partly because traders changed their behavior. The IMF study treats this as an endogeneity problem: its preferred specifications use statutory tariff changes as an instrument for effective tariff changes, so that the estimated response is anchored to policy-driven variation rather than to the ratio alone.
The Federal Reserve household study faces a related but different measurement problem. It needs to compare retail categories that were more exposed to the tariff shock with categories that were less exposed. For that purpose, it constructs an effective tariff rate at the four-digit Harmonized System level by dividing collected duties by import value. It then multiplies that rate by import penetration, using a fixed measure based on 2022 trade and domestic shipments. In plain language, the treatment is larger when a category depends more heavily on imports.
This adjustment is important because the same customs rate can imply different consumer exposure. Consider two categories that each face a 10 percent tariff on their imports. If foreign sources supply nearly all of domestic consumption in one category, the tariff reaches a large share of the goods households can buy. If domestic production supplies most of the other category, the same import tariff is applied to a much narrower part of the consumption base. The Federal Reserve paper’s illustrative calculation gives electronics with import penetration near 0.9 an exposure of about 9 percent, compared with about 1 percent for a category with import penetration near 0.1. The point is not that those examples predict a particular store price. It is that the denominator changes the meaning of the rate.
The resulting measure is a treatment intensity, not a line item on a household bill. The study maps retail classifications to HS4 trade categories through a hierarchy rather than matching every retail item to a ten-digit customs code. That choice makes broad coverage possible, but it also means that the measure describes expected category exposure, not the tariff content of each individual product. Import penetration is held at its 2022 level, so the measure does not automatically update as sourcing changes during the tariff episode.
This is why the Federal Reserve reports both a tariff-rate specification and a tariff-exposure specification. In the first, a one-percentage-point increase in the realized tariff rate is associated with roughly a 0.15-percentage-point increase in retail prices. In the benchmark exposure specification, which gives more weight to import-dependent categories, the coefficient is about 0.20. These are pass-through coefficients in a comparison across categories and time. They do not mean that a 15 or 20 percent tariff appeared on every shelf, or that every household experienced the same increase.
They also answer slightly different questions. The tariff-rate estimate asks how retail prices move with the customs rate when categories are treated more evenly. The exposure estimate asks how prices move where the tariff is more likely to touch the goods consumed domestically. For the household-incidence question, the second is conceptually more direct. For comparisons with other tariff research, the first has a more familiar pass-through interpretation. Neither should be detached from its treatment definition.
From exporter price to retail shelf
The price that an exporter quotes before U.S. duties is not the price an importer pays after customs. The IMF study calls the former a duty-exclusive or FOB price. It constructs the latter by adding collected duties to the import value and quantity. At the exporter–product variety level, the study finds that the 2025 tariff changes produced essentially no response in the duty-exclusive unit value in its baseline estimates. The duty-inclusive price, by contrast, rises one-for-one with the effective tariff.
A simple illustration shows what that means. Suppose an exporter’s pre-duty price is $100 and a 10 percent duty is collected at the border. If the exporter continues to quote $100, the duty-inclusive import price becomes $110. The exporter has not absorbed the tariff through a lower FOB price; the buyer at the border faces the added charge. This is only a mechanical illustration, not an estimate of a particular product. Its purpose is to separate exporter incidence from border incidence.
The IMF result is therefore not in conflict with the Federal Reserve’s smaller retail pass-through estimate. “One-for-one” belongs to the duty-inclusive price of a detailed imported variety. “About 15–20 percent” belongs to the relationship between tariff exposure and retail price indexes across household categories. Between those observations lie sourcing, aggregation, domestic production, distribution, and the way retailers adjust prices. A border charge can rise fully while the eventual retail-price response is partial.
The difference between a variety and a category is especially important. A variety in the import data is an exporter–HS10 product combination: the same detailed product from a particular source country. A Federal Reserve panel unit is a retail hierarchy linked to an HS4 trade category. The IMF can ask whether one source’s import unit value changed relative to another source’s price for the same detailed product. The Federal Reserve can ask whether the price index for a household-facing category changed more after its measured exposure increased. These are different comparisons, not competing versions of the same statistic.
The IMF study also finds that higher tariffs reduce import values and quantities at the variety level. An eight-percentage-point increase in the effective tariff—similar to the average increase between February and December 2025—implies an average import decline of about 3.6 percent. That result helps explain why price incidence cannot be inferred from exporter pricing alone. If a source keeps its pre-duty price but loses sales, the adjustment may be occurring through trade flows rather than through a lower quoted price.
The sequence can be represented as follows:
The arrows show a possible transmission order, but they should not be read as equal-sized effects. The evidence says something specific at each point: statutory and effective rates can diverge; exporter FOB prices were muted in the IMF baseline; duty-inclusive variety prices rose with the tariff; and retail prices rose only partially and gradually in the Federal Reserve estimates. The diagram does not claim that every retailer follows the same sequence or that the two studies observe the same transaction.
The timing of retail adjustment reinforces the need for separate layers. The Federal Reserve event-study estimates show little systematic price movement before the tariff implementation, followed by positive price coefficients that rise gradually after February 2025. The authors interpret this as consistent with lagged and incomplete pass-through. The relevant price response is therefore not necessarily visible in the first month a policy is announced. A retailer may be comparing goods purchased under different border conditions, and a category index may combine products whose prices adjust at different times. The data establish the timing pattern, but they do not identify one operational cause for every delay.
The same event study shows the spending and quantity response becoming sharply negative later, around June and July, and remaining negative through the end of the sample. That timing does not mean the retail price response caused every later purchase decision in a simple one-step fashion. It means the policy shock has a temporal structure: border charges, retail prices, and household behavior do not move as if they were one instantaneous variable. The next interpretive question is therefore not merely how much of the tariff reached the shelf, but how the imported goods behind that shelf changed.
There is also a limit to what “retail pass-through” can tell us. The Federal Reserve’s price index covers the physical-goods categories represented in the transaction panel, not housing, utilities, most services, financial services, or the full healthcare budget. Its estimated coefficient is identified from differences in exposure across mapped categories, with time and category-by-calendar-month controls and price-specific pre-trend adjustments in the benchmark. That design helps distinguish the tariff episode from common time shocks, but it does not turn the coefficient into a universal rule for all consumer prices.
A price index also compresses a changing choice set into one number. If the available sources or products change, an average retail price may reflect not only what incumbent sellers charge but also which varieties remain for sale. That issue is measured explicitly in the import evidence and becomes important for interpreting a lower product-level average. A category can appear to have modest price growth while consumers face a different mix of suppliers or products. Conversely, a household can pay a lower unit price only by accepting a less preferred variety or by giving up a purchase altogether.
For the household question, then, the most defensible reading is layered. The policy rate is an instruction. The effective rate is a realized customs ratio. Import penetration scales the part of domestic consumption likely to be touched by that ratio. The exporter’s FOB price records what happens before U.S. duties. The duty-inclusive import price records the border cost after duties. The retail index records only the portion that appears in observed store and online prices, and it may appear with a lag. The household’s eventual cost includes the decisions made in response to all of those conditions.
That vocabulary resolves the first apparent contradiction without resolving the whole story. Muted exporter repricing does not mean the tariff disappeared. Full duty-inclusive pass-through at the border does not imply a one-for-one retail increase. Partial retail pass-through does not imply that households were largely unaffected. To see why spending could contract much more than prices rose, we next have to examine a margin that a single price index cannot show: the changing mix of suppliers behind the average product.