The Price Increase Is Only the Beginning
The subject of this series is the effect of the 2025 U.S. tariffs on household purchasing and welfare. This opening focuses on a puzzle that a price-only account cannot explain: the Federal Reserve’s household study finds that prices in affected categories rose by roughly 1–2 percent, while real spending on those categories fell by about 4 percent. The useful understanding is not that one of those numbers must be wrong. It is that a tariff can reach people through several margins at once—what they pay, which goods remain available, how much they buy, and how they rearrange a limited budget.
That distinction matters whenever a policy is judged by a single inflation number. A small retail-price increase can coexist with a large loss of purchasing activity. A lower average import price can coexist with a more expensive, lower-appeal mix of goods. A household that cuts spending may be protecting essentials rather than consuming less uniformly. To understand the 2025 episode, we therefore need to follow the tariff from the border to the shopping basket, while keeping each layer of evidence in its proper place.
The household question behind a tariff
A tariff is charged when an imported good enters the country. But the customs charge is only the first location at which its economic effects appear. The importer may pay the duty directly, an exporter may alter a quoted price or shipment, a retailer may change a margin or delay a repricing, and a household may switch products or stop buying a discretionary category. “Who pays?” is consequently a question about incidence across a chain, not merely about whose name appears on a customs form.
The Federal Reserve study is unusually suited to the household part of that question because it observes prices and quantities together. It follows 126,448 households that remained active from January 2024 through December 2025, recording purchases across physical-goods categories such as groceries, apparel, home goods, electronics, and health and beauty. The panel contains about $3.8 billion in expenditure across more than 29 million items. That scale does not make the sample a complete national consumption account: housing, utilities, financial services, most services, and vehicle purchases are outside its main coverage, and the resulting basket represents roughly 20 percent of the PCE and CPI consumption baskets. Its value is different. It captures frequent purchases that households notice and that are plausibly exposed to traded-goods costs.
The study links each retail category to a measure of tariff exposure. The measure combines the effective tariff rate with import penetration, so a tariff receives more weight where foreign sources supply a larger share of domestic consumption. This is not a household-specific bill, and it is not a claim that every item in a category carries the announced tariff. It is a way to compare categories that should have felt more or less of the shock. In the benchmark estimates, a one-percentage-point increase in exposure is associated with about a 0.20-percentage-point increase in the retail price index. At the average change in exposure during the sample, that translates into about a 1.2 percent price increase.
The same calculation implies a real-spending decline of about 3.9 percent. Spending and quantity coefficients are close because spending is deflated by the PCE goods price index: most of the measured contraction is associated with fewer units or less participation, not simply with paying more for exactly the same basket. These are differential estimates for covered categories before and after the tariff changes, not a forecast that every household’s total consumption fell by 4 percent. Even with that qualification, the gap is large enough to change the question. The issue is no longer whether tariffs raised prices. It is why households adjusted their purchasing more strongly than the observed retail-price movement would suggest.
One tariff, several places where cost can appear
The answer becomes clearer when “price” is treated as a sequence of related but different objects. An announced statutory tariff is the policy written into official measures. An effective tariff is what customs data show being collected after exemptions, implementation lags, shipment timing, and behavioral responses. An exporter’s price before duties is different from the duty-inclusive price at the border. The retail price is another step downstream. Finally, the price a household pays is only one part of the household’s cost: the household may also buy fewer units, accept a less preferred variety, search longer, or give up a non-essential purchase.
The import study by JaeBin Ahn, Lorenzo Rotunno, and Michele Ruta examines the upstream part of this chain using U.S. Census data at the exporter–product level. The Federal Reserve study examines the downstream part using household transactions and survey responses. Their evidence can be connected because both ask how the 2025 tariffs changed economic choices, but their coefficients cannot simply be added or multiplied. One study follows varieties and source countries; the other follows households and retail categories. Treating them as one dataset would erase the very distinctions that make the overall mechanism understandable.
The diagram is a map of possible transmission, not a claim that every arrow has the same estimated strength. Import dependence changes how much a tariff matters to a category, while supplier decisions can change which variety represents the category’s average price. Retailers may pass through only part of the border cost and may do so with a lag. Households then face a choice set that is not fixed: they can buy fewer units, shift from a non-essential category toward essentials, or choose a cheaper variety while continuing to buy the category. The resulting welfare effect depends on income and on what has been lost, not just on the percentage printed on a price tag.
This also explains why apparently opposing observations can be true at the same time. At a detailed exporter–product level, a tariff may leave the exporter’s pre-duty price nearly unchanged while raising the duty-inclusive border price. At a broader product level, the measured average may fall if higher-priced or more heavily taxed sources lose share and lower-priced sources enter or gain share. Downstream, the retail price may rise less than the border price because the supply chain absorbs part of the change or adjusts its timing. None of these statements cancels another; they refer to different stages and different averages.
A better question than “did tariffs raise prices?”
The more useful question is: through which margin did the tariff become a cost for this household, and how much room did the household have to respond? A family with little discretionary spending may continue buying necessities at higher prices. A household with more slack may postpone or abandon non-essential purchases. A worried shopper may keep the number of units unchanged but trade down within a category. These choices can produce similar spending totals with very different implications for comfort, variety, and welfare.
The Federal Reserve evidence makes the distributional issue visible even before the behavioral details are considered. In its income-group estimates, price pass-through is highest for low-income households and lowest for high-income households, while total spending responses are broadly similar. The apparent similarity therefore hides different burdens. A dollar of extra cost consumes a larger share of a low-income household’s resources, while a larger-income household may be able to absorb a spending cut by giving up a discretionary purchase. The dollar size of an adjustment is not the same as its hardship.
Nor is the relevant cost exhausted by realized price pass-through. The Federal Reserve’s price-anchored welfare measures capture the burden associated with differential price changes, while the larger net-spending response also reflects behavior such as precautionary reallocation. The IMF evidence adds a related warning upstream: a lower nominal import price may partly reflect movement toward varieties with lower appeal, which can matter for consumers and for firms using imported inputs. These are not interchangeable welfare estimates, but they point to the same interpretive discipline. A policy can look modest in one price index while imposing costs through quantities, composition, quality, or unequal capacity to adjust.
The evidence also has a defined perimeter. The household panel covers a selected portion of goods spending and follows a short period around a volatile policy episode. The import analysis uses different data, definitions, and model assumptions. Both studies are preliminary research products. Their findings are strong enough to reject a simple story in which tariffs merely raise retail prices by a fixed amount, but not strong enough to settle the long-run national welfare effect or to identify every household’s exact supplier and product choice.
Before deciding who paid, then, we must establish which number describes which layer. The announced tariff, the duty collected at the border, the exporter’s price, the product-level import average, the retail price, the quantity purchased, and the household’s welfare loss are related—but they are not synonyms. Keeping them separate is the first step toward explaining why a tariff can raise prices modestly and still make households buy less.