The Full Tariff-to-Welfare Chain

The subject of this series is the household cost of the 2025 U.S. tariffs, and the final question is why spending on affected goods could fall much more than retail prices rose. The most defensible answer is a chain rather than a ratio: policy changed border charges; suppliers and importers changed the mix of varieties; retail prices passed through partly and with a lag; households reduced participation in discretionary categories, protected essentials, and traded down within categories; and the resulting burden differed sharply by income. That sequence gives the reader a way to evaluate tariff claims without treating one price index as the whole story.

The chain also explains why the evidence can look contradictory when its layers are collapsed. Exporter prices can remain nearly unchanged while duty-inclusive border prices rise. A product-level import average can fall because its source mix changes. A retail price index can rise by about 1–2 percent while spending falls by roughly 4 percent. Low-income households can make a smaller dollar adjustment yet lose more relative to income. Each statement refers to a different margin, and the relationship among the margins—not any single coefficient—is the central finding.

The chain is longer than the price tag

Start with the policy variable. A statutory tariff is the rate prescribed in official measures; an effective tariff is the duty actually collected relative to import value. The two can diverge because of exemptions, implementation lags, shipment timing, and changes in what traders import. The Federal Reserve household study uses an effective HS4 tariff rate and scales it by fixed 2022 import penetration to form a consumer-exposure measure. The IMF import study uses statutory changes as an instrument for effective rates in its preferred estimates because the effective ratio can itself respond to trade behavior.

This is the first reason a headline rate cannot be carried straight to a household bill. A 10 percent tariff on imports does not mean that 10 percent is added to every domestic purchase. The relevant exposure depends partly on how much of domestic consumption relies on imports, on which products are covered, and on how the policy is implemented. The Federal Reserve measure is designed to compare more-exposed and less-exposed categories; it is not a product-by-product invoice and not a complete cost-of-living index.

At the border, the IMF study identifies another distinction. A variety is a detailed product from a particular exporter, while a product aggregates varieties from multiple source countries. For the 2025 episode, exporter duty-exclusive or FOB unit values show essentially no baseline response to tariff changes, while duty-inclusive import prices rise one-for-one with the effective tariff. An eight-percentage-point increase in the effective tariff is associated with about a 3.6 percent average decline in import values and quantities at the variety level. The border charge therefore reaches buyers without requiring exporters to lower their quoted prices enough to offset it.

But importers do not necessarily keep the same supplier mix. Higher tariffs raise the probability that existing varieties exit and reduce the probability that varieties at risk of entering appear. Among the varieties that do enter in the product-level decomposition, entrants are relatively low-priced in products facing larger tariff increases; exiting varieties tend to be relatively high-priced. The result is that a product-level before-tariff average can decline even while continuing exporters barely reprice. About 65 percent of the negative 2025 product-level response is attributed to extensive-margin changes, with entry and exit making up the dominant part of the composition shift.

This upstream change does not mean the tariff made the product intrinsically cheaper. It means that “the product” now represents a different weighted set of suppliers. The distinction resembles the difference between comparing the price of a fixed shopping list and comparing the average price of whatever remains on the shelves. The latter can fall because expensive options disappear or cheaper ones gain weight, even if no continuing seller changes its own price.

The quality question appears at this point. The IMF study estimates variety appeal from pre-tariff data in a CES framework. Appeal combines information about quality and preference heterogeneity; it is a model-based demand shifter, not a physical inspection score. When the authors adjust for appeal, the 2025 tariff coefficient on duty-exclusive product prices is about −0.357 instead of −0.642 in the unadjusted index. The nominal decline is therefore substantially smaller once the lower-appeal mix is recognized. For duty-inclusive prices, the unadjusted coefficient is small and statistically insignificant at about 0.059, while the appeal-adjusted coefficient is positive and significant at about 0.375.

That pattern means that a nominally cheaper import mix can be a poor guide to welfare. Some of the price reduction can reflect efficient movement toward lower-cost sources. Some can reflect movement toward varieties that buyers value less or that firms find less productive as inputs. The evidence establishes that appeal composition matters, but it does not fully resolve the balance between search efficiency and quality loss. Nor does it show that a particular household purchased a particular lower-appeal item.