Which states are most exposed to tariff increases? A new measure based on the consumption channel

A new Federal Reserve FEDS Note by Nick Heyman, Colin J. Hottman, and Ryan Monarch examines how exposed different U.S. states are to tariff increases through the consumption channel. The authors construct a state-level measure of direct consumer goods imports using 2023 data, building on earlier work by Hottman and Monarch that measured import exposure for U.S. income deciles and demographic groups. The finding is stark: import shares of consumption vary substantially across states, with California at 13.8 percent and New Jersey at 12.7 percent on the high end, while Nebraska and Missouri sit at just 2.6 percent each. This heterogeneity matters because tariffs on imported consumer goods would translate into very different cost-of-living increases depending on where a household lives.

The methodology combines two datasets. First, the authors use the Consumer Expenditure Survey (CES) for 2023 to generate average annual expenditure for representative households in 31 U.S. states, covering 86 percent of the U.S. population. Second, they use state-level import data for 2023 from the U.S. Census Bureau, filtered to consumer goods using the United Nations Statistics Division's Classification of Broad Economic Categories. The authors concord product categories between the two datasets using an LLM-based approach starting from the concordance in Furman et al. (2017). The resulting measure shows a simple average import share of consumption of 5.6 percent across the 31 states, with a population-weighted average of 7.1 percent. The authors note these figures are lower than prior national estimates of around 10 percent, largely because earlier work relied on a proportionality assumption that combined national import penetration rates with category-level consumption differences, whereas the new measure allows both import and consumption data to vary directly across states.

The authors investigate what drives the large cross-state variation. On measurement, they note the CES is only designed to be nationally representative, so they include only states with more than 300 responding households, with 80 percent of the 31 included states having more than 500 households. They also scale CES spending within each state to match the Bureau of Economic Analysis's Personal Consumption Expenditures by state totals for 2023, and they drop product codes appearing in fewer than two-thirds of included states. On substantive drivers, the authors find that differences in the mix of products explain much of the variation. Items such as cellphones explain about 24 percent of the variance in state-level import shares, while categories like household equipment each explain about 14 percent. Collectively, those three categories alone explain more than 50 percent of the variance. Differences in source countries also matter: China explains the largest share of variance in import shares, the EU explains about 19 percent, and Mexico about 15 percent, with those three countries collectively explaining almost 60 percent of the variance. By contrast, differences in spending from Canada explain only about 2 percent of the variation.

The authors translate these import shares into a back-of-the-envelope tariff incidence. They find that a tariff increase would produce a cost-of-living increase of about 1.4 percent in the most exposed state and about 0.8 percent in Pennsylvania, while Nebraska would experience an increase of only about 0.3 percent. This implies large differences in the real income effects of tariffs across states, with the burden falling most heavily on states that consume a higher share of directly imported consumer goods. For markets and policymakers, the implication is that a uniform tariff policy would have geographically uneven effects, potentially amplifying regional disparities in consumer purchasing power even if the aggregate national impact appears modest.

The analysis has clear limitations. The measure covers only direct consumer goods imports, not intermediate goods that feed into domestically produced consumer products, so it may understate total tariff exposure through the consumption channel. The state-level CES estimates carry sampling uncertainty despite the authors' mitigation efforts, and the import data are measured at wholesale prices while consumption is measured at retail prices, which the authors acknowledge likely lowers the computed import shares relative to prior household-level estimates. The note also does not model behavioral responses, substitution away from imported goods, or general equilibrium effects. What to watch is whether future work extends this state-level measure to include indirect import exposure through domestic supply chains, and whether the geographic pattern of tariff incidence shifts as trade flows and sourcing countries evolve.