Chicago Booth economists estimate U.S. tariffs moved earnings from services into tradable industries

The inference runs backwards. You observe a tariff schedule, you estimate how imports respond to it, and then you ask what set of social values would make that schedule the one a policymaker would choose. Services is the comparison group, because its output is not taxed at the border ^[4]. Twenty-one tradable industries sit on the other side of that line ^[4]. The team estimated how tariffs affected imported products, built a theoretical model on those estimates, and checked it against real data, where it reproduced the actual effects of the Trump administration's 2018 tariff policy changes ^[7].

That check validates the model. It does not test the premise. The premise is that the change in real imports between 2017 and 2019 came from policymakers wanting to favor some constituencies over others, and that fundamental changes in the U.S. economy did not drive it ^[8]. Whatever the model does not assign to economics, it assigns to preference. "Society values transfers to some individuals much more than others," Adao said ^[9].

"U.S. tariffs end up representing a transfer from those employed in services to those employed in tradable sectors," Adao said ^[5]. The size of that transfer is where the paper gets concrete. Without the redistribution, people at the top of the estimated gain distribution would have done worse, by $538 at the 90th percentile, $1,727 at the 95th and $2,861 at the 99th ^[11]. At the 10th percentile the sign flips: that person would have earned $621 more ^[12]. The distance from one end to the other is $3,482 in real earnings ^[18]. The percentiles rank people by how much tariffs moved their earnings. They are not a ranking of income.

The favoritism the model finds is sectoral. Income in apparel, metals and vehicles came out worth 140% more to policymakers than the U.S. average ^[13]. The three states named as above-average beneficiaries, Florida, Vermont and Wyoming, clear the average by 7% ^[14], so the sectoral premium is twenty times the state premium ^[19]. "The variation in tariffs across goods creates larger or smaller gains across workers employed in different tradable sectors," Adao said ^[6]. The data are assembled state by state, across all 50 states and the District of Columbia ^[3].

Lobbying is consistent with the estimates and accounts for about one-fifth of the overall variation in gains and losses across sectors and regions ^[15]. That leaves roughly four-fifths of that variation unaccounted for by lobbying ^[20]. Individuals from swing states did not appear to receive significantly higher benefits. The researchers treat that null as a problem for the view that courting politically potent voters drives U.S. tariff policy ^[16].

What these numbers cover is the earnings channel, by sector and region, over a two-year window ^[3]. The working paper carrying them, "Why Is Trade Not Free? A Revealed Preference Approach," is dated June 2026 ^[17].