ÃÛ¶¹ÊÓƵ

That is almost exactly the dollar amount of reciprocal tariffs proposed on April 2, which were then subsequently paused for 90 days (all countries at 10%) and to which tech product exemptions were added last week. The escalating tariffs on China have fully offset the tariff relief from those other items. Interestingly, China was the primary beneficiary of the tech exemptions: it accounted for 25% of the USD 400bn in exempted tech products, but 81% of the reduced tariff revenue for the US, as China was being tariffed at 145% and everyone else at 10%. Also note that, with the lower tariff rates on autos/steel/aluminum and the tech exemptions, the weighted average tariff China faces is now 97%, not 145%.

Although the headline tariff amount is largely unchanged vs April 2 (a 24% effective tariff rate), the compositional shift towards China likely greatly reduces the tariff revenue collected. Our standard rule of thumb is that in the first year of the tariff, demand falls by the same amount ('unit elasticity'). That is what happened (in real terms) in 2018/2019, and then over time that 'elasticity' climbed from 1 to 2 (losing even more tariff revenue). With the current level of tariffs on China, trade could largely cease. Although the hit to the US import sector is unchanged (a 2½% GDP tax hike in the form of tariffs, the revenue collected on all tariffs falls from 1.7% GDP to 0.9% GDP (China is only 3% of that revenue number despite its tariff being 63% of the total).

We have revised global growth three times in the past two weeks, but with the current vintage of the tariffs we forecast it to slow from 2.9% to 2.5% in 2025 and an additional 20bp in 2026 (vs our prior forecast; global growth would stay close to 2.5%). The GDP changes are roughly proportional to the tariffs export exposure. The level of global GDP would be 2½pp lower than a no-tariff baseline, or about USD 2.75 trillion.

It is clearly possible that we end up with lower tariffs through negotiation, in which case we would need to revisit our forecasts again, but there is also some risk of higher tariffs as pharma and semiconductor tariffs have not yet been announced. At this stage we have decided to have our baseline reflect the full effect of the tariffs (those that are in the Federal Register and for which duties are being collected), without trying to guess what the US administration might do next.