Why has the U.S. economy been resilient to the largest increase in tariffs since the Great Depression?

So much for the textbook tariff shock.

U.S. import duties imposed in early 2025 abruptly tripled the average effective tariff rate. Following court challenges and trade deals, average tariffs remain more than twice the level Americans had known for decades. In standard economic analysis, this would constitute a sharp supply shock; prices should jump for American consumers and imports should fall. Instead, U.S. imports initially spiked in 2025, then sagged, then surged again through 2026. In real dollars, imports are now almost 5 percent higher than before the first new tariff announcement.

Does this mean traditional trade models are wrong on tariffs? Not wrong, but incomplete—they are lacking essential features to capture the crosscutting economic effects of this episode. These effects are modeled out in new research from Minneapolis Fed Research Director Andrea Raffo and Monetary Advisor Michael Waugh (Staff Report 686, “Tariffs, Investment, and the Missing Trade Collapse,” with Francesco Ferrante and Andrea Prestipino of the Federal Reserve Board). With some intuitive enhancements to a standard trade model, calibrated with U.S. data, the economists can closely replicate the experience of the past two years.

Three economic facts drive their design and interact within the model. One is that many U.S. tariffs have been announced in advance. Given ample notice of a tariff hike, many importers respond by stocking up while they can. The front-loading of inventories of soon-to-be-tariffed goods drives up imports at a time of generally rising tariffs, especially for intermediate inputs.

Another fact is that the application of U.S. tariffs has been far from uniform. The tariffs feature thousands of exceptions and deviations, with the cumulative effect that tariffs fall differently across broad categories of goods. Tariffs on imports directly consumed by Americans rose the most, an average of 13 percentage points in 2025 to a whopping 18 percent. Tariffs on intermediate inputs—that is, raw materials or unfinished components—rose 9 percentage points. And tariffs on capital goods—machinery or technology invested in American production—rose only 6 percentage points.

The paths of imports in each category reflect this varying tariff treatment (Figure 1).

These varied rates are only the first part of the story. Raffo, Waugh, and co-authors advance modern trade theory by exploring how tariffs on these three categories have different macroeconomic effects. Tariffs on consumption and intermediate goods have the traditional, “stagflationary” effect of a supply shock, pushing prices up and economic output down. But tariffs on capital goods play out like a deflationary demand shock, depressing output, labor demand, and prices.

The aggregate economic effects of tariffs depend on the different rates, these different category effects, and the degree to which each sector depends on imports. Crucially, imports comprise a much higher share of U.S. capital goods (30 percent) than of intermediate inputs (9 percent) or consumer goods (6 percent). So, even though tariff rates on capital are low relative to these other categories, the contractionary effect is a powerful force restraining GDP growth and providing some offset to the price pressures from tariffs.

Along with inventory front-loading and tariff heterogeneity, the model from Raffo, Waugh, and co-authors incorporates a third economic phenomenon playing out alongside the tariff surge: investment in artificial intelligence.

As Waugh has documented, AI-related imports increased more than 70 percent between 2023 and 2025, driving up the U.S. trade deficit despite the broader effects of tariffs. The economists model the rise of AI as a positive shock to the marginal efficiency of investment (MEI), essentially the bang-for-each-buck of capital expenditure. Tariffs on capital act like a negative MEI shock. But, the economists write, “because the AI sector’s import-intensive capital faced little or no tariff, the investment boom could pull in imports without being choked off by the trade-cost shock hitting the rest of the economy.”

The AI effect is substantial, offsetting the contractionary effects of tariffs elsewhere. By toggling the MEI component off in the model, they estimate that without the impact of AI-related trade, U.S. imports would have fallen by 10 percent and GDP would have contracted by 0.7 percent.

Further analyses underscore a key finding of the research: the asymmetric effect of tariffs on capital investment versus other types of goods. One counterfactual exercise supposes that the average tariff rate is the same as what the U.S. actually experienced, but this rate is identical across investment, intermediate, and consumption goods. Under this uniform tariff (and ignoring, for this exercise, the AI boom), inflation from tariffs is lower. But U.S. investment declines nearly 9 percent by the end of one year, and GDP contracts almost 50 percent more (Figure 2). “The heterogeneous distribution of the 2025 announced tariffs,” they conclude, “contributed to the resilience of the U.S. economy.”

At the same time, however, the changing composition of U.S. imports makes the economy more vulnerable to tariffs. Since the 1940s, the U.S. has become more reliant on trade in general, as imports rose from a low-single-digit percentage of GDP to more than 15 percent in recent years. And the mix of imports shifted heavily during this period: Capital goods imports rose from essentially zero to more than 30 percent of U.S. imports today, while the share of intermediate inputs declined. Through the lens of the model, this dependence on capital imports makes tariffs today less inflationary for the U.S. economy, but more recessionary.

That holds unless, of course, capital goods are exempt—as they largely have been in the case of AI-investment goods like microprocessors. But as Raffo, Waugh, Ferrante, and Prestipino illustrate, even a 6 percent tariff rate on capital goods overall has been a drag on U.S. economic growth. One lesson for policymakers from our tariff experience so far: “The composition of tariff protection is a first-order consideration for macroeconomic stabilization.”

Read the Minneapolis Fed staff report: “Tariffs, Investment, and the Missing Trade Collapse”