The anti-tariff document everyone cites and what it doesn’t actually show
Well, here we go again.
A new NBER Working Paper is making the rounds, touted as a definitive argument against tariffs. It references a controversial New York Fed study which suggested that the burden of recent tariffs fell on Americans, while also claiming that higher tariffs would lead to reduced trade, production, and manufacturing.
This paper by economists Tamer den Besten and Diego R. Kenzig is quite thorough. It charts a series of events: Major U.S. tariff changes from 1840 to 2024 and then estimates how these tariff spikes relate to economic outcomes. Unlike the New York Fed’s isolated case, this offers a long-term outlook rather than focusing on a single presidency or a limited time frame.
That said, the findings might be a bit exaggerated. In the postwar period, the authors’ “clean” episodes mostly consist of tariff reductions, neglecting significant increases that occurred due to economic conditions. Essentially, their conclusion about modern “tariff hikes” is largely inferred from periods of liberalization and a couple of increases during President Trump’s first term. So they seem to draw conclusions about the effects of increasing tariffs without fully investigating what raising tariffs entails.
This reflects a recurring issue among economists studying tariffs. The post-World War II push for “trade liberalization” was so effective that there’s scant evidence on the effects of raising tariffs. With that in mind, their narrative turns on its head when examining tariff reductions instead.
The problem for trade economists: Too many rate cuts, not too many rate hikes.
The authors have identified 35 major tariff occurrences from 1840 to 2024, labeling 21 as “exogenous.” This means they weren’t enacted in response to fiscal issues or economic downturns. Post-war, most of these exogenous events involve tariff reductions, which encompass eight rounds of GATT liberalizations. The only recent tariff hikes included are from Trump’s administration in 2018 and 2019. Notably, they disregard the 1971 Nixon levy, which led to Section 122 tariffs that Trump later reinstated, and the 1975 Ford oil import charge, since they view these as “endogenous” reactions to external conditions.
This stance is defensible. The aim is to spotlight hikes occurring as part of protectionist strategies, framed around national security or redistributing wealth to neglected Americans. However, it poses some challenges. From the end of World War II until 2018, conditions were primarily favorable for lowering tariffs. The general trend of history has pointed toward reduced U.S. tariffs. Additionally, it’s curious that Trump viewed his tariffs as reactions to external imbalances, with many indications suggesting they were “endogenous” by the authors’ criteria. But let’s set that aside for now.
This context makes the sample from den Besten and Kenzig quite limited. Consequently, discussions about tariff increases tend to rely heavily on “tariff increases” which assume symmetry: that tariff reductions parallel the effects of tariff increases. While this symmetrical view is common in econometrics, it doesn’t equate to actually investigating current hikes—an assumption the authors haven’t rigorously tested.
A helpful analogy comes to mind. Imagine using rate cuts predominantly to predict economic responses to rate hikes, then confidently asserting that “rate hikes always lead to X.” You could use a linear model for that. But it wouldn’t provide real insight if you lack sufficient data on actual hikes.
Additionally, the paper fails to analyze the Trump tariff episode. Although the 2018 and 2019 tariffs are marked as two distinct “exogenous” events, the paper doesn’t evaluate them independently or compare them against model predictions. Instead, they’re folded into an overall estimate that is mostly based on post-war tariff reductions.
The authors’ own robustness checks reveal that postwar gains hinge on a limited number of events. Removing specific episodes from the postwar context increases variability and uncertainty sharply. There simply haven’t been many significant tariff events since the war. However, it’s still unclear if the Trump tariffs, the only notable modern hikes in the sample, validate or challenge the model.
Reality check: 2018, 2019, 2025 do not match the pattern
The paper estimates a pattern of “tariff shock” where imports are expected to drop, exports will lag, production and manufacturing activity will wane, consequently lowering GDP. Yet records from recent years indicate that such predictions failed to materialize.
Take 2018. Tariffs on aluminum, steel, and some Chinese goods were raised. Despite fears of protectionism signaling the end of the post-war order, the economy showed resilience. Real GDP growth remained steady throughout the year. Non-residential investment was robust, industrial production increased year-over-year, and imports didn’t plummet; they continued to rise. Exports even cooled off towards year-end but didn’t collapse. The anticipated massive contraction simply didn’t happen.
Moving to 2019, after escalating tensions with China in May, Real GDP maintained positive growth through the year. Nominal growth hit 4.1%, with real growth at 2.3%. While non-residential investment did slow compared to 2018, it stayed positive. Industrial production showed slight decline year-over-year, which was the most evident weakness. Imports weakened somewhat in the latter half, while exports remained stable to slightly positive. In essence, 2019 painted a mixed picture, with continued aggregate growth along with a mild recession in manufacturing, and investment remained unscathed.
Now we find ourselves in 2025. The previous year’s tariffs are only applicable until 2024, and the last tariff episode discussed is from 2019, leaving it out of this paper. However, if the paper’s conclusions hold weight, we should be seeing expected shrinkage. But guess what? Even following the “Liberation Day” tariffs in April, real GDP stayed strong throughout the year, showing two quarters of unexpected growth. Real final sales to domestic buyers were steady. Imports saw a drop in the latter half, aligning with the traditional narrative, yet exports remained on an upward trend, contradicting the standard “tariff shock” story. Manufacturing hours grew year-on-year, capacity utilization aligned well, and core goods inflation hovered at a moderate level.
The failure of these recent tariff hikes to produce the foreseen consequences raises critical questions. If Trump’s tariff hikes are indeed the only clear modern hikes accounted for, why not demonstrate how these instances align with the model’s pattern? Or could it be that the evidence we’ve encountered suggests otherwise, and that the notion of retrenchment stems from reversing the thesis of tariff reductions?
New tariff system?
Crucially, the authors themselves stress that “systems matter.” They don’t treat the effects of tariffs as constant over time, pointing out that tariff interaction changes in relationships based on monetary and trade systems. They explicitly highlight monetary policy responses and foreign retaliation patterns as factors that can alter how tariffs affect the economy.
This framework is theirs, not ours. And once it’s accepted, a singular historical average can’t be treated as a universal principle.
If this mechanism is predicated on retaliation and monetary response, then tariffs used as economic levers differ significantly from those raised within retaliatory contexts. When negotiation is prioritized and retaliation is minimized, the potential for an “export collapse” diminishes. As substitutions become faster and investment shifts toward reshoring, the narrative of “output decline” weakens. Tariff policies aimed at rebuilding productive capacity and accessing markets aren’t equivalent to shifts in a multilateral liberalization context.
It’s likely there will be more papers on tariffs coming in the following months, each rehashing the same debates. Perhaps eventually the economic establishment will produce something insightful addressing Trump’s tariffs. However, that clarity hasn’t surfaced as of February 2026.
