New York Fed research puts the pass-through rate at about a quarter, says tariffs added 2.9 points to goods inflation, and finds the full effect takes a year to arrive
NEW YORK, Oct. 9, 2026. Economists at the Federal Reserve Bank of New York have put a number on one of the most contested questions in American trade policy, and the number is more nuanced than either side of the argument has been using.
About a quarter of any tariff increase passes through to consumer goods prices after one year, according to research published this week on the New York Fed’s Liberty Street Economics blog by Mary Amiti and Sebastian Heise of the New York Fed and David E. Weinstein of Columbia University. The post, which summarizes New York Fed Staff Report 1201, was published Oct. 6 and revised Oct. 8.
The headline estimate is that each one percentage point increase in the average tariff rate raises consumer goods prices by roughly 0.25 percent after twelve months. Applied to the policy that actually ran, the authors calculate that by February 2026 tariffs had added 2.9 percentage points to goods price inflation. Without the duties, they find, goods prices would have fallen slightly rather than risen.
“In sum, we estimate that about a quarter of a tariff increase is passed through to consumer goods prices,” the authors write.
That conclusion will be read two ways, and both readings are defensible. A quarter is far less than the dollar-for-dollar pass-through that tariff critics often assume, which means American consumers did not absorb the full cost of the duties. A quarter is also far more than zero, which is roughly what proponents of the policy predicted when they argued that foreign exporters would eat the charge.
Three stages, three different speeds
The value of the paper for anyone running an import business is less in the headline figure than in its decomposition, because the authors trace the tariff through three distinct stages of the supply chain and find that each behaves differently.
At the border, pass-through is nearly complete and nearly immediate. Tariffs move into import prices at roughly 90 percent, close to one for one. Foreign exporters cut their prices only marginally in response. This is the finding that most directly contradicts the proposition that trading partners would absorb the duties to preserve market share, and it has held across the sample.
At the factory gate, the picture changes. Domestic producer prices rise through two channels. The larger is a marginal cost channel: American manufacturers buy imported inputs, those inputs cost more, and finished goods prices follow. The smaller is a strategic complementarity channel, the economists’ term for the simple competitive fact that when the imported rival gets more expensive, the domestic producer can raise its own price without losing volume. Crucially, this stage builds slowly. The effect roughly doubles between the sixth and twelfth months after a tariff takes effect.
At the retail shelf, the increase is substantially diluted. A 10 percent rise in import and producer prices translates into about a 5.6 percent rise in retail prices, the authors find, because distribution margins account for roughly half of consumer spending and those margins are domestic. Warehousing, trucking, store rent and labor do not carry a tariff, so the duty is spread across a larger base by the time it reaches a shelf tag.
The timing matters for anyone forecasting. The direct effect, meaning the imported goods channel, is about half complete after three months and fully in place by six. The indirect effect, meaning the domestic production channel, more than doubles between months six and twelve. The full effect of a tariff change therefore takes approximately a year to work through, which means that price data collected in the first quarter after a tariff action systematically understates its eventual impact.
As a reference scenario, the authors calculate that a 10 percent tariff applied to all imports would raise consumer goods prices by about 2.6 percent after twelve months.
What the study does and does not cover
The methodology deserves attention because it defines the boundaries of what can be claimed from the results.
The sample covers 67 non-oil goods categories, weighted by 2022 consumer expenditure. Services, which make up roughly two thirds of the consumer basket, are excluded entirely. The estimates run on data through February 2026, with a forecast extending through September 2027.
The identification strategy compares goods with higher tariff exposure against goods with lower exposure, holding economy-wide factors constant, including exchange rates, wages, aggregate demand and monetary policy. That design is what gives the estimate its credibility. It is also what limits it. The authors are explicit that a relative comparison across goods cannot establish how much of the overall price level moved because of tariffs, and they note that tariffs could themselves influence the very economy-wide factors the method holds fixed. A tariff program that moves the dollar, or that changes the Federal Reserve’s reaction function, has effects this design is not built to capture.
Readers should hold the 2.9 percentage point figure with that caveat in mind. It refers to goods price inflation, not to headline inflation. Because goods are roughly a third of the consumer basket, the implied contribution to overall inflation is considerably smaller, on the order of six tenths of a percentage point by the same arithmetic. Several secondary accounts of the research have blurred that distinction.
Quartz, which covered the research on Oct. 8, summarized the finding by noting that prices on many everyday goods “would have actually dipped modestly without the levies.”
The forecast is the surprising part
The backward-looking estimates have been broadly anticipated by practitioners. The forward-looking path is where the paper contains news.
The tariff effect on the goods price level peaked near 3 percent in February 2026, at which point the indirect channel accounted for about 30 percent of the total. The authors forecast that the level effect eases to roughly 2 percent by August 2026, and they attribute the decline to a specific policy event: the Supreme Court ruling that ended the emergency-powers tariffs, followed by the substitution of a 10 percent surcharge.
The contribution to the twelve-month inflation rate follows a sharper path. It is forecast to fall to approximately zero by August 2026, then turn negative, then return to slightly positive by the middle of 2027 as tariffs on Canadian goods pass through.
That last clause is the one importers should underline. The forecast does not describe a return to a pre-tariff baseline. It describes a trough produced by the removal of one tariff regime, followed by a renewed climb as a different and more narrowly targeted set of duties works through the same twelve-month pipeline the paper documents. The negative contribution in late 2026 is arithmetic, not relief: it reflects the comparison of current prices against a period when the emergency duties were still in force.
For 2027 planning, the implication is that the measured inflation contribution of tariffs will look benign at precisely the moment when the newest actions, including the Section 338 measures against Canada, the Section 232 pharmaceutical duties, the polysilicon regime that begins Dec. 4 and whatever emerges from the open overcapacity investigations, are only beginning their six to twelve month climb.
A wider deficit, not a narrower one
The research landed in the same week as trade data that complicated the policy’s other central claim.
The U.S. trade deficit widened 13.7 percent between July and August to $105.6 billion, the Bureau of Economic Analysis reported Oct. 6, the highest reading since March 2025, when the gap hit $140 billion in the pre-tariff import surge. Imports rose 4.3 percent to $420.8 billion. Exports rose 2.2 percent to $205.7 billion.
Much of the widening traces to artificial intelligence hardware. The Federal Reserve Bank of Minneapolis estimated that AI-related demand added roughly $200 billion to the deficit as of April, a category of import that tariffs were never designed to suppress and that the domestic economy cannot presently supply at the required scale.
The number carries political weight because the deficit was the stated justification for the original program. The April 2025 action characterized the trade gap as a national emergency, and the policy’s central promise was that duties would shrink it.
Economists interviewed by Fortune pointed to policy churn rather than tariff levels as the binding constraint. “There’s so much volatility that, essentially, companies are not going to change their behavior,” said Tarek Hassan of Boston University, who also warned that a deficit closing rapidly would more likely signal a crisis of foreign confidence than a manufacturing revival. Donald Boudreaux of George Mason University offered the opposite framing, reading the deficit as evidence of a healthy economy importing the inputs it expects to need.
The White House pushed back on the premise. Spokesperson Taylor Rogers said record capital goods imports reflect a manufacturing renaissance and pointed to trillions of dollars in announced manufacturing investment.
Tariffs on Chinese goods, which spiked to 145 percent in the weeks after the April 2025 action, now sit near 30 percent following the Supreme Court ruling and a series of bilateral arrangements. Modelled estimates from Global Trade Alert’s U.S. tariff engine put the trade-weighted applied rate on Chinese imports at about 25.5 percent as of this week, with India near 18.7 percent, Turkey near 17.6 percent, Brazil near 16.6 percent, Japan near 12.5 percent, South Korea near 12.3 percent, Germany near 10.6 percent, Canada near 7.6 percent and Mexico near 6.7 percent.
The policy architecture the forecast assumes
The paper’s forecast rests on an assumption about what the tariff landscape looks like through September 2027, and that assumption is worth spelling out because it is the most fragile part of the exercise.
The emergency-powers duties ended in February. The 10 percent surcharge that replaced them, imposed under Section 122 of the Trade Act of 1974, was itself struck down by the Court of International Trade in May and expired by operation of its own statutory clock on July 24. What took its place was not a single measure but a lattice of sector-specific and partner-specific actions, each resting on a different legal authority and each with its own timetable.
Section 301 duties tied to forced labor findings took effect July 24 at rates of 10 percent and 12.5 percent across roughly 60 trading partners, with extensive exemptions. An additional 25 percent Section 301 duty on Brazilian goods took effect July 22. Section 232 actions have multiplied: 25 percent on advanced computing semiconductors from January, 100 percent on patented pharmaceuticals and their active ingredients phased in across July and September, 100 percent and 25 percent tiers on unmanned aircraft systems from September, and minimum import prices plus a 15 percent duty on polysilicon and downstream solar products beginning Dec. 4. Section 338 measures against Canada escalated to 50 percent on specified products from Aug. 22, with outright import prohibitions on certain goods from Sept. 29, stacking on top of Section 232 and applying regardless of USMCA qualification.
Several of those measures entered force after the February 2026 data cutoff in the paper’s estimation sample. They appear in the forecast as assumptions rather than as observations. If the open Section 232 investigations into robotics and industrial machinery, semiconductor manufacturing equipment, personal protective equipment and medical consumables, wind turbines and critical minerals produce duties in the coming months, or if the Section 301 excess capacity investigations covering more than a dozen trading partners conclude with new tariffs, the forecast path understates the 2027 contribution.
There is also a procedural change that makes the forecast harder to maintain. Commerce and the Office of the U.S. Trade Representative can now add derivative products to existing Section 232 actions at any time by Federal Register notice, without a new investigation or a fixed inclusion window. Scope therefore expands on no predictable schedule, which is a difficult feature to model and an even more difficult one to budget against.
Who paid
The pass-through literature has become the empirical backbone of the distributional argument over the refund program now moving through the Court of International Trade, and the New York Fed numbers sit at its center.
A Center for American Progress report published Oct. 1 draws on New York Fed research to estimate that roughly 26 percent of the 2025 tariff increase was passed on to consumers, and concludes that “nearly 90 percent of the tariffs’ economic burden fell on U.S. firms and consumers.” The report’s argument is that because importing firms absorbed most of the cost, refunds of unlawfully collected duties flow back to those firms rather than to the households that paid the remaining share at the register.
That is a defensible reading of the pass-through evidence, and it is also an uncomfortable one for every side of the debate. If firms absorbed roughly three quarters of the duty, then the tariffs functioned largely as a tax on domestic business margins rather than on foreign producers or American consumers. If consumers absorbed roughly a quarter, then there is no mechanism by which the refund program returns that portion to them.
The report estimates that about $35 billion of $115 billion in refunds paid between May and August may reach the roughly 234,000 identified small-business importers, an average near $149,000 each, while small-business importers paid an average of $426,000 more in duties over the eighteen months from March 2025. An Atlanta Fed analysis it cites estimates that only 34 percent of refunds will reach the most financially constrained small and mid-sized firms.
What it means for importers and exporters
The practical lessons from the research are specific enough to act on.
First, the ninety percent border pass-through finding should end a planning assumption that has survived longer than the evidence supports. Buyers negotiating with overseas suppliers on the theory that the supplier will absorb a meaningful share of a new duty are negotiating against the data. Exporters to the United States, correspondingly, have less pricing flexibility than the political rhetoric around tariffs implies, because the margin to give has largely not been given.
Second, the twelve-month lag is a budgeting instruction. A duty that takes effect in the fourth quarter of 2026 will not show its full cost until the fourth quarter of 2027. Companies that priced the pharmaceutical, drone and polysilicon actions on the basis of first-quarter observations have almost certainly understated them. The same applies in reverse to relief: a tariff removed this quarter continues to show up in cost structures for several quarters afterward.
Third, the retail dilution finding explains a persistent puzzle for consumer-facing businesses. The 5.6 percent retail response to a 10 percent upstream increase means that a brand selling through a long distribution chain sees less of the shock reach its shelf price, but also that the shock has been absorbed somewhere in that chain, by distributors, wholesalers and retailers whose margins are thinner than a manufacturer’s. Importers negotiating with channel partners in 2027 should expect those partners to arrive with this arithmetic.
Fourth, the domestic production channel means that buying American is not a complete hedge. One third of the estimated consumer price effect comes from domestically produced goods, operating mostly through imported input costs. A manufacturer sourcing steel, chemicals, semiconductors or components from abroad carries tariff exposure regardless of where final assembly happens, and the strategic complementarity channel means that even a fully domestic competitor may raise prices when imports become costlier.
Fifth, and most important for the year ahead, the forecast path implies that the quietest period in the tariff inflation data is arriving just before the next climb. Trade teams that use the late-2026 numbers to argue internally that the tariff problem has resolved will be making that argument at the worst possible moment in the cycle.
The argument the numbers do not settle
It is worth being precise about what this research resolves and what it leaves open, because both sides of the tariff debate have already begun quoting the same paper for opposite conclusions.
The paper settles the border question. Foreign exporters did not absorb the duties in any meaningful degree. At roughly 90 percent pass-through into import prices, the proposition that trading partners would pay for American tariffs is not supported by the data, and that finding is consistent across the goods categories in the sample.
The paper does not settle the burden question. Knowing that a quarter of the duty reaches consumer goods prices tells you where three quarters did not go, but it does not tell you precisely who absorbed it. Importers’ margins, distributors’ margins, retailers’ margins and wage bills are all candidates, and the aggregate price data cannot separate them. The Center for American Progress estimate that nearly 90 percent of the burden fell on U.S. firms and consumers combines the consumer share with an inference about the firm share, and that inference is doing substantial work.
The paper also does not speak to the policy’s stated objectives. Nothing in a pass-through estimate indicates whether tariffs induced domestic investment, reshored production, improved bargaining leverage in bilateral negotiations or strengthened supply chain resilience. Those are the arguments on which the program’s proponents principally rest, and they operate on multi-year horizons that a twelve-month price study is not designed to evaluate.
What the research does supply, and what has been conspicuously missing from the public argument, is a defensible magnitude. For nineteen months the debate has run between a claim that consumers pay everything and a claim that consumers pay nothing. The answer appears to be about a quarter, arriving over about a year, with a third of it coming through American factories rather than through the ports.
The research was also published on VoxEU, and the underlying Staff Report 1201 contains the category-level detail the blog post omits.