What the 2025 Trade War Actually Cost in Tariffs and Inflation

What Actually Happened Last Year

When the Trump administration announced sweeping tariffs on April 2 2025 dubbed “Liberation Day” the immediate reaction was panic. Markets sold off hard. Economists ran their models and got big hits on GDP. Reading those forecasts again now I'm struck by how much they missed. What actually happened was more confusing and more instructive about how tariffs work in practice than any of those models predicted

Companies bore most of the initial cost. Not consumers. The United States collected $187 billion more in tariff revenue in 2025 than in 2024 an increase of nearly 200%. Companies covered about 80% of that bill themselves choosing to absorb it with their own margins rather than immediately raise prices. One large grocery supplier laid out the problem clearly: It didn't know how to account for thetariffs on thousands of SKUs each with different country-of-origin rules so it simply postponed price increases entirely. That kind of operational paralysis was common not an exception

Who Actually Pays a Tariff

Before we continue it is worth being precise about what a tariff is. It is a tax that US Customs charges an importer at the border at the moment a merchandise enters the country. The importer writes that check not the exporter or the buyer at the cash register. That is mechanical. The question that really matters and the one that took me the longest to feel comfortable with is who ends up assuming the financial cost once the check clears

Economists call this incidenceThe legal burden of a tariff and its economic burden are two different things and the gap between them is decided by elasticities that is how sensitive buyers and sellers are to price. If the foreign exporter has nowhere else to sell and cannot easily reduce its own price most of the impact is left on the importer or exporter. If domestic buyers have no good substitute and pay almost any price the cost will fall on them

The empirical record of the 2018 to 2019 tariff round which I will return to later generally found that most of the cost was passed on to domestic buyers rather than being absorbed by foreign exporters who reduced their prices

A Worked Example: Following One Import to the Shelf

Numbers make this concrete faster than any description so let me build one from scratch. Each figure below is illustrative made up for this example not taken from any real company or product

Suppose an overseas factory sells a stand mixer to a U.S. importer for $100. That's the factory price sometimes called the FOB price. Let's call the tariff rate 20 percent applied at the border. The importer now owes $20 in tariffs 20 percent of $100 so the destination cost that is what the importer actually has in the product once it passes customs is 120 dollars

From there the mixer travels two more brands before reaching a shelf. Suppose the wholesale distributor increases the landed cost by 50 percent to cover its own costs and profits: $120 times 1.5 is $180. Then the retailer increases the wholesale price by 40 percent: $180 times 1.4 is $252. That $252 is the selling price

Now let's compare that to the world without tariffs. Factory price $100 wholesale 100 times 1.5 is $150 retail 150 times 1.4 is $210. The tariff turned a $210 mixer into a $252 mixer an increase of $42. As a proportion of the original price 42 divided by 210 is exactly 20 percent.cent the same as the tariff rate. That's not a coincidence. When each stage in the chain represents a fixed percentage of its own cost a percentage increase in cost at the border passes unchanged in percentage terms in the register but grows in dollar terms at each stage. A tariff of $20 becomes a price increase of $42 because both the wholesaler and the retailer are marking a base that now includes the tariff. Call this margin stacking. It's a real mechanismunderestimated and is one of the reasons why general tariff rates tend to underestimate the effect of retail prices rather than exaggerate it

That was a full pass-through: The foreign factory didn't change its price at all and every dollar of cost moved up. Now let's assume a partial pass-through. Suppose the exporter worried about losing the American account reduces its factory price by $10 from $100 to $90 to remain competitive. The tariff still applies at 20 percent but now at the lower base: 20 percent of $90 is $18 so thelanded cost is 90 plus 18 or $108. The wholesale price becomes 108 times 1.5 or $162. Retail becomes 162 times 1.4 or $226.80.The tariff rate never changed. The exporter absorbed part of the cost through a lower price and the consumer absorbed the rest through a lower price increase. That is incidence worked out in dollars instead of described in the abstract

There is a third case and it is the one that actually describes the year 2025. Suppose the importer and retailer decide to keep the selling price at $210 absorbing the full $20 tariff themselves instead of passing it on. The delivery cost remains $120 the same as the full pass case. If the wholesaler continues to increase the profit margin by 50 percent to $180 the retailer now has to sell at$210 at a cost of $180 a profit margin of only 16.7 percent instead of the usual 40 percent. The margin was reduced from 40 percent to 16.7 percent and the buyer never saw a price change. This is what "companies absorbed about 80% of the tariff bill" looks like at the level of a single product repeated across an entire portfolio of SKUs. Here's the same math side by side.another

Case Study: What the 2018 to 2019 Tariff Round Taught Economists

The 2025 tariffs were not the first modern test of these ideas. The first term of the Trump administration carried out a smaller version of the same experiment starting in 2018 first with tariffs on washing machines and solar panels then on steel and aluminum and then a much broader set of tariffs on Chinese imports under Section 301. That round has been studied for years and is the closest thing this issue has to a controlled experiment

Washing machine tariffs are my favorite part of that literature because the finding is almost funny. Economists who studied the episode found that retail prices for washing machines rose by about the amount that would be expected from a near-total transfer of the tariff. That part is not surprising. What is surprising is what happened to clothes dryers which were never imposed any tariffs. Dryer prices rose almost as much as washing machine prices. The main explanation is that dryers were almost alwaysThey are purchased along with a washing machine so manufacturers and retailers used the tariff as a cover to increase the price of the non-tariffed product sitting right next to it in the showroom. No one was legally required to pay a penny more for a dryer. Many people did it anyway

The broader research on that round including widely cited work by trade economists who study data on customs and prices generally found that the pass-through to U.S. import prices was nearly complete meaning that Chinese and other foreign exporters did not significantly reduce their prices to absorb the tariff. The cost manifested itself in U.S. import prices and then in retail prices not in a lower price from the exporter. Several studies estimated the added cost to U.S. households at several hundred dollars a yearalthough the exact figure varies by study and by what type of after-effects are counted and I would treat any point estimates there as a rough order of magnitude rather than a hard fact

That's the empirical backdrop against which the 2025 figures should be read. An 80% share of the cost by companies in the first year reverting to 80% share by consumers as the above model would predict once margin compression runs out is broadly consistent with what the 2018 to 2019 round also ultimately showed. Companies can slow the pace. They generally can't help it toalways

The Counterargument: Why Headline Tariff Costs Get Overstated

I just devoted four sections to making tariffs look like a clean mechanical tax that goes directly to consumers. That's too neat. There is a real serious case where blanket tariff cost estimates like the ones above systematically exaggerate the real pain and it deserves a fair hearing rather than a token paragraph before moving on

Start with substitution. The stand mixer example assumed that the buyer keeps buying the exact same tariff mixer at whatever price it arrives at. Real buyers don't behave that way. When a tariffed product becomes more expensive some buyers switch to a non-tariffed brand a different country of origin a generic alternative or simply a cheaper model. Each of those substitutions shows up in the data as a smaller measured price increase than the math offees alone because the buyer who would have paid full freight is no longer in the sample. Standard inflation statistics are also not good at capturing changes in quality and variety which can hide some of the true cost

Currency movements go in the same direction. If a tariff makes a country's exports less competitive that country's currency can weaken against the dollar lowering the dollar price of its products even before any additional tariffs are applied. A weaker currency partially offsets a stronger tariff and the net price movement at the US border ends up being less than the tariff rate alone suggests. This is a real channel although it is really difficult to isolate it from everything.everything else moves currency markets at the same time and I wouldn't lean too heavily on any single estimate of how much of a specific tariff it offsets

Supplier switching is the third leg and it shows up directly in the 2025 to 2026 data discussed below: the shift in import volume to Mexico and Vietnam. If an importer can obtain a similar good from a country that is not subject to tariffs the tariff applied to the original country becomes almost voluntary. The buyer simply stops buying from the tariffed source. Compared to the original supplier it seems like a completely avoided cost. Compared to the world beforeIf tariffs exist it is actually a cost too it just manifests itself as longer shipping distances longer delivery times and loss of specialization rather than a line item on a receipt

My honest read is that the "tariffs cost less than the headline number" camp is right that the sticker shock estimates overstate the pain felt by any individual buyer who successfully substitutes. It's less obvious that the total cost to the economy is small because the change itself is not free. It just moves the cost somewhere that the customs data doesn't show as clearly

The Shift Happening Now: 2025 Into 2026

That 80/20 split with companies absorbing 80% and consumers 20% is being reversed. JPMorgan research published in early 2026 estimated that consumers' share of tariff costs could rise to 80% essentially reversing the ratio as companies that delayed price increases until 2025 ran out of room to continue absorbing margin pressure. Goldman Sachsprojected an additional 0.3 percentage point increase in inflation in the first half of 2026 alone due to that pass-through effect on top of the roughly half a point already added in 2025

Food is the most exposed category and it's also the one that households notice most quickly since grocery prices are one of the few numbers people check each week. The Yale Budget Lab estimated that food prices rose 2.8% of all 2025 tariff actions combined and fresh produce rose 4%.Tariffs among thousands of SKUs do not remain paralyzed forever. In the end the price increases that occurred delayed all lands in the same period of months instead of being distributed evenly over time

The Supply Chain Is Actually Moving

The history of trade diversion is not just a talking point. Harvard Business School research using Census Bureau trade data found that U.S. imports from China have fallen to levels close to 2001 before China joined the World Trade Organization. This is no small adjustment. It is a reversal of about a quarter-century of integration between the world's two largest economies at least as measured by direct import flows

Mexico and Vietnam have been the main beneficiaries of that diversion. Part of that is a true relocation of manufacturing capacity. Part of this by most accounts is transshipment meaning that goods substantially manufactured in China are shipped through a third country for final assembly or paperwork before entering the US which counts as an import from that third country even though the underlying value chain has not moved as much as the data implies.customs data. Disentangling how much of the "big reallocation" is an actual capacity shift versus a reorientation of paperwork is one of the most difficult open questions in this data and I doubt anyone outside customs agencies has a precise answer. What does seem structural rather than temporary is the direction of travel: away from China toward a more diversified set of manufacturing bases in a way that will take years to completely undo even if tariff policy were to change.tomorrow

What This Means for the Fed

This is where things get really difficult for monetary policy and I say this as an actual description of the problem not as a cover phrase. The FOMC's core problem in 2026 is that tariff-driven inflation behaves like a supply shock.The problem is not that people are spending too much. It is that with the same expenditure now they buy less

That puts the Fed in an awkward position. Cutting rates to support a weakened labor market risks re-accelerating the inflation that the tariffs caused. Maintaining rates to combat inflation risks pushing an already weak economy into something worse. Neither lever clearly solves a supply-side problem which is exactly why this is difficult not because the Fed is being indecisive

The March 2026 FOMC minutes described goods inflation as "concentrated in sectors where tariffs apply" and projected that pressure would "reduce after the first quarter of 2026." That decline has been slower than the committee expected. If the Fed under its new leadership can communicate the distinction between tariff-driven inflation and demand-driven inflation clearly enough for markets to actually value it differentlyis in my opinion one of the most consequential open questions of the year and I don't feel confident addressing it in any way yet

The Timing Mismatch Inside the Data

There's a more subtle dynamic beneath the headline inflation numbers which is easy to miss if you just look at the annual data. Companies that delayed price increases until 2025 for exactly the operational reasons described at the beginning of this article are now bringing those increases forward until 2026. That creates a temporal mismatch: Inflation data in a given month partly reflects decisions made months earlier lumped together rather than a smooth reading of current cost pressure

In practice that means the current inflation figure will probably look worse than the underlying trend actually is because a year of delayed pass-through is being compressed into a shorter window. It also means the opposite will likely happen on the way out. Once the lag in price increases clears month-over-month inflation readings tied to tariffs should look better than the trend not because anything has actually improved but because the batch effect is over. Read a single month of this dataNot taking that lag into account is in my opinion one of the easiest ways to draw the wrong conclusion about where inflation is really headed

How I Actually Think About This

My reading and I want to clearly point out that this is my reading and not established fact is that the most useful habit in following this story is to separate the two things that are lumped together as "inflation": the tariff-driven part and the core demand-driven part. They behave differently respond to different policy tools and combining them is how you end up surprised by the patience or impatience of the Federal Reserve

The way I would actually use the incidence framework mentioned earlier in this article is as a filter for the company's statements not as a commercial signal. When a company says it is "managing tariff impacts" I now wonder which of the three practical example scenarios it is actually describing. Is it reducing cost through margin compression like the grocery supplier did in 2025? Is the cost being passed on entirely as the mixer example shows whenexhausting margin? Or are you substituting suppliers as the shift toward Mexico and Vietnam that some companies are making suggests? Those are three very different stories hiding behind the same phrase in an earnings call and only one of them the full pass is a genuine Margin compression is a sign that it doesn't at least not yet and supplier substitution says almost nothing about pricing power only about supply chain flexibility

I admit I was wrong when the 2025 numbers first appeared. I read the 80% business share of the tariff bill as evidence that companies were simply choosing to protect customers which sounded almost generous. The problem with the grocery supplier's SKU convinced me otherwise. Much of that 80% wasn't so much a choice as an inability to change price quickly enough in a catalog with thousands of items and different country of origin rules.for each item. Operational friction not generosity. That rethinking also changed the way I read the 2026 reversal. The increase in the share of consumers towards 80% is not due to companies giving up and passing on costs out of impatience. It's friction that is finally becoming clear

None of this is a recommendation to buy or sell anything. It is a lens to read company disclosures and the Federal Reserve's language more carefully than the headline suggests

The Bottom Line

Tariffs are a tax collected at the border but who actually bears the cost - the exporter importer or consumer - is decided by the economy more than who writes the check. In 2025 American businesses absorbed about 80% of a $187 billion increase in tariff revenue mostly through margin compression and operational paralysis rather than generosity. This will reverse in 2026 and JPMorgan estimates that shareof consumers could rise to 80% as delayed price increases accelerate in advance. The 2018 to 2019 round showed the same final pattern: approval is almost complete cushioned in the meantime by substitution currency movements and changing suppliers that make the overall cost appear lower than the full picture. The Federal Reserve is stuck treating a supply shock with tools designed for demand shocks and the time lag between the momentwhen the costs were incurred and when they hit the record makes any month of inflation data difficult to read clearly. My own approach is to keep the tariff-driven part of inflation analytically separate from the rest and to read "managing tariff impacts" as a specific verifiable statement about margin transfer or supplier switching rather than a vague guarantee