A new economic study puts US steel prices up 33.9 percent and aluminum up 58 percent since the Section 232 tariffs began, with downstream manufacturers carrying the cost
WASHINGTON, Oct. 22, 2026
The Montreal Economic Institute published an economic note on Wednesday that attempts to answer a question American manufacturers have been asking for nineteen months: what have the Section 232 metals tariffs actually bought.
The answer the study arrives at is uncomfortable for the policy. US steel prices, measured by the producer price index for steel mill products, have risen 33.9 percent since February 2025. US aluminum prices, measured by the index for aluminum mill shapes, have risen 58.0 percent over the same window. US steel production rose 7.6 percent and steel employment rose 0.7 percent. US aluminum production fell 0.1 percent, while employment in aluminum and other nonferrous metals rose 0.6 percent.
Put plainly, prices moved by tens of percentage points and employment moved by fractions of one.
“All that the tariffs have accomplished since being put in place is to raise prices for U.S. manufacturing companies,” said Joseph Steinberg, a professor of economics at the University of Toronto and a senior fellow at the MEI, in the release accompanying the note. “Trump’s tariff policy has not helped revitalize the industry as promised. The promised jobs and production have not materialized.”
The study is advocacy research from a free market think tank and should be read as such. Its central comparison, however, is harder to dismiss than its framing.
The comparison at the center of the study
The authors set the US price series against Canadian equivalents over an overlapping period. From December 2024 to June 2026, Canadian steel prices rose about 1 percent and Canadian aluminum prices rose about 26 percent.
The gaps are large. More than thirty percentage points separate the two countries on steel. Roughly thirty two points separate them on aluminum. The authors are careful to note that these gaps give an indication of the tariff effect but are not formal econometric estimates, and that caution is appropriate. The two markets differ in currency, in energy input costs, in industrial composition and in exposure to third country demand. Canada has also been subject to its own trade disruption over the period, which cuts in both directions.
Still, the direction and magnitude are consistent with what trade economists predicted when the measures were introduced and with what US purchasing managers have reported since. A tariff on an input raises the domestic price of that input toward the tariff inclusive price of the import. That is not a side effect. That is the mechanism by which the policy is intended to work. The debate has never been about whether input prices rise. It has been about whether the production and employment gains justify the input price increase.
On the study’s numbers, they do not.
How the tariff structure got here
The current architecture is the product of repeated expansion and one significant structural change.
The Trump administration imposed 25 percent tariffs on steel and aluminum under Section 232 of the Trade Expansion Act of 1962 in March 2025. Within months the rate was raised to 50 percent. In June 2025 the 50 percent rate was extended to derivative products from effectively every trading partner. Copper was added in July 2025 at the same rate. In August 2025 the Commerce Department added 407 product categories to the steel and aluminum derivative lists, pulling in goods as far from a steel mill as wind turbines, mobile cranes and bulldozers.
The structural change came on April 2, 2026, in a proclamation effective April 6. It replaced a single headline rate with a tiered schedule. Commodity grade steel, aluminum and copper carry 50 percent applied to full article value. Derivative products substantially made of those metals carry 25 percent, also on full value. Certain metal intensive industrial and electrical grid equipment carries 15 percent through 2027. Products manufactured abroad using at least 95 percent US melted and poured steel or US smelted and cast aluminum carry 10 percent, a threshold the June 1, 2026 proclamation lowered to 85 percent. Articles with 15 percent or less metal content by weight were removed from the program, except for goods classified in HTSUS Chapters 72, 73, 74 and 76. Goods falling under more than one metal’s provisions pay one rate rather than stacking.
The April proclamation also terminated the public petition process for adding derivative articles. Commerce and USTR retain the ability to add products, now on a rolling basis determined by cabinet officials rather than through the prior inclusions windows. The first round, in August 2025, covered 407 HTSUS subheadings. A second window ran from Sept. 15 to Sept. 29, 2025. A proposed addition published Aug. 6, 2026 would bring in fourteen further categories, among them aluminum powder, fire extinguishers and certain trailers.
Product exclusions, meanwhile, have been largely wound down. All general approved exclusions terminated on March 12, 2025. Previously granted product exclusions remain valid only until they expire or until their excluded volume is imported, whichever comes first. The exclusion request process itself has been terminated.
The rate cut that raised the bill
The most consequential element of the April 2026 restructuring is the one that received the least attention, and it is the reason many importers saw duty costs rise in a quarter when the headline rate on their goods fell.
Before April 6, the duty on a derivative article was assessed on the value of the steel or aluminum content. After April 6, it is assessed on the full value of the article. The Cato Institute illustrated the arithmetic with a worked example that has since circulated widely among customs practitioners. A $1,000 product containing $200 of steel owed $100 under the old 50 percent metal content rule. Under the new 25 percent full value rule, the same product owes $250.
The rate was halved. The bill multiplied by two and a half.
The reason is that full article value includes fabrication, machining, labor, overhead and margin. For any derivative article whose metal content is less than half its value, which describes the overwhelming majority of manufactured goods, the base expansion outweighs the rate reduction. The Cato analysis characterized the change as a tariff increase presented as a simplification.
There is a related enforcement dispute running underneath this. An internal Customs and Border Protection memorandum circulated in December 2025, never published, reportedly redefined non steel content to exclude fabrication, machining and labor costs. The practical effect was to expand the dutiable base even under the prior content based rule. An Illinois fastener importer, Express Fasteners, followed CBP’s published guidance and paid 50 percent on steel content plus reciprocal tariffs on the remaining value. CBP instead applied the 50 percent steel tariff to the full product value. The company learned of the December memorandum from another importer and sued.
A separate challenge filed at the Court of International Trade on Jan. 27, 2026 contests how CBP calculates Section 232 duties on articles made entirely of steel. Trade counsel have advised importers to file protests on that issue to preserve refund rights while the case is pending. That advice remains current.
Where the cost lands
The MEI note includes an estimate of downstream exposure that deserves attention from anyone modeling input costs.
Fabricated metal manufacturers require roughly 31 cents of steel and aluminum per dollar of output. Automakers require roughly 16 cents. If input requirements were held fixed and the observed price increases passed through fully, automaker production costs would rise 6.5 percent and fabricated metal production costs would rise 13 percent.
Those are large numbers against manufacturing margins that typically run in the mid single digits. They are also upper bound estimates, since firms substitute, renegotiate, redesign and absorb. But the direction is clear, and it describes a transfer rather than a creation. As Steinberg put it, “each dollar of additional profits earned by protected industries because of tariffs comes from increased prices,” and “it is American manufacturers of processed goods and their customers who are paying the bill.”
The historical precedent the authors cite is the 2018 tariff episode. Downstream production losses from that round reached $3.5 billion by 2021, against steel and aluminum production gains totaling $2.25 billion. The ratio was unfavorable then at a 25 percent rate on a narrower product scope. The current program operates at 50 percent across a derivative list that has expanded repeatedly.
On aggregate employment, the study is appropriately cautious. US manufacturing shed roughly 60,000 jobs between February 2025 and July 2026. The authors link this to the metals tariffs but describe the relationship as small and not statistically significant. That is an honest finding and the study should be credited for reporting it rather than overstating it. Manufacturing employment responds to many things, and attributing a 60,000 job swing to one policy instrument is not defensible on this evidence.
The supply response, such as it is
Defenders of the program argue that the relevant time horizon is years, not quarters, and that capacity investment takes time to appear in production statistics. There is substance to that argument. The MEI note acknowledges several large announcements made since the tariffs took effect, including a Hyundai steel mill in Louisiana and an aluminum smelter in Oklahoma, and notes that both will take years to come online. Century Aluminum restarted idled capacity at Mount Holly, South Carolina, adding roughly 50,000 metric tons of annual production.
Against that, 50,000 metric tons is a rounding error in a US primary aluminum market that consumes millions of tons annually, and the two greenfield projects will not supply a single ton before the end of the decade. The aluminum production figure in the study, down 0.1 percent, is the relevant number for anyone purchasing metal in 2026.
The administration has also moved to address the supply side directly. A July 20, 2026 directive instructed Commerce to establish an incentive program for new or expanded US primary aluminum production. That is an implicit acknowledgment that the tariff alone has not produced the capacity response the policy anticipated. Primary aluminum smelting is electricity intensive, and US industrial power costs and interconnection queues are a larger constraint on new smelter capacity than import competition has been for some years. A tariff does not fix an electricity problem.
Steinberg’s framing on this point is blunt. “High tariffs and uncertain trade policy are the worst-case scenario,” he said. “Building a strong economy requires favourable investment conditions, not ever-higher taxes.”
What importers and US manufacturers should do now
The study is a diagnosis, not a policy change. Nothing in it alters any duty owed on any entry. But it arrives at a moment when several practical decisions are live for companies that buy metal or metal intensive goods.
Re model the April 2026 change if that has not already been done. Any company that treated the move from 50 percent to 25 percent on derivative articles as relief has an error in its landed cost model. The relevant comparison is not rate to rate but duty to duty, and for most derivative articles the duty rose. Finance teams working from headline rates rather than entry summaries are likely understating fourth quarter and full year duty expense.
Audit the metal content calculation. The 15 percent by weight exemption threshold is the single most valuable line in the current structure for manufacturers of lightly metallic goods, and it is a weight test, not a value test. Companies near that threshold should have documented, defensible content calculations prepared before a CBP inquiry rather than after one, and should understand that the Chapter 72, 73, 74 and 76 carve out removes the exemption entirely for goods classified there regardless of content.
Preserve refund rights. With the full value calculation methodology under challenge at the Court of International Trade and a separate dispute over the unpublished December 2025 CBP guidance, importers paying Section 232 duties on derivative articles have a live reason to protest. Protests are inexpensive relative to the exposure and they preserve a position that cannot be recovered retroactively if the litigation succeeds. Entries that liquidate without protest are substantially harder to reach.
Treat the derivative list as expanding. The termination of the formal petition process did not reduce the rate of additions. It removed the notice and comment rhythm that gave companies advance warning. Products adjacent to those already listed should be assumed to be candidates. A compliance review that was current in August 2026 is not current now.
Reconsider US content sourcing economics. The 10 percent rate for products manufactured abroad with US origin metal, now available at an 85 percent threshold rather than 95 percent, is a meaningful structural option for companies with offshore fabrication. The threshold reduction in June 2026 made that pathway reachable for a wider set of supply chains than before. It requires melt and pour or smelt and cast documentation that many suppliers do not currently provide, which is a procurement problem rather than a legal one.
Finally, build contract terms that assume volatility. The current program has been modified substantially in March 2025, June 2025, July 2025, August 2025, April 2026, June 2026 and July 2026, with a further proposed expansion in August 2026. That is not a stable regime. Long term supply agreements covering metal intensive goods should allocate duty risk explicitly rather than relying on a force majeure clause or on the parties’ shared assumption about what the rate will be.
The wider reading
The MEI note lands in the middle of an active political argument. Tariffs on steel and aluminum are being defended in the current US election cycle as a manufacturing jobs policy, and the study’s central finding is that the jobs have not appeared at anything like the scale of the price increase. That finding will be contested, and some of the contesting will be fair. A free market institute in Montreal publishing a critique of US metals tariffs during a US midterm campaign is not a neutral actor, and the Canadian comparison it leans on is affected by Canada’s own trade position.
What is harder to argue with is the structure of the result. A 33.9 percent steel price increase and a 58.0 percent aluminum price increase against a 0.7 percent and 0.6 percent employment response is not a close call about measurement technique. It is a very large number next to a very small one. Even allowing substantial error in both directions, the ratio does not reverse.
For the US companies that buy metal, that ratio is not an abstraction. It is the quote they received this month.
The uncertainty cost nobody invoices
There is a second cost in the MEI note’s argument that does not show up in any price index, and it is the one manufacturers describe most often when asked what the tariffs have done to their businesses.
Count the modifications. March 2025 brought the original 25 percent rate. A few months later it doubled to 50 percent. June 2025 extended the 50 percent rate to derivative products across effectively all origins. July 2025 added copper. August 2025 added 407 derivative categories. A second inclusions window ran through the end of September 2025. December 2025 brought an unpublished CBP memorandum that changed how non metal content was calculated. April 2026 restructured the entire program into four tiers and moved the duty base from metal content to full article value. June 2026 modified rates for agricultural equipment and residential HVAC, extended the 15 percent tier to mobile industrial equipment from preferential trade agreement partners, added aluminum lithographic plates and steel racks to the 25 percent list, and lowered the US content threshold from 95 percent to 85 percent. July 2026 directed Commerce to build an aluminum production incentive program. August 2026 proposed fourteen more derivative categories.
That is eleven material changes in nineteen months, several of them retroactive in practical effect because they altered how entries already in transit would be assessed. A purchasing manager setting a twelve month contract price in that environment is not making a forecast. They are making a bet.
The investment consequence follows directly. Capital committed to a steel intensive production line in the United States today is committed against a duty structure that has changed roughly every two months. The projects the study cites as evidence of a supply response, the Hyundai mill in Louisiana and the Oklahoma smelter, were announced in that environment and will be delivered into a different one. Whether they pencil out depends on a tariff level nobody can currently commit to.
This is the point Steinberg returns to. The complaint is not simply that tariffs are high. It is that they are high and unpredictable at the same time, which raises the hurdle rate on exactly the domestic investment the policy was designed to induce.
Reading the employment number honestly
The study’s finding that US manufacturing lost roughly 60,000 jobs between February 2025 and July 2026, and that the link to metals tariffs is small and statistically insignificant, is worth dwelling on because it will be quoted in both directions.
Tariff supporters will note correctly that the study cannot attribute those losses to the metals program. Tariff opponents will note correctly that the program’s stated purpose was to add manufacturing jobs and that aggregate manufacturing employment went the other way.
Both readings miss the more useful point, which is about distribution rather than totals. Steel employment rose 0.7 percent. Employment in aluminum and other nonferrous metals rose 0.6 percent. Those are the protected sectors, and they grew, slightly. The 2018 precedent the authors cite suggests where the offsetting losses sit: downstream production losses of $3.5 billion by 2021 against upstream gains of $2.25 billion. Protection worked as designed for the protected industries. The cost appeared one step downstream, in the much larger set of firms that buy metal rather than make it.
Fabricated metal manufacturers at 31 cents of metal per dollar of output and automakers at 16 cents are not a small constituency. They employ multiples of the steel and aluminum workforce. A policy that raises their input costs by 13 percent and 6.5 percent respectively in order to add fractions of a percent to upstream employment is making a trade, and the study’s contribution is to put approximate numbers on both sides of it.