Trade & Sourcing
Fifty-Two Import Families Had Higher Seven-Month Duty Rates. Only Thirteen Rose in Every Matched Month.
For January–July 2026 versus the same months of 2025, 52 of 57 families have higher pooled observed-duty rates, but only 13 are higher in all seven year-over-year month pairs.
Editorial evidence cutoff: September 9, 2026. Published September 29, 2026. Observation periods are stated throughout; older figures are retrospective evidence.
The monthly records available by September 9, 2026 run through July. They allow a matched seven-month comparison with 2025, but do not describe September shipments or establish their legal treatment. For a manufacturer assessing whether an apparent increase has been persistent, the useful question is how much of the result survives a month-by-month examination.
Higher duty rates were widespread across the tracked manufacturing families in the first seven months of 2026. They were much less uniform from month to month. Fifty-two of 57 families have a higher pooled January–July rate than in the same period of 2025. Only 13 have a higher rate in every matched month.
Those are different descriptions of the same records, and neither should stand in for the other. A higher cumulative burden can coexist with months in which the year-over-year rate is lower. A run of positive comparisons can also end when the reference period catches up with an earlier increase.
The distinction changes the question a tariff dashboard should answer. Is the buyer looking at the burden across a completed purchasing window, the direction of the latest comparison, or an uninterrupted pattern? One number cannot identify all three.
THE SEVEN-MONTH BILL IS BROADLY HIGHER
For the complete 57-family panel, calculated duties divided by customs value rise from 9.30% in January–July 2025 to 13.30% in January–July 2026. Each window contains 399 family-month observations. The combined comparison uses 798 records, with the same calendar months and families on both sides.
The calculation pools the dollar numerators and denominators before dividing. It does not average seven published monthly percentages. That distinction matters when a high-value month and a low-value month have different rates: each should contribute in proportion to the customs value actually represented.
Fifty-two families have positive pooled changes. The median family increase is 3.33 percentage points. These are descriptive results from the saved USITC DataWeb monthly observations, not estimates of a uniform increase applied to every product or origin.
THIRTEEN FAMILIES PASS THE STRICTER TEST
The persistence test compares January with January, February with February, and so on through July. A family passes only if its observed rate is higher in all seven pairs. Thirteen families meet that condition.
The list includes copper semi-finished products, aluminum semi-finished products, several steel families, tanks and vessels, lifting equipment, power transmission, transformers and plastics resins. This is a more specific result than saying manufacturing duty rates kept rising. It identifies which tracked families had an unbroken sequence of positive matched-month comparisons.
The persistence test concerns year-over-year differences. It does not mean rates increased sequentially in every month of 2026. A July rate can be lower than June’s and still be above the previous July. Seven positive comparisons establish consistent elevation against the prior-year months, without establishing a continuously rising path within 2026.
THE CONTRARY CASES BELONG IN THE STORY
Five families have lower pooled January–July rates. Semiconductor equipment falls from 3.93% to 1.72%. Specialty metals fall from 10.36% to 8.62%, semiconductors from 2.53% to 2.16%, industrial lighting from 14.66% to 14.48%, and turbines from 4.09% to 3.97%.
Those cases prevent the broad increase from becoming a universal statement. They also illustrate why counting families and weighting by dollars answer different questions. A small number of large categories can influence the pooled panel rate far more than their count suggests.
The complete monthly record and the pooled family changes answer complementary questions: how consistently the sign held, and how large the cumulative burden was relative to customs value. A family can have several negative monthly comparisons and still finish the seven-month window with a higher pooled rate if larger-value months carry higher rates.
REMOVE JULY AND THE PERSISTENCE COUNT CHANGES
An endpoint can make a run look more or less durable. Removing July increases the number of families that are higher in every remaining matched month from 13 to 16. Three families satisfy the January–June test but fail once July is added.
That result is worth publishing because it bounds the claim. There is clear breadth in the seven-month pooled increases, but the exact number with uninterrupted positive comparisons depends on the observation window. A phrase such as relentlessly rising would erase that sensitivity.
The pooled result was also tested on the long-running 36-family panel. Its January–July rate rises from 10.21% to 15.76%. The higher pooled burden therefore does not depend on including the 21 additional families available for the more recent comparison. The two panels have different levels and must remain separately labeled.
A POSITIVE CELL DOES NOT IDENTIFY A LEGAL CAUSE
Each monthly rate combines the goods and origins represented in that month’s imports. A changing mix of products, preferences or entry conditions can change the average. The heatmap cannot isolate a particular legal action, and it should not be presented as an implementation calendar for statutory measures.
The underlying customs values are nominal amounts. The records do not establish whether a higher dollar value reflects more units, higher prices or different specifications. They cannot, by themselves, prove that companies accelerated orders, delayed shipments or changed suppliers in response to policy.
Matching calendar months avoids comparing January–July with a differently composed part of the year, but it is not a full seasonal or causal model. The persistence count is also not a statistical significance test. Its role is to describe how often the sign holds across a clearly specified set of comparisons.
MAKE THE WINDOW VISIBLE BEFORE DRAWING THE CONCLUSION
A useful monitoring table would keep three fields together: the latest matched-month difference, the pooled year-to-date rate and the count of positive monthly comparisons. Each should carry its observation window and panel definition.
That design makes revisions easier to interpret. A newly negative latest month may weaken the persistence claim without undoing the higher pooled burden. A larger year-to-date increase may reflect the weights of high-value months, even when the number of positive comparisons is unchanged.
The records support a strong but specific conclusion. Higher observed duty intensity was broad across the 57 families over the matched seven-month window. An uninterrupted sequence was much less common. Publishing both results gives the reader a usable picture of breadth and variation, without asking one headline to do the work of the entire calendar.
Source window: January–July 2025 and January–July 2026. Monthly data were fetched September 9, 2026. Rates use reported calculated duties and covered customs values, consistent with the Census statistical definitions.
Sources and evidence
Evidence period: January–July 2025 versus January–July 2026. The frozen evidence record lists the source files and verified hashes available September 9, 2026. Source revision: 5e4fb7726c3d40060c0151c086c903baae856cab. Later live-data updates do not alter the historical evidence in this article.
census.gov/foreign-trade/guide/sec2.html
Published 2026-09-29.