Trade & Sourcing
July’s Tracked Tariff Average Fell While Copper and Steel Rates Rose
The 36-family monthly panel’s observed effective rate fell from 16.03% in July 2025 to 14.66% in July 2026. Copper semi-finished products rose from 0.89% to 45.46%, flat-rolled steel from 37.23% to 47.39%, while semiconductors fell from 3.90% to 1.92%.
Editorial evidence cutoff: September 9, 2026. Published September 29, 2026. Observation periods are stated throughout; older figures are retrospective evidence.
The manufacturing tariff average can fall while the categories a buyer actually needs become more heavily burdened. The July observations available by September 9, 2026 show that divergence clearly: the tracked 36-family effective rate fell from 16.03% in July 2025 to 14.66% in July 2026, while copper semi-finished products and flat-rolled steel moved sharply higher.
The comparison is between matched calendar months. It is not a report of a new tariff announcement and does not establish the legal rate on a September shipment. Its usefulness is analytical: it tests whether a single manufacturing average can represent the products inside it.
For companies translating broad economic signals into quotes, the answer is consequential. An accurate aggregate can still provide an inaccurate guide to a particular purchasing basket.
THE AVERAGE FELL, AND THE UNDERLYING BILL FELL TOO
The constant monthly panel recorded $24.79 billion in customs value and $3.97 billion in calculated duties in July 2025. In July 2026, customs value was $25.48 billion and calculated duties were $3.74 billion. Dividing duties by value produces the lower observed rate.
This is a value-weighted result: total calculated duties divided by total covered customs value. It is not the arithmetic mean of 36 product percentages. Families with larger import values contribute more weight to the aggregate.
The observations come from the saved USITC DataWeb series. The September 9 snapshot already contained both Julys. It therefore supports a cutoff-date comparison, while the distinction between observation month and data retrieval date remains essential to understanding its timing.
COPPER AND STEEL TELL A DIFFERENT STORY
Copper semi-finished products had an observed duty-to-value ratio of 0.89% in July 2025 and 45.46% in July 2026. Flat-rolled steel rose from 37.23% to 47.39%. Fasteners rose from 51.29% to 53.41%.
Semiconductors moved in the other direction, from 3.90% to 1.92%. These contrasts show why the aggregate cannot be interpreted as a common experience shared by every family. A purchasing basket weighted toward metals may encounter a different historical pattern from one weighted toward semiconductors.
The percentages are observed ratios, not a list of statutory changes. Product classifications, country composition and entry characteristics can affect each family's result. The comparison establishes that their measured burdens diverged; it does not isolate which legal provision or sourcing decision generated each movement.
WEIGHTS CAN MOVE EVEN WHEN THE PANEL DOES NOT
Keeping the same 36 families fixes membership, but it does not fix their shares of import value. A family's weight can increase or decrease between months. The aggregate can therefore change through shifts within families, shifts between families, or both.
This is a general problem in reading weighted averages. A category's percentage can rise while the total falls if other categories become cheaper on the measured basis or gain sufficient weight. The headline and the exceptions do not contradict each other. They describe different levels of aggregation.
The current comparison does not attribute the whole fall to a particular composition effect. That would require an explicit decomposition using every family's matched values and rates. Without it, claiming that buyers switched toward lower-duty products would move beyond what the aggregate alone demonstrates.
ONE MATCHED MONTH CANNOT ESTABLISH DURABLE RELIEF
July-to-July comparisons avoid mixing different calendar months, but they still compare only two points. Shipments can be lumpy. A concentration of entries in a particular product or origin can move a family ratio even when the commercial conditions facing an individual manufacturer have not changed in the same way.
The broader seven-month record gives an instructive counterweight. For the same long-running 36-family panel, the pooled rate was 10.21% in January through July 2025 and 15.76% in the corresponding 2026 window. Thus the lower July endpoint coexists with a higher seven-month burden.
Neither window invalidates the other. The single month is useful for observing recent entries, while the pooled window describes a larger set of transactions. Calling the July result a general easing trend would ignore the contrasting accumulation over the year to date. Calling the seven-month result the rate on every current entry would make the opposite mistake.
SEPTEMBER QUOTES NEED A DIFFERENT KIND OF PRECISION
For a buyer quoting work in September 2026, the historical ratio can be a reason to examine assumptions. It cannot determine the duty on the quote. That requires the actual product classification, origin, entry date and relevant eligibility facts, together with the applicable schedule and additional provisions.
The USITC classification guidance explains the role of the tariff schedule. The statistical series answers a different question: what calculated duties and customs values were recorded for a group of entries. Confusing those questions produces unjustified precision even when the arithmetic is correct.
A company can also have a very different product mix from the national panel. A narrow fabricator may care primarily about one metal family. A broad equipment producer may need several. Applying 14.66% to either without matching the purchasing weights would discard relevant information already visible in the family breakdown.
THE USEFUL SIGNAL IS THE DIVERGENCE
The Census definitions distinguish calculated duties from an audited cash-payment measure. They also separate customs value from transport-related import charges. The effective rates here therefore describe one defined layer of the import burden, not a fully delivered cost.
All comparisons retain the stated 36-family monthly panel. The site's broader annual coverage includes 57 families and should not be substituted silently into the same chart. Keeping the panel, denominator and calendar window visible makes the conclusion reproducible rather than dependent on a convenient headline.
Viewed at the September cutoff, the evidence supports a specific warning for interpretation: the July aggregate eased while important metals families rose, and the longer matched window remained higher. The practical gain comes from recognizing those simultaneous facts. They allow a manufacturer to ask a more precise question about its own inputs than whether tariffs, in general, went up or down.
Sources and evidence
Evidence period: July 2025 to July 2026, matched monthly dates. 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.