Edition 2: October 2026

This edition covers producer prices through September 2026 for six construction-material groups, with five years of history behind each, and reports what changed since the first edition. It is written for anyone revisiting the material assumptions behind an estimate or weighing an explanation for a proposed increase. It reports market evidence. It does not estimate any contractor's cost and draws no conclusion about any contract.

The read

The pattern from the first edition held. Copper, steel and aluminum products remain far above a year ago; gypsum and concrete remain close to flat. In the producer price index, copper wire and cable is 30.8% above September 2025, steel mill products 27.6% and aluminum mill shapes 25.2%. Gypsum products (+1.4%) and ready-mix concrete (+2.3%) moved little over the same year, and the cement index is 1.2% lower. Softwood lumber is 14.2% above September 2025, but that comparison is with a weak September: lumber is still 1.0% below where it stood five years ago.

September itself was quiet for most of the six. Steel mill products rose 2.2%, the largest monthly increase among them. Copper wire and cable rose 0.8% and set another five-year high. Gypsum products rose 0.8%, also to a five-year high. Aluminum mill shapes slipped 0.1%, ready-mix concrete was unchanged, and softwood lumber fell 2.0%, after a decline in August. BLS marks all six September values preliminary.

A claim that "materials are up 25% or more" still fits the metals and still does not fit gypsum, concrete or cement. Which part of an estimate it bears on depends on what the work is made of; the comparison table is arranged so that a reader can check the categories that matter to a particular estimate rather than a general claim.

The comparison

Context measures, each with its own latest period:

Index levels are not comparable across rows; each series has its own base period. Two-year comparisons, against September 2024: lumber +12.6%; copper wire and cable +46.3%; steel +43.4%; aluminum mill shapes +57.6%; gypsum products +2.3%; ready-mix concrete +2.7%.

Historical charts

Each core index over 5 years, on its own scale, then the raw-metal benchmarks, then the six together rebased to a common start. Charts are drawn from series.csv; every value is in the table under each chart.

Material notes

Standing notes on what each measure covers and what it is not are on the materials page and below the charts. What is new this month, by material:

Softwood lumber fell 2.0% in September and is 15.4% above last December. Its year-over-year figure of 14.2% is measured against a September 2025 that was itself 1.3% below the September before, and the Septembers before that show −2.2%, −15.2% and +12.9%. The index is 1.0% below where it stood five years ago and far below its March 2022 peak of 540.7.

Copper wire and cable set another five-year high in September, at 607.7, and is 19.0% above December. The raw metal also set a five-year high: LME copper averaged 14,474 dollars per metric ton in September, 45.0% above a year ago. The product has risen less than the metal over the year, by 14.2 points (30.8% against 45.0%). In September the two moved almost together, +0.8% for the product and +1.0% for the metal.

Steel mill products rose 2.2% in September and are 24.4% above December and 43.4% above September 2024. The index is still 10.8% below where it stood five years ago, under its December 2021 peak of 449.7, and has climbed from a January 2025 low of 261.6.

Aluminum mill shapes were nearly unchanged in September (−0.1%) and remain below the June high of 394.4. LME aluminum rose 1.0% in the month and remains below its own May high of 3,666 dollars per metric ton. Over the year the product and the metal show nearly the same change, +25.2% and +23.7%, a gap of 1.5 points; over two years the product has risen more, +57.6% against +34.0%.

Gypsum products rose 0.8% in September to a five-year high of 488.9 and are 1.4% above September 2025. The narrower gypsum building materials index, which includes wallboard, was unchanged in the month and is 0.4% below a year ago, 1.8 points below the broader index's yearly change.

Ready-mix concrete was unchanged in September and is 2.3% above a year ago and 2.4% above December, just below its June five-year high of 396.0. The cement index was also unchanged and is 1.2% below a year ago; a year ago, in September 2025, cement stood at its five-year high of 360.5. Over the year, concrete's change exceeds cement's by 3.5 points. Concrete moved without cement.

Standing notes

What each measure covers, why it matters, and what it does not represent. These notes are the same in every edition; the paragraphs above say what is new.

Context

What the evidence shows: for copper, the raw metal and the manufactured product rose in the same direction over the year, and the metal rose further. For aluminum, the two rose by close to the same amount over the year. Cement shipments from U.S. plants for January through June 2026 were 7.2% above the same months of 2025, and June alone was 10.4% above June 2025. Total housing starts in September were 7.4% below a year earlier at a seasonally adjusted annual rate of 1,221 (thousands of units), and 11.4% below December; the five-year low in the series was 1,182, in May 2026. Single-family starts were 2.6% above a year earlier at 859.

What the evidence does not show: why any of these moved. A higher raw-copper price is a plausible reason for a higher wire-and-cable index, and the timing is consistent with it, but the record does not establish that the one caused the other or how much of the product's rise the metal accounts for. The record contains prices, not policy, and does not attribute the rise in steel. Cement shipments rising faster than a year ago while the cement index sits slightly lower is consistent with a market that is busy without being tight, and equally consistent with other explanations the record cannot test. Lower housing starts alongside higher shipments describe two different things: starts measure new residential units, not all construction, and the shipments figure describes an earlier month.

What the evidence cannot connect: the shipments figure describes June; the price indexes describe September. The three-month gap is the source's publishing lag, and no inference across it is made.

Uncertainty

All six September index values are preliminary. BLS revises the most recent four months of a producer price index. This run's August values for several series differ from the preliminary August values the first edition quoted; the revised August levels are in this edition's numbers.json as each series' prior-month value. The first edition stands as published and is not changed by them.

Census also revised its August housing-starts estimates. The August levels in this run are 1,288 for total starts and 900 for single-family starts (thousands, SAAR), both different from the figures the first edition quoted. The single-family series in particular moves widely from month to month; Census starts are survey estimates with wide sampling error and are revised.

The USGS shipments figures are marked estimated by USGS and are published about three to five months after the month they describe. The latest month in this record is June 2026. Shipments are strongly seasonal, so a change from December or from one spring month to the next says little on its own; the comparison with the same months of 2025 is the more useful one, and it relies on prior-year figures USGS may itself have revised. The five-year window for this series is limited to what the current workbook holds, and its chart shows 18 months, not 60.

The World Bank benchmark is a monthly average of daily exchange settlements in dollars per metric ton. It is a world price, not a U.S. delivered price, and it includes no regional premium or fabrication. The gap between the metal and the product is an observation about two series, not a margin.

The edition page states which timestamp chain entry covers this edition's files and which authorities' tokens it holds. A timestamp shows the files existed unchanged by a date; it does not show that a source's figure was correct.

What the run flagged

The collection step flags conditions for the editor before the text is written: stale series, large moves, missing comparisons, preliminary values. They are listed as the run produced them.

The record

Four sources were fetched on 29 October 2026 and retained byte for byte: the BLS public API response for the eight PPI series (ten years each), the World Bank's monthly commodity workbook, the USGS cement workbook for June 2026, and the Census housing-starts time-series workbook. Every figure above appears in numbers.json; the formulas are in dictionary.md; the rows the figures were computed from are in data.csv, and the charted windows are in series.csv. No correction has been made to this edition.