Manufacturing Costs

Steel took 151 months to regain its 2008 price peak. Lumber took 156. A rebound is not a reset.

Steel regained its August 2008 peak in March 2021 after 151 months; lumber regained August 2004 in August 2017 after 156. As of July 2026, the 2021 peaks remained unrecovered after 55 and 62 months.

Editorial evidence cutoff: September 9, 2026. Published September 29, 2026. Observation periods are stated throughout; older figures are retrospective evidence.

Steel prices rebounded long before they regained their old high. The steel mill products producer-price index peaked in August 2008, fell sharply and did not reach that level again until March 2021. The elapsed interval was 151 months.

Lumber supplies another long example. Its August 2004 high was not regained until August 2017, a 156-month interval. These histories do not predict how long any current retreat will last. They show why a rebound, a recovery to a previous high and a stable price environment are different claims that need different evidence.

For September 2026 contract and inventory discussions, these older recoveries help distinguish a recent rebound from a return to the selected historical reference.

The first rebound does not close the episode

The steel index fell 40.47% between August 2008 and May 2009. To climb from that trough back to the old peak required a 67.97% increase. A large positive move after the trough could therefore coexist with a price still well below its previous high.

Lumber's August 2004 to May 2009 decline was 34.38%. Regaining the old high from that trough required a 52.40% increase. The recovery arithmetic is asymmetric because the increase is calculated from a smaller base than the decline.

That distinction matters whenever a discussion moves from “prices are up” to “the market has recovered.” Up from which observation? Recovered to which reference? A purchasing team looking at a recent change and one looking at a historical contract base can describe the same period very differently without either calculation being wrong.

Define the clock before measuring the wait

This analysis tracks each index's running high within the saved history beginning January 1990. An episode begins when the index falls below that high and ends at the first subsequent month that equals or exceeds it. Where a high is repeated, the most recent occurrence anchors the interval.

The reported duration is the number of calendar months from the peak to that first recovered observation. It is not the number of days a buyer held inventory, a measure of losses, or a count of months when prices were continuously falling.

That definition produces the 151-month steel interval and the 156-month lumber interval. Prices could rise and fall several times within either one. Calling such a period a long descent would erase those movements. The precise statement is that the index had not yet regained the chosen previous high.

The 2008 crisis is not the only long wait

An investigation built around one exceptional financial-crisis episode would be vulnerable to the charge that it generalizes from a historic disruption. The archive contains other prolonged intervals under earlier highs.

Steel's July 1995 high was not regained until February 2004, an elapsed interval of 103 months. Paper's October 1995 high was not regained until July 2006, an interval of 129 months. These examples do not have identical causes, and they should not be pooled into a universal commodity cycle.

They do show that the distinction between a short rebound and a complete nominal recovery is not unique to 2008 steel. Looking across several source-defined products makes the analytical point stronger while also exposing its limit: different materials have different paths, and a previous waiting time is not a scheduling rule for the next purchase.

The 2021 episodes are still unfinished

As of July 2026, the steel index remained 16.79% below its December 2021 high, 55 months after that peak. Lumber remained 37.01% below its May 2021 high, 62 months later.

Those durations are unfinished observations. In statistical language, they are right-censored: the archive ends before a recovered peak is observed. Treating them as completed recovery times would silently assume an event that had not happened.

Ignoring them is also a problem. An average calculated only from completed episodes would tend to leave out some of the longest waits still in progress. That is why this investigation does not offer a mean time to recovery. It reports identified episodes, their depth and whether the previous high had actually been regained by the final observation.

A nominal high is not a fair-value estimate

Producer-price indexes measure changes in selling prices for defined product groups. They do not provide the value of a particular stockholding or a benchmark for what any buyer ought to pay. An old high is an identifiable reference point, not an economic promise that the market should revisit it.

Inflation, product mix, quality, supply conditions and demand can change during a 12-year interval. The comparisons here are nominal. Regaining an old nominal level is therefore different from regaining an inflation-adjusted level or matching the economics of an earlier purchase.

Likewise, remaining below a peak does not prove that a market is calm. The steel history and lumber history can display considerable movement within an unrecovered interval. A plateau or stable regime requires evidence about recent variation, not merely a negative percentage relative to an old maximum.

Choose the reference that matches the decision

For contract administration, the meaningful reference may be the specified base month. For inventory review, it may be the acquisition cost of a particular lot and its realizable value. For a new quote, it may be a current supplier price and the period during which that quote remains valid.

The historical peak can provide context, but it should not displace those transaction-specific measures. A buyer does not gain a realized saving simply because today's national index is below an old peak, nor does a supplier's price become unreasonable merely because an earlier year was cheaper.

The deeper finding is about disciplined language. Steel's first strong rebound after 2009 was not the same as regaining the August 2008 high; that took until March 2021. Keeping the distinction visible prevents a dramatic percentage from doing the work of a complete market history. It also keeps the investigation from turning a measured past interval into an unsupported prediction.

Sources and calculation

The study uses saved BLS producer-price histories, January 1990 to July 2026, including steel mill products, lumber, resins and paper. Drawdowns exceeding 10% are retained for episode review. Required rebound equals peak divided by trough minus one. Running highs are defined only within this archive, and monthly observations miss intramonth extremes. Unfinished episodes are explicitly retained as censored observations.

Sources and evidence

Evidence period: January 1990 to 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.

data.bls.gov/timeseries/WPU1017

data.bls.gov/timeseries/WPU081

data.bls.gov/timeseries/WPU066

data.bls.gov/timeseries/WPU0913

Published 2026-09-29.