Manufacturing Costs

Paper and lumber share a raw-material story. Their monthly prices barely moved together in 36 years.

Across January 1990 to July 2026, paper/lumber level correlation was 0.69, monthly-change correlation -0.11. Direction agreed in 205 months and opposed in 204, with 29 zero-containing comparisons.

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

Paper and lumber make an appealing shared story. Both are associated with forest products, both can appear in a manufacturer's purchasing discussions, and both price indexes have risen over long periods. Yet their monthly price changes barely moved together across 36 years of observations.

From January 1990 through July 2026, the correlation between their price-index levels was positive 0.69. The correlation between their monthly percentage changes was negative 0.11. That difference challenges the use of one as a convenient short-term price proxy for the other. A common origin does not establish a common invoice adjustment.

A September 2026 packaging review can use this history to test whether one input benchmark reliably describes another before applying a shared escalator.

The long chart makes a persuasive first impression

Two nominal price series can rise over decades and produce a strong correlation between their levels. That calculation records an association in where the lines sit through time. It does not answer whether increases in one occur during the same months as increases in the other.

For a buyer reviewing a recent price adjustment, the second question is often more relevant. If the underlying materials changed very differently during the contract period, pointing to the similar shape of a 36-year chart does not explain the adjustment being requested today.

The paper and lumber comparison illustrates that distinction with unusually little ambiguity. Its positive level relationship becomes slightly negative when measured as monthly percentage changes. The change correlation is close enough to zero that it should not be portrayed as a useful inverse pricing rule either. It mainly undermines the claim that the two move together reliably at that frequency.

Count the agreements, not just the correlation

There are 438 matched monthly changes in the saved histories. Paper and lumber moved in the same nonzero direction in 205 months and in opposite directions in 204. In the remaining 29 months, at least one of the indexes was unchanged.

These counts make the result easier to interpret than a coefficient alone. A buyer using the direction of one series as a description of the other's monthly movement would repeatedly encounter disagreement. The result does not show that either series is uninformative about its own defined market.

Nor are those months independent trials from an unchanging process. A sustained movement can last across several observations, and economic conditions differ across decades. The counts are a description of the archive, not an estimated probability that the next month's directions will agree or a trading strategy waiting to be applied.

The pandemic does not explain away the mismatch

Lumber's extraordinary pandemic movements raise an obvious objection: perhaps a few extreme observations obscure an otherwise consistent relationship. Removing 2021 leaves the monthly-change correlation at negative 0.10 across 426 observations.

Another check limits the influence of unusually large moves without removing a whole year. Capping each series' changes at its own 5th and 95th percentiles leaves the correlation at negative 0.08. Taking nonoverlapping calendar-quarter endpoint changes instead of monthly changes produces negative 0.13 across 145 observations.

The era comparisons point in the same broad direction. Monthly-change correlations were negative 0.20 in 1990 to 2007, negative 0.09 in 2008 to 2019 and negative 0.09 in 2020 to July 2026. The result is not confined to a single crisis or an isolated lumber spike. That strengthens the case for treating the two price histories separately in short-horizon purchasing analysis.

The product definitions are part of the finding

The paper series is WPU0913. It measures paper, not paperboard, corrugated boxes or the full delivered cost of packaging. The lumber series is WPU081, which measures lumber rather than all products made from wood. Calling either measure a general packaging index expands its scope beyond the source.

That matters because a finished packaging purchase may contain material conversion, manufacturing labor, printing, hardware, recovery content and delivery. A paper producer's selling-price index cannot identify the movement of all those components. A pallet specification can also differ substantially from the lumber mix represented by the broad index.

The paper history and lumber history remain useful where their actual product scope fits the question. Their legacy route names should not be read as a substitute for the underlying series definition. The correct label is a basic requirement for an honest benchmark comparison.

A weak correlation does not sever the supply chain

The analysis does not prove that paper and lumber markets have no economic connection. Related industries can respond to some common influences while product-specific supply, processing requirements, inventories and demand dominate monthly price changes.

Those are plausible explanations, not effects identified by this calculation. Establishing their contribution would require additional data: relevant fibre markets, energy costs, output, trade, inventory or product-specific demand. A correlation cannot select among them, and adding a persuasive supply-chain narrative would not solve that evidence gap.

The narrower result is sufficient for the purchasing question. The saved record does not support treating the monthly direction of paper prices as a reliable description of lumber prices, or the reverse. It also does not identify a useful lagged forecasting relationship; that would require a separate design with performance evaluated outside the period used to choose it.

Ask for a cost bridge that fits the purchase

When a supplier cites a broad forest-products story, the constructive response is to ask how the relevant cost reaches the quoted item. Which material is covered? What share of the price changes with it? Which dates define the adjustment? Which processing and delivery costs are separately supported?

A buyer can then compare the stated exposure with the appropriate observations, rather than assume that common raw-material ancestry is enough. The packaging cost calculator can help organize a specific bill once its own inputs are established. It cannot turn a broad paper index into a carton quotation.

The deeper value of the 36-year check is its restraint. Two rising lines looked related, but the month-to-month evidence did not sustain a shared pricing shortcut. Preserving that negative result makes the data more useful. It tells a reader exactly where a plausible explanation stops being a dependable measurement.

Sources and calculation

This analysis uses saved BLS paper and BLS lumber histories, January 1990 to July 2026: 439 levels and 438 monthly changes. Correlations are Pearson coefficients. Quarter changes use March, June, September and December endpoints. Zero changes remain in correlations but are identified separately in directional counts. No causal or predictive result is claimed.

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/WPU0913

data.bls.gov/timeseries/WPU081

Published 2026-09-29.