Manufacturing Costs
Equipment prices fell 42.18% relative to factory hourly earnings over three decades. The automation case is still more complicated.
Annual machinery PPI relative to hourly earnings fell 42.18% from 1990 to 2019. The 2025 ratio remained 1.07% below 2019; July 2026 relative to December 2019 was 3.07% higher.
Editorial evidence cutoff: September 9, 2026. Published September 29, 2026. Observation periods are stated throughout; older figures are retrospective evidence.
Equipment and factory labor both became more expensive in nominal terms over the past three decades. They did not become more expensive at the same pace. Between 1990 and 2019, the ratio of the machinery and equipment producer-price index to manufacturing production-worker hourly earnings fell 42.18%, using complete annual averages.
That finding gives the automation debate a useful historical reference. It does not prove that machines can replace workers cheaply, that automation caused any employment change, or that a particular investment earns its cost. The relative-price history is real. The operational conclusions still need to be demonstrated.
For September 2026 equipment decisions, the long history provides context for relative prices while leaving the actual investment case to the plant and process.
Two rising lines can conceal a falling ratio
The annual-average machinery index divided by hourly earnings was 11.20 in 1990, 8.66 in 2000, 7.05 in 2010 and 6.48 in 2019. Those raw ratios mix index points with dollars per hour, so their levels should not be presented as the price of a machine measured in work hours. Rebasing the ratio to 1990 equals 100 is the clearer display.
The resulting decline means the broad machinery selling-price index rose more slowly than the average hourly earnings measure across that period. This is a relative comparison, not a statement that machines became cheaper in dollars.
That distinction is easily lost when a project discussion switches between the cash cost of equipment, a wage rate and the abstract cost of automation. The first two are inputs to a decision. The third depends on what the equipment actually does, how much labor it changes and what additional costs it introduces.
The historical pattern is not one selected year
The intermediate annual observations show a decline across several decades rather than a result resting entirely on the beginning and end of the sample. The machinery-to-earnings ratio fell between 1990 and 2000, again by 2010, and again by 2019.
Using the earliest and latest individual months also produces a large decline: January 1990 to July 2026 is down 41.93%. That does not make endpoints preferable, but it shows that the long-run direction is not created solely by averaging the first and last calendar years.
There are still important changes inside those decades. The machinery basket evolves, product quality changes, and the composition of manufacturing employment shifts. The ratio describes the movement of two published aggregates. It cannot hold a specific machine model and an identical workforce constant across 36 years.
The apparent reversal fails a full-year test
The latest monthly comparison initially looks like a different story. Between December 2019 and July 2026, machinery prices rose 39.25% while hourly earnings rose 35.10%. Dividing one by the other puts the machinery-to-earnings ratio 3.07% above its December 2019 level.
But the claim that the entire post-2020 period reversed the earlier pattern does not survive a simple check. The 2025 full-year ratio was still 1.07% below the 2019 full-year ratio. The complete-year record and the latest monthly endpoint answer different questions, with 2026 contributing to the apparent change.
An honest assessment keeps both. There is evidence of a recent relative increase using those monthly anchors. There is not evidence from the annual comparison of a sustained reversal covering the whole period since 2019. Publishing that limitation is more informative than declaring a new era from one endpoint.
Hourly earnings are not the full cost of labor
The wage series covers manufacturing production and nonsupervisory employees. It is neither a measure of all manufacturing employees nor a complete compensation cost including every employer expense. Average hourly earnings can also change when the mix of workers and industries changes, even if an individual employee's rate does not.
A plant's labor case therefore needs its own payroll, benefits, staffing, training and scheduling information. A machine that removes a difficult bottleneck may add value without reducing headcount. Another may reduce a task's direct labor while requiring maintenance, programming or inspection elsewhere.
The national hourly measure cannot determine those effects. Calling the ratio a labor-replacement cost would imply a production relationship the data never measures. Its proper role is to describe one part of the environment in which plants considered equipment purchases, not to assign an economic result to those purchases.
The machinery index is not an installed system
A producer-price index describes selling-price movement within its defined product basket. A working automation project includes much more than the purchased machine. Integration, tooling, commissioning, space, utilities and the period before stable output can materially alter the cost and timing of the investment.
Performance matters just as much. A system producing less usable output than expected can have poor economics even if its purchase price compares favorably with wages. Conversely, improved quality or capacity can justify equipment without a simple reduction in paid hours.
This is why the machinery price history and the hourly earnings history should be read as context. The earnings link uses legacy wording in its path; the analysis here uses the production-and-nonsupervisory population actually measured by the source.
Use the long history to sharpen the local question
The relative-price decline makes a worthwhile question visible: how did a plant's own equipment options change relative to the labor and output requirements of its process? That question can be investigated with successive supplier quotations, installed costs and measured production performance.
The capital equipment payback calculator can organize an investment case after those inputs are known. A national ratio should not be inserted as an assumed productivity improvement or an automatic savings rate.
The historical finding remains substantial without those embellishments. The relative machinery index fell 42.18% between 1990 and 2019 annual averages. The recent monthly comparison has moved upward, while the complete 2025 comparison had not. That combination provides context, identifies a development worth watching and sets a clear boundary around what the evidence establishes.
Sources and calculation
The calculation joins saved monthly BLS machinery PPI and manufacturing production-worker earnings, January 1990 to July 2026. Complete calendar-year averages are calculated separately before division. The 36 complete annual pairs end in 2025. The analysis measures relative nominal indexes, not installed equipment cost, total compensation, productivity or a causal effect of automation.
Sources and evidence
Evidence period: Annual 1990 to 2025; matched monthly data through 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/CES3000000008
Published 2026-09-29.