LEARN / REINDUSTRIALIZATION
The Manufacturing Workforce Constraint
Capital can buy a building in eighteen months and cannot buy a journeyman machinist in eighteen months. Where the trades bind, the constraint on new capacity is a training pipeline whose duration no amount of investment compresses.
Key points
- The constraint is occupational rather than sectoral. Naming which trades bind is more useful than citing an aggregate shortage.
- Training duration for the skilled trades runs in years and cannot be shortened by spending, which makes workforce a lead time rather than a cost.
- A trained worker can leave, so the firm that paid for the training does not capture all of its value. This is why firms underinvest in skills they all depend on.
- Wage-driven and pipeline-driven explanations of a shortage predict different remedies, and the evidence needed to distinguish them is wage data rather than vacancy data.
- A job posting and a BLS job opening are different objects. Shortage claims built on posting counts are not measuring what the federal series measures.
Which trades bind
Manufacturing employs roughly 12.6 million people, and a shortage in that aggregate would be a different problem from the one firms actually report. What binds is narrower: machinists and tool and die makers, welders working to code on pressure vessels and structural assemblies, industrial electricians and controls technicians, and the process engineers who qualify a line before it can run at rate.
These occupations share a property that distinguishes them from most labor. Competence in them is verified by demonstration rather than by credential alone. A welder is qualified to a procedure on a material at a thickness, and that qualification does not transfer automatically to a different procedure. The resulting labor market is a set of narrow, partially non-substitutable pools rather than a single supply of workers, which is why national employment totals reveal so little about whether a particular plant can be staffed.
Data$56,150 · −2%
Median annual wage for machinists, May 2024. Tool and die makers, $63,180. The Bureau of Labor Statistics projects overall employment of machinists and tool and die makers to decline 2 percent between 2024 and 2034. Source: BLS Occupational Outlook Handbook, Machinists and Tool and Die Makers.
Series and provenance
- Agency
- U.S. Bureau of Labor Statistics
- Program
- Occupational Outlook Handbook and Occupational Employment and Wage Statistics
- Series
- Machinists and Tool and Die Makers
- Measure
- Median annual wage and ten-year employment projection
- Units
- Dollars per year; percent change
- Adjustment
- Median, not mean
- Period
- Wages May 2024; projection 2024 to 2034
- Latest
- Machinists $56,150; tool and die makers $63,180; overall employment projected to decline 2 percent
- Source tier
- Primary
- Retrieved
- August 9, 2026
The projection is for the occupation as a whole and reflects expected productivity gains from CNC and automation. It does not describe whether individual employers can fill a given position, which is a separate question that this series was not built to answer.
Why capital cannot compress the timeline
A registered apprenticeship in the skilled trades typically combines several thousand hours of supervised work with classroom instruction, structured so that the trainee accumulates experience across the range of situations the work presents. The duration exists because judgment under varied conditions is what separates a competent tradesperson from someone who has been shown the procedure once.
Money accelerates almost every other input to a new plant. It buys steel, permits expediting, overtime on construction, and equipment from a second supplier at a premium. It does not shorten the interval between a person starting an apprenticeship and that person being able to run a job unsupervised, because the binding input is supervised hours rather than dollars. A firm that decides in 2026 that it needs additional journeymen is making a decision whose effect arrives several years later regardless of what it is willing to pay.
Institute analysisWorkforce belongs in capacity planning as a lead time, alongside transformers and long-lead castings, rather than as an operating cost. Treated as a cost it looks adjustable. Treated as a lead time it determines the earliest date a facility can reach rate, and that date is often later than the construction schedule implies.
The commons problem
Training produces a worker who can leave. The firm that funds an apprenticeship bears the full cost and captures only the portion of the benefit that remains with it, while a competitor that trains nobody can hire the finished tradesperson at market wage and capture the rest. Every firm in a region faces that calculation, and the rational individual response is to train less than the industry collectively needs.
The outcome is a shortage that no single firm caused and no single firm can fix. Managers are behaving sensibly given what they face, which is an investment whose returns are partly captured by competitors, and the shortage persists precisely because each individual decision is sound. Arrangements that change it work by changing who bears the cost: joint apprenticeship programs across employers, public training capacity, or contractual structures that let a firm recover some of the investment if the worker leaves early.
Measuring a shortage
Claims about labor shortages usually rest on counts of unfilled positions, and the federal measure of those is the Job Openings and Labor Turnover Survey. JOLTS is more restrictive than the word “opening” suggests.
Measurement noteBLS counts a position as a job opening only if it satisfies all three conditions: a specific position exists and there is work available for it, the job could start within 30 days, and the establishment is actively recruiting from outside for it. Openings are counted on the last business day of the month. A job posting satisfies none of these tests by default, so a shortage argument built on posting counts is using a broader and less disciplined series than the one BLS publishes.
Even the disciplined series cannot settle the question on its own. A position that stays open is consistent with a pipeline that produces too few qualified people, and equally consistent with an offered wage below what the available people will accept. The two explanations predict different things about compensation, and wage data rather than vacancy data is what distinguishes them.
Measurement gapNo federal series connects an occupational vacancy to the wage offered for it. JOLTS counts openings by industry, the Occupational Employment and Wage Statistics program reports wages by occupation, and the two are not published as a pair. A claim that a trade is short of people at the wage being offered therefore cannot be checked against any single published table, which is why the shortage argument has run for a decade without resolving.
Common misconceptions
That automation resolves the constraint. Automated equipment requires controls technicians, maintenance electricians and programmers, which substitutes one scarce trade for another rather than removing the requirement. The occupations change; the dependence on trained people does not.
That a national employment total describes the labor market a plant faces. Qualification is specific to procedure, material and equipment. A region with ample manufacturing employment can still lack the twelve people a particular line needs, and the aggregate will not show it.
That the shortage is uniform. Some manufacturing occupations face genuine scarcity while others face weak demand. Treating manufacturing labor as one market produces policy aimed at an average that describes nobody.
What the evidence says, and where it is contested
ContestedWhether the constraint is pipeline or price is genuinely disputed. One position holds that training capacity and enrollment have not kept pace with demand for the trades, so the people do not exist at any wage in the short run. A second holds that a shortage at a given wage is not a shortage, and that persistent vacancies alongside flat compensation indicate an unwillingness to bid rather than an absence of workers. The federal projection cuts against the first: BLS expects employment of machinists and tool and die makers to fall 2 percent over the decade to 2034, on the expectation that CNC equipment raises output per worker. An occupation that federal projections show shrinking, and that employers describe as impossible to staff, is either two different labor markets sharing one occupational code or a projection that has not absorbed the buildout. The distinction matters because the first calls for training capacity and the second calls for nothing at all. Resolving it requires occupation-level wage series read against occupation-level vacancy series, and much published commentary presents the vacancy half without the wage half.
Related Institute research
Who Trains the Next Generation? The Workforce Training Commons Problem
Why firms underinvest in the skills they all depend on.
What Is Reindustrialization?
Capacity as a system in which the scarcest input sets the ceiling.
Why Reindustrialization Requires More Electricity
The other lead time that determines when a facility can reach rate.
Sources
- U.S. Bureau of Labor Statistics, Job Openings and Labor Turnover Survey. bls.gov
- U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics. bls.gov
- U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, Machinists and Tool and Die Makers. bls.gov
- U.S. Bureau of Labor Statistics, Current Employment Statistics, series CES3000000001. fred.stlouisfed.org
- U.S. Department of Labor, Registered Apprenticeship program data. apprenticeship.gov
- National Center for Education Statistics, completions data. nces.ed.gov
Reference entry maintained by the Institute for American Manufacturing & Technology. Figures are drawn from primary sources and cited above. Where the Institute states a position rather than a finding, it is marked as such.