Published: June 10, 2026

A new working paper built on U.S. Census Bureau microdata makes a claim that should change how the defense manufacturing community argues for workforce investment.  American firms are not running out of ideas.  The economy is getting worse at converting ideas into growth.  If the authors are right, the chokepoint in the defense industrial base is not inventing new technology.  It is putting new technology to work on the shop floor.  And that is a workforce problem.

What the Study Found

The paper, Growth Is Getting Harder to Find, Not Ideas, comes from Teresa Fort, Nathan Goldschlag, Jack Liang, Peter Schott, and Nikolas Zolas, a team with deep roots in Census Bureau research.  They built a 45-year panel covering the universe of U.S. private-sector employer firms from 1977 through 2021, linking Census business records to patent grants, R&D spending, employment, payroll, and sales.  The starting puzzle is familiar.  R&D inputs roughly tripled over the period while sales-per-worker growth stayed flat.  The standard reading of that divergence, from a widely cited 2020 paper, is that ideas are getting harder to find.

The new study splits the question in two.  First, are research inputs still producing ideas?  Yes.  Patents per real R&D dollar rose from 1977 to 2017 across the universe of firms, and the responsiveness of patenting to R&D spending nearly doubled, meaning that a 10% rise in R&D spending produced about a 2.6% rise in patents in the late 1970s but about a 4.8% rise by 2017.  In the study’s terms, the elasticity of patent grants rose from 0.26 to 0.48.  Quality-adjusted measures, including citations and breakthrough patents, show the same pattern. 

Second, are ideas still producing growth?  Here the news is worse.  The link between a firm’s patent growth and its productivity growth remains positive and steady.  But average firm growth, after accounting for innovation, has declined.  The research engine is fine.  What has weakened is the conversion of invention into expanded, scaled production.

The 73-12 Problem

Manufacturing firms produced 73% of breakthrough patents in the 1970s.  By the 2010s their share was 12%.  Manufacturing’s share of patenting overall fell from a majority to less than a third, while information technology, professional services, and management-related sectors took up the slack.  The firms that build physical things now invent a shrinking share of what they will be asked to produce.  For a defense industrial base that must turn designs into hulls, airframes, and munitions at rate, that widens the distance every idea must travel before it becomes output.  Someone has to carry it across.  That someone is a workforce.

The Supplier Base Feels It First

Here is what that means in practice.  The technologies our small and mid-sized defense suppliers are asked to adopt are mostly proven.  Federal manufacturing programs mature them.  Primes specify them.  Equipment vendors will gladly sell them.  A 40-person machine shop asked to qualify a new welding process or stand up an additive cell does not have an idea problem.  It has nobody certified to write the weld procedure, nobody trained to run the new equipment, and no slack to send its best technician to a two-week course while deliveries are due.  The scale of that problem is documented.  A 2024 Deloitte and Manufacturing Institute talent study projects a net need for 3.8 million new manufacturing workers between 2024 and 2033, and finds that 1.9 million of those jobs could go unfilled if the skills and applicant gaps are not closed.  Diffusion does not fail in a strategy document.  It fails at the level of a single unfilled technician seat in a single supplier trying to qualify for new work.

This is the study’s quiet message for our field.  The authors find that promising technologies do not generate broad gains when firms cannot expand production, find skilled workers, or spread adoption across supply chains.  Every one of those failure modes runs through people.  Workforce development is not a substitute for innovation investment.  It is what makes innovation investment pay.

The Part Nobody Photographs

The hard part is that diffusion is invisible.  A research center gets a ribbon cutting, a congressional visit, and a press release.  A supplier that finally qualifies a process because two technicians completed a credential gets a purchase order and silence.  Public investment tends to reward what photographs well, and this paper says the growth payoff has moved to the work that does not: training aligned to specific production processes, supplier qualification support, and the patient work of staffing thousands of small firms to take on new processes.  So, the question this research leaves on the table is the one worth arguing about.  When the next defense production-scaling initiative is scoped, will the workforce line be sized to the technology it is expected to put into production, or will it remain a rounding error at the bottom of someone else’s budget?

---