Manufacturing’s Stagnation: The Sector That Stopped Growing

There is a story that American manufacturing tells about itself. It is a story of relentless efficiency gains, of robots replacing assembly lines, of computerized supply chains squeezing waste out of every process. In this story, manufacturing is the sector that shows the rest of the economy how to do more with less. Output per hour goes up. Unit costs go down. American factories compete with the world by being smarter, faster, and leaner.

For two decades, the story was true. Between 1990 and 2010, manufacturing productivity — measured as output per hour worked — more than doubled, rising from an index of 48.4 to 102.9 (2017=100). This was the golden era of American manufacturing efficiency. Automation was sweeping through factories. Computer-controlled machinery replaced manual processes. Enterprise resource planning systems revolutionized inventory management. The sector was, by any measure, getting dramatically more efficient.

Then something happened. Or more precisely, something stopped happening.

Since 2010, manufacturing productivity has been essentially flat. The index that doubled in twenty years has barely moved in fifteen. In Q4 2025, it stood at 99.7 — almost exactly where it was in 2017, and actually lower than its 2010 reading of 102.9. The sector that once embodied American ingenuity has spent a decade and a half running in place.

To understand why manufacturing’s stagnation matters, you first have to understand how extraordinary its earlier growth was. In 1990, the manufacturing productivity index stood at 48.4. The sector was already being reshaped by automation — computer numerical control (CNC) machines had been transforming metalworking since the 1970s, and the first generation of industrial robots was well established. But the real revolution was just beginning.

Three forces converged in the 1990s to supercharge manufacturing productivity. The first was information technology. Enterprise resource planning (ERP) systems, pioneered by companies like SAP and Oracle, gave manufacturers real-time visibility into their entire supply chain for the first time. Inventory that once sat in warehouses for weeks could now be ordered just in time. Defects that took days to detect could be caught in minutes. The entire production process became faster, leaner, and more predictable.

The second force was trade liberalization. NAFTA, implemented in 1994, and China’s entry into the World Trade Organization in 2001 opened vast new markets for American manufacturers — and simultaneously exposed them to fierce competition. The survivors were the ones who invested in technology and efficiency. Those who couldn’t keep up simply disappeared, and their market share flowed to the more productive firms. This compositional effect — less-efficient plants closing, more-efficient plants expanding — was itself a powerful source of measured productivity growth.

The third force was advanced automation. Industrial robots became cheaper, more capable, and more widely deployed. The automotive industry led the way, but by the early 2000s, robots were welding, painting, assembling, and packaging goods across every manufacturing subsector. Each robot displaced multiple workers while maintaining or increasing output. Productivity per remaining worker soared.

The numbers tell the story. From 1990 to 1995, the index rose from 48.4 to 57.2 — a gain of 18%. From 1995 to 2000, it jumped to 70.8, another 24% gain. The pace was extraordinary. By 2005, the index had reached 90.2. By 2010, it hit 102.9. In two decades, output per hour in American manufacturing had more than doubled.

Compare this with the broader economy. Nonfarm business productivity — the headline measure covering the entire private sector — rose from 57.9 to 95.0 over the same period, a gain of 64%. Impressive, but nowhere close to manufacturing’s 113% surge. For twenty years, manufacturing was the economy’s efficiency leader, pulling the national averages upward and demonstrating what technology could accomplish when deployed at scale.

This was the era that made “American manufacturing” synonymous with “American productivity.” Policy papers celebrated the sector. Business school case studies featured it. Economists pointed to it as proof that technological progress could deliver sustained, compounding efficiency gains. It was, by all accounts, a success story.

It was also about to end.

The chart below shows two lines that once moved in rough parallel and then sharply diverged. Manufacturing productivity (amber) raced ahead of nonfarm business productivity (slate) from 1990 through 2010, reflecting the factory automation revolution. Then manufacturing flatlined while the broader economy kept improving.

Look at the right side of the chart. In 2010, manufacturing productivity (102.9) was well above nonfarm business productivity (95.0). Manufacturing was the leader. By 2025, the positions had reversed completely: nonfarm business productivity stands at 119.6 while manufacturing remains stuck at 99.7. The leader became the laggard.

This reversal is one of the most important structural shifts in the American economy over the past generation. The sector that defined productivity growth for decades is no longer contributing to it. The torch has passed to services — software, finance, healthcare technology, logistics — sectors that are themselves being transformed by a new wave of digital tools. Manufacturing, the sector that once embodied that transformation, has been left behind.

The divergence is stark in the numbers. Between 2010 and 2025, nonfarm business productivity grew by 25.9% (from 95.0 to 119.6). Manufacturing productivity declined by 3.1% (from 102.9 to 99.7). These are not rounding errors. Over a fifteen-year span, the entire manufacturing sector produced less output per hour worked than it did at the start of the period.

What makes this particularly troubling is that the 2010–2025 period was not a recession. The economy expanded almost continuously (with a brief COVID contraction in 2020). Interest rates were historically low for most of the period, making capital investment cheap. Yet manufacturing firms, despite favorable conditions for investment, could not improve their productivity. The problem was not cyclical. It was structural.

To understand why manufacturing productivity soared and then stalled, you have to look at what happened to manufacturing employment. The chart below tells the other half of the story — and it is not a comfortable one.

In 1990, the manufacturing employment index stood at 140.8 (2017=100). American factories employed roughly 17.7 million workers, down from a peak of 19.6 million in 1979 but still a massive employer. The factory floor was a path to the middle class for millions of Americans without college degrees. It was, alongside construction and mining, one of the pillars of blue-collar prosperity.

Then the workforce was decimated. By 2000, the employment index had edged down to 138.1 — a modest decline, and one that could be attributed to normal efficiency gains. But between 2000 and 2010, the floor fell out. The employment index collapsed from 138.1 to 93.4 — a staggering loss of more than five million jobs. In raw terms, manufacturing employment fell from about 17.3 million to 11.5 million in a single decade.

Two shocks hit simultaneously. The China shock — the surge of imports following China’s WTO accession in 2001 — devastated labor-intensive manufacturing sectors like textiles, furniture, and electronics assembly. Academic research has documented that communities exposed to Chinese import competition lost not just manufacturing jobs but also retail, construction, and service jobs as the local economic multiplier collapsed. And the Great Recession of 2007–2009 delivered the coup de grâce, eliminating another 2.3 million manufacturing positions in less than two years.

Here is where the two lines tell a deeply revealing story about how productivity actually works. Notice the inverse relationship from 1990 to 2010. As employment plunged (red line falling), productivity surged (green line rising). This was not a coincidence. It was the mechanism.

When factories close and their workers are laid off, the remaining factories tend to be the more productive ones. When a firm invests in a robot that replaces three assembly workers, the firm’s output stays the same (or rises slightly) while its labor input drops sharply. Output per hour — productivity — jumps. Across an entire sector, this process of job destruction and capital substitution shows up as rising productivity. The denominator shrinks faster than the numerator.

This is the uncomfortable truth about manufacturing’s productivity miracle: much of it was productivity through subtraction. The sector did not simply learn to produce more with the same workforce. It learned to produce roughly the same amount with a much smaller workforce. The millions of workers who lost their jobs were not retrained for higher-productivity roles within manufacturing. They were simply gone — to lower-paying service jobs, to disability rolls, to early retirement, or to long-term unemployment.

And once the subtraction was complete, the productivity gains stopped.

Since 2010, manufacturing employment has stabilized. The index rose from 93.4 to 101.6 by 2025, a modest recovery of about one million jobs. But notice what happened to productivity during this period of employment stability: it went nowhere. In fact, it declined slightly. Without the powerful “denominator effect” of massive job losses, manufacturing could not generate the productivity gains it once had. The sector was not actually becoming more efficient. It was simply shedding workers. And when it stopped shedding workers, it stopped appearing to improve.

This is the paradox at the heart of the manufacturing productivity story. The gains that looked so impressive on paper were partly an artifact of destruction. When five million workers disappear and output holds roughly steady, productivity doubles. When the workforce stabilizes, the illusion of progress evaporates.

The flatline in manufacturing productivity is not a mystery. It has multiple, well-documented causes that reinforce each other.

The low-hanging fruit has been picked. The first generation of automation — replacing manual assembly with robots, paper records with digital systems, batch production with lean manufacturing — delivered enormous one-time gains. But each subsequent round of automation delivers smaller improvements. The tenth robot on a production line adds less than the first. The second ERP upgrade is less transformative than the initial implementation. This is the natural trajectory of any technology: explosive early gains followed by diminishing returns.

The compositional effect reversed. During the 1990s and 2000s, the least productive plants closed while the most productive ones survived. This survivorship bias inflated measured productivity growth. By 2010, the culling was largely complete. The remaining manufacturing base was already relatively productive, and there were fewer weak plants to close. Some of the post-2010 employment recovery came from re-opening or expanding less-efficient facilities, which actually lowered average productivity.

Investment slowed. Despite low interest rates, manufacturers were cautious about capital expenditures after the Great Recession. Many firms had been burned by overcapacity in the 2000s and were reluctant to invest aggressively. Corporate strategy shifted toward share buybacks and dividend payments rather than new equipment. The capital intensity of manufacturing — the amount of plant and equipment per worker — grew much more slowly in the 2010s than in the preceding decades.

The product mix shifted. Higher-productivity manufacturing subsectors like semiconductors and computers, which had driven much of the aggregate productivity growth, saw their domestic production share shrink as fabrication moved offshore. The manufacturing that remained in the United States was disproportionately in lower-productivity subsectors: food processing, beverages, chemicals, and durable goods assembly. This compositional shift dragged down the overall productivity index.

Measurement issues may play a role. Some economists argue that BLS productivity statistics do not fully capture quality improvements in manufactured goods. A car assembled in 2025 is more sophisticated than one assembled in 2010, with advanced electronics, safety systems, and materials. If quality improvements are undercounted in output measures, productivity growth would be understated. However, measurement problems alone cannot explain a fifteen-year flatline — they can only partially mitigate it.

When productivity stalls, costs surge. This is a mathematical certainty. Unit labor costs (ULC) — the amount of labor compensation required to produce one unit of output — are the mirror image of productivity. If workers are paid more (or the same) and produce the same amount per hour, each unit of output costs more to make. The table below traces this dynamic through manufacturing’s entire arc.

The ULC trajectory is a story in three acts. In the first act (1990–2005), unit labor costs in manufacturing actually fell — from 92.2 to 85.1. This was the golden era. Productivity was rising so fast that it more than offset wage increases. American manufacturers were becoming more cost-competitive with every passing year. This was the period when “Made in America” was becoming economically viable again, at least for high-value goods.

In the second act (2005–2015), the cost advantage eroded. ULC rose from 85.1 to 96.2 — a 13% increase in a decade. Productivity was stalling, but wages continued to rise (albeit slowly). The math was no longer working in manufacturing’s favor. Each year, producing a unit of goods cost a little more than the year before. The sector was losing its cost edge.

In the third act (2015–2025), costs exploded. ULC surged from 96.2 to 137.3 — a 43% increase in just ten years. The pandemic accelerated the trend, as labor shortages forced manufacturers to raise wages while supply chain disruptions constrained output. But the post-pandemic numbers have not retreated. ULC in 2025 stands at 137.3, up 20% from even the pandemic-elevated level of 2020.

To put the full arc in perspective: in 2005, it cost 85.1 cents of labor (indexed) to produce a unit of manufacturing output. In 2025, it costs 137.3 cents. That is a 61% increase in unit labor costs over twenty years. And because manufacturing productivity is flat, there is no efficiency offset. Every penny of cost increase flows directly into either higher prices or lower margins.

The arc of American manufacturing over thirty-five years can be read as a series of turning points, each reshaping the sector’s trajectory. What follows is not a comprehensive history but rather the key inflection points that show up in the BLS productivity data.

Manufacturing’s productivity stagnation is not just a sectoral story. It is the sector-level version of a pattern that has haunted the American economy for decades: the exhaustion of easy productivity gains.

In Episode 1 of this series, we documented the Great Divergence — the decoupling of productivity growth from wage growth that began in the 1970s. In Episode 2, we explored the 1973 break that ended the postwar productivity boom. Manufacturing’s arc from 1990 to 2025 is a compressed version of the same story. The sector had its own productivity miracle, driven by technology and globalization. And then the miracle ended.

The parallels are striking. Just as the broader economy saw rapid productivity growth from 1948 to 1973, manufacturing saw rapid growth from 1990 to 2010. Just as the broader economy hit a wall after 1973, manufacturing hit a wall after 2010. And just as economists have debated for fifty years why aggregate productivity growth slowed, the same debate now surrounds manufacturing.

The implications extend well beyond the factory floor. Manufacturing accounts for roughly 11% of GDP but punches above its weight in terms of R&D spending, export revenue, and supply chain effects. When manufacturing productivity stalls, it drags on the entire economy. It means higher prices for goods. It means reduced competitiveness in global markets. It means the sector that was supposed to benefit most from robotics and AI has instead become a case study in what happens when the easy gains run out.

The obvious question is whether a new wave of technology — artificial intelligence, advanced robotics, 3D printing, the Industrial Internet of Things — can restart manufacturing productivity growth. The optimistic case is plausible. AI-driven predictive maintenance could reduce downtime. Collaborative robots (“cobots”) could augment workers rather than replace them. Additive manufacturing could reduce waste and enable customization at scale. Digital twins could optimize production processes in ways that were impossible even five years ago.

But the optimistic case must contend with stubborn realities. First, these technologies are still in early deployment. Adoption is slow, expensive, and uneven. Second, the manufacturing firms that remain in the United States are disproportionately in subsectors — food, chemicals, large-scale assembly — that are harder to automate than the electronics and computing subsectors that drove earlier productivity gains. Third, the workforce challenge is acute. Manufacturing workers are aging, retirements are accelerating, and younger workers are not entering the sector at replacement rates. A shortage of skilled technicians and engineers constrains the pace of technology adoption.

There is also the reshoring question. The current policy environment — tariffs, industrial policy, supply chain security concerns — is encouraging some manufacturing to return to the United States. But reshored production is often less efficient than the overseas production it replaces. Building new semiconductor fabs in Arizona or battery plants in Georgia is strategically valuable, but these facilities are starting from scratch and will take years to reach optimal efficiency. In the short term, reshoring may actually lower measured manufacturing productivity by adding output at below-average efficiency levels.

The honest answer is that no one knows whether manufacturing productivity will recover. The technology exists to drive a new wave of gains. But the technology existed in the 2010s, too, and it did not move the needle. The gap between technological potential and actual deployment is vast — and closing it requires not just better machines but better management, better workforce training, and better capital allocation. None of these are guaranteed.

The BLS data on manufacturing tells a clean, uncomfortable story that resists easy narratives. Here are the facts, stripped of interpretation.

Productivity doubled, then stopped. The manufacturing output-per-hour index went from 48.4 in 1990 to 102.9 in 2010 — a compound annual growth rate of 3.8%. From 2010 to 2025, the index went from 102.9 to 99.7 — a compound annual growth rate of negative 0.2%. There is no ambiguity in this reversal.

Nonfarm business surpassed manufacturing. In 2010, manufacturing productivity (102.9) exceeded nonfarm business productivity (95.0) by 8%. By 2025, nonfarm business (119.6) exceeds manufacturing (99.7) by 20%. The gap has swung from +8% to -17% — a 25-point reversal.

Employment collapsed, then stabilized. Manufacturing employment fell 34% from 1990 (index 140.8) to 2010 (93.4), then recovered modestly to 101.6 by 2025. The massive job losses of 1990–2010 mechanically inflated productivity statistics. The employment stabilization of 2010–2025 removed that statistical tailwind.

Unit labor costs exploded. ULC fell from 92.2 (1990) to 85.1 (2005) as productivity outran wages, then surged to 137.3 (2025) as wages grew while productivity stalled. The 61% increase in ULC from 2005 to 2025 is among the largest in any major sector over that period.

The productivity-through-subtraction pattern is clear. Plot employment and productivity together, and the inverse relationship from 1990 to 2010 is unmistakable. Plot them together from 2010 to 2025, and both are flat. This is not a coincidence.

Manufacturing’s story is a warning for anyone who believes technology automatically delivers sustained productivity growth. It does not. Technology delivers a burst of productivity — sometimes a dramatic burst, as the 1990–2010 data shows — and then the gains plateau until the next genuinely transformative innovation arrives. Between waves, the sector is stuck. Costs rise. Competitiveness erodes. And the workers and communities that depend on the sector bear the consequences.

The current moment is particularly fraught. Manufacturing ULC at 137.3 means the sector is substantially less cost-competitive than it was twenty years ago. Global competitors in Asia and Eastern Europe are not standing still. If American manufacturing cannot find a new source of productivity growth, the cost pressures will intensify. Prices for manufactured goods will continue to rise, contributing to the persistent inflation that central banks have struggled to tame. And the reshoring projects that policymakers are counting on for strategic independence will struggle to be economically viable without productivity improvements to offset their higher labor costs.

In the next episode, we turn to unit labor costs across the entire economy — not just manufacturing — to examine the relationship between productivity, wages, and inflation. Manufacturing’s soaring ULC is a sector-level symptom of a broader pattern. The story of what happens when labor costs outrun productivity is, ultimately, the story of inflation itself.