Unit Labor Costs: The Inflation Signal

Every quarter, the Bureau of Labor Statistics publishes a number that most financial journalists skip, most investors ignore, and most policymakers study obsessively. It is called the unit labor cost, and it answers a question more fundamental than CPI, more structural than PPI, and more predictive than any inflation expectations survey: How much does it cost, in labor terms, to produce one unit of economic output?

The Consumer Price Index tells you what prices are. Unit labor costs tell you what prices will be. If it costs more labor to make a widget this year than last year, the price of that widget is going up — not because of supply chain disruptions or commodity spikes or tariff games, but because the fundamental cost structure of the American economy has shifted. That kind of inflation does not resolve itself. It is sticky. It is structural. And it is exactly what the Federal Reserve fears most.

As of Q4 2025, the BLS unit labor cost index for the nonfarm business sector stands at 122.6, where 2017 equals 100. That means it costs 22.6% more in labor, per unit of output, to produce goods and services today than it did eight years ago. Since 2019, unit labor costs have risen 18.5% — an annualized rate of 2.9%. That is nearly three times the 1.0% annual growth rate that prevailed during the two decades before the pandemic, the era of low and stable inflation that the Fed considered normal.

This episode traces the 78-year history of unit labor costs, from the postwar industrial economy of 1947 to the post-pandemic service economy of 2025. The data reveals five distinct eras of ULC behavior, each with its own inflation story. And the current era — the one we are living through right now — looks nothing like the “Goldilocks” period that preceded it.

The concept behind unit labor costs is elegant in its simplicity. It is the ratio of what workers are paid to what they produce. If a factory worker earns $30 an hour and produces 10 widgets per hour, the unit labor cost is $3 per widget. If that worker gets a raise to $33 an hour but still produces 10 widgets, the unit labor cost rises to $3.30 — a 10% increase that, sooner or later, gets passed on to the customer.

This formula is what makes ULC so powerful as an inflation predictor. It captures both sides of the wage-price equation simultaneously. Wages can rise without creating inflation — as long as productivity rises by the same amount. A 3% raise with 3% productivity growth means unit labor costs are flat, and there is no cost-push inflation pressure. But a 5% raise with 1% productivity growth means unit labor costs are rising at 4% — and that 4% has to come from somewhere. Either profit margins shrink or prices go up. In practice, prices go up.

This is why the Federal Reserve watches ULC more closely than headline wage growth. Average hourly earnings rising at 4% sounds inflationary, but if productivity is also rising at 3%, the actual cost pressure is only 1%. Conversely, modest 2% wage growth sounds benign, but if productivity is falling at 1%, the cost pressure is actually 3%. Unit labor costs strip away the noise and reveal the signal.

The BLS tracks unit labor costs for the nonfarm business sector, which covers roughly 75% of GDP. It also tracks ULC for manufacturing separately — and as we will see, manufacturing ULC has risen even faster than the nonfarm aggregate, reaching 137 on the same index where nonfarm sits at 122.6. The manufacturing sector, which was supposed to benefit most from automation and productivity gains, has instead become a leading indicator of cost pressure.

The chart below traces the unit labor cost index from 1947 to 2025, using Q4 readings for each year. The trajectory is unmistakable: an eightfold increase over 78 years, from 15.6 to 122.6. But the rate of that increase has varied enormously across eras, and those variations map almost perfectly onto America’s inflation history.

The early postwar period, from 1947 through the mid-1960s, was an era of remarkable stability. Unit labor costs barely moved, hovering in a narrow band between 15.6 and 21.6. Workers received raises, but productivity was growing at a roughly equal pace — the great expansion of American manufacturing capacity was absorbing labor cost increases as fast as they occurred. Inflation during this period averaged less than 2% per year. The ULC index tells you why.

Then came the rupture. Beginning in the late 1960s and accelerating through the 1970s, unit labor costs began rising at a pace the economy had never experienced. The index went from 21.1 in 1965 to 59.0 in 1982 — a 180% increase in 17 years, or 6.3% per year. This was the era of stagflation, and the ULC data shows exactly why it happened. Wages were surging — driven by powerful unions, cost-of-living adjustment clauses, and an overheated labor market — while productivity growth had stalled. The result was relentless cost-push inflation that the Fed ultimately had to break with the most severe monetary tightening in its history.

The Volcker disinflation and the subsequent decades brought a return to order. From 1982 to 2000, ULC rose from 59.0 to 85.5 — a 45% increase over 18 years, or about 2.1% annually. This was a period of moderate wage growth and modest productivity gains, producing the low-inflation environment that defined the “Great Moderation.”

But the real golden age of ULC behavior came in the 21st century. From 2000 to 2019, unit labor costs rose from 85.5 to just 103.5 — a mere 21% over 19 years, or 1.0% annually. This was the Goldilocks era of price stability: wages rose modestly, productivity grew modestly, and the net cost pressure was almost negligible. Core inflation averaged 1.8% during this period. The Fed’s 2% target was within reach, sometimes too far below it. Central bankers worried more about deflation than inflation.

Then the pandemic struck, and the Goldilocks era ended.

The four annotations on the chart tell the entire story. The 1960s plateau at roughly 21 on the index is the visual representation of an economy where wages and productivity moved in lockstep — a world of stable prices and broad prosperity. The 1970s surge from 27 to 47 is what happens when that balance breaks: wages run ahead of output, costs spiral upward, and inflation becomes entrenched. The long, gentle slope from 1995 to 2019 — 78 to 104 over 24 years — is the Great Moderation made visible, the era when globalization, technology, and labor market slack combined to keep unit costs in check.

And then there is the post-pandemic surge: the steep climb from 108 to 123 in just five years, a rate of ascent not seen since the early 1980s. This is the portion of the chart that keeps Fed governors awake at night.

The 78-year history of unit labor costs divides naturally into five eras, each defined by the balance between compensation growth and productivity growth. The table below summarizes each era with its starting and ending ULC level, total growth, and annualized rate. The patterns are striking.

The first era, postwar stability (1947–1965), was defined by two forces working in tandem: strong productivity growth from the ongoing industrialization of the American economy, and moderate wage increases that roughly matched that productivity growth. Unit labor costs rose just 1.7% per year. Manufacturing was expanding rapidly, new technologies were being deployed across every sector, and the American worker was becoming dramatically more productive. This productivity dividend allowed wages to rise without creating cost pressure. CPI inflation averaged about 1.8% during this period — and ULC growth of 1.7% explains almost all of it.

The second era, the stagflation disaster (1965–1982), was the mirror image. Productivity growth slowed sharply in the late 1960s, falling from roughly 3% per year to under 1%. But wage growth did not slow. Powerful labor unions negotiated automatic cost-of-living adjustments that tied wages to CPI, creating a feedback loop: higher prices led to higher wages, which led to higher unit labor costs, which led to higher prices. The annual rate of ULC growth reached 6.3%, and the index nearly tripled from 21.1 to 59.0. This was the root cause of stagflation — not oil prices, not government spending, but a fundamental breakdown in the relationship between compensation and productivity.

The oil shocks of 1973 and 1979 amplified the crisis, but they did not cause it. Unit labor costs were already accelerating before the first oil embargo. What the ULC data reveals is that stagflation was a domestic phenomenon at its core: American workers were being paid more to produce less. The oil shocks were gasoline on an already-burning fire.

The Recovery

The third era, the Great Moderation (1982–2000), began with Volcker’s brutal disinflation. By raising the federal funds rate to 20% and accepting a recession that pushed unemployment above 10%, the Fed broke the wage-price spiral. Union power declined. Cost-of-living clauses disappeared from contracts. Globalization and deregulation introduced competitive pressures that held down both wages and prices. Productivity growth recovered modestly. The result: ULC grew at 2.1% per year, less than a third of the stagflation rate. Inflation fell from double digits to the 2–4% range.

The fourth era, the Goldilocks period (2000–2019), took the Great Moderation to an extreme. Unit labor costs rose at just 1.0% per year for two full decades. This was the era of China’s integration into the global economy, the rise of automation, the explosion of information technology, and the steady erosion of worker bargaining power. Wage growth was tepid — average hourly earnings rarely exceeded 3% — and productivity growth, while not spectacular, was sufficient to absorb most of the increase. The Fed spent much of this period worrying that inflation was too low, struggling to hit its 2% target from below.

Note the extraordinary contrast: the stagflation era produced 180% total ULC growth in 17 years. The Goldilocks era produced just 21% in 19 years. The entire inflation history of the United States over the past half-century can be read in these two numbers.

The New Regime

The fifth era, the post-pandemic surge (2019–2025), marks a clear break from the Goldilocks period. Unit labor costs have risen 18% in six years — nearly as much in six years as they did in the previous 19. The annualized rate of 2.9% is not catastrophic by historical standards — it is less than half the stagflation rate — but it represents a threefold acceleration from the prior trend. And crucially, it has shown no sign of returning to the 1.0% norm.

The question that haunts the Federal Reserve is simple: Is this a temporary post-pandemic overshoot that will normalize, or is it the beginning of a new structural regime? The answer lies in the components of ULC — wages and productivity — and neither is sending reassuring signals.

To understand the current ULC surge, it helps to decompose it into its two components. Between 2019 and 2025, hourly compensation in the nonfarm business sector grew at roughly 4.5% per year. Labor productivity grew at roughly 1.5% per year. The gap — about 3 percentage points — is the unit labor cost increase.

Compare this to the Goldilocks era: hourly compensation grew at roughly 2.5% per year, and productivity grew at roughly 1.5% per year. The gap was only about 1 percentage point. Same productivity growth, much faster wage growth — that is the entire story of the post-pandemic economy in two numbers.

The year-by-year ULC changes tell an even more dramatic story. The chart below shows annual percentage changes in unit labor costs for selected years that marked turning points in the inflation story. The volatility is striking. In any given year, ULC can swing from double-digit increases to outright declines, driven by the interaction of the business cycle with the slower-moving productivity trend.

The bar chart reveals the extremes that the era-average numbers conceal. 1974 saw a 20% surge in unit labor costs — the single worst year in the dataset, driven by the combined effect of the oil shock, the wage-price spiral, and a sharp recession-induced productivity decline. When output falls but workers are still being paid (because layoffs lag economic downturns), unit labor costs spike mechanically. The 1974 reading captures both structural cost pressure and cyclical distortion.

1980 brought another 10% surge, the echo of the second oil shock and the continued failure of productivity to keep pace with compensation. By this point, the wage-price spiral had been running for 15 years, and the Fed was on the verge of its most dramatic intervention.

2009 stands out as the only significant decline: a 5.1% drop. This is the Great Recession effect in reverse. Mass layoffs reduced the numerator (total compensation fell as workers were let go), but the denominator (output per remaining worker) held up better because firms shed their least productive workers first. The result was a temporary improvement in unit labor costs — a “productivity miracle” that was actually just a brutal culling of the workforce.

The post-pandemic years of 2021 and 2022 each saw roughly 4% ULC increases. These are not 1970s-level surges, but they are four times the Goldilocks-era average. More importantly, they came in consecutive years, establishing a trend rather than a one-off spike. The 2023 through 2025 period continued this pattern, with ULC growing at 2–3% annually — slower than the initial burst but well above the pre-pandemic trend.

Another way to read the ULC history is decade by decade. The table below shows average annual ULC growth for each decade since the 1950s, along with the corresponding average CPI inflation rate. The correlation is unmistakable: decades with high ULC growth have high inflation, and decades with low ULC growth have low inflation. The relationship is not one-to-one — other factors like energy prices, import costs, and monetary policy also matter — but ULC growth is the dominant structural driver.

The decades of the 1950s and 1960s established the baseline: ULC growth in the 1.5–1.8% range produced inflation of 2.0–2.3%. This was the postwar norm, and it was a period of genuine prosperity. Wages rose, prices were stable, and the American middle class expanded dramatically.

The 1970s shattered this equilibrium. ULC growth of 6.8% per year — more than three times the prior decade — produced CPI inflation of 7.1%. The near-perfect correspondence between ULC growth and inflation during this decade is not coincidental. When the cost of producing everything rises by nearly 7% per year, prices follow. There is nowhere else for the cost to go.

The 1990s and 2000s saw a return to low ULC growth — 1.6% and 1.2% respectively — as globalization, technology, and deregulation held down costs. The 2010s were the most benign decade on record, with ULC growing at just 0.8% per year. This was the era of persistent below-target inflation, when the Fed experimented with unconventional monetary policy to push inflation higher.

The 2020s so far average 2.6% annual ULC growth — more than three times the 2010s rate. This is not the 1970s, but it is a clear departure from the trend that prevailed for the previous three decades. And the CPI data confirms the signal: inflation has averaged roughly 4.0% in the 2020s, double the 2010s rate.

The unit labor cost equation has two components, and the one that drives long-term outcomes is not compensation — it is productivity. In the long run, workers will be paid in proportion to the value of their output. If productivity grows at 2% per year, wages will eventually grow at roughly 2% per year, and ULC will be approximately flat. If productivity grows at 0.5% per year, wages may grow at 3% (because labor markets can overshoot), and ULC will rise at 2.5%. Productivity is the structural anchor of unit labor costs.

This is why the current slowdown in productivity growth is so alarming for the inflation outlook. During the Goldilocks era, nonfarm business productivity grew at roughly 2.1% per year. Since 2019, it has averaged about 1.5%. That half-percentage-point difference may sound trivial, but over time, it compounds into a significant gap between compensation growth and output growth — which is exactly what the ULC data is showing.

There is a common misconception that higher wages automatically cause inflation. They do not. Higher wages cause inflation only when they outpace productivity growth. Germany and Japan have demonstrated this repeatedly: they maintained moderate wage growth with strong productivity growth, keeping unit labor costs and inflation in check for decades. The United States did the same thing from 2000 to 2019. What changed after the pandemic was not that wages got “too high” — it is that productivity did not keep up.

While the nonfarm ULC index stands at 122.6, the manufacturing sector tells an even more troubling story. Manufacturing ULC has risen to approximately 137 on the same 2017=100 index — 12% higher than the nonfarm aggregate. This is counterintuitive. Manufacturing was supposed to be the sector where automation, robotics, and lean production would keep costs permanently in check. Instead, it has become the leading edge of the ULC surge.

Several factors explain this divergence. First, manufacturing wages have risen faster than average, driven by tight labor markets in skilled trades and the reshoring of production that requires experienced workers. Second, manufacturing productivity has actually declined in several recent quarters, as firms ramped up production to rebuild supply chains without achieving the efficiency of pre-pandemic operations. Third, energy and materials costs, while not directly part of ULC, have forced manufacturers to invest in redesigning processes and supply chains — a transition period that depresses measured productivity.

The manufacturing ULC surge matters beyond the factory floor because it feeds directly into the prices of physical goods. When it costs 37% more in labor to produce a manufactured product than it did in 2017, that cost eventually appears in the prices consumers and businesses pay. The recent stabilization of goods prices in the CPI data has been driven more by the normalization of supply chains than by any improvement in manufacturing productivity. The underlying cost structure remains elevated.

The Federal Reserve has a dual mandate: maximum employment and price stability. Unit labor costs sit at the exact intersection of both. Rising ULC means either wages are growing faster than productivity (a labor market running hot) or productivity is declining (an economy becoming less efficient). Either way, the result is upward pressure on prices.

The current dilemma is acute. The labor market has cooled from its 2022 peak but remains tight by historical standards. The unemployment rate sits around 4.2%, below the long-run average of 5.2%. Wage growth has moderated but still runs above 4%. And productivity growth — the variable that could solve the equation painlessly — has not accelerated sufficiently to close the gap.

For ULC growth to return to the Goldilocks-era rate of 1.0%, one of two things must happen. Either wage growth must fall to roughly 2.5% (which would require a significantly softer labor market and would be politically and socially painful), or productivity growth must accelerate to roughly 3.5% (which would require a technological revolution comparable to the late 1990s internet boom). Neither outcome is currently in sight.

The Fed has a third option: accept that ULC growth of 2.5–3.0% is the new normal and tolerate inflation modestly above the 2% target. This is, in effect, what the Fed has been doing since 2022 — allowing inflation to run above target while waiting for either productivity growth to accelerate or wage growth to moderate. But the longer ULC growth persists at current rates, the more likely it is that inflation expectations become anchored at a higher level, making the problem self-reinforcing.

The current ULC environment most closely resembles the late 1960s — not the 1970s. In the late 1960s, unit labor costs began accelerating from the stable postwar norm, rising from about 21 to 25 between 1965 and 1970. Inflation ticked up from 1.3% in 1964 to 5.9% in 1970. The Fed was slow to respond, distracted by the Vietnam War fiscal stimulus and reluctant to raise rates aggressively. By the time it acted decisively (under Arthur Burns, and later Paul Volcker), ULC growth had become entrenched.

Today’s situation shares several features with that period. ULC is accelerating from a long stable base. The Fed is reluctant to tighten further because the labor market, while tight, is not overheating. Fiscal policy remains expansionary. And there is a widespread belief — as there was in the late 1960s — that the acceleration is temporary and will resolve itself without aggressive intervention.

The late 1960s should serve as a caution. The ULC acceleration that began in 1965 did not resolve itself. It intensified for 17 years, and it took a deliberate, devastating recession to end it. The current acceleration is far milder — 2.9% annual growth versus 6.3% — and may indeed prove transitory. But the lesson of history is clear: ULC accelerations, once established, tend to persist rather than self-correct.

For those who want the complete picture, the table below presents the ULC index at five-year intervals from 1947 to 2025, along with the five-year growth rate and annualized rate. The data shows how ULC growth has varied across nearly eight decades, with the 1970s standing out as the extreme and the 2000s and 2010s as the calm before the current storm.

Several entries in this table deserve particular attention. The 1977 and 1982 readings show the most extreme five-year acceleration in the dataset: 44.7% and 49.4% total growth, or 7.7% and 8.4% annualized. These numbers are not merely academic curiosities — they describe an economy in which the cost of producing everything was rising by nearly 8% per year, year after year, for a decade. That is the kind of structural cost pressure that makes 2% inflation physically impossible.

Equally striking is the 2007–2012 period: zero ULC growth over five years. This was the aftermath of the Great Recession, when the economy was operating so far below capacity that there was no cost pressure whatsoever. Workers had no bargaining power. Wages were stagnant. Productivity held up because the least-productive jobs and workers had been eliminated. The Fed was desperately trying to generate inflation and failing.

The 2017–2022 interval tells the post-pandemic story: 15.9% total growth, or 3.0% annualized. This is the fastest five-year acceleration since the early 1990s — and unlike the early 1990s, there was no subsequent deceleration. ULC continued rising at an elevated pace through 2023, 2024, and 2025.

History offers three mechanisms for slowing ULC growth, and all of them have costs.

The first is recession. When the economy contracts, firms lay off workers, and the remaining workforce becomes more productive per hour (because the least productive workers are the first to go). Compensation growth slows as the labor market loosens. Both effects push ULC down. This is what happened in 2009: a 5.1% decline in ULC, bought at the price of 8.7 million jobs and the worst financial crisis since the 1930s. It is effective but brutal.

The second is a productivity boom. If output per worker accelerates — through technological innovation, capital investment, or structural reform — unit labor costs fall even if wages continue rising. This is the best-case scenario, and it is what happened in the late 1990s when the internet revolution drove productivity growth above 3% per year. ULC growth fell to under 1% even as the labor market was the tightest in a generation. This is the outcome the Fed is hoping for from artificial intelligence, but so far, the AI productivity dividend has not materialized in the aggregate data.

The third is wage suppression. If compensation growth is reduced through policy (immigration, trade liberalization, deregulation) or market forces (declining union power, increased labor supply), ULC can fall without a productivity increase. This is essentially what happened from 2000 to 2019: globalization and the erosion of worker bargaining power kept wage growth below 3%, and ULC growth stayed at 1%. This was effective for price stability but created the political backlash that produced populism on both sides of the political spectrum.

None of these mechanisms is painless. A recession destroys jobs. A productivity boom requires investment and innovation that may take years to materialize. Wage suppression creates inequality and political instability. The Fed is currently hoping for door number two — a productivity miracle, perhaps driven by AI — while preparing for the possibility that door number one may eventually be necessary.

Unit labor costs have direct and measurable effects on corporate profit margins and, by extension, on equity valuations. When ULC rises faster than firms can raise prices, margins compress. When ULC is stable or declining, margins expand. The relationship is mechanical: labor is the single largest cost for most businesses, accounting for 60–70% of total production costs in the nonfarm sector.

The Goldilocks era of 1.0% annual ULC growth coincided with the greatest expansion of corporate profit margins in American history. S&P 500 net margins rose from roughly 6% in the early 2000s to over 12% by the late 2010s. A major reason was that labor costs were barely rising relative to output. Firms could give modest raises, invest in automation, and still see their margins expand year after year.

The post-pandemic ULC surge threatens this dynamic. If unit labor costs continue growing at 2.9% per year while revenue growth averages 5–6%, the margin available for profits shrinks significantly. Firms have two responses: raise prices (which contributes to inflation) or accept lower margins (which depresses earnings and stock prices). In practice, most firms attempt a mix of both, which is why corporate earnings growth has been sluggish relative to revenue growth since 2022.

For bond markets, the implications are even more direct. Unit labor cost growth of 2.9% implies a structural inflation floor of roughly 2.5–3.0%, well above the Fed’s 2% target. If the market comes to believe that ULC growth will persist at current rates, long-term bond yields will adjust upward to incorporate a permanent inflation premium. The 10-year Treasury yield, which averaged 2.3% during the Goldilocks era, may need to remain in the 4.0–4.5% range as long as ULC growth stays elevated.

The most important variable for the future of unit labor costs is something that no one can forecast with confidence: the productivity impact of artificial intelligence. If AI delivers a genuine productivity revolution — comparable to the internet in the late 1990s or electricity in the 1920s — it could solve the ULC problem without any labor market pain. Workers could receive 4% raises while producing 3.5% more output, and ULC growth would fall to 0.5%.

The precedent is encouraging. The internet revolution drove productivity growth from roughly 1.5% in the early 1990s to 3.0% in the late 1990s and early 2000s. Unit labor costs fell from 2.0% growth to under 1.0% growth. Inflation fell to the lowest levels since the 1960s. And the stock market boomed because margins expanded even as wages grew.

But the precedent also offers a caution. The internet productivity boom took roughly five to seven years to materialize after the technology became commercially available (roughly 1993 to 1997–2000). AI became commercially significant with the launch of large language models in late 2022. If the internet analogy holds, the productivity dividend may not appear in the aggregate data until 2027–2029 at the earliest. That is a long time for ULC to keep growing at 2.9% per year.

Moreover, the productivity impact of AI may be concentrated in sectors that have a limited weight in the ULC index. If AI dramatically improves productivity in software development, legal research, and content creation but does not affect manufacturing, construction, healthcare, or retail — which together account for the bulk of nonfarm employment — the aggregate effect on ULC could be modest. The BLS will continue measuring what happens on factory floors, in hospitals, and in warehouses, not what happens in AI research labs.