For three decades after WWII, American workers' pay rose in lockstep with what they produced. Then, around 1979, the two lines split. An honest look at how wide the gap really is — and what actually caused it.
There is a chart that has launched a thousand arguments. It shows two lines — the productivity of the American economy and the pay of a typical worker — rising together from the late 1940s, then splitting apart around 1979 and never rejoining. To one side it is proof that workers have been cheated of their fair share. To the other it is a statistical illusion built on mismatched measures. The truth, as usual, is more interesting than either.
Let's start with the numbers that aren't really in dispute. According to the Economic Policy Institute's long-running analysis of federal data, net productivity of the U.S. economy grew about 90% between 1979 and 2025, while the hourly compensation of a typical (production and non-supervisory) worker grew roughly 33%. Before 1979, the two had climbed together — productivity up around 108%, pay up around 93% across 1948–1979. After 1979, they diverged and stayed that way.
Put in human terms, EPI estimates that if the typical worker's pay had kept rising with productivity, they would earn roughly $16.40 more per hour today — about $13.50 of it in straight wages. Over a year of full-time work, that is real money missing from real households.
The gap is real. The fight is over what it means.
Here is where a careful analyst earns their fee. "Productivity rose and workers were robbed" is a slogan, not an explanation. When economists decompose the gap, it breaks into three distinct pieces — and they don't all point at the same villain.
First, a falling labor share: over these decades, a smaller slice of each dollar of output went to workers and a larger slice to capital. Second, rising inequality: even within the labor share, the gains concentrated near the top, so the typical (median) worker fell behind the average. Third, measurement: productivity is deflated by the prices of what the economy produces, while pay is deflated by the prices workers actually consume — and those two price baskets drifted apart, widening the headline gap for reasons that have nothing to do with bargaining power.
EPI's own decomposition is telling: for the 2000–2014 period, it attributes roughly 80% of the gap to the two inequality channels — a shrinking labor share plus growing pay inequality — rather than to measurement quirks. In other words, the divergence is mostly a story about who captured the gains, not a mirage.
A credible study has to steelman the other side, and here it is genuinely strong. In a widely cited 2017 paper, Harvard's Anna Stansbury and Lawrence Summers asked whether the link between productivity and pay is actually broken. Their answer: no. Across 1973–2016, they found that each additional percentage point of productivity growth was associated with roughly 0.7 to 1.0 points of higher median and average compensation growth. Productivity still lifts typical pay — the engine works.
Measurement matters too. Swap the consumer-price deflator economists often use (CPI-U-RS) for an alternative (the PCE index) and median compensation growth over the period roughly doubles — from about 12% to about 26%. The choice of yardstick alone moves the story substantially. And when productivity is measured at output prices and compared against total compensation rather than wages alone, the two series track far more closely than the famous chart implies.
The provocative implication of the Stansbury–Summers work is that boosting productivity is, for the typical worker, at least as powerful a lever as reducing inequality. They estimate that holding inequality at its 1973 level would have left the median worker about 33% better off by 2016 — but that restoring the faster productivity growth of the 1949–1973 era would have lifted pay by around 41%. Both matter. Neither is the whole story.
Since ~1979 the typical worker's pay has clearly lagged economy-wide productivity. That divergence is in the data, and it is large.
How much is "decoupling" versus rising inequality versus measurement — and therefore whether the fix is stronger worker power, faster growth, or both.
Strip away the slogans and a defensible reading emerges. The middle class's gains did not simply vanish — they were redistributed. The link between productivity and pay never fully broke, but two forces pulled typical paychecks off the productivity trend: a larger share of output flowing to capital and to top earners, and a productivity slowdown that left a smaller pie growing more slowly. A portion of the raw, eye-popping gap is also an artifact of comparing two different price yardsticks.
For a policymaker or a business leader, that reframing changes the prescription. If you believed pay had "decoupled" entirely, you might give up on growth and focus only on redistribution. The evidence says don't: productivity growth still reaches workers' wallets, so reviving it is pro-worker. But it also says growth alone won't restore the old bargain while so much of the gains concentrate at the top. The honest answer uses both levers.
That is the moral this dataset actually tells — not the one that fits neatly on a protest sign or a rebuttal blog, but the one the numbers support when you let them.
Methodology note: "Net productivity" is output of goods and services, less depreciation, per hour worked, for the total economy. "Typical worker pay" is hourly compensation (wages plus benefits) of production and non-supervisory workers — roughly 80% of payroll employment. Era figures are cumulative within each window and depend on base-year and deflator choices; alternate but defensible measures narrow the headline gap. This study summarizes published findings and does not reproduce source charts or text.