Ask why your pay went up and you get familiar answers. You worked hard, the company had a good year, employers were desperate to hire. Those explain who got a raise. None of them explain how a whole country pays everyone more without prices rising to cancel it out. That question belongs to one measurement most people never hear about.
What Productivity Actually Measures
Labor productivity is output per hour worked. The Bureau of Labor Statistics divides an index of real output (output with inflation stripped out) by an index of hours worked by everyone doing the work, employees and the self employed alike. The headline figure covers the nonfarm business sector, the economy minus farms, government and nonprofits.
GDP tells you how big the pile is. Productivity tells you how much each hour of work adds to it. A country can grow its GDP by adding people or by working them longer, and neither makes anyone better off per hour spent.
Why It Is Counted Per Hour, Not Per Worker
Output per worker would be easier to compute, and it would mislead you, because workers do not all put in the same hours. Part time jobs, overtime and vacation length move hours per worker without changing what gets done in an hour.
In 2023 the average American worker put in about 1,800 hours. The average German worker put in about 1,340. Judged per worker, Germany looks far behind. Per hour the two are much closer, because most of that difference is time off, not output.
The same trap sits inside your own pay. An extra shift raises what you earn, but not your productivity, and a week only holds so many hours.
The Ceiling On Sustainable Pay
Here is the mechanism at the size of one coffee shop. A barista paid $18 an hour makes 20 drinks, so labor costs 90 cents per drink. Raise the pay to $20 with nothing else changing and labor costs $1.00 per drink. The owner charges 10 cents more or takes 10 cents less profit on every cup.
Now change something else. A faster machine and a better counter layout let the same barista make 25 drinks an hour. At $20, labor costs 80 cents per drink, less than it cost before the raise. The barista is paid more and the coffee does not get more expensive.
Cost per drink is the unit labor cost, and the coffee shop is the whole economy in miniature. Run the first version everywhere at once and the price increases are what you already know as inflation. Run the second and the raises stick. Output per hour is the ceiling, and pay pressed against it for long enough comes back as higher prices.
What Actually Raises Output Per Hour
Four forces do most of the work.
Capital deepening means more and better equipment behind each hour of labor: machines, software, vehicles, buildings. One person with a backhoe moves more dirt than one person with a shovel.
Better technology is a different thing from owning more of the same equipment. Somebody had to invent the backhoe.
Education and skills change what a given hour can do. BLS calls this labor composition, because an hour of skilled work is not the same input as an hour of inexperienced work.
Better organization is the one people forget. Same workers, same machines, arranged differently: shorter handoffs, less waiting, fewer defects to redo.
Total Factor Productivity, The Leftover
You can count capital and you can count hours. Add up what those two contribute and they still do not explain all the growth in output. The leftover is total factor productivity, also called multifactor productivity: output per unit of all inputs combined.
TFP is a residual, defined by subtraction rather than measured directly. BLS treats it as capturing technological change, efficiency gains, better management and the shifting of resources toward better uses. That is real, and it is also the weakness. Anything mismeasured on the input side lands in the leftover too, so TFP carries the errors along with the ideas.
Why A Fraction Of A Percent Decides A Career
Productivity growth gets reported in small numbers that are easy to shrug at. Compounding is what makes them matter.
Take a $50,000 salary. In an economy where output per hour grows 1% a year, that salary supports about $74,400 of today's buying power after 40 years, roughly a working life. At 2% growth it supports about $110,400. One percentage point, run for four decades, is a gap of about $36,000 a year.
There is a shortcut. Divide 70 by the growth rate for the years to double. At 2%, living standards double in about 35 years. At 1%, it takes about 70.
US nonfarm business productivity grew about 2.8% a year from 1947 to 1973, then slowed to roughly 1.5% a year from 1973 to 1996. That slowdown is much of why getting ahead felt different for one generation than the next.
The Argument About Where The Gains Went
For the postwar decades, productivity and the pay of a typical worker climbed together. The Economic Policy Institute puts productivity up 108.1% from 1948 to 1979 and hourly pay for production and nonsupervisory workers up 93.4%. After 1979 the lines separate and productivity pulls ahead. That chart is famous, and the fight over it is about method.
The first objection is deflation. Productivity is adjusted using the prices of what gets produced. Pay is adjusted using the prices of what people buy. Those baskets rise at different rates, so part of the gap opens before anything happens in the real world. Martin Feldstein argued that using two different deflators makes the comparison wrong.
The second is what counts as pay. Health insurance premiums and retirement contributions are compensation, and they never show up in a wage figure. Compare total compensation instead and the gap narrows.
The third is the labor share, the fraction of national income going to workers rather than to owners of capital. That share fell, and no accounting choice removes it.
Anna Stansbury and Lawrence Summers studied 1973 to 2016 and found the link intact: one extra percentage point of productivity growth came with 0.7 to 1 extra points of median and average compensation growth. Their reading is that productivity still lifts pay, and that other forces pushed typical pay down at the same time. EPI reads the same decades as a story about bargaining power. Productivity growth is necessary for rising pay, clearly not sufficient, and the rest is still argued over.
Where The Numbers Get Blurry
Counting cars per hour is easy. Counting therapy sessions, legal briefs or classroom hours is not, because there is no obvious unit and the quality moves. Some service output ends up estimated from what went into it, which quietly assumes zero productivity growth.
Free digital goods are a second blind spot. GDP counts what people pay for. A mapping app that replaced a $15 road atlas carries a price of zero, so an improvement can register as measured output going down.
New technology also shows up late. Robert Solow wrote in 1987 that you could see the computer age everywhere but in the productivity statistics. Paul David's answer in 1990 was history. Electric power was available in the 1880s, and factories did not capture the big gains until the 1920s, once they rebuilt around one motor per machine instead of a central shaft. The technology arrived decades ahead of the reorganization that made it pay, which is why economists are slow to call anything a productivity revolution before the statistics say so.
