What’s Going On with U.S. Worker Pay and Productivity

In economic theory, in a perfectly competitive market, worker pay should rise at the same rate as worker productivity. Obviously, in the real world, this doesn’t hold, and we shouldn’t expect it to be exactly the same. But there are some economic doomsayers that purport to show massive differences in pay and productivity in the past few decades in the U.S.

A new report from Scott Winship of the American Enterprise Institute looks at the data to get the real story. He finds that the doomsayers are wrong, but the relationship between pay and productivity has changed in the past few decades, just not in the way many people might think.

He looks at three examples from the doomsayers: One from American Compass, one from the Economic Policy Institute, and one from the Hamilton Project at the Brookings Institution. They each show, to varying degrees, worker pay lagging far behind worker productivity over roughly the past 50 years.

Winship demonstrates that these comparisons are largely the result of comparing apples and oranges:

• American Compass says productivity tripled between 1964 and 2022, but hourly pay rose only by 15 percent. Winship writes that they get there by comparing productivity in the nonfarm business sector with pay for private production and nonsupervisory workers. These aren’t the same parts of the economy. By excluding the 20 percent of American workers who are supervisors or non-production workers from its pay measurement, it understates pay increases. By excluding farms, public-sector employment, and some of the nonprofit sector from its productivity measurement, it overstates productivity growth. The pay measure also doesn’t include the self-employed, who have become a larger proportion of workers over the span measured, or nonwage income, which has become a larger proportion of total worker compensation over the span measured. It adjusts productivity for inflation using the price deflator for the nonfarm business sector, but it adjusts wages for inflation using the consumer-price index, which overstates inflation. “Part of the reason, then, that pay lags productivity in the American Compass analysis is that inflation wrongly eats away too much of the increase in pay relative to the increase in productivity,” Winship writes.
• The Economic Policy Institute says productivity rose by 65 percent between 1975 and 2022, but pay rose only by 15 percent. (It is worth noting that the Economic Policy Institute is basically organized labor’s think tank.) It uses the same pay measurement as American Compass, but it measures productivity for the entire economy, simply dividing GDP by hours worked. Winship points out that GDP includes gross housing value added, which workers are not responsible for creating. “One reason that economy-wide productivity has increased faster than compensation is that gross housing value added has increased more than the parts of GDP that involve goods and services primarily produced by workers,” Winship writes. “The housing sector of the economy should be left out of analyses comparing productivity and pay.” It uses the same inflation adjustment for both productivity and wages.
• The Hamilton Project tracks pay and productivity from 1948 to 2022 and shows that they were in line until 1971; over the entire span, productivity has increased by five times, but pay has increased only by three times. It is better than the first two because both of its measures are just for the nonfarm business sector, Winship writes. But it mishandles self-employed workers and fails to accurately account for depreciation, which has become a larger share of the national income over the span measured. It should have used net productivity, which subtracts depreciation, a mistake American Compass also made. It also uses the same inflation adjustment for productivity as American Compass while using a different version of the CPI for pay, still overstating inflation for the pay component and therefore understating pay increases.

Winship performs several different analyses that are more theoretically sound, and they demonstrate that worker pay has tracked pretty closely with productivity:

• Just sticking with the nonfarm business sector for both pay and productivity to avoid the housing complications and the sector problems, Winship finds that both pay and productivity have risen by about four times since 1948. He uses net productivity, to account for depreciation, and real hourly compensation, to include nonwage benefits.
• Winship then looks at the nonfinancial corporate sector, which excludes finance, government, and the self-employed. In that measure, pay growth has exceeded productivity growth consistently since the 1960s.
• Net value added and total compensation have much longer datasets, going back to 1929. When comparing those measures for the nonfarm business sector, Winship finds that they both have risen by a factor of 23.
• “Comparing real net value added in the nonfinancial corporate sector to real total compensation in the sector, the former rose by a factor of 29 from 1929 to 2023, while the latter grew by a factor of 28.5,” Winship writes. “The growth rates were essentially the same as of 2019.”
• “In the corporate sector (including the financial sector), we can compare nominal net value added to nominal total compensation,” Winship writes. “The former was 268 times its 1929 level in 2023, while the latter was 263 times higher (“nominal” means they are not adjusted for inflation, which is why the numbers are so much larger). “As of 2020, total compensation had risen slightly faster than net value added.”
• The last way Winship measures is by looking at the entire nonfarm business sector, including proprietors. “From 1929 to 2022, productivity rose by a factor of 7.6 and real hourly compensation by 7.4. In 2020, those figures were 7.6 and 7.7.”

Partly this just shows that data are not self-explanatory. As the great Robert Lucas explained, economists are storytellers. That isn’t a bad thing. “We do not find that the realm of imagination and ideas is an alternative to, or a retreat from, practical reality,” Lucas said. “On the contrary, it is the only way we have found to think seriously about reality.”

Thinking seriously about economic reality means comparing the stories economists tell and deciding which ones make more sense. Winship’s story on productivity makes a good amount of sense.

He says in his report that the growth in median compensation has lagged behind the growth in mean productivity. He argues that the U.S. has seen growing productivity inequality, with the most productive workers become more productive at a faster rate than other workers. Winship cites economic research that has found greater productivity inequality across industries, across firms, and within firms. Because pay tracks productivity, this would also help to explain why pre-tax income inequality, excluding government transfers, has grown. (Post-tax income inequality, including government transfers — i.e., what people actually make — has hardly budged.)

The bigger story in the trends for pay and productivity has to do with social changes in gender roles. Women’s pay has increased at a much higher rate than men’s pay as women have entered more productive sectors of the economy at higher rates.

Crucially, Winship argues that when men were more commonly seen as breadwinners and fewer women were formally employed, men were overpaid relative to their productivity. Government policies, such as the New Deal, World War II price controls, and the Wagner Act, which supercharged labor unions, contributed to wage growth above and beyond productivity growth. The U.S. also faced less international competition because most of the rest of the industrialized world had to spend a decade or two rebuilding after World War II. It would make sense that in this environment U.S. workers, mostly men, would be paid more than they would in a competitive market.

“Looking at trends since 1948 or 1973 is like choosing a year with a housing bubble as a starting point from which to look at homeownership trends,” Winship argues.

“If median pay was excessively high in 1973 relative to productivity levels, then it should not necessarily have grown as quickly as productivity over the next 49 years,” he writes. “Instead, we might expect that it would have grown more slowly until productivity growth could catch up—that pay growth might have been sluggish for some time so as to rationalize pay levels that had become unanchored to productivity growth.” The data seem to indicate that productivity caught up in the mid 1990s, and consequently, men’s pay then began to rise more quickly than it had since the mid 1970s.

Trends in marriage and family life have also contributed to lower wage growth for men relative to women, Winship argues. Married men today feel less pressure to be the sole income-earner for their households, so they have the ability to take a lower-paying, less productive job, focus more on child care, and still live comfortably. They are also likely to have fewer children on average than in the past, so there are fewer mouths to feed.

There has also been a decline in marriage rates in general, which means there are more single men who do not feel the pressure to provide for a family at all. “Research suggests that marriage has a causal impact on men’s earnings, raising them by as much as 25 percent,” Winship writes. Fewer married men means lower pay for men on average than in a population with more married men.

Winship is not saying everything is great. He is saying that the problems the U.S. faces are different from the ones the economic doomsayers purport to describe. “Rather than take seriously claims that the American economy is broken, policymakers should look for ways to raise economy-wide productivity and the productivity of working- and middle-class earners specifically,” Winship writes. This would include removing disincentives to marriage, still found throughout the welfare system, and improving education such that middle- and lower-class children are better prepared to be productive workers in today’s economy, not the economy of the 1950s.

Pay and productivity still track with each other when you look at apples-to-apples comparisons. Parts of Winship’s story may be wrong, but on the whole, it makes a lot more sense than the alternative stories that some cabal of greedy capitalists is conspiring to keep the little guy down, or that something about markets just stopped working in 1973 and has never been fixed despite decades of different government policies and economic trends since then. Policy-makers need to be informed by facts, not vibes, and populist agitation is a poor substitute for economic reasoning.