American Enterprise Institute economist Michael Strain has a new paper looking at the relationship between wages and worker productivity and concludes that, contrary to common belief, the link between wages and productivity is strong.

I say “contrary to common belief” because the prevailing view is that the once-tight link connecting wages to productivity has been severed (illustrated by a nice chart). In other words, somehow wages for the last few decades have not kept up with productivity growth and, as a result, there is today a real productivity-pay gap. The gap has been used to make the case, on both the left and the right, for allegedly corrective interventions, such as higher minimum wages and wage subsidies.

As Strain shows in his paper, however, the alleged gap is an artifact of methodological choices made regarding how to calculate the relationship between productivity and earnings. A more careful and comprehensive analysis of real worker pay and productivity data shows that worker compensation does indeed remain closely linked to worker productivity.

Strain starts his paper with a nice discussion about how we should think about wages. He presents the simple economic model and then notes that “In reality, the labor market for an industry or a geographic area—to say nothing of the U.S. labor market as a whole—likely never reaches equilibrium.” In part that’s because nominal wages are sticky, but also playing roles are minimum wages, international-trade patterns, and technological changes. In addition, employers do have more power at setting wages than the simple model states.

Strain then goes into which workers should be included in the calculation of the productivity-pay gap. He thinks that most workers should be included, but he is also open to arguments made on the left that managers should be excluded when studying the relationship. He nevertheless concludes that:

In addition to the typical worker’s wages, it is also of interest to study the relationship between productivity and the average wage of all workers in the economy. The logic here is straightforward: If you are using economy-wide productivity to study the relationship between productivity and wages, then you should use economy-wide wages as well. While it is true that wages have been growing relatively faster for high wage workers over the past several decades, it may also be true that the productivity of those workers has been growing relatively faster. Excluding them from the analysis may leave a key piece of the puzzle missing. In addition, if the underlying reason for interest in the relationship between productivity and wages is not to see how workers’ standards of living have evolved with productivity, but instead to study how firms compensate workers in their role as a key input to production, then it’s desirable to study the average wage of all workers, not just of production and non-supervisory workers.

Next, Strain looks at “which measure of inflation should be used to convert nominal wages into real wages.” He concludes, as have many others, that,

…when investigating the relationship between wages and productivity, a strong case can be made that wages should be deflated using a measure of the change in the prices of goods and services produced by businesses, not those consumed by workers. Economic theory predicts that workers are paid according to the marginal product of what they produce, not what they consume. Thus, an output price deflator is most appropriate.

He also correctly makes the case that to capture the relationship between worker pay and productivity correctly, investigators should use total compensation rather than just wages. This point is particularly important since about one third of employee compensation today is paid in the form of benefits. Benefits, of course, are compensation paid to workers no less than are money wages. Here’s what I wrote a while back for Reason magazine:

According to the Federal Reserve Bank of St. Louis, inflation-adjusted wages have grown by just 2.7 percent in the last 40 years. But inflation-adjusted total compensation—wages plus fringe benefits, such as health insurance, disability insurance, and paid vacation, along with employer-paid Social Security and Medicare taxes—increased by more than 60 percent in the same period.

Wages still make up a significant share of your total compensation: 68.3 percent, according to 2017 data from the Bureau of Labor Statistics, vs. 31.7 percent that goes to benefits. But that latter piece has grown significantly, in no small part due to the rising cost of health insurance. And that trend is only going to get worse.

Strain’s final argument is that a better way to measure productivity is to use net output rather than gross output. The main reason is that “gross output includes capital depreciation, while net output does not. Since depreciation is not a source of income, net output is the better measure to use when investigating the link between worker compensation and productivity.”

He puts it all together and concludes:

When properly measured, with variable definitions based on the most appropriate understanding of the relevant underlying economic concepts, trends in compensation and productivity have been very similar over the past several decades. Of course, it is also the case that two variables can evolve similarly over time without necessarily being related. But this chart, combined with the statistical evidence in the Stansbury and Summers paper and economic theory, provides compelling evidence that productivity and compensation are strongly related.

Here is the Anna Stansbury and Larry Summers paper.

I couldn’t copy the chart he has in the paper, so I am adding this chart I made a few years ago making roughly the same case. Here it is:


The Strain paper is here and well worth reading.


Veronique de Rugy is a Senior research fellow at the Mercatus Center and syndicated columnist at Creators.