Should Median Worker Pay Be $102,000? Gullible Reporters Think So
A recent report from Rand Corporation researchers has been making headlines with dramatic claims like median worker pay should be $102,000.
Although the paper is technical, the $102,000 calculation is simple. The median full-year, full-time, prime age (FYFTPA) employee made an inflation-adjusted $47,000 in 1975 and $50,000 in 2018. But real per capital GDP rose almost 120% over that period, from around $26,000 to $57,000. If real median FYFTPA income had risen with per-capita gross domestic product, the median income would have been $102,000.
The authors are clear they are not making a normative point—what workers should earn—merely investigating a counterfactual—what would the world be like if median FYFTPA workers’ real income had risen with real GDP? But many commentators have taken the claim as normative; that median workers should make $102,000. Even before this study, many people pointed to the near-zero growth in real wages over the past five decades as evidence something is unfair in the economy.
Therefore it’s worth pointing out that a closer reading of the paper demonstrates that the counterfactual is a much less fair world than we have today, and that wages should not increase at the same rate as per-capita GDP in the mechanical fashion of the counterfactual.
At first blush, the counterfactual seems absurd. How could worker real pay double, holding inflation and GDP constant? Worker compensation is 61% of business revenue. You can’t double everyone’s wages without raising prices, or businesses would run out of money before paying wages, not to mention all other expenses. Doubling wages would seem to require higher prices—inflation.
Also, if workers earned more, they would spend and consume more. Since GDP—the total value of all goods and services produced—is held constant, someone else must consume less.
To make this tangible, just consider cars. Using 2018 consumption patterns and the wage increases from the counterfactual, the 25th to the 95th percentile of households would demand 43 million more cars. But if GDP stays the same, where would those cars come from? The top 5% only own 19 million cars, so even taking all of them wouldn’t cover the shortfall. The bottom 25% could supply 36 million cars, but this would not make the world more fair. Also, the lower-cost and older cars owned by poorer households are not the cars workers making $102,000 per year want to own.
And the newfound income wouldn’t just be used for cars, there would be substantial increased demand for almost all goods and services. So if total goods and services stay the same, and someone gets more, someone else has to get less. The top 5% of households don’t consume enough to cover the extra the 25th to 95th percentile households would want to buy in the counterfactual. So even if we take everything from them, we still have to steal from the poor to make the counterfactual work.
To be clear, the top 5% have enough income to cover the excess payments to the 25th to 95th percentile. But much of that income goes to taxes and investment—the top 5% don’t spend enough to cover the extra spending the 25th to 95th percentile could be expected to do with the extra income. So for the counterfactual to work, you need to reduce goods and services consumed by the bottom 25% of households.
But the counterfactual does not say the median FYFTPA worker earns $102,000 in wages, it’s $102,000 in pretax income. The average household with $102,000 of pretax income in 2018 had only $86,000 in wages—the other income came from investments and other sources. Moreover in addition to the $50,000 in salary the median FYFTPA worker earned in 2018 came an average of $16,000 in benefits—mainly health insurance and tax-deferred retirement plans—which were much smaller in 1975. Therefore an apples-to-apples comparison should be $66,000 in total employee compensation in 2018, versus $86,000 of counterfactual wages; a smaller gap than $50,000 of salary versus $102,000 of counterfactual pre-tax income in 2018.
Finally, it’s important to consider the composition of FYFTPA workers. In 1975 they were 20% of the population, in 2018 they had grown to 27%. It’s not consistent to compare the median from the top 20% of a population to a median from the top 27%. Moreover, the composition of FYFTPA workers changed from 1975 to 2018. White men declined 5% relative to population, but white women increased 59%, black men 58% and black women 109%. Including more lower-wage female and non-white workers brings down the median without making anyone worse off. White men median inflation-adjusted wages were nearly flat, but white women grew 52%, black men 7% and white women 33%.
If we adjust for these differences, the overall gain in wages drops from 30% ($66,000 to $86,000) to 5% ($66,000 to $69,000). White males still do well in the counterfactual, from the actual 2018 median of $73,000 in total compensation to $115,000. Black males lose, from $61,000 to $53,000. White women get crushed, going from an actual 2018 median of $63,000 to $38,000, and black women lose half their total employment compensation, from $56,000 to $28,000.
After accounting for these factors, the median FYFTPA worker is $3,000 better off in the counterfactual. The massive gains to white males—who make up most of the above-median FYFTPA workers—offsets the losses to other groups. Without a more detailed counterfactual, we can’t tell if the $3,000 gain comes at the expense of people without full-time jobs, from investors or from changes within the distribution of pre-tax income among FYFTPA. But it does come at the expense of 45 years of progress in reducing sex and racial disadvantages in pay and employment opportunities.
Using this counterfactual to make normative statements about wage rates, against the authors’ warning, makes little sense. There’s no reason to think 1975 had optimal wages (in fact, the degree of racial and sex discrimination is a strong counterargument), nor that median FYFTPA wages should be a constant fraction of per-capita GDP. Moreover, an important consideration ignored in the counterfactual is that distributions of income and consumption affect GDP growth, so there may be trade-offs between equality and growth.
Unfortunately, much of the coverage of the paper has ignored these points. The $102,000 figure that dominates reporting is never discussed in the text of the paper. Among 43 tables and charts, and an additional 98 in appendices, each with many lines—someone looked through every line of every table to pull out one from appendix C with the most dramatic wage increase. $102,000 is represented as the figure median wage earners should have gotten, if not for some nebulous unfair “inequality.”
It may well be true that median wages should be higher, and that unfairness has slowed median wage growth since 1975. But this paper has little to say about either of those points. It does provide a wealth of detail and analysis about wages and inequality, and uses simple counterfactuals to make the points easy to grasp. If you read it, you’ll have a more nuanced understanding of the issues. But it does not support the simple message that someone has stolen half the income from median workers, and if you read it carefully you’ll see that most of the alleged stealing has been partial correction of racial and sex discrimination, removing tax on 30% to 40% of workers’ compensation, increases in the FYFTPA population and increasing non-employment income for middle-income households—all good things.
