I was reading this piece on The Economist recently, which showed a graph of median real wages over time to make the point that take-home pay has been increasing. Indeed, from 2010 to 2025, real wages have gone up about 10% –but over that time, the median American has gone from 37 to 39 years old (income is expected to increase ~3-5% per year). On top of that, the percentage of Americans with a college degree has increased from 28.2% to ~38.5% in 2025. It seems we have a shifting population that is going to affect the median's values.
Am I wrong in thinking that median wages are just what they say, the wages of the 50th percentile, not adjusted for things like age and education? If so, what would an apples-to-apples trendline for median wages over time look like? Are there other variables that should also be accounted for?
Do measures like median wages account for Americans getting older and more educated?
byu/Bill_Nihilist inAskEconomics
Posted by Bill_Nihilist
2 Comments
Plenty of wage data (such as Atlanta Fed’s wage growth tracker) allows you to condition on age, education, sex, …. You can do it in the CPS, but that is incredibly time and computation intensive.
You can also look at income deciles (like in EPI [here](https://www.epi.org/publication/strong-wage-growth-for-low-wage-workers-bucks-the-historic-trend/)).
So, no, median wages don’t condition on other variables as well. But why should they? We want to see how the middle person is doing, regardless of how that middle person is defined.
Edit: here is an example of median wages by age range and sex. https://www.bls.gov/charts/usual-weekly-earnings/usual-weekly-earnings-current-quarter-by-age.htm
But the finer you make the median, the fewer datapoints it contains, and the more variability that the measure has
median wages are the wages of the median, so they won’t account for compositional shifts. the clearest way to see this is to notice the massive spike in wages during COVID; this is because low-wage workers were disproportionately laid off, not because of any sudden wage increase.
if you want, you can look at wages for different subgroups (see link), although there are issues with this, as well. if you do this, there isnt much difference in wage growth between high school and college workers, although there is a large levels difference. 35-44 year old workers saw the largest growth increases; 25-34 is about the same as 45 to 54.
on the education point, though, it’s not clear to me that this is something you actually want to adjust for. more people going to college is an expected result of the fact that the returns to college have increased over time, and “adjusting” for this introduces its own compositional issues.
this is related to the fact that the data are repeated cross-sections; you’re following “high school graduates” but who is a high school graduate has changed over time. so looking at within subgroups isn’t actually an “apples to apples” comparison in the way one might want.
– https://imgur.com/a/cJCi9TP