Thank you for writing this comment so that I don't feel like a crazy person after reading this article. It came across to me as so obviously flawed that it's hard to even apply Hanlon's Razor to it.
Not once does he mention of any correction for the obvious bias that people make different amounts at different points in their careers.
Using his exact same process, you could also show that the average height of Americans has increased significantly in the past thirty years, possibly up to a foot! Because, duh, people are shorter when they're kids.
If you're trying to measure how the economy is doing over time, you don't want panels, you want to compare equivalent people at different points in time. Comparing a 20 year old in 1980 to a 50 year old in 2010 confounds the differences between 1980 and 2010 with the differences between being 20 and being 50.
There are reasonable criticisms that identifying representative "equal" people over time is hard, but the author's solution seems dramatically worse than all of them.