Back in 2013, you see, I first argued that conservatives need to grapple with how the American economy was leaving significant segments of society behind (right here at National Review). I wanted to underscore the severity of our labor-market challenges but knew that just citing raw wage trends was insufficient. A prolific and well-respected policyman named Scott Winship had challenged the adequacy of the government data and especially its inflation adjustment and a debate was raging. How could I depict an accurate picture without stepping into the crossfire?
I landed on the idea of using median income relative to the poverty threshold. Because the Census Bureau reports both figures annually in current dollars, inflation concerns fell by the wayside. I could look at the 1970 ratio of income to the 1970 poverty threshold for a family of four in 1970 dollars, and the 2012 ratio in 2012 dollars, and have a relevant comparison. The result was also striking: In 1970, the median male high school graduate’s income was more than double the poverty threshold, while by 2012 he exceeded it by barely 30 percent.
Fast-forward to today and I find myself in conversation with Scott on these very same topics! But lo and behold he rejects this new measure too. He raises two objections, which I think are fair but insufficient.
The first objection is that the population of high school graduates has changed and now represents a lower subset of the income distribution. That’s true. But it doesn’t come close to explaining the decline: Even high school dropouts were doing much better on this metric in 1970 (exceeding the poverty threshold by more than 80 percent) than high school graduates are today.
The second objection is that this data includes all men over age 25, including retirees, of whom we have a higher share today and who tend to have less income. That’s true, too. But we could just as well observe that the high school-only population still working also skews older than it did in 1970, so we are now seeing relatively more people in their higher-earning late-career years. Which of these effects dominates? I don’t know. But once again, Scott has only highlighted the adjustment that paints a rosier picture, whereas the full range of adjustments don’t point obviously in one direction or the other. No measure is perfect! But each can tell us something, and we should take note when they all tell us the same thing.
As it happens, the Census Bureau has also tried to provide a measure that accounts perfectly for both of Scott’s concerns, looking at median income specifically for men between the ages of 25-34 and regardless of education: “More young men are falling to the bottom of the income ladder. In 1975, only 25 percent of men, aged 25 to 34, had incomes of less than $30,000 per year. By 2016, that share rose to 41 percent of young men. (Incomes for both years are in 2015 dollars.)”
I could offer some Winshipian objections to this too . . . Perhaps more people in that age bracket are still in school. The $30,000 cut-off is arbitrary and doesn’t include fringe benefits. And so on.
But at some point the burden of proof has to shift. The standard government data shows that the labor market is failing a substantial share of the population. So do the data after being subjected to every favorable adjustment imaginable. So do more nuanced slices of the data. So do surveys of public opinion. So do secondary measures like labor force participation. So do tertiary measures like family formation and mortality. So do hyper-targeted analyses of children’s earnings as compared to their own parents.
On the other side of the ledger, Scott offers, as far as I can tell, no alternative view from any field that would suggest things are not going poorly. With enough adjustments he can show median male earnings rising a barely perceptible 10 percent over forty years. If we think of all these perspectives as efforts at sampling reality, his estimate anchors the high end of the distribution, which ranges from that merely bad down to the truly terrible.
Remember also that the American experience cannot be aggregated to a single reality. Whatever measure we use is averaging or choosing the median from a much wider set of outcomes that range from terrible to wonderful. When the best we can say is that overall things look not terrible, we have to remember that for some segments of society things will look much worse. One of the reasons that stagnation is unacceptable is that it means by implication some group must be in decline. Only with steady progress should we have any confidence that even those relatively disadvantaged can keep their heads above water.
And so when Scott says that I have conceded we are talking about stagnation and not decline, I hear, “Ha, you said he sank to the bottom of the pool. But now you’re just saying he stopped paddling.” These are not mutually exclusive!
Likewise, when Scott translates my concern about relative outcomes into a concern about income inequality, I have to say, “yes, but.” Yes, rising inequality is one of the dynamics at play. But inequality, per se, is not the issue. One could envision rising inequality and none of our problems.
Rather, we are experiencing a particular flavor, in which a substantial segment of society isn’t just seeing fewer gains, it’s seeing no gains. Many people are seeing losses. The problem isn’t a “top 1%” pulling away from everyone else, it’s a bifurcation along roughly the college divide. That means the unfortunate half is not just stuck, but left behind and left out, with a rising safety net overtaking what work might offer and a culture denigrating those who still do it. That same segment faces the greatest declines in labor-force participation, family formation and stability, and health. Whether we think the best median wage trendline is flat, slightly up, or slightly down doesn’t change the picture.
The labor market is the mechanism in our society that bears responsibility for those economic outcomes and so I would describe it as failing, broken, and, yes, in decline. The Once and Future Worker tries to explain how we allowed this to happen, and how we can reverse course.
Thanks to all who have read, and especially to Scott for participating in this dialogue. I’ve learned a lot, his arguments have pointed me to new lines of research that I want to pursue, and he has persuaded me in particular that I too casually use “high school graduate” as a stable demographic group when it has changed over time. And as he has been gracious to note, while we disagree on how to understand many of these challenges, we actually agree on a lot of policy prescriptions. I hope we will be working on them together in the future.