Like other reformocon wonks (and wonks generally), I’ve been thinking a lot the past few weeks about Oren Cass’s impressive and important new book, The Once and Future Worker. TOFW has received deserved attention from a number of commentators, including Yuval Levin here at the Corner. It has also been criticized by AEI’s Jim Pethokoukis and Michael Strain as being anti-growth, or at least not sufficiently pro-growth. I took the same impression from Oren’s American Interest essay teasing TOFW, but having heard him talk about the book and having worked through it, I feel less strongly than I did about that. But despite the fact that I agree with most of his policy proposals, I’m still uncomfortable with Oren’s analysis of the roots of the discontent that we both see among the working class/Trump voters/the “forgotten Americans.” We’ve gone back and forth a bit on Twitter — the worst medium for debate since semaphore — and took up the Corner on the chance to find a better forum.
As far as I can tell, Oren sees 45 years of economic decline for Americans without a college education, characterized by declining real wages, falling labor-force participation, and declining economic mobility. These changes stem from a labor market that is broken, thanks to too much policy emphasis on overall economic growth and consumption at the expense of more balanced growth and more empowerment of less-skilled workers to be productive. The economic crisis, in turn, has driven various social and public-health crises, including the breakdown of the family, generally declining social capital, rising “deaths of despair,” and the election of Donald Trump.
My beef is that this analysis overstates the extent to which the labor market is broken and attributes too much causal force to changes in the economy. While there are few economic stats on non-college Americans to inspire celebration, the most important indicators have either not worsened or have explanations that point to labor-supply problems rather than problems of labor demand. Generally, “no change” can’t explain change. Why are we experiencing various crises now rather than in, say, 1973?
Instead of economic decline, I see other factors behind the elements of cultural and social crisis we face. Economics is, of course, not unimportant, but culture and society have changed more dramatically than labor markets. In some cases, the causality probably runs in the other direction, with cultural change affecting labor-market outcomes. This debate is important because it should dictate both the kind of appeal conservatives make to voters and the policies that they champion. We’ll also be disappointed when social and cultural problems remain if we “fix” the labor market and it’s not really the thing that matters. And we may inspire people with truly bad ideas who like the narrative of decline.
Perhaps the best indicator of how the labor market is doing by workers is to look at trends in hourly pay. Let’s stipulate off the bat that pay rose faster in the 1940s, 1950, and 1960s than it has over the past 45 years. There is no mystery about the primary cause — labor productivity has grown more slowly since the 1960s. That is true not just in the United States, but across the industrialized world. In part, that reflects the shift of such economies toward more service occupations. Barbers do not get more efficient at cutting hair over time at nearly the same rate as manufacturing workers aided by ever-improving technology.
Why the economy shifted toward services is a question too big for this debate, but I would submit that if we want to go back in time, all we need to do is (1) convince Americans to pay higher prices for domestically produced manufacturing goods instead of buying imports, (2) restrict the economic opportunities of women so that they can perform at home the services that we’ll no longer be able to afford after buying expensive manufactured goods, and (3) pay working-class men a premium so that they can raise a family on one income.
Let’s also stipulate that pay higher up the earnings ladder has risen faster than pay lower down the ladder (especially at the very top). It is entirely possible that white working- and middle-class frustrations have increased because either they expected the growth in pay previous generations saw or because they resent rising earnings inequality. I’ll try to come back to these hypotheses in future posts, but for now, I’ll just say that I don’t think these are Oren’s hypotheses, and that they rarely are the hypotheses invoked by declensionists generally.
What has happened to the pay of working- and middle-class employees? I modify Economic Policy Institute numbers using data from the Bureau of Economic Analysis (Tables 1.1.9 and 2.1) and the Bureau of Labor Statistics to improve the inflation estimates and to incorporate non-wage benefits that employers pay to workers. With these adjustments, real median hourly compensation rose by about 25 percent between 1973 and 2017. Most of that growth occurred between 1996 and 2003; growth has been slower the past 15 years and increased at roughly that same slower rate from 1973 to 1996. (Note that 1973 was a cyclical peak — albeit one artificially created by Nixon’s price controls — while 2017 was not. Ideally, we’ll want to compare the eventual peak of the current expansion to 1973, and that will show higher growth.)
So what’s wrong with a 25 percent increase in pay, given that jobs are almost surely more agreeable and safer than in 1973? Pay rose by less lower down the ladder, but even for the worker at the 20th percentile — worse-paid than all but 20 percent of workers — it rose by 19 percent.
Now, things look worse if we focus on men, whose median hourly pay rose by just 10 percent. It actually fell between 1973 and 1997, but it has risen a bit over the past 15 years. Strikingly, at the 20th percentile, male pay was the same in 2017 as in 1973. To some extent, that is too dour a picture of trends lower down, because rising immigration added more and more lower-paid workers to this picture. In fact, pay at the 20th percentile rose by 6 percent among non-Hispanic white and black men and by 16 percent among Hispanic men. Each group does better than the three groups combined, because over time there are more lower-paid disproportionately immigrant Hispanics (even as Hispanic pay has risen).
At this point, I hope the reader will have asked why we should just ignore women if we are concerned about how the labor market is serving workers? Their story is much better. At the median, women’s hourly pay rose by about 50 percent, and at the 20th percentile it rose by 32 percent.
How is it that the economy has “failed” men but not women? In ongoing research, I am assessing whether the answer lies in my three-point plan above for going back to “the good old days.” If men received breadwinner premiums through the first part of the 20th century, then as more and more wives entered the labor force, it killed the normative basis for paying working-class men enough to raise a family on one income. With their productivity levels not otherwise high enough to justify these “rents,” the result was stagnation in men’s pay while their productivity caught up during the 1970s and 1980s. Since the 1990s boom, men’s and women’s pay have been rising and falling together, suggesting that we’ll not see anything like that extended period of stagnation in the future.
The timing of changes in men’s pay should lead us to question whether wage trends can account for any of our current crises. “Deaths of despair” began rising after 1998, at which point the 20th percentile of male pay had been rising for five years after falling for two decades. And among nonwhites, only drug overdoses rose; suicides and alcohol-related disease did not; Pay was 9 percent higher in 2017 than in 1998, but deaths of despair were skyrocketing (due to the opioid crisis, which is more about prescribing rates than economics).
Male labor force participation — about which I’ll have more to say — began declining in the 1930s or 1940s and has kept going down since, through the glory days of the 1950s and 1960s and the 1990s boom.
Family breakdown indicators tend to start heading south around 1960. Unwed birthrates rose primarily before the early 1990s (see here, here (Table 16), and here (Table 10)). Teen pregnancy has dropped dramatically since the early 1990s. Donald Trump was elected 43 years after 1973, and 23 years after male wages at the 20th percentile hit rock bottom. Male pay at the 20th percentile is 19 percent higher today than in 1993, when Bill Clinton took office.
The fact that male pay has stagnated rather than fallen is important for assessing the causal role that should be attributed to it. For instance, it becomes hard to see how the rise in single motherhood can be attributed to male pay, which didn’t change.
Okay, that’s more than enough for the first volley. Over to you, Oren!