A central question in our assessment of economic conditions, and the one on which Scott and I perhaps disagree most squarely, is how to interpret stagnation.
In my view, 40 years of either little or no wage growth in an economy where others are experiencing strong gains is a disaster that has great explanatory power for the problems we see, both in the labor market itself (labor-force participation) and in society beyond (family formation, substance abuse and suicide, dissatisfaction with the country’s direction, etc.).
I understand Scott’s view as being that, while not good, stagnation per se cannot explain other things getting worse. After all, if wages stay the same, shouldn’t labor-force participation, and general attitudes, stay the same too? Thus, via Twitter, he wrote recently: “Stagnant is bad, but it’s not ‘worse’. Can’t explain why Trump wins but Wallace couldn’t have.” In his first post of this exchange, he wrote, “The fact that male pay has stagnated rather than fallen is important for assessing the causal role that should be attributed to it.” In his second, he wrote, “male pay has not declined. That strengthens the case that the drop in labor-force participation among men is more about labor supply.”
So why do I think people’s relative position is what matters — that being stuck and falling behind have such consequential effects for attitudes and behavior? The easy answer would be to say “because it just does”; rational or not, that’s how people feel and we have to deal with it. But I think I can do better, in both conceptual and substantive terms.
Conceptually, it’s actually quite straightforward to show that the relationship between labor-market outcomes and attitudes/behaviors is based on relative rather than absolute outcomes. The way to see this is to recognize that the point at which wages began stagnating in the 1970s is actually an arbitrary one to use for this discussion. “People were happy with 1970s wages in 1970, so why wouldn’t they be in 2018?” sounds coherent only because we take for granted that people can and do live on 1970s wages today. But if absolute wages are what matter, why look to 1970? We could just as well look back to 1910 and ask whether we think it would work to have less-skilled Americans today earning 1910 wages, which were fine in 1910.
The response to that question might be that the problem is not the 1910 wage per se but rather the decline from the 1970 to 1910 level. But that is precisely the point. If one takes the “absolute” view seriously, declines can’t be important either. What is a decline, after all, but a shift from one absolute level to a different absolute level, both of which proved perfectly adequate at various times in the past. To claim that “decline” would be cognizable is to already concede that it is relative trajectory, not absolute level, that matters. And if that’s the case, then stagnation — while better than decline — can still trigger all manner of ills.
Still, we should want a substantive explanation for why this intuition is correct. One explanation is the easy one: that humans care about status and no amount of adjustment to our inflation indices will reverse countless millennia of cultural and biological evolution. That it’s an easy explanation doesn’t make it any less true. If your work isn’t respected, if you can’t provide your family the things they see that others around them have and that the cultural mainstream takes for granted, if the restaurant you can afford for a special occasion is a sitcom punchline, that matters.
A second explanation, which I think Scott acknowledges as well, is the problem of our safety net. As society becomes wealthier, we raise the baseline standard of living that we strive to provide for everyone. If that baseline approaches or surpasses what people can earn in the labor market, major problems will arise in both economic incentives and cultural attitudes toward work.
Third, being left behind economically can have very tangible effects if society begins to orient itself around incomes and living standards beyond your reach. Businesses focus their attention where the consumer-spending power is and segments that do not offer profitable opportunities at scale are out of luck. I’m not aware of good analysis that has been done on this issue, but I suspect that the type of family vacation one might have planned in 1970 on a $35,000 salary would be hard to replicate today. Certainly, in health care, few alternatives exist to the gold-plated insurance model. Many business models now presume access to a smartphone.
Scott’s last post focused in the weeds (his words!) on the nuances in calculating wage trends. I had raised a number of factors that his own calculations disregard — commute times, income volatility, marginal value of health expenditures, payroll taxes, low-income labor force dropout — and as far as I can tell, he by and large acknowledges that those things have gotten worse, but sees them as less important than his own adjustments. Our main disconnect, and likely a result of confusing wording on my part, is with regard to health benefits. I don’t worry that we are miscalculating health-cost inflation, I worry that high-cost health insurance has much less value to low-income households than the cash it is replacing. The Oregon Medicaid study, for instance, found that Medicaid recipients got 20 to 40 cents of value per dollar spent on their coverage. I don’t think a 2016 family with $30,000 of wages and a $20,000 insurance policy is as well off as a 1975 family with $40,000 of wages and a $10,000 policy.
My point in raising such factors was not to suggest that I had found the One Adjustment to Rule Them All but rather the opposite, that all of the possible adjustments in both directions don’t change the overall picture of stagnation. If there’s some magical line where 0 percent wage growth is okay but a 3 percent decline disastrous, then we’d need to keep haggling. But if relative standing is what matters then 10 percent up or 0 percent up or 10 percent down over 40 years still gets us back to the same mess — one confirmed by the other available lines of evidence.
So I’ll end with a question to Scott: Whatever your preferred estimate of wage growth, over how long a period of time do you think that such stagnation would be unlikely to cause problems? If we’re comfortable with wages up 10 percent since the 1970s, how about since 1960, or since 1910? And if there is a point at which stagnation or minuscule gains would be unsustainable and would lead toward concerning behaviors and attitudes, and if we agree that we are in fact seeing those concerning behaviors and attitudes, then why can’t the breaking point be the decades of stagnation we’ve experienced?