The Job Market Doesn't Care About COVID Anymore
The Job Market Doesn't Care About COVID Anymore
And other insights from the jobs report. Plus: Lies, damned lies, and issue polling.
Happy Friday!
As promised, here’s a special Friday issue of Very Serious with some thoughts on the surprisingly good jobs numbers, and also on the uses and abuses of public opinion polls.
Jobs report surprise: Everyone was braced for the January jobs number to be terrible, due to Omicron, and yet the Bureau of Labor Statistics report showed 467,000 jobs added in January on a seasonally adjusted basis. November and December job gains were also revised sharply upward, with over 300,000 additional job gains attributed to each of those months.
The original, weaker job growth numbers for November and December were a bit puzzling at the time because of other economic data that came in much stronger — both GDP growth data and unemployment data from the household survey.1 But this month, the data points are in agreement. The report doesn’t just show robust job growth, it also shows strong wage growth (0.7% in one month) and it shows an uptick in the fraction of Americans participating in the workforce.
So this is great news. And while I wouldn’t rule out future revisions of this month’s surprisingly good numbers, there’s less reason to scratch our head at them than in some previous months. The household survey data this month, including a still-very-low unemployment rate of 4.0%, is consistent with the strong data on job growth.
A major thing this report tells you is the labor market is no longer fazed by COVID waves, and has been unfazed for approximately as long as vaccines have been available. As the economist Jason Furman pointed out to me, if you look back at the as-revised job creation numbers by month in 2021, you can’t even tell where the COVID waves were.2 It’s just a pretty smooth upward trend in employment. And however Omicron disrupted life this past month, it doesn’t seem to have stopped that trend.
Furman points this out to note that people should be a little less pat about the idea that good things happen together. The idea that fighting the virus effectively is key to growing the economy has been a commonplace for the last two years. People want to believe if you reduce infections you’ll be rewarded not only with less sickness and death but also with stronger growth, or that if infections rise that will hurt the economy as people become afraid to go out and do things. But that’s not necessarily true. Of course, lots of things are worth doing even if they don’t enhance economic growth or job growth. But people like to tell stories where all the good things move in line because they relieve us of the need to make trade-offs, and not everything in life works out that way.
All of this is not to say that the job market is back to normal. It’s more that it’s marching steadily back toward normal, with a long way still to go. Labor force participation is still significantly lower than pre-pandemic (62.2% now, versus 63.4% in February 2020) and that’s one reason there’s still a mismatch between consumer demand (strong) and availability of labor (limited), which is a driver of shortages and inflation. It’s another reason for the Biden administration to focus on processing the backlog of visa applications and getting the annual pace of immigration back to the US back to normal. We need the workers.
Survey says: If you watched Family Feud this Wednesday, you saw a round where the prompt was “name an activity that brings men closer together.” The correct answers were sports/fishing, “cold brewskis,” poker/games, and one that stumped both teams: “gay lovemaking,” which had been cited as a key male bonding activity by 5 of 100 survey respondents.
A few weeks ago on the Very Serious podcast, I talked with pollsters Kristen Soltis Anderson and Brian Stryker about how the way you ask questions influences the answers you get. If Family Feud’s question had been “name an activity that happens on ‘boys’ night,’” you wouldn’t have gotten five respondents to answer with variants on gay sex.3 But because the Family Feud producers love a double entendre, they went with the “closer together” language, got “gay lovemaking” onto the board, and launched a meme that I’m pretty sure will endure.
It goes to show that survey design matters.
Here’s a more consequential example. In an editorial, Science Magazine Editor-in-Chief H. Holden Thorp pushed back against legal efforts to prohibit affirmative action policies that have the practical effect of imposing a cap on Asian American enrollment at prestigious private universities, citing the statistic that “70 percent of Asian Americans support affirmative action.” Do they, though? To support this claim, he cites a column from Columbia University Sociologist Jennifer Lee, who in turn cites surveys sponsored by progressive groups that asked Asian American registered voters “Do you favor or oppose affirmative action programs designed to help Blacks, women, and other minorities get better access to higher education?”
This is obviously a question designed to elicit an answer of “yes.” Who could be against better access? Plus, it says the programs improve access for “other minorities,” which would presumably include Asian Americans. But this doesn’t describe the actual policy question at stake. Placing explicit weight on race in admission decisions is one of the least popular aspects of the Democratic Party agenda when you ask about it directly. When California voters were asked to vote on bringing the practice back to government institutions, including the state’s public universities, they rejected the idea soundly in a 2020 referendum, 57% against. Heavily Asian American communities such as those in the San Gabriel Valley of Los Angeles County rejected affirmative action when they were actually given the opportunity to vote on it.
Like the Family Feud question about male bonding, the question design that produced the 70% result was no mistake. One of the problems I discussed with the pollsters was that so much of the opinion research the public gets to see isn’t conducted with the objective of learning what the public thinks and wants; the objective is to convince politicians that the public wants what the group sponsoring the survey wants.4 So you get groups like Asian Americans Advancing Justice sponsoring survey research that progressives can wave around to say Democrats don’t need to worry about affirmative action and the Asian American vote because Asian American voters are overwhelmingly on board with it.
The only problem is it’s not true, and there are significant signs that Democrats are losing ground not just in Hispanic communities but also in Asian American communities, as unpopular progressive messaging on both education and policing has become increasingly salient. You can think affirmative action is worth pursuing even if it’s part of a suite of issues costing you Asian American votes, but don’t let biased survey research fool you into thinking it’s popular where it’s not.
That’s all for this week, but you’ll be hearing from me on Tuesday.
Wishing you a very serious weekend,
Josh
The monthly jobs report is based on two large surveys: A survey of employers (the “establishment survey”) conducted by the Bureau of Labor Statistics, and a survey of individuals (the “household survey”) conducted by the Census Bureau. The establishment survey is used to measure job creation and wage growth, while the household survey is used to measure the size of the labor force and the unemployment rate. In theory, the two surveys are measuring closely related matters and they should move in line. In practice, they don’t always do that, and measurement challenges related to the pandemic have made them noisier than before.
Until this report, we thought job growth had cratered from 1.091 million last July to 379,000 last September as Delta spread, but BLS has revised those numbers and now says we actually only added 689,000 jobs last July, falling much more modestly to 424,000 in September, on a seasonally adjusted basis. The seasonal adjustments are key here: BLS usually expects lots of summer layoffs from schools and big seasonal retail hiring before Christmas, but the shortage of workers appears to have reduced labor market seasonality, meaning roughly that June and July got too much credit for job growth, and November and December didn’t get enough credit, in the original measures.
I think.
Sometimes, you don’t even need a push poll to get reporters to incorrectly treat the opinion of young, left-wing Asian American activists as representative of broader community opinion.
Reading over this with a bit more time to digest the numbers, my understanding is that this initial take may have been a bit of a miss. The job market absolutely did care about COVID, January job numbers were hit pretty hard by Omicron, it's just that other corrections hid that fact.
https://www.slowboring.com/p/the-real-story-behind-last-weeks
https://apricitas.substack.com/p/omicron-and-the-labor-market
I'm not sure what the correct take is of course. Omicron appears to be passing fast enough that the impact will soon pass, so it seems likely that there's still even more room for improvement in jobs. But I think it is clear that if our vaccines were more effective vs Omicron, the January numbers would have been even better which directly contradicts the headline.
I understand from Kevin Drum (https://jabberwocking.com/we-are-pretty-close-to-full-employment/) that we're pretty much back on the trend line for labor force participation. That is, it has been trending down for years and the current numbers are simply reflecting that trend. There's not so much reason to expect that the *current* numbers should reflect, or match, those of two or more years ago.