Economists widely expected Friday’s jobs report to be a dud thanks to the omicron wave, which left millions of Americans sick at home throughout the month. It turns out that they needn’t have worried. The U.S. added a strong 467,000 jobs in January as hiring mostly plowed through the effects of the virus, according to the government’s monthly data release. COVID did sideline more than 3.6 million workers during the month, more than at any other point in the pandemic, but not enough to derail the labor market recovery.
Beyond the surprisingly good headline numbers, however, Friday’s jobs data told a much broader story about the COVID recovery. It starts with a sort of frustrating fact: On a month-to-month basis, the jobs report was essentially misleading everybody for all of 2021.
One of the most important things to always keep in mind about the jobs report is that the initial numbers released by the Labor Department are very much a rough draft. These first estimates are revised in each of the two following months. In 2021, those revisions were often quite large. That became increasingly apparent as the year wore on, and made the whole exercise of breathlessly waiting for the preliminary figures feel a little bit absurd—especially since, due to the news coverage they get, the monthly reports help form our impression of how the economy is progressing.
But the Labor Department also does a much larger, comprehensive round of annual revisions for the entire past 12 months, where it benchmarks the total number of jobs based on comprehensive payroll data from the previous March. We finally got to see those revisions on Friday, and they were huge—big enough to fundamentally change the story of the job market last year in some important ways. Overall, the economy only added about 217,000 more jobs than previously believed. But the pattern of job growth was vastly different from what the old data showed.
Throughout much of 2021, it looked as if hiring was oscillating wildly, recovering after a weak winter, booming in the summer with more than 1 million jobs added in July, and tailing off again in the fall and winter, possibly due to the virus. In fact, none of that happened. Job growth just trucked along, consistently adding somewhere between 400,000 and just over 700,000 a month, with the exception of April. So why did job growth explode in the summer? It actually didn’t—we added 807,000 fewer jobs than we thought in June and July. Why did it collapse in the winter? It didn’t—we added 709,000 more jobs than we thought in November and December.
The most striking thing about these newly revised numbers is that the labor market appears to have been more or less insensitive to the state of the pandemic last year. Job growth tailed off modestly in September when the delta wave was peaking, but otherwise, it’s basically impossible to spot COVID’s impact with the naked eye. Harvard economist Jason Furman told me he thinks the same holds for most of the economic data from last year. “Hand anyone a major data series for 2021 and they would have no way to tell you when the virus was surging or when the virus was retreating,” he said. While I don’t want to draw too many strong conclusions based on one morning skimming data, it feels like you could mostly say the same about government policy, like the expiration of enhanced unemployment benefits.
Why were the initial jobs reports so far off in 2021? It seems to have to do with the adjustment process the Labor Department uses to smooth out the impact of seasonal hiring and layoffs in the data. That process appears to have gone utterly haywire, thanks to the unprecedented job swings of 2020 and 2021, and only with this last round of revisions have the government’s statisticians been able to fix it. That explains why, while the month-to-month data was badly off, the department was pretty on the mark when it came to the overall level of job growth over the year. Here’s how the issue was described in this month’s report:
Now that there are more monthly observations related to the historically large job losses and gains seen in the pandemic-driven recession and recovery, the models can better distinguish normal seasonal movements from underlying trends. As a result, some large revisions to seasonally adjusted data occurred with the updated models; however, these monthly changes mostly offset each other.
A lot of economists and journalists suspected that seasonal adjustments had been messing with the numbers, but maybe not quite to this extent. It’s as if the entire economics profession spent 2021 reading labor market data while wearing extremely strong eye glasses, and the whole time it turned out to be the wrong prescription.
So this is yet another reminder that all knowledge is tentative and economic narratives are often fragile artifacts of bad data. On the other hand, the labor market appears to have been brushing off the pandemic for months now. You win some, you lose some.