Deciphering the trajectory of the US labor market has never been harder—and it’s never been more important.
It’s harder than ever for at least two reasons:
There is more volatility in the data: numbers are bouncing around with a speed and magnitude we have not seen before. As large parts of the economy were suddenly shut down to halt Covid-19 contagion, initial jobless claims climbed by a cumulative 44 million over the 12-week period to June 6th. Non-farm payrolls dropped by about 20 million in April and unexpectedly surged by 2.5 million in May--markets expected a 7.5 million drop.
There is more noise in the data. Because the shutdown was expected to be temporary, many layoffs were temporary, some workers were placed on reduced working hours, some were “employed but absent from work”, a category normally meant to capture people who are on leave or temporary absence, but that has been heavily impacted by coronavirus-driven business closures. Estimating how many people are out of a job and how many have left the labor force has become harder. The Bureau of Labor Statistics noted that close to 5 million “employed but absent from work” people had been classified as employed, whereas in its view they should have been classified as unemployed. This would have placed the May unemployment rate at over 16% rather than the official 13.3%--but would still have marked a significant decline, since the April numbers suffered from a similar classification issue.
It’s more important than ever for at least three reasons:
Unless the bulk of these staggering job losses can be reversed quickly, the economy will be condemned to a very slow recovery;
The longer people remain unemployed, the harder they will find it to get back into the workforce; this will cause a substantial loss of human capital which will exacerbate the skills gap and undermine productivity and growth for years to come;
The job losses have been skewed against the most vulnerable segments of the labor force: people with lower education and skill levels, lower wages and savings, as well as against minorities. Protracted high unemployment will exacerbate the inequality in incomes and opportunities that has already contributed to rising social tensions.
The higher degree of uncertainty, noise and volatility in the data extends to much of the economy and has highlighted the importance of gathering real-time data. We know that current economic trends can change quickly—in fact, we hope so: as parts of the economy begin to reopen, we hope that business activity and jobs can move fast towards their pre-crisis levels. Recent analyses have looked at Google mobility data and OpenTable restaurant reservations to assess the pick-up in dining, retail shopping, and other economic activities.
For the labor market, a valuable source of real time information is the data on job postings available on this website set up by Greenwich.hr and OneModel. I participated in the kick-off webinar last month, and had the opportunities to start delving into the data. The website has a number of interactive graphics that allow you to see the level of job postings compared to the pre-covid-19 norm.
Here are a few observations and sample charts:
Job postings confirm the improvement in the May BLS jobs report. Most industries bottomed out in April and then begun to stage a recovery. It’s especially encouraging that in many industries the recovery in jobs postings has accelerated in the first ten days of this month. The chart below highlights three of the sectors that have been most directly impacted by the shutdowns. Hospitality was the hardest hit: in March and April job postings had plunged to 80% below pre-covid levels; they have now recovered to about 50% below norm. Retail has recovered to 13% below norm from a low of -55%, and construction is at about -20% from a low of -40%.
Between retail and hospitality and entertainment, there is still great variability in the degree to which different industries have recovered, as the next chart shows.
The downturn and the latest recovery have been highly correlated across geographical regions, but the West lags behind, still 29% below norm compared to -19% for the Midwest. There is greater variation within each region than across regions: in the West, for example, Nevada is still reeling from the hit to its hospitality and entertainment industry, with job postings close to 40% below norm compared to just -5% for Montana.
Different states are reopening their economies at different speeds, so it’s interesting to compare states that have adopted very different approaches. For example, Georgia and Florida have been singled out for being among the fastest in reopening (some judged them to be dangerously hasty); their job postings are 25% and 27% below norm respectively, not too far from California and New York state (both about -31%), which have taken a much slower approach. I don’t want to read too much into these numbers because there are several factors at play, such as how diversified these states’ economies are and the fact that within each state different counties have reopened at different speeds—but it does suggest scope for some interesting analysis.
And when we look at the trends by job family, it’s interesting to find health care and IT at the bottom of the distribution, both close to 90% below norm, with no sign of improvement yet.
Before the pandemic, job postings have proved to be a useful leading indicator for jobs growth. How much can we trust them now? To the extent that the bulk of layoffs are temporary, we could see a rebound in employment not reflected in job postings, to the extent that employers call their old workers back; However, the data above indicate there is still a robust correlation between job postings and employment trends, which suggests job postings maintain a very important signaling value. Moreover, job postings might be even more valuable in identifying and predicting shifts in the labor market in terms of both industries and skills: sectors that recover more quickly might see a weak labor market as an opportunity to attract more qualified workers; and industries affected by structural changes—driven either by post-pandemic changes in work organization or by technology—might need different skill sets.
The jobs market is evolving at unusually high speed in an unusually dense fog of uncertainty and statistical noise; it’s worth keeping an eye on these real-time job postings data to get an additional line of sight into the recovery.
Comments