A comparison of two sets of jobs data shows that labor demand may actually be strengthening despite fears about machines taking our jobs, according to Brad Hershbein and Nathan Sotherland of the W.E. Upjohn Institute for Employment Research.
The U.S. has added 15 million jobs over 79 months, the longest recorded consecutive streak of job gains. However, the pace through September has weakened and wage growth, has slowed. In fact, the Conference Board’s Help Wanted OnLine (HWOL) Index, which has tracked electronic job postings since 2005, has shown a decline in want ads of 15 percent since the November 2015. The only previous time a drop of that magnitude occurred was during the Great Recession itself (See figure 1).
With growing debate whether machines are replacing humans, is the HWOL evidence that the machines are beginning to win? Has the demand for labor begun to fizzle?
NOTE: The HWOL index represents a three-month moving average. The shaded area indicates the Great Recession.
The short answer: probably not. Job growth is still more than fast enough to increase the share of the population with jobs. Real wage growth has slowed not because employers are giving out smaller raises, but because energy prices have mostly stopped falling and putting downward pressure on inflation.
A curious pattern emerges when the HWOL Index is compared with the Bureau of Labor Statistics’s Jobs Opening and Labor Turnover Survey (JOLTS), which tries to measure the stock of all job openings that employers intend to fill in the next 30 days, regardless of how (or even whether) they advertise the opening. While HWOL and JOLTS technically track different things, they both measure job openings.
There’s an odd deviation that started in 2014, when JOLTS showed rapid increases in the number of job openings while HWOL indicated more muted growth, shown in the chart below. The indicators seem to be telling different stories. Why?
NOTE: Both indices represent three-month moving averages. The shaded area indicates the Great Recession.
As economists at the Federal Reserve Board of Governors have pointed out, the culprit may be Craigslist. Craigslist began charging $25 for each job ad between 2012 and 2015 in metro areas for which it had previously been free to post vacancies. According to the Fed economists, this higher posting price for a single job board site—albeit a very large site—can account for one-quarter of the deviation between HWOL and JOLTS over the 2013–2015 period.
However, that still doesn’t explain what’s driving the reversal in direction between HWOL and JOLTS around the start of 2016, or in other words, the apparent decrease in job postings.
The Fed researchers document an additional price increase for Craigslist job ads at the end of 2015, to $35 or $45 in almost every major metropolitan area. It’s possible that this latest price increase did not just slow the growth of the HWOL index, it actually contracted the index as fewer employers decided to post on Craigslist. For this to occur, the employers could not simply have posted the ad to a competing online job board, as such substitutions would presumably still get counted in HWOL.
Instead, we suspect that some employers are simply posting fewer jobs in total now. Most likely these are employers that used Craigslist exclusively for jobs that did not justify posting on multiple boards— jobs that are likely temporary, casual, or relatively unskilled.
In fact, another source for online job postings, labor market analytics firm Burning Glass Technologies (BGT), which apparently puts less weight on Craigslist ads, shows trends much more in line with JOLTS.
This pattern of increasing demand is also echoed in another online postings aggregator, LinkUp, which includes only ads on corporate web sites.
When these pieces of evidence are stacked together, it’s just one suggestion that labor demand is not weakening, but may actually be strengthening. Job postings are moving away from casual work more likely to be found on Craigslist and toward formal—and more-skilled—jobs advertised by companies and larger employers. So, at least not for the humans who have the right skills, no, the machines aren’t coming yet for all our jobs.
(Top photo: Courtesy Getty Images.)
Brad Hershbein is an economist at the W.E. Upjohn Institute for Employment Research.
Nathan Sotherland is a research assistant at the W.E. Upjohn Institute for Employment Research.
All views expressed are those of the authors.