How easily can firms find workers? How long does it take to hire them? These are crucial questions for economists who study unemployment. Unfortunately, the available data are very bad, but for very good reasons.
The main workhorse models of unemployment include, at their core, “search frictions”—forces that prevent willing workers from matching up with available jobs. The models also rely on the following premises: The more willing workers out there, the more likely an available job is filled. And the more available jobs out there, the more likely a worker finds one. But how does an economist define an available job? Is it a posted job vacancy? In the stylized world of economic models, a worker who is hired fills a vacancy that was posted; the posted vacancy is necessary for the hire. However, as we see in the graph, hires almost always outnumber posted vacancies. Clearly, then, many hires occur without an explicit posting. Elsewhere in the labor statistics world, this reality is acknowledged: Unemployment is calculated every month by asking would-be workers how they searched for a job. Responding to a vacancy is only one of a dozen other methods of searching, including asking friends and relatives. The vacancy posting measure clearly undercounts the number of available jobs.
How this graph was created: Search for and select “Hires: Total Nonfarm, Level in Thousands” (first the seasonally adjusted and then the not seasonally adjusted series) and add them to the graph. To create the ratio, we must add the job openings series. In the “Add Data Series” section, search for and select “Job Openings: Total Nonfarm, Level in Thousands, Seasonally Adjusted” and select “Modify existing series” for series 1 (the smoother blue line, which is seasonally adjusted). Then enter the formula a/b in the “Create your own data transformation” section. Now do the same for series 2 (the rockier red line) with the job openings series that is not seasonally adjusted.
Suggested by David Wiczer.
View on FRED, series used in this post: