Shifting through the implications of the Beveridge Curve

The U.S. Bureau of Labor Statistics last week released new data from its Job Openings and Labor Turnover Survey, better known as JOLTS. The data set gives us information on the amount of movement in the labor market: how many workers left a job, how many were hired, and how many jobs are employers offering. One way to digest this data is by charting a relationship known as the Beveridge Curve.

Named after British economist William Beveridge, the curve shows the relationship between the unemployment rate and the jobs vacancy rate, or the number of job openings as a share of the labor force. If the curve shifts right and upwards, like we’ve seen during the economic recovery in the wake of the Great Recession of 2007-2009 (See Figure 1),  most conclude  that employers are posting more job openings but don’t like what they see among possible new employees.

Figure 1

But is that interpretation of the shift correct? Well, the increase in unemployment during and the high levels after the Great Recession is uniquely affected by high levels of long-term unemployment, so that might have something to do with the shift. The Heritage Foundation’s Salim Furth points out that may indeed be the case. And sketching out the Beveridge Curve with the long-term unemployment rate– instead of the overall rate—confirms this notion that most of the shift is driven by long-term unemployment (See Figure 2.)

What might be behind this somewhat surprising relationship? A working paper by economists Kory Kroft of the University of Toronto, Fabian Lange of Yale University, Matthew Notowidigdo of Northwestern University, and Lawrence Katz of Harvard University posits that most of the increase in the long-term unemployed is because many unemployed workers who otherwise would drop out of the labor force have stayed in due to extended unemployment insurance. As a result, the unemployment rate remains high because workers keep looking for jobs.

There’s also evidence that the job-openings rate portrays employers’ search for workers as higher than it actually is. A paper by economists Steven J. Davis and R. Jason Faberman at the University of Chicago, alongside John C. Haltiwanger at the University of Maryland looks at the jobs opening rate and finds a decline in “recruiting intensity.” This means that employers are increasing postings but that they aren’t super interested in hiring yet. So again, the shift might be overstated.

But let’s pull back a bit here and look at the historical record. The JOLTS database only goes back to December 2000. Looking at the shift of a curve with 15 year-old data can be problematic. If we think there might be a structural shift in the labor market, then we should look at data that covers a much longer period.

That’s just what a report published last year by Peter Diamond, a Massachusetts Institute of Technology economist and a Nobel laureate, and Ayşegül Şahin of the Federal Reserve Bank of New York did. They constructed a data set that spans back to 1951. They show that the Beveridge Curve has shifted out several times over the past six decades, but these shifts weren’t indicative of major structural changes in the labor market or a rise in structural unemployment.

What’s the lesson here? Shifts in the Beveridge Curve should be interpreted with caution. A shift out of the curve isn’t necessarily a sign of a rise in structural unemployment. Simple curves and graphs can be clarifying and instructive, but they aren’t the end of the analysis—particularly when the underlying data is constrained to a narrow time frame. Taking the time to dig in more is a worthwhile investment.