Author: Andrew Berger-Gross
Back in November 2014, LEAD published an article describing the mismatch between unemployed job seekers and the record number of job openings in North Carolina and the broader United States. Job vacancy rates (the percent of all nonfarm jobs that are currently vacant) have fully surpassed their prerecession highs, demonstrating an increase in labor demand as the economic recovery has chugged along. However, the unemployment rate has remained stubbornly higher than its prerecession low, suggesting that the labor market is having trouble matching unemployed workers to available jobs.
The following graph shows the relationship between unemployment and job vacancy rates in North Carolina from 2006 through the most recent 12-month period as a continuous, sequential line.i This relationship is known as the Beveridge curve.
By fitting a dotted curve through the “normal” period from 2006 to 2009, we can clearly see what we’ll call (nonscientifically) a “gap” in North Carolina’s labor market, represented by a dotted arrow in the graph below. This gap represents a form of what economists call “structural unemployment” — joblessness that persists despite an improving economy. North Carolina’s unemployment rate should be several points lower than it currently is given the large number of job openings we are seeing.
So why are we seeing a gap emerge in North Carolina’s labor market despite several years of continuous economic growth? Following the lead of a 2012 article from Rand Ghayad and William Dickens at the Federal Reserve Bank of Boston, we disaggregated the Beveridge curve to see whether any explanations lingered beneath the headline numbers.
We performed this disaggregation by graphing job vacancy rates with unemployment rates that only included certain segments of the unemployed, allowing comparisons of North Carolinians from different age groups, levels of educational attainment, and the industry of the last job they held (e.g., manufacturing.) We found that whether a job seeker was older or younger, more or less educated, or whether they had previously worked in manufacturing made little difference in their contribution to the state’s job vacancy-unemployment gap.
However, comparing the Beveridge curves for the short- and long-term unemployed did provide some important clues. The long-term unemployed (commonly defined as job seekers who have been without work for 27 weeks or more) are still weighing heavily on the state’s unemployment rate, even as job vacancy rates have declined:
Meanwhile, unemployment rates have, for the most part, recovered in tandem with job vacancy rates among the short-term unemployed, yielding a relatively smaller “gap” for this subgroup:
The contribution of long-term unemployment to North Carolina’s labor market mismatch is made even clearer by plotting individual Beveridge curves across more detailed unemployment durations. The following graph shows that there is hardly any job vacancy-unemployment gap at all for unemployed job seekers who have been without work for less than five weeks. The gap grows wider and wider as we consider longer jobless spells:
Figuring out how to assist the long-term unemployed can be difficult because there is little agreement about the causes of protracted joblessness. It might be that the long-term unemployed are (on average) less employable than other workers for some reason. On the other hand, employers might (correctly or incorrectly) presume that long-term unemployed job applicants are poor candidates for employment; a recent experiment showed that long-term unemployed job applicants were less likely to get callbacks than identical job candidates who were more recently jobless. This is despite the fact that the long-term unemployed largely resemble the short-term unemployed along most observable traits, a finding that LEAD has independently verified using North Carolina-specific data.
Although the precise causes of long-term unemployment are contested, the Beveridge curve provides compelling evidence that the long-term unemployed continue to create a significant drag on North Carolina’s economy. Previous research has shown that this dynamic is occurring nationwide, prompting the federal government to commit additional resources to resolving the problem and encourage states to develop new programs focused on putting the long-term unemployed back to work.
The Current Population Survey (CPS), Current Employment Survey (CES), and Job Openings and Labor Turnover Survey (JOLTS) estimates are based on surveys and are subject to sampling error. All data sources cited in this post are also subject to nonsampling error. Any mistakes in data management, analysis, or presentation are the author’s.
i Unemployment rates for these graphs are constructed using data from the Current Population Survey. The job vacancy rates use nonfarm employment estimates from the Current Employment Statistics program and a job vacancy series derived by the author. We use state-level data from the Conference Board's Help Wanted OnLine® (HWOL) and perform an adjustment to account for structural shifts in the relationship between the HWOL and the Bureau of Labor Statistics’ Job Openings and Labor Turnover (JOLTS) series. This adjustment is accomplished by dividing the U.S. JOLTS job vacancy estimate by U.S. HWOL job postings, then multiplying by the number of HWOL job postings in North Carolina: (JOLTSUS / HWOLUS) × HWOLNC. This simple pro rata approach is consistent with research authored by several other economists who have used aggregate vacancy measures as a proxy for — or an adjustment to — subnational job postings data.