This article appears in the May 20, 2013, issue of National Review.
In a perfect frictionless economy, there would never be any unemployment. If a number of workers were expected to lose their jobs tomorrow, firms would anticipate an increase in the supply of workers, everybody’s wage would drop, and higher demand for the newly available workers would emerge instantly. In the real world, of course, there are many frictions that make non-zero unemployment possible. Most important are skill mismatches and geographic mismatches. A surge in the demand for engineers might not reduce the unemployment rate today if it takes time to train engineers. An energy boom in North Dakota can create job listings that remain unfilled until workers decide to move to North Dakota.
In one of the more important papers in economic history, Christopher Dow and Louis Dicks-Mireaux showed that there is a regular relationship between job openings and unemployment. When there are many unemployed workers, openings disappear relatively quickly, and when there are many openings, unemployment tends to be low. Economists later began referring to this curve as the “Beveridge curve” after William Beveridge, an economist who studied unemployment in the first half of the 20th century.
Subsequent research has demonstrated that economies tend to move up and down a relatively stable Beveridge curve over the business cycle. In recessions, there tends to be high unemployment and few openings.
The nearby chart portrays one of the most striking shifts in U.S. labor-market data on record. Data on U.S. openings and unemployment from January 2001 to February 2013 are shown, and non-linear estimates of the Beveridge curve are provided for the periods from January 2001 to August 2009, and from September 2009 to February 2013.
In an economy with low friction, the Beveridge curve would be very close to the origin. Unemployment would tend to be low, and openings filled quickly. In an economy with large matching problems, high unemployment and high job openings could coexist, and the curve could be farther from the origin. The chart indicates that the Beveridge curve has shifted out sharply during the Obama administration.
It has been almost four years since the end of the recent recession, but the U.S. has yet to return to its previous levels of unemployment. The shift in the Beveridge curve suggests that it may never do so. The points labeled A and B illustrate why. In February 2013, the job-openings rate (unfilled jobs as a percentage of total jobs) was 2.8, a rate that would have corresponded with an unemployment rate of about 5.25 on the Beveridge curve from 2001 through August 2009. The unemployment rate in February, however, was 7.7—almost two and a half points higher.
What explains the shift in the Beveridge curve? The biggest factor is likely the massive increase in the number of workers who have become long-term unemployed. The long-term unemployed made up only 17.3 percent of all unemployed workers in December 2007; today they constitute fully 39.3 percent of all unemployed workers in the labor market. In December 2007, the average length of unemployment was 16.6 weeks; today, it is much higher, at 37.7 weeks. There is growing evidence that employers are extremely reluctant to make a job offer to persons who have been out of work for more than six months. A recent study by Rand Ghayad involved submitting fake résumés in response to job postings and varying the levels of experience and lengths of unemployment on them. He found that the chances of receiving a callback from an employer dropped off significantly for the long-term unemployed; a worker who had been unemployed for only a short time and had no relevant experience in the industry to which he was applying was more likely to receive a callback than a long-term-unemployed person who did have relevant experience. Providing 99 weeks of unemployment insurance may not have helped matters, since it encouraged workers to stay out of jobs for longer and they then became mired in unemployment for the long run. (There are large geographical differences in unemployment as well, and workers reluctant to move to find jobs may have become stuck as their area floundered while others, such as North Dakota, flourished.)
Employers make that choice for rational but unfortunate reasons: Workers who go through longer periods of unemployment have a heightened risk of substance abuse and suicide, a shorter life expectancy, and a higher likelihood of personal problems, including divorce and troubled children. The costs of long-term unemployment are high, and the shift of the Beveridge curve implies that they may stick around for a long time.
-Kevin Hassett is the John G. Searle Senior Fellow and Director of Economic Policy Studies at AEI