The Beveridge Curve and Structural Unemployment

Written by Dean Baker   
Tuesday, 01 January 2013 17:12

There is a whole industry of economists and policy types who are incredibly anxious to find evidence of structural unemployment. The reason is that structural unemployment implies a mismatch of the available jobs and the skills of the unemployed.

This is important because if our unemployment problem is structural then simply using macroeconomic policy to generate more demand won't be of much help. If we want to get people back to work we have to get them the right skills or in the right place (there can be a locational mismatch as well) for the jobs that are available. That is a very difficult and much more complicated process than just spending money.

Recently some proponents of the structural unemployment view of the economy have been highlighting an outward shift in the Beveridge Curve. The Beveridge Curve relates unemployment to the vacancy rate (the number of job openings divided by the number of jobs). In general, higher rates of unemployment are associated with lower vacancy rates. However, in the last couple of years there has been some increase in the vacancy rate without as large a drop in unemployment as would ordinarily be expected. This is taken as evidence that employers are having difficulty finding workers with the necessary skills in spite of the large number of unemployed workers. This is the story of structural unemployment.

A new paper from the Boston Fed by Rand Ghayad and William Dickens looks at this shift in the Beveridge Curve more closely. It finds a very interesting story. If we look at the long-term unemployed (people who have been out of work for more than 26 weeks) we see this shift clearly.

The current vacancy rate of around 2.5 percent used to be associated with rates of long-term unemployment of around 1.0 percent. Now it is associated with rates of long-term unemployment of close to 3.0 percent. That is a substantial outward shift that would be consistent with a large rise in structural unemployment. However if we just look at the short-term unemployed, workers who have been unemployed for fewer than 27 weeks, we don't see the same story. There is no evidence of any shift whatsoever. In fact, a 2.5 percent vacancy rate appears to be assoicated with a slightly lower rate of short-term unemployment than was true before the downturn. This pattern should be troubling for the proponents of the rising structural unemployment view.

Some folks may look at these curves and claim that the long-term unemployed are the misfits, so it should not be surprising that we only see the shift in the Beveridge Curve among this group. There are two problems with this story. First, when we saw shifts in the Beveridge Curve in the past it occurred among both the long-term and short-term unemployed. The other problem is a simple logical one. The long-term unemployed had all previously been short-term unemployed. If this group has a chronic mismatch with the available jobs then it should also have shown up as an outward shift in the Beveridge Curve for the short-term unemployed when they had been unemployed for less than 27 weeks. Clearly this story doesn't work.

There is an explanation having to do with extended unemployment benefits. There are two ways this can work. First, because workers know that they can get benefits for much longer than the normal 26 weeks they are being more choosy about accepting jobs. While there is likely some truth to this, it cannot be too large a factor because this should also affect behavior among workers in their first 26 weeks of unemployment. In other words, if workers willingness to accept a job offer is influenced by the fact that they can get benefits for a longer period of time, then we should also expect to see an outward shift in the Beveridge Curve for the short-term unemployed. Since we don't, this explanation doesn't fit the data.

There is another way in which extended benefits can affect the behavior of workers. In order to qualify for benefits workers must continue to look for work. This could mean that many workers who would have otherwise given up looking for jobs after 40-50 weeks of unemployment keep looking in order to continue to qualify for benefits. Insofar as this is the case, extended benefits are not keeping people from being employed, but rather keeping them from leaving the labor market and therefore raising the unemployment rate.

The findings of Ghayad and Dickens would be consistent with this view. In this case we are not looking at structural unemployment in any meaningful sense. The extended duration of unemployment benefits are not keeping people from working, they are just increasing the number of people counted as unemployed. (Jesse Rothstein has found that the vast majority of people who see their benefits expire leave the labor force; they do not find work.) 

It is also possible that the shift in the Beveridge Curve could be due to a change in employer behavior. If employers see a large pool of unemployed workers and see little urgency in hiring due to weak demand, then it is plausible that they would spend more time choosing among job applicants. It doesn't make sense to be picky if there is little reason to believe that you can find better applicants, however with a large pool of unemployed it is more likely that a better applicant will be available if an employer waits.

This employer behavior story is also not consistent with a structural unemployment explanation for the shift in the Beveridge Curve. While a full explanation for this outward shift will require more research, Ghayad and Dickens' work shows clearly that there is not a simple structural unemployment explanation for the shift.