I had a radio interview not too long ago on cyclical versus structural unemployment, and in rereading some old posts on the topic I came across this statement from Brad DeLong:
Let us remember what structural unemployment looks like. The economy is depressed and unemployment is high not because of slack aggregate demand generated by a collapse in spending, but instead because “structural” factors have produced a mismatch between the skills of the labor force and the distribution of demand.
That reminded me of a point I’ve been meaning to make. With all the talk about whether our unemployment problem is cyclical or structural (it’s mostly cyclical), many people are looking at measures of mismatches to assess how much of the problem is structural. But care needs to be taken in the interpretation of mismatch numbers. Here’s why.
Suppose that you run a business in Town A and you need someone to run a complicated piece of equipment. Unfortunately, the size of your town is relatively moderate, and there are no qualified job applicants available. You have advertised the job for weeks, but no takers. This sounds like a classic case of structural unemployment — there is a need for workers with a different skill set — but it may not be a structural problem.
Suppose also that the economy is in a recession, and business has not been good. Because of that, you can’t offer a very high wage. It turns out that in the very next town, Town B, there is a qualified worker who was laid off due to a business failure caused by the recession, but at the wage you are offering the worker is not willing to move. The worker has a job and is surviving, though the pay is much less than before and the worker is underemployed — the worker is mismatched — but the family is getting by.
However, if things were better — if the economy was humming away at full employment — the employer in Town A could offer a higher wage and induce the worker in Town B to move. If so, then this unemployment is cyclical, not structural. There is a mismatch, but the mismatch is driven by lack of demand.
The point is that when we talk about structural unemployment, we assume aggregate demand is not the problem. Thus, structural unemployment must be measured under a full employment assumption (i.e. how many mismatches would persist at full employment). Structural unemployment is driven by changes in tastes, technology, etc. that produce geographic, skill, or other mismatches that prevent reemployment. For example, if there is a change in tastes that causes the demand for hula hoops to fall and the demand for skateboards to increase, or a change in technology that causes the demand for typewriters to fall and the demand for word processing software to increase, then we have to move workers and other resources out of hula hoop and typewriter production and into the skateboard and word processing businesses. That will take time if there are geographic, training, or other barriers that prevent the quick translation of resources from one use or one place to another. Note, however, that the problem is not lack of demand. People want more skateboards and word processers than they currently have, so that unlike the example above where higher demand and the higher wages that come with it cause the worker to move and eliminate the mismatch, an increase in demand won’t fix the problem. If an increase in demand will fix the problem, as in the example above, then it’s not a structural problem.
The bottom line is that to measure structural unemployment in a recession, it’s not enough to simply survey the labor market and count the mismatches. You have to know if those mismatches would persist at full employment. To the extent that the mismatch problem is due to lack of demand, and wages and prices that are too low to induce resource movements to their best use, the problem is cyclical, not sructural.