Cyclical vs Structural Unemployment part N

Robert Waldmann

Kevin Drum stresses the very sound point that even if part of current unemployment is structural, we should stimulate to get rid of the part which is cyclical. I don’t have a serious disagreement and choose to debate his guess as to the level of structural unemployment for the sake of debating.

“The “normal” unemployment level is about five points less than it is today. I wouldn’t be surprised if perhaps three of those points are cyclical and two are structural.”

He agrees with Annie Lowrey who presents the following analysis

The unemployment is cyclical and structural. Most sectors have suffered from the turndown, but job losses are concentrated in some industries: In residential construction, they are down 38 percent since 2006. (Between Aug. 2007 and Dec. 2009, unemployment in construction quintupled from about 5 percent to about 25 percent.) In health care and education, however, jobs are up.

This analysis is accidental theory. Between the first sentence and the second, there is a theoretical argument that the cycle has the same effect on log employment in each sector. This argument makes no sense. More after the jump.

I know it is not wise to argue with an “I wouldn’t be surprised” but their approach to measuring cyclical unemployment is completely incorrect. Lowrey identifies the cyclical component by assuming that around the cycle percent changes are equal. Therefore a much much larger percent change in construction than in health is not cyclical.

In fact, the amplitude of the cycle in log employment is not the same in each sector. Some sectors are very cyclical (always go down a lot in recessions) others are almost acyclical. A correct estimate of how much unemployment can be eliminated with fiscal and monetary policy requires an estimate of the effect of fiscal and monetary policy on log-employment by sector. The assumption that this is necessarily equal is an accidental theory — a very strong and plainly false assumption which is made by people who think they can just look at the data without theory.

This is the topic of one of my very rare contributions to the actual economics literature

The practical relevance of the decomposition is due to the fact that it is argued, that, if unemployment is due to miss-match, then a general stimulus will lead to labor shortages in some sectors. This would lead to inflation. As Drum notes this is not a problem at the moment. However, let’s imagine a stimulus powerful enough to drive unemployment down to 7%. Does anyone really think this would create (much more of) a shortage of workers in health care ?

Why would demand for health care increase much (as a percent of current demand for health care). 85% of people are insured. Much of the effect of the recession is people moving from private insurance to Medicaid. There isn’t a big cycle in the number of uninsured and there is little reason for demand for care by the insured to shift with aggregate demand. Even the uninsured demand health care in ERs, then go bankrupt. Recall the HCR debate.

Now how about residential construction. Does anyone really think that no construction workers can shift from residential to commercial construction ? Is there any reason to think that an increase in employment which were to drive the unemployment rate down to 6% would cause labor shortages anywhere ?