The first puzzle about unemployment when thought about from within the search-matching framework is that unemployment rates are highest among the least skilled, i.e. among those worker/jobs with the easiest matches. It’s hard to believe that it takes a year to match a construction worker to a job.

the second puzzle is that uncertainty should matter most when hiring and firing costs are high and once again these costs are lowest for those workers with the greatest unemployment rates.

Read the whole post. The overall puzzle is why the unemployment rate is highest for low-skilled workers. I will get to that puzzle shortly.

I recommend looking at employment from a ten-year perspective. In December of 1999, total nonfarm payroll employment was 130.5 million. To keep up with population growth, employment should have increased by about 15 percent from then until now, to about 149 million. Instead, total payroll employment today is a tiny bit lower than it was in December of 1999. So there is an “employment shortfall” of 18 or 19 million.

If you limit yourself to the periods that the NBER calls “recessions,” they only account for about 2/3 of the shortfall. Another 1/3 took place during “recoveries.” For example, even though population has increased since the recession officially ended in June of 2009, total payroll employment is slightly less today.

At this point, I am not sure how much of the decline in the employment/population ratio is cyclical and how much is secular. To put it another way, when the economic cycle reaches its next peak, it is hard to predict how much lower the employment/population ratio will be relative to the glory days of 1999.

With that in mind, let us turn to Alex’s puzzle that the unemployment rate is highest among low-skilled workers. Here are some possible explanations:

1. Health insurance costs. When the employer provides health insurance, this cost represents a “wedge” between the take-home pay of workers and the compensation costs incurred by employers. The size of this wedge has increased over the past ten years. If your output is worth \$30,000 a year, and your health insurance costs \$15,000 a year, then you have to agree to accept \$15,000 a year as your nominal pay, which is about \$7.50 an hour. I suppose that if workers valued their health insurance at \$15,000 a year and were flexible about taking wage cuts, this could happen. But it doesn’t, and instead we get unemployment.

2. The story of skill-biased technological change and globalization would say that low-skilled workers face more downward pressure on wages than high-skilled workers. If there is the same resistance to wage cuts at all skill levels, there will be more unemployment among the low-skilled.

3. What we call “highly skilled workers” may be people who are better at self-deception. The superfluous high-school educated worker admits to being unemployed. The superfluous college-educated worker shows up as “enrolled in law school.” This leads to a misleading difference in measured unemployment by skill level, at least if we measure skill level as education.