Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Thursday, October 17, 2013

Employment Gaps

Is the level of employment in the U.S. currently too low? To many people, the answer to this question seems obvious: of course it's too low, you moron.

But "too low" relative to what? Relative to historic averages? Employment seems low relative to recent history, but high relative to more distant history; see here. Moreover, secular employment dynamics across demographic groups often move in different directions, making the question even more difficult to answer. (Marcela Williams and I talk at length about the "many moving parts" of the labor market here.)

Maybe we can learn something by comparing the U.S. experience with Canada. As far as different countries go, Canada is about as "close" to U.S. as one can get. Moreover, as I've pointed out before, the Canadian economy experienced a great slump in the 1990s, a phenomenon that appears to be playing out now in the U.S.

Let me start by looking at the employment-to-population ratios across these two countries. (In Canada, the population constitutes those aged 15+, in the U.S., those aged 16+). Here is what the picture looks like for prime-age males:

Employment is similar early in the sample, but a gap emerges in the 1980s, growing even larger during the "great Canadian slump" of the 1990s. But for most of the 2000s, up to 2008, the employment gap appears to have vanished. Since 2008, the employment gap has reversed itself: the employment rate among prime-age American males is now significantly lower (2 percentage points) than their counterparts in Canada for the first time in about 40 years.

Can we use these employment gaps to infer something about the slowness of the U.S. recovery? I'm not sure. Well, we have to be careful. But this picture might make one more sympathetic to the idea that there is an "output gap" in the U.S. that's at least as large as the value-added associated with increasing prime-age male employment by 2 percentage points. (Of course, this says nothing about what the source of the gap is.)

What does this data look like for other age groupings? Let's take a look. Here's the picture for "adult" teens:

A lot of this employment must be in the form of part time work. The employment ratios are low relative to other demographic groups, as one would expect, but the two countries are quite similar here until about 2000. What happened?

Here we have young adult men:

The picture here looks similar to the one for prime-age males. Together, the two pictures above show that the recent recession hit younger men in the U.S. harder than their counterparts in Canada, and also relative to older men in general.

As for older men:

Evidently, older men are immune from negative aggregate demand shocks. Interesting.

Let me now report what the same data looks like for females. For prime-age females, the picture is this:

For most of the sample, the employment ratios track each other fairly closely, with the Canadian ratio slightly below its American counterpart. Again, as with teenage men, something appears to have happened in 2000. The female employment rate appears to be in secular decline while, in Canada, it has remained elevated and stable. What are the implications of this recent divergence? And how should it be evaluated by policymakers? We need more data to answer these questions.

Here's the picture for teenage women. Again, a large cross-country gap emerges around 2000.

It is interesting to note that the upward trend in female employment is absent in this age category. It is also less apparent in young women:

But once again we see a significant divergence across these two countries beginning at around 2000. The recession in 2008 served to enlarge these differences.

Finally, for older women:

As with older men, older women seem largely impervious to the business cycle.

What is it that is leading older people to devote more time to market work -- seemingly at the expense of younger people? It is tempting to argue that the financial crisis, by wiping out retirement portfolios, compelled older people to work more to rebuild their lost wealth. But the trends here appear to have been in place since before 2000.


  1. Good post David.

    "Evidently, older men are immune from negative aggregate demand shocks. Interesting."

    But that is (roughly) what we would expect. If firms are hit with a negative aggregate demand shock, they would first respond by hiring fewer (or no) new workers, and only if the shock was a big one would they lay off workers. It's the younger workers, entering the labour market, who will be worst affected.

    Aside from that, I am wary of data on employment rates for 55+ men. Because it aggregates the 55 year olds, who are mostly working, with the 99 year olds, who mostly aren't. And with the baby boomers now hitting that age-group, the age-composition of the 55+ group has been changing a lot recently. My *guess* is that the increase in employment rates of 55+ men we have seen since 2000 is partly due to the early boomers turning 55. So the 55+ group is "getting younger" on average.

    1. Thanks, Nick.

      Good point. But if what you say is correct (and it sounds plausible), then any adverse shock is likely to age groups in the manner you describe. A standard search model would deliver this result in response to any shock that lowered the expected return to recruiting, together with the assumption that young workers are generally less productive.

      With respect to the age-composition of the 55+, I will have my RA look inot it.