The Mulligan and Mankiw challenge to extreme Keynesians

To quote Greg:

University of Chicago economist Casey Mulligan offers a challenge to that view.  Casey points out that there is a regular surge in teenage employment during the summer months because more teenagers are available to work (that is, the supply of their labor has increased).  That is no surprise: It is normal supply and demand in action.  But if aggregate demand were the main constraint on employment, this increase in supply should not translate into higher employment during deep recessions such as this one.  But it does!

The link is here.


Yeah, but teens don't have a long-run relationship with the firm. They'll be gone soon. Hiring a real employee means benefits, difficulty firing or getting rid of them etc. If there's a possibility of hiring these short-lived workers, treating them pretty poorly, then sending them on their way, it's low cost compared to taking on a 40-year-old who just got dumped from a mid- to low-skilled job.

So this observation, while cute, is neither here nor there, since labor is not homogenous in this context.

Wow. Those NYT commenters sure didn't like the Professor's hypothesis. He must be wrong.

"Demand could also surge temporarily in summer, thus 'solving the paradox' without "rejecting Keynesianism"."

Seems to me that a seasonal demand surge both bolsters and belies the Keynesian position. On one hand, seasonal demand might be correlated with seasonal employment gains, which suggests an AD shift will increase employment. However, it sort of puts a wrench in the sticky wages/non-fluidity of AD assumption they require to build most their models.

I am not enough of a macro guy to know if Keynsesians allow for any amount of rational expectations or not, but the summer glut of teen availability is totally expected by everyone. After all, many summer jobs only exist during the summer.

Why on Earth does Mankiw think that the summer vacation exists in the first place?

It exists because there was a demand for children and students to help out on farms. The longer hours of daylight mean there is more demand for outside and evening labour.

It's ridiculous to use this to suggest that supply creates the jobs. If he got out of his office a bit more he might notice that the sun is in the sky for longer. But if it's not in his models, I guess it doesn't exist...

The increase in employment is lower in the three years in which AD is low (08-10) and higher in the years where AD is high (06-07).

"Demand could also surge temporarily in summer, thus 'solving the paradox' without "rejecting Keynesianism"

Your thinking is totally backwards. If there were a surge in demand in the summer, that would be even more cause to reject Keynesian explanations. If demand is higher in the summer and supply still outstrips it, then it's even more clear that demand is not the source of the problem during the rest of the year. Now, a sudden large drop in demand could explain increased unemployment, but that's far-fetched.

Jason- you're missing the point. It's not a question of degrees. It's a question of whether aggregate demand is the main constraint to hiring. Clearly it's not.

One problem with this graph is that it shows relative employment, not overall employment. Total employment for 16-24 year olds could be flat overall, but the graph would show a spike if the 16-19 yr old cohort takes jobs from the 20-24 yr olds. The chart shows (16 to 19 employment) / (20-24) employment, but the article completely ignores the denominator.

"More teens available to work" is equivalent to "more teens demanding goods and services." The root cause here is "teens have more time to spend in the summer," which they can spend working or consuming. If teens primarily patronize businesses which are easily able to employ other teens (movie theatres, malls, water parks), the whole phenomenon could be self-contained.

It's likely a bit more diffuse than that, but might there be a corresponding decrease in the total employment of, say, SCHOOL TEACHERS?

"Your thinking is totally backwards. If there were a surge in demand in the summer, that would be even more cause to reject Keynesian explanations."

Indeed, that is one of the points nobody seems to want to address.


I respectfully say - "nonsense". I did get a chuckle out of this statement - "This effect is amplified by policies like the extension of unemployment insurance, means-tested mortgage and tax-debt relief, and various other supply-side policies." I am sure all those policies will work like a charm in getting people to produce more!

After reading the original article, the chart therein shows the seasonal "bump", but also that teen employment has steadily declined each year, even as a proportion of a declining labor market. Has the supply of teens dramatically decreased over the past few years? Have teens become less willing to work since 2006? Most teens aren't eligible for unemployment benefits, so surely that isn't a factor. What is the "supply-side" explanation for this?

I don't understand Mulligan's argument at all. Let's accept Mulligan's premises completely: It would disprove Keynesian economics if aggregate employment increases over the summer. OK... his graph doesn't say anything about aggregate employment. His graph shows relative employment levels, and the relative employment level of any subgroup should of course be proportional to its relative labor supply.

Say you have a fixed 1000 unskilled jobs on the demand side. In the winter, there are 250 unskilled high schoolers willing to work plus 1000 unskilled high school graduates on the supply side. So you'd expect about 200 employed high schoolers and 800 employed graduates. If over the summer you now have 500 high schoolers looking for work and the same 1000 graduates, you'd expect 333 employed high schoolers and 667 employed graduates. So: relative employment shifts with supply, but aggregate employment is fixed.

Alex F-

Your first sentence is correct. The proof of that is in the first sentence of your second paragraph - "Say you have a fixed 1000 unskilled jobs on the demand side." Since your assumption is that Mulligan is wrong and demand is fixed, your "demonstration" is of no value.

In any case, your example assumes that employers fire 133 graduates and employ 133 summer students. Presumably, when September comes around, they reverse the process. Now I ask you - do you have any evidence that this happens? I do not recall taking the place of a full time worker when I worked in the summers and then having him return in the fall when I hit the books again.

I'm having a hard time understanding this argument, but it might be due to my age. I grew up in a small town in the Central Valley of CA. During the summer, there was an explosion of recreational opportunities for kids. There were summer film series, the Fair, sports, the pool ( which was a big employer during the summer ), etc. For whatever reason, most of the jobs for these activities went to kids. In high school, a number of my friends were lifeguards and taught kids how to swim.

Some of my other friends did farm work, a lot of which varied by the month. August and September were the big months for that. It's certainly true that adults could have performed these jobs, but, since most of them had full time jobs, that wasn't likely. In any case, adults doing these jobs would have seemed strange in those days.

The bottom line is that the employment opportunities had to do with school being out, and the fact that adults wanted their kids time taken up with something outside of the house, and they demanded that it be well supervised. A couple of summers, I got paid to announce Babe Ruth and American Legion games.

Some are demand induced, some are supply induced, most generally both blades of the scissors are operating.

There are 20 million people in the 15-19 year old cohort. If the employed fraction increases by 10%, at $10/hour for 500 hours (1/4 year), that only represents

(20,000,000*0.10)*500*10 = 10 billion dollars.

That is 0.07% of the national economy, or ~1% of the stimulus.

Is economics a 1% accuracy discipline yet? No.

This is like saying hyperfine structure is a big challenge to pre-quantum physics.

Mulligan is full of it. He knows very well that both labor supply and labor demand are seasonal, but chooses to pretend that only one of them is. If, as he claims, only labor supply is seasonal, then wages should be higher in winter than in the summer for the same jobs, and they're not.

There's plenty of examples out there of dramatic seasonal labor changes: So much that there are countries where one has to adjust for seasonal elements for almost every piece of economic activity. Take Spain, Greece or Italy, who put together children vacation, paid vacation for adult, which is typically used in July or August, and an influx of tourists that peaks in the same summer months. In many of those places, said employment fluctuations have very little to do with extra teens available for work, as there isn't much a tradition of summer jobs at those ages: Very few students of any age spend much time working on the side. And yet, the effect remains.

The US, having very different cultural norms and being a far larger country, must have areas where this effect is major, while others where it is not. The data must be broken down into smaller pieces, both geographically and by age, to be able to make any serious conclusion.

Minimum wage was raised from 2008-2009 and again from 2009-2010. Teenage employment is almost perfectly correlated with the real minimum wage, leaving an explanation for why teenage employment is falling summer after summer.

Comments for this post are closed