Saturday, December 27, 2008

Rebutting Mulligan's argument that the rise in unemployment is primarily due to decreased desire to work

Casey B. Mulligan had an article published in the New York Times last Wednesday strongly implying that for the current recession, "the decreased employment is explained more by reductions in the supply of labor (the willingness of people to work) and less by the demand for labor (the number of workers that employers need to hire)."

I left a strong critique of his argument in the comments section of the article, and on his blog, here and here. I think these are good responses, and I encourage you to read them later, but first I'd like to go through the basics of this issue and some of the key arguments very clearly and carefully.

Here is what I see as Mulligan's argument, based on his description of it in his New York Times article:

1) If there is a drop in supply of workers (that is, for whatever reason, workers want to work less), then the firms will be short handed, and will have to work the remaining workers harder (less downtime, less sitting around waiting for something to do), and so worker productivity will go up. There will be more goods produced per worker.

We have seen a recent rise in productivity, thus, there is strong evidence that the supply of workers is down, that is people want to work less (at least the way Mulligan wrote his New York Times article, he either intentionally or unintentionally -- and it looks like intentionally to me -- made it really sound like he thought this was strong evidence.)

2) If, on the other hand, there is a drop in demand for companies products, and thus for workers to produce those products (people don't want to work less in this case; it's just many can't find jobs), then the firms will not be short handed. On the contrary, they will have more workers than they need, and so there will be more downtime, more sitting around waiting for something to do, and so worker productivity will go down. There will be less goods produced per worker.

Because we have not seen a recent drop in productivity, instead it's risen, there is strong evidence that it's not the demand for products and workers that's down, rather it's the supply of workers that's down. That is, people want to work less.

And in two previous severe recessions, the 1930's and early 1980's, we did see a drop in productivity, but in this recession we see a rise in productivity (um, what about all of the recessions besides the two Mulligan chose to mention?).

So, that appears to be his argument. I think it's very weak because, for one, there are other very plausible ways productivity can go up in a recession without there being a decrease in workers desire to work. And for two, the other evidence and logic regarding recessions is overwhelmingly against Mulligan's hypothesis that in the current recession unemployment is due mostly to people's desire to work less (see Nobel Prize winning economist Paul Krugman's current book, The Return of Depression Economics and the Crisis of 2008, for details). This is why the vast majority of top economists don't agree that a decrease in the supply of workers, the desire to work, is the primary reason for the recent high unemployment (for example, see here and here).

In the previous comments I left on Mulligan's blog I gave several reasons why productivity could go up in a recession, but I'd like to go into one of those reasons in a little more detail:

Consider again 1) and 2) above. We have a measuring period for unemployment and productivity, say a quarter, or a year. Now, in that, say, year, suppose there is a decrease in demand for products and thus workers (so it's not that workers want to work less). There will then be two major effects: Effect 1; Immediately, the workers have less to do and so there's more sitting around and less productivity. But, Effect 2: Eventually, some of the workers are let go, and this increases productivity, partly because now there is less sitting around, with more to do per worker, partly because those let go will tend to be the least experienced and productive, and partly because there is now more capital per worker.

Now, if employers are very reluctant to lay off workers, because there is a strong culture of concern for workers and not just the bottom line – a culture like we used to have a lot more of in the 1930s and early 1980s – then for most of the year, or measuring period, we would see much more of effect 1 than effect 2. There'd be a lot more extraneous workers sitting around, and a lot less laying off. With effect 1 dominating, we'd see a productivity drop.

But now let's fast forward to the late 2000s, where there is much less of a culture of concern for workers and much more of a culture of concern for the bottom line. In this culture extraneous workers may not be allowed to sit around with little to do, producing little, for very long at all. They may be swiftly laid off, so that for only a very small portion of the measuring period the company has too many workers, and for the vast majority of the measuring period, it is running lean, with only the most experienced and productive workers remaining, and each having more capital to work with. In this case effect 1 would be small and effect 2 would be large, and productivity would go up.

By contrast, let's again go back in time to the 1930s or early 1980s, where there was much more of a culture of concern for workers and much less of a culture of concern for the bottom line, the extraneous workers may then be allowed to sit around with little to do, producing little, for a long time before managers reluctantly lay them off. They may not be swiftly laid off at all, so that for a very large portion of the measuring period, the company has too many workers. It may only start to get lean at the end of the measuring period, if at all, laying off enough workers so that effect 2 is stronger than effect 1 leaving the firm lean and with only the most experienced and productive workers remaining, each having more capital to work with. In this case – where firms are very reluctant to lay off workers -- effect 1 would be big and effect 2 would be small over the measuring period, and productivity would go down.

Thus, we see one mechanism (and there are others) which could lead to productivity going either way in the face of a decrease in the demand for goods and workers.

A decrease in the demand for goods and workers does not have to lead to a decrease in productivity as Mulligan intones.

1 comment:

Anonymous said...

"(um, what about all of the recessions besides the two Mulligan chose to mention?)"

You can check that yourself at http://highered.mcgraw-hill.com/sites/0073125679/student_view0/chapter17/graphing_exercise.html. But the short answer is: Mulligan's right--this is pretty unusual:
1907: down
1913: down
1916: down
1919: down
1929-33: way down
1938: down
1944-46: way down
1949: down
1953: flat
1958: flat
1960: up (slightly)
1970: down
1973-74: way down
1980: down
1982: down
1990: down
2000: up (slightly)

So that's 13 downs (3 dramatically so), 2 flats, a 1 up (slightly so) up until today.

Mulligan's point seems pretty spot on--a sharp up tick in productivity during a recession is unprecedented.

Perhaps you want more evidence and less cheekiness in future posts.