Did the Stimulus Create Jobs? Not Always for the Unemployed

Yesterday, Tyler Cowen linked to a new report from Mercatus on the employment impact of the stimulus.   Dan Rothschild and Garrett Jones interviewed and surveyed companies that had accepted stimulus funds, and asked them about how those funds had impacted their hiring.  The result: only a minority (though a large minority) of the people hired were hired from unemployment.

But wait!  Jonathan Chait says that this assumes that labor isn't fungible:

Let's ponder this. Suppose Company A creates a new job, and fills it by hiring a worker away from Company B. What do the authors suppose think happens to the job at Company B? Do they say, well, that's it, we lost Joyce, nothing we can do about that in an economy with unemployment at only 9%.

I read the paper in the vain hope of finding the authors' explanation of what they think happens when a new job is filled by moving a worker from another job. I did not see one. Nor did I see any attempt to demonstrate, or even suggest, that the newly-opened jobs of workers moving into stimulus-created jobs were going unfilled. They genuinely seem to assume that "job shifting" is simply the opposite of job creation. (They also seem to think that hiring half of new workers from the ranks of the unemployed, when some 90% of the potential workforce has jobs, is a wildly low figure.)

Mmm . . . sort of.  Chait is refuting a claim that I don't think either Jones/Rothschild, or Cowen, are making: that the result of the stimulus spending was a 0% increase in employment.  What they're talking about is the relative effectiveness of employment.  And contra Chait, the distinction between hiring from the unemployed, vs. poaching from competitors, is only irrelevant if you assume that there is no labor hoarding.

Labor hoarding is just what it sounds like: employers keep on workers who aren't really worth what they're being paid, because it's traumatic to fire them.  I don't just mean traumatic for the boss, although it is, to an extent that employees tend to underestimate.  But it also hurts morale among the employees who remain.  (This is why employers tend to do a big layoff all at once rather than trickle them out--it minimizes the blow to morale.)  This matters, both because it can hurt productivity, and because firms that do layoffs often see their best, most productive employees leave soon thereafter.  There are also things like unemployment, severance, and the possibility of a discrimination suit to consider. In union shops, there's seniority to worry about as well--if you have to lay off by "Last hired, first fired", you may lose your most productive employee.

Now, I think the mild consensus among economists is that labor hoarding has fallen since the seventies.  But I don't think anyone believes that it has fallen to zero.  In fact, I have empirical proof of this, in the form of Jonathan Chait's employer, which I'm pretty sure has dealt with its financial woes in part by not replacing the people who depart for greener pastures.  (As have many other publications).  If you gave a liberal publication $100,000, and they used that money to hire a writer away from the New Republic, there seems to be a very good chance that this would have zero impact on the labor market for journalists.

Note that Chait is only right that labor is essentially infinitely fungible if the economy is at full capacity--really booming, so that no one's labor is being hoarded.  This is obviously not the case now.

In the construction industry, there's another wrinkle; many of the specialties in heavy construction are, at least as I understand it, not overfull with qualified applicants; finding young people who have the math skills and other academic talents necessary to be a modern skilled construction worker, and also want to skip college and apprentice with an outfit like the operating engineers, is something that a lot of the skilled trades worry about.

I think a lot of people assumed that doing infrastructure construction projects would be a great way to soak up excess labor from the homebuilding industry, but there's not actually that much overlap; knowing how to install drywall or do framing work does not qualify you for a job that requires sandhogs and specialty welders.  And it can take a long time to make journeyman in many of these professions.  This is also true of certain kinds of civil engineers and so forth.

Because the supply of these workers is fairly inelastic, how many jobs you create for those kinds of specialties mostly depends on how high the unemployment rate in that profession is.  Yet Jones and Rothschild say that the employers they surveyed were finding it no easier to get qualified people than they had before the crisis.  This suggests, though certainly does not prove, that at least for some categories of workers, the hiring was neutral, not additive.

What about knock-on spending?  Even if poaching from company A to Company B does not create a job, one presumes it comes with a higher salary; perhaps that spending creates jobs in less skilled areas.  But we should be careful about this.  One of the very problems we're trying to fight is that when people get their hands on money, they stick it in the bank and don't spend it, while the banks don't want to lend it; though this is great for balance sheet repair of both households and banks, as far as stimulus is concerned, that money has just died.  So the multiplier might be pretty weak.  (Might not, of course--good empirical estimates are not really available.)

So while I certainly don't think we have evidence that the stimulus did nothing for employment at all, I do think there's reason to worry that it didn't do nearly as much as we'd want, or as advocates promised.  The purpose of the stimulus was not to raise salaries for construction managers.  It was to keep people from suffering the hell of long-term unemployment.