How to Get a Bigger Keynesian Multiplier

Garett Jones -- Macroeconomist at George Mason University.  You can follow him on Twitter: @GarettJones

In a normal economy, 40% of new hires are people who switched right over from another job: Friday at Walmart, Monday at Target. The other 60% come from some mix of the unemployment lines and the ether.

In this normal economy, a slowly-growing pool of workers play a high-speed game of musical chairs with a slowly-growing number of jobs: Workers are constantly switching in and out of employment, sometimes voluntarily, sometimes not. Yes, workers who switch jobs often get replaced, eventually, but those new hires are rarely net hires: This is the world of churn, John Haltiwanger's world of endless job creation and destruction.

When Rothschild and I sent out surveys to organizations that received ARRA stimulus funding --the first and unfortunately still only study of its kind -- we asked newly-hired workers a simple question: What were you doing right before this job? 47% said they came directly from another job -- perhaps slightly higher than the amount of "job shifting" or "poaching" in a normal economy. In a recession, you should do even better than that, since there are so many unemployed workers to choose from.

Why does that matter? Because it's one sign -- and not the only sign -- that some of the stimulus was targeted at sectors of the economy where it was hard to find good workers on short notice, and that means a smaller multiplier effect of government spending. In Brad DeLong's apt expression, it was one sign that the stimulus had run into some bottlenecks.

Of course "job shifting" or "poaching" by itself doesn't axiomatically prove that the multiplier is small -- but I'm rarely one for axiomatic proofs. But ask yourself: What would happen if 100% of the stimulus went toward, say, hiring anesthesiologists to work for the Military Health System for two years? It's not like we can make new anesthesiologists instantly: the government would mostly poach them away from other hospitals. Hospitals would have to get by with whoever had a little anesthesia training back in med school, they'd have to pay some kind of overtime to the few remaining anesthesiologists, and a few would come out of retirement or speed through med school. There'd be a small multiplier at best.

The likely results of anesthesiologist stimulus:

  1. Hospitals would complain that "it's harder to find good people these days." 
  2. Wages would rise for anesthesiologists: A demand boost for a scarce good. 
  3. And almost all of the new anesthesiology hires, private or public, would be poached from somewhere else. 
This is an extreme hypothetical, a parable. But in our study, we found evidence that all 3 occurred to some degree at ARRA-funded organizations, and and it happened more for some jobs than for others.

Examples: Among people with grad degrees who answered our survey, 63% were poached, but among those with high school degrees, only 38% were: Skilled workers were harder to replace. Raises were bigger for the poached than for the previously unemployed -- a sign of DeLongian bottlenecks. And while half of organizations said it was easier to find good people than before the Great Recession, 12% said it was harder (health clinics? solar energy engineers?) and the rest said there was no change--that latter group might be living in Robert Hall's normal economy.

I hope that pro-fiscal-stimulus economists draw some lessons from our work: If you want a big Keynesian multiplier:

  1. Focus on projects that hire less-skilled workers (remembering that "skilled" includes pipefitters, electricians and other specialties that mercifully don't yet require a college degree). 
  2. Focus on projects that use workers from sectors with temporarily high unemployment rates.
  3. Make some kind of effort to target parts of the country with high unemployment rates; this time around they didn't quite succeedPublic choice (or was it the tendency to spend ARRA dollars in college towns?) for the win. 
One way to get all 3 would be to follow Martin Shubik's plan: An independent agency that plans for ways to spend money next time there's a recession. I'm not crazy about the idea --- I prefer
automatic stabilizers plus Taylor-rule monetary policy myself, since I see strong benefits to predictable rules -- but it'd be better than the hodgepodge Congress offered up in January 2009.

If the Shubik plan suspended Davis-Bacon prevailing wage laws for the duration of an economic emergency, and if it were mostly a way of saying "let's do more of our government construction projects during recessions, not during boom-times," it might even get bipartisan support. I said "might."

Let me close with a plea for better data: Dan and I shouldn't have had to be the ones collecting micro-level survey and interview data on whether and when the stimulus did (and didn't!) run into DeLongian bottlenecks. The federal government, with its extensive ARRA reporting requirements, could have easily collected that data. But they didn't: And when Elaine Povich of the Fiscal Times asked GSA and how many of their workers were poached, she received shoulder shrugs. The wonks should insist on better data collection next time around.