Stimulus Package Analysis: Which Type of Spending Created The Most Jobs?

Bruce Sacerdote and James Feyrer, both of Dartmouth, have produced a paper that looks at how the stimulus package (American Recovery and Reinvestment Act) affected employment, and which type of spending had the most (and least) amount of impact. It’s among the first detailed analyses of employment and earnings effects from the stimulus that uses actual employment outcomes. Here’s the abstract (full version here):

We use state and county level variation to examine the impact of the American Recovery and Reinvestment Act on employment. A cross state analysis suggests that one additional job was created by each $170,000 in stimulus spending. Time series analysis at the state level suggests a smaller response with a per job cost of about $400,000. These results imply Keynesian multipliers between 0.5 and 1.0, somewhat lower than those assumed by the administration. However, the overall results mask considerable variation for different types of spending. Grants to states for education do not appear to have created any additional jobs. Support programs for low income households and infrastructure spending are found to be highly expansionary. Estimates excluding education spending suggest fiscal policy multipliers of about 2.0 with per job cost of under $100,000.

The authors split federal spending into three types:

  • Agencies providing block grants to fund local government employment, including spending by the Departments of Education and Justice used to fund teachers and police;
  • Support to low-income families, including spending by the Departments of Agriculture, Health Education and Welfare, and Housing and Urban Development, a large component of which was food stamps, Medicaid, and rental assistance
  • Paying for new infrastructure projects, especially through building projects funded by the Departments of Transportation and Energy.

Their results show that certain types of spending are more effective at creating jobs than others.

Perhaps most intriguing is our analysis of how the impacts on employment appear to differ by type of spending. Transfers to the states to support education and law enforcement appear to have little effect. This is consistent with a model where the states consider the grants to be temporary and therefore avoid making permanent changes based on the transfer. States may have used the money to lower borrowing or limit tax increases. Cogan and Taylor (2010) find that this is the case.
On the other hand, support for low income households appear to have been extremely effective with Keynesian multipliers of over 2 and a cost per job of under $100,000. This is consistent with low income individuals having a high marginal propensity to consume. Infrastructure spending such as highway projects had impacts that were nearly as large. This all suggests that a stimulus package that did not include state level grants for local services would have been more effective per dollar than the actual stimulus package.

Though they conclude that the stimulus was generally effective, the authors point out the fundamental problem with evaluating its effectiveness: The lack of a counterfactual. We’ll never know what would’ve happened if we hadn’t had the stimulus. Which is why the “things would’ve been much worse…” story has been difficult for the Obama administration to sell, particularly with unemployment still above 9%.

caveat bettor

No null hypothesis, huh? The best dollar was the one that wasn't spent to stimulate ... no, MUST AVOID that possibility.


I am largely under the impression that "Infrastructure spending such as highway projects" are highly temporary jobs, am I incorrect in that understanding? The impression I have is that these guys get picked up to do task X (for a week or a month) and then get released back onto unemployment until the boss picks them up for another job.

If I'm wrong, someone please let me know (especially with factual evidence/data to back it up). If I'm right, how much "double-dipping" is going on in those "Infrastructure spending such as highway projects" stats?

Joshua Northey

I am pretty sure the ARRA reporting is just for a time period. So how many FTE did you create/preserve this quarter?

Joshua Northey

As someone whose expertise partially overlaps with this field, and who had to do some ARRA reporting myself I would say the following.

I would be highly suspicious of the reported data. Many of the reporters had little idea what they were doing and the system was rushed, opaque, and asinine.

Counting the jobs created by consumption betrays a ridiculously short time horizon for outcomes.

Surely you should care less about jobs created and more about if the money was spent well? Printing money to hire workers to dig and fill in holes will sure help employment in the short term, but it will also destroy the economy in the long run. You cannot mal-invest your way out of a problem.

Definitely giving the money to low income people will have a quicker immediate effect on the economy but if that effect is that 300 more X-boxes are purchased that seems like a catastrophe as far as government spending goes. On the other hand if you are preventing neighborhoods from collapsing and decaying into blighted criminal spawning grounds, well then it was money well spent. The devil is in the details.

I am not sure if the ARRA money was misspent, it is too complicated a question to tackle here, but the very methods of measuring economic activity bias economists towards the same wrong answers that led to this mess. Valuing totals exchanges and churn over actual production and utility.



One thing I don't see mentioned is whether a job that was not cut counts as a generated job. How many job cuts for local services were prevented by stimulus grants? I know that locally our government used a stimulus grant to avoid some of the layoffs in emergency services, law enforcement and education due to property tax revenue shortfalls.


Not that this will help, but try reading the paper before commenting. Pages 10 & 11 describe how they figured spending, jobs, etc. Try reading it. They break spending down to the county level. They identify $85B that was not block grants because they wanted to avoid double counting. They use real employment surveys.

Top of p.13 they break out how much money went into government employment - not much, btw, but most of it comes from government jobs. They also analyze starting on p.19 why the effect on education employment was minimal. It looks like other factors were overwhelmingly more important in those areas.

The way they derived spending makes sense. My state used stimulus money to reduce state university and college fee increases so they don't analyze that because it couldn't be allocated without double counting.

The limits of the analysis, beyond the narrow time period, is that it doesn't address most of the money. They note that $288B was tax cuts and much else was program support - like unemployment benefits. Much else, notably block grants to states, provided cushioning. The spending they can analyze shows Keynesian multipliers.

Read the bleeping paper. It's not long.



Please edit the prior post by me to say that most of the government job gain is from local government. I left out a word.


Studying stimulus without looking at the costs of financing is a common mistake of analytical work on the effects of stimulus. Academics tend to forget the implications of debt-based financing of stimulus and the interest costs involved, and are even more ignorant of the inflationary impact of financing through monetization, as well as the opportunity costs to taxpayers on their earnings that were transferred to the government for the purposes of stimulus. What a sad state of affairs.


Investing on infrastructure is how the US raised from the 1929 crash, it seems to work. Today I think that only a package of stimulus in different fronts would be effective. If you stimulate production, and on the other hand consumption it is more likely to be sustainable. But that would require a certain stimulus in infrastructure, because the demand for transport, housing and so forth will also increase.

Eric M. Jones

Joshua Northey has it exactly right. The calculation is perilous and the window is too short. A bridge collapse is so expensive to experience and so relatively cheap to avoid. Money spent on infrastructure is usually money well spent.

Then there's Boston's Big Dig that cost $16 BILLION, which is one-million dollars per foot--enough to build 2.5 Panama Canals today from scratch--all so that politically-well-connected real estate developers could buy cheap property overlooking the elevated highway and then sell luxury condominiums overlooking the "Rose Kennedy Greenway" that replaced it.

Calculate that.

Chris P.

Gotta love people. I hate that damn saying "it's better than doing *nothing.*"

John B

The problem with this kind of "stimulus" spending is that the money is now spent. And we need to have endless discussions trying to figure out where it went.

And when you really need it, the money won't be there. Because you already spent (wasted) it.

Jim S

Where is your proof that it was wasted? Is unemployment wasted when the private sector is not offering nearly enough jobs? Sorry, but equating the kind of spending that was done with "waste" is intellectually vacuous.


This is much ado about nothing.

The fact that anybody thinks stimulus spending works better than the alternatives is amazing to me. If we are doing science, then a hypothesis has to be disprovable though predictions. The Keynesian model predicts inflation should move directly with employment. During the late 1970's the economy experienced "stagflation," which is short for high inflation and high unemployment. Ergo, the Keynesian model is falsified because employment moved inversely to inflation.

Keynes, like Ptolemy, was a great scientist, but like Ptolemy, he was wrong. I wonder whether faith has replaced science.

It is not terribly controversial premise that increasing purchasing power raises prices. This is after all essential to the demand curve. When the government borrows to spend, it increases demand. This tends to shift prices higher, which makes it more difficult for businesses to purchase the goods and services they use to make a profit. This means that they will be less likely to employ new people, slowing economic growth. Ergo, stimulus spending creates both a positive and negative effect on the economy.

The first law of economic is that there is no free lunch.


Jim S

The first law of economics is that in the real world economic "laws" are never absolute but dependent on many variables, including psychological ones. When the economy is in the dumpster and trying to dig its way through the bottom of the dumpster, doing nothing should not be an option. The private sector just doesn't care. They are told over and over again that compassion is not their job. Management in publicly held companies has been told for decades that firing people is a good thing and Wall Street will reward you for it. That kind of habit is largely unbreakable.


Your last paragraph reeks of advocacy rather than journalism. You minimize the authors' conclusion by restating a premise upon which the authors argue their paper is necessary.

The authors say, "Even in the absence of a counterfactual, we conclude the stimulus was generally effective." You say "even though the authors conclude the stimulus was generally effective, we'll never know because there's no counterfactual." Shady, shady, shady.