No, CBO Didn’t Say That Stimulus Created 2.8 Million Jobs

CBO doesn’t do that. They just measure dollars spent. We know we’re out $800 billion. What we got for it is anyone’s guess.

FILED UNDER: Economics and Business, Quick Takes,
Dodd Harris
About Dodd Harris
Dodd, who used to run a blog named ipse dixit, is an attorney, a veteran of the United States Navy, and a fairly good poker player. He contributed over 650 pieces to OTB between May 2007 and September 2013. Follow him on Twitter @Amuk3.


  1. Andy says:

    As opposed to traveling to a stimulus-free parallel universe and measuring the effects that way? Of course they start with money spent, use the models they have available to translate the extra demand into job creation, and report the results. It doesn’t change the fact that that’s the best estimate that a non-partisan accountability agency can come up with.

    You do the same thing when you say the stimulus has “no effect”, albeit with a much shallower (zero) jobs-per-dollar curve. Unless you have a parallel universe machine you’re not telling us about.

  2. Dave Schuler says:

    Andy, Mr. Elmendorf also said they hadn’t recalibrated their models using the current data. I doubt that’s a formula for accurate prediction.

  3. Andy says:


    You’re right about that, and since we’re living in exceptional times it’s doubly true. But the CBO put the job creation figure at 1.2 to 2.8 million… with goalposts that wide I think it’s likely that the “real” number is somewhere in that range, don’t you?

    Imagine if actual job creation was something like 300,000. It’s hard to believe the CBO would be off by a factor of 10 with their top estimate. Especially because a force of economics both bizarre enough to be unaccounted for in the previous models and dramatic enough to have such an overwhelming effect would probably have been noticed, and widely remarked upon, by many people smarter than myself.

  4. Dodd says:

    Even if the upper bound was a reliable figure (it’s not), 800 billion for 2.8 million jobs comes to $285,714.29 per job ‘saved or created.’ I think my mildly hyperbolic “no effect” is a generous description given that we could have simply given 1/6 that amount to 16.8 million people and they’d all have had more than a year’s worth of about the median income.

    And that’s ignoring the Harvard study linked in my other Quick Take today about how gov’t spending crowds out private money.

  5. Andy says:

    The problem with just writing checks- illustrated by Bush’s $600 tax rebates- is that people won’t often spend the money in stimulative ways. Paying down credit card debts or padding the savings account doesn’t really help the overall economy the way buying a new television does. But it’s what the individual is inclined to do in uncertain times.

    Thus, to get the most job-creation bang for its buck, the gov’t is well-advised to spend the money itself, and taxpayers get roads, buses, etc. instead of rebate checks. (Not that we aren’t getting checks: unemployment benefits were extended again, which is a wise way to target the money at people who are likely to actually spend it. But a comprehensive stimulus shouldn’t be entirely money transfers.)

    Furthermore I’m not entirely certain that $300K/job is unreasonable. I know my boss considers my total overhead, including health insurance and all, to be about double my salary. So, for the portion of the stimulus comprised of demand-side outlays, all the first figure shows is that the gov’t is buying products for which about 25% of the cost is labor (assuming median salaries). Which doesn’t sound entirely unreasonable, depending on what they’re buying.