According to this (as yet unpublished, it appears) manuscript, the American Recovery and Reinvestment Act may have saved/created up to 450,000 government sector jobs at the expense of 1,000,000 private sector jobs. The best-case scenario, the authors claim, it that it saved/created a net of 659,000 jobs, mostly in the government sector.
Since the Act cost around $800 billion, at the good end that comes to about $12,000 per job saved, which would be a pretty decent value (although having most of them be government jobs is not encouraging). At the bad end we would have paid about $14,500 for each job destroyed. (Remember what I said in a recent post about the great disparity in calculating the multiplier on fiscal stimulus?)
The gist of the argument is that because much of the stimulus was funneled through the state governments they used the federal funds to replace lost tax revenue, so that state spending was kept from diminishing (or states having to raise taxes), but only remained constant, rather than increasing. Meanwhile, the stimulus seems to have crowded out private sector jobs, although the authors can’t yet explain why.
To be fair, Krugman has effectively made the “off-setting state spending reductions” argument–more speculatively and with less data, I think–and used it as the basis for his argument that the stimulus wasn’t large enough. But the private sector job-cost argument goes against his “of course federal spending stimulates” argument. There doesn’t seem to be any crowding-in of investment effect here, and unless I’ve missed it, Krugman doesn’t argue that crowding-in of investment only occurs after a certain crucial amount is spent.
Is this research more accurate than research finding a more positive effect? I have no idea. But for a policy that is supposedly beyond doubt, I think the uncertainty and discrepancies in the findings are rather troubling.