In early 2008, with the economy beginning to slow down, Larry Summers famously laid out three criteria for any fiscal stimulus: it should be “timely, targeted and temporary.” Targeted, meaning that the funds should flow to “those with low incomes and those whose incomes have recently fallen for whom spending is most urgent,” i.e. low-income and, especially, unemployed people. If these three conditions weren’t met, Summers warned, the stimulus coud “have worse side effects than the disease that is to be cured.”
A pair of studies performed by Garett Jones (who has been featured in the Spectator in the past) and Daniel Rothschild of the Mercatus Center provide new, concrete evidence that the spending in the 2009 stimulus bill was not targeted, and created few jobs. Jones and Rothschild simply surveyed and interviewed organizations that received stimulus funds, and found that fewer than half the workers hired by those companies were previously unemployed. Almost half of them were hired away from other companies. In other words, far from employing unused resources, the stimulus crowded out other areas of the labor market.
One reason these results reflect poorly on the stimulus is that the purpose of the bill was to help mitigate the harmful effects of mass unemployment in the short term. When President Obama introduced the American Recovery and Reinvestment Act, the Congressional Budget Office estimated, using models similar to the administration’s, that the bill would increase output and employment in the short term, but hurt the economy in the longer run. Specifically, the CBO projected that ARRA would slightly decrease GDP overall by the end of 2014, relative to the no-stimulus baseline. In other words, the administration decided it would be worthwhile to sacrifice a tiny bit of long-run growth to alleviate the near-term suffering of the unemployed.
Unfortunately, the Jones and Rothschild studies provide evidence that spending measures in the stimulus bill did not work in the way that the administration thought they would (other parts of ARRA might well have, such as the tax cuts). The basic Keynesian idea behind the administration’s thinking was that, when aggregate demand collapsed, productive workers and companies became needlessly unemployed. Government spending could replace private spending, and prevent them from remaining idle for a long time, thereby creating a free lunch — gainful employment for the workers, and increased aggregate demand for the whole economy.
The fact that organizations receiving stimulus funds hired more workers away from other firms than they did from unemployment suggests that the spending did little to put underutilized resources to work — as would happen in the administration’s Keynesian model. As Jones and Rothschild note, the proportion of hires that were previously unemployed roughly matches the proportion that obtains during normal economic times. Under Keynesian assumptions, that should not be the case: the presence of idle but productive workers during the recession should have made it easier for the companies receiving stimulus funds to hire directly from the ranks of the unemployed. As Jones and Rothschild write, “In other words, we find little evidence that stimulus spending was particularly effective at moving the unemployed into work. During the worst recession in generations, the ARRA-receiving organizations in our sample hired away employed workers at roughly the same rate as in normal economic times.”
The authors also anticipated one question about the high number of workers hired from other firms, namely that their jobs might have been filled by unemployed workers (The New Republic‘s Jonathan Chait, apparently without having read the paper, makes that exact criticism). Jones and Rothschild note that, in that case, the stimulus would fail Summers’s “timely” criterion, as firms cannot quickly find skilled workers — “the process of hiring good workers takes time, and that eats up the short time wherein free-lunch Keynesianism can work.” Furthermore, they write,
if, as is the case of the construction engineering firm discussed above, only a third of a company’s new hires come from unemployment, then it is quite a lot to hope that some other firm will actually hire the unemployed.
Moreover, job switching is not costless. When companies lose workers to stimulus-funded firms, they lose valuable skills and experience, what economists call -organizational capital.‖ So when a mid-level manager who understands the company’s database program switches jobs, or when an engineer with valuable contacts moves to the ARRA-funded engineering firm, the old firm is left weaker.
The surveys also revealed that, by a two-to-one margin, funds went to organizations that reported that “things had been busy” rather than “things had been slow” prior to receiving the stimulus money. This result provides further evidence that the stimulus did not put unemployed resources to work, but rather merely redirected private activity.
These two studies strongly suggest that the stimulus bill was poorly targeted, and that it did little to aid the unemployed or boost aggregate demand. Furthermore, they also indicate it’s impossible for stimulus spending to be targeted very effectively. The government does not have the diffuse knowledge needed to identify firms that would otherwise be needlessly idle and people who need jobs.