(Hint: It depends on what you mean by “work”).
The most recent article being bandied about in the economics blogosphere is this Washington Post op-ed by E.J. Dionne. In his article, Dionne argues both that there is a normative role for government to play in maintaining employment during crises (and, indeed, a place in the overall operation of the economy during non-crisis times) and that the evidence overwhelmingly shows that the US government’s stimulus worked. Dionne claims that during the 2008 Financial Crisis, ARRA—“The American Recovery and Reinvestment Act” (often referred to colloquially as “the stimulus,” although it is simply one part of a larger government aid package)—helped to lower the unemployment rate. For evidence he provides the following:
“It was thus salutary that Douglas Elmendorf, the widely respected director of the Congressional Budget Office, told a congressional hearing last week that 80 percent of economic experts surveyed by the University of Chicago’s Booth School of Business agreed that the stimulus got the unemployment rate lower at the end of 2010 than it would have been otherwise. Only 4 percent disagreed. The stimulus, CBO concluded, added as many as 3.3 million jobs during the second quarter of 2010, and it may have kept us from lapsing back into recession.”
Russ Roberts responds with a number of points. I wrote in to Russ to offer my thoughts. I like the Booth School IGM forum and check it regularly. For those who are unfamiliar with the IGM Forum it is a survey of roughly 40 leading economists in the United States. Every couple of weeks the panel is asked a particular question about an economic issue, many of them policy oriented. The question about ARRA actually had two parts:
Question A: Because of the American Recovery and Reinvestment Act of 2009, the U.S. unemployment rate was lower at the end of 2010 than it would have been without the stimulus bill.
Question B: Taking into account all of the ARRA’s economic consequences — including the economic costs of raising taxes to pay for the spending, its effects on future spending, and any other likely future effects — the benefits of the stimulus will end up exceeding its costs.
Dionne’s op-ed, however, only considered Question A. I have three main issues with his argument. The first is that we need to be cognizant that economics is now very specialized and as such every economist cannot possible be well-versed in every subspecialty of the field. Richard Thaler (a behavioral economist) is undoubtedly brilliant, but I’m not sure how much he knows about fiscal policy effects. I’m sure this can be said of many of the respondents (a casual glance at their credentials reveled there are several environmental economists on the panel, for instance).
Second, and even more jarring, is a glance at the comments section of Question A. Here are a few comments from those who actually agreed (or strongly agreed) with the premis in Question A (that ARRA lowered the unemployment rate):
“Subsidizing employment leads employment to go up, other things equal. Adverse impacts through growth incentives might take time.”
“All reasonable models have this implication, but there’s enough the models are missing…”
“But this is an incredibly low bar. ”
“Caveat: how much was it offset by less agressive (than otherwise) unconventional monetary policy?”
“At the very minimum the transfer to the states prevented them from firing state employees. But I do not know how big the overall effect was.”
Lastly, if you consider Question B—that is, what is the long run net effect on the economy—the percentage of economists plumets from 80% who agreed that ARRA lowered unemployment to only 47% who believe ARRA will end up having net benefits over the longterm. Veronique de Rugy makes many of my same points (a boost to my ego), but in a much more elegant and cogent way (a blow to my ego).
Russ responded that he felt the conversation should be limited to Question A since that is the focus of Dionne’s article. Fair enough, but I think Russ is being too generous. Though Dionne focused on unemployment his real argument was that government has an ongoing role in the economy and that ARRA was simply a sign of its success in that role. I don’t mind the argument on its face, maybe the government should have a large role in helping to steer the economy, reasonable people can disagree. But in conflating government’s effect on unemployment during the crisis and government’s broader role in general economic management, one might infer from his column—as I did—that Dionne’s argument was about the broader evidence that government intervention has been successful. Even if we limit that stament to, “government intervention during the financial crisis,” Dionne’s own evidence is mixed. That is easily seen both by the respondents’ comments to Question A and by their answer to Question B about the long-term net effects of government stimulus, which Dionne conveniently ignored.
Russ sums it up nicely after pointing out the large number of confounding factors whose effects are impossible to measure independently:
“No one has a model of the independent impact of these different factors or a way of measuring them accurately and reliably in a way that can be tested and confirmed or rejected. No one. That means everyone, on the left or the right, who claims to have evidence for the impact of one of them or who cherry-picks one of those out of the myriad to choose from and blames that one factor for the lousy pace of the recovery is either fooling himself or fooling you. Don’t be a fool. So when the E.J. Dionnes of the world tell you that government creates jobs, just ask them how they know. Their answer will be that someone with exemplary credentials says so. But there are those with exemplary credentials who say otherwise. Where does that leave us? It should leave us in ignorance and doubt. No certainty. No exclamation points. More humility.”