While I am no defender of fiscal stimulus — “Government Spending is Bad Economics” —, I call it like I see it: Anthony Davies’ recent empirical case against fiscal stimulus is not particularly persuasive. Here is the short video, taken from a lecture,
The data Davies uses is available in .pdf format. Rather than reproduce the graphs here, I’ll refer to pages within the .pdf.
In a nutshell, Davies argues (p. 2) that a +∆ in fiscal stimulus should have a positive impact on ∆ in GDP. This is, under the condition of ceteris paribus, fair and non-controversial. But, in the real world there is no such thing as “ceteris paribus.” Nobody disagrees with me, no doubt, but I feel Davies doesn’t fairly consider the implications of this truth.
Take a look at the graphs on pp. 4, 6, 8, 10, 12. I’m not a good statistician, or maybe even a good interpreter of scatter plots, but the lack of clear correlation on the first three graphs (of those listed in the previous sentence) is not particular damning. A fiscal stimulus advocate could just as easily argue that a lack of fiscal stimulus would have achieved a lower growth rate in %∆ in GDP. In fact, as displayed, there might be some empirical support for this kind of defense: countries with –%∆ in fiscal outlays have comparatively lower %%∆ in GDP than others. We don’t even know what countries each point plotted represents (although, I suppose we can look at the relevant BEA databases). The graphs on pp. 10 and 12 show a negative trend, but this could be caused by changes in economic fundamentals — the kind of fundamentals that take 1–2 years to impact the economy —, unrelated to fiscal stimulus. Finally, again, without knowing the countries it’s impossible to interpret the data.
To put the ambiguity of “faceless” data into perspective, suppose that that the point on the graph on p. 10 corresponding with, roughly, (1.25, –0.9) represents the fictional country of “Spreece.” On average, Spreece’s government practices relatively high levels of fiscal stimulus. But, inter alia, it also has restrictive labor laws, a dysfunctional financial sector (private credit freeze), and domestic violence issues (e.g. rioting) and thus a relatively high degree of regime uncertainty. It’s conceivable that the effects of fiscal stimulus can be outweighed by relevant factors outside the State’s direct control (Ilzetzki, et. al., ).
[Edit: Thanks to commenter “M.H.” I now realize that the points don’t represent countries, but different historical episodes in the United States. My general argument in the above two paragraphs still applies, however. Change “countries” for “episodes.” The impact of fiscal stimulus may be contingent on other factors.]
The bar graph on p. 18 is no less misleading. Which recession matters. Most were “solved” (smoothed over) by changes in monetary policy — viz. loose credit policies —, and most recessions between 1945 and 2007–08 have been “mild” (what some like to call “garden variety”). To put it straight, one could argue that the need for fiscal stimulus is contingent on the magnitude of the fluctuation, the effects of countercyclical monetary policy, and the status of the banking sector (whether it’s in a position to respond to Fed stimulus). But, I think this graph is better than the others.
Like I wrote above, I’m no fiscal stimulus (or monetary, for that matter) advocate. But, we should avoid weak arguments, because these are the ones that many well-known advocates focus on when they try to discredit their opponents. I think the best case against stimulus is theoretical, but few people tend to be persuaded by pure theory. If we must go to the evidence, then I think the best argument against fiscal stimulus is to look at past recessions with similar traits and make an exhaustive empirical case for what did push the recovery. A good example is the work, over the past two decades or so, on the impact of World War II on the U.S. economy. Hasn’t persuaded everyone, but I think the skeptic’s case against World War II as fiscal stimulus has gained a lot of ground.