A Response to a Good Question from Russ Roberts

Yesterday Russ Roberts posed this important question:

"So my challenge to Paul Krugman and Brad DeLong and Matt Yglesias and Daniel Kuehn and others who didn’t like my claims about ideology and truth-seeking is this: what evidence could you imagine that would dissuade you from supporting massive government spending?"

Upon further consideration, the question is a little more ambiguous than I thought as to whether he's refering to a specific policy position or simply to a Keynesian orientation. I will be dissuaded from "supporting massive government spending" when there's no need for "massive government spending" (or when debt levels pose bigger problems). Sometimes Russ writes as if Keynesians think the economy needs government spending all the time (see his claims in the past and in this post about the post-WWII period). That's not my understanding of things at all, and I've never been taught anything even resembling that. So when interest rates start to increase, when inflation gets particularly substantial, when unemployment starts to decline, then I will yield on government spending as a policy position.

But based on the discussion in the rest of the post I think what Russ is really getting at is "what would dissuade you from supporting massive government spending as a potential policy response" - in other words, what would make you abandon your Keynesian disposition? I can think of a couple things:

1. No more plausible empirical studies showing multipliers substantially greater than one during periods when we expect them to be greater than one. As Russ notes, we have a range of multiplier estimates. I haven't had the pleasure of doing a detailed literature review (although we've discussed in detail many of the recent papers on this blog), but Russ says that they go from 0.5 to 2.0, and that seems right based on what I have seen. The point I've raised with Russ multiple times now is that the estimation strategies for most of these papers along this range are convincing to me. There are better studies and worse studies, but even the best ones show this sort of range. Russ interprets that as a problem and even goes as far as claiming "No anti-stimulus economist is convinced by a multiplier of 2 from a regression. No pro-stimulus economist believes the multipliers of less than one", which of course he knows is not true because I've noted many times on his blog and this blog my acceptance of low-multiplier estimates like Barro's (much to Andrew Bossie's chagrin). The question for me is "do we have an explanatory framework that provides a reason why we see low multipliers in some situations and higher multipliers in other situations?", and the answer is most emphaticaly "yes". Keynesian and New Keynesian models do a very good job explaining the range of values we see. The most important consideration for policy today, of course, is that multipliers are high when the economy is depressed and low when it is strong. Here are estimates from Auerbach and Gorodnichenko's SVAR models that illustrate the point beautifully:



It would seriously shake my acceptance of Keynesianism if we stopped seeing this pattern in the empirical literature of high multipliers during depressed periods and low multipliers during growth periods. I know Russ thinks that people's views aren't substantially shaped by these sorts of empirical findings, but in my case he's simply wrong. I don't have a head for high-level theory - try as I may. Maybe readers aren't fully aware of this, but the theory I peddle here is very basic stuff. As an undergrad, I excelled in the stats and econometric courses. In my early career at the Urban Institute my job was to figure out ways to empirically estimate program impacts. I'm an empirical guy. Will one study sway me? No. The subject of study here is too complex. But if the body of evidence shifts, that will sway me. The body of evidence on the multiplier right now strongly suggests higher multipliers in depressed periods and low multipliers outside of depressed periods. And the best theory I have at my disposal for explaining why that is a Keynesian theory of the economy. Which leads me to the second thing that would change my mind:


2. A better theory that explains what we see in the world. Ultimately what's attractive about Keynesianism is the same thing that's attractive about any scientific theory: it seems to explain observed phenomena better than any alternative. It helps me understand why inflation isn't sky-rocketing right now despite what's been happening with the money supply. It helps me understand the behavior of interest rates despite what's been happening with government deficits. It helps me understand why a lot of resources would simply go unused. It helps me understand the multiplier estimates that these studies keep producing. And ultimately it makes a lot of sense. Micro-founded versions of it and non-micro-founded versions of it are relatively consistent with each other, which is reassuring.

Other theories can do parts of that, and I feel strongly that other processes (including the mechanism underlying Austrian business cycle theory) are also operating in the economy. That's great. A grand unified theory is probably not a reasonable expectation for a complex phenomenon like the macroeconomy, so I am willing to entertain all of that. But until I am informed of a theoretical framework that can explain all of this better and can still incorporate other important insights, my own theoretical framework is going to be basically Keynesian. I'm not alone on this point. Many people who are against fiscal stimulus as a policy position (because of other reservations or perhaps political views) also consider themselves within the Keynesian paradigm.

*****

It's probably worth noting a couple things that won't change my mind.

1. Additional evidence that politicians are self-interested and that they act on those interests. I already know this. Nothing in my understanding of Keynesianism relies on this not being true. This is something I've always taken to be true, and something which I've always understood the science of economics to have accepted.

2. Failure of politicians to take Keynesian advice. If you thought that the government shouldn't do something, and they demonstrated a strong capacity to go ahead and do that thing anyway, would you consider that a reason to stop telling the government they shouldn't do it? I didn't think so. Let's please stop pretending this argument has any shread of logic or respectability to it.

3. Paul Samuelson and the post-WWII economy. I've been reading up on this in detail and am still forming my reaction to it, but I'm tiring of this argument too. People have explained to Russ Roberts and David Henderson many times why it's not surprising from a Keynesian perspective that the post-WWII economy would do relatively well. It's not a convoluted argument either - it's pretty straightforward. Russ writes that "they explained it away" as if there is something underhanded or insincere about what we've said. If he's going to react to counter-arguments that way there's not much I can do - he's clearly made up his mind on this. And it's also clear he thinks a valid test of Keynesianism is whether the economy can do well without government stimulus - which is an absurd test. Of course it can. Keynesian theory doesn't say it can't. Paul Samuelson was indeed worried about the post-war economy. You know who argued against these fears? (1.) John Maynard Keynes, (2.) Nicholas Kaldor, (3.) the CED (institute which promoted Keynesianism in business circles), (4.) The Brookings Institution (quite Keynesian at the time), (5.) Alvin Hansen (a.k.a. "The American Keynes"). Samuelson was swimming against the tide on this one. I think Abba Lerner disagreed with him too, although I'm still hunting that source down. William Beveridge is unclear on what he thinks in the book of his that I have and was reading last night, but he seems like he might disagree with Samuelson too. You know who agreed with Samuelson and who also expected a post-war depression? Friedrich Hayek. I don't blog at "Cafe Samuelson", but Russ still blogs at "Cafe Hayek" - yet for some reason we still have to deal with this Samuelson/WWII nonsense as apparently definitive of what we think every single time we have this discussion. Can we bury this please?