Friday, April 27, 2012

Keynesians on NGDP-growth paths: Confusing arguments about policies and arguments about the goal

Bob Murphy recently suggested that a DeLong post expressing shock at Bernanke's policy decisions struck him as "B.S." because DeLong didn't endorse Scott Sumner's plan back in 2008. As I've noted here before, this kind of thinking bungles what Keynesians thought in 2008/09 (...10/11/12), and what the real differences are between Sumner and Keynesians. Scott Sumner's own spin on things has had an impact on the way people talk about this. Scott is an important voice in this crisis, but he also carefully manages the image of himself as a voice in the wilderness, and in a lot of ways I think that image is misleading. His position is unique (although getting less so every day) in his singular faith in the monetary authority's ability to successfully implement NGDP level targeting. He is not unique at all in (1.) thinking that expansionary monetary policy until we're back on trend is wise, and (2.) thinking that Bernanke of 2008-12 is not the Bernanke we used to know.

Let's get the easy part out of the way. I seem to remember Brad talking a lot about the gap between trend nominal GDP and actual nominal GDP in the early days of the crisis. I seem to remember lots of graphs of the gap on his blog (like this) that he would update on a regular basis and keep on a little banner at the top of the blog. I remember that it was Brad DeLong that challenged Ben Bernanke directly on the point of nominal growth in January 2010. I remember DeLong explicitly stating the problem in terms of nominal spending growth in 2009 when he contrasted the pragmatists (us) with the purists (the Fama gang), and I remember that from the beginning he explicitly pointed out that Fisher and Friedman and other quantity-of-money enthusiasts were on our side precisely because they recognized these points. Oh, and in March 2009 Bryan Caplan certainly seemed to be under the impression that Brad DeLong cared about NGDP trends.

So I think it's abundantly clear that NGDP was on Brad's radar from the beginning. It's also abundantly clear he had concerns about the Fed doing enough from the beginning. Which raises the question - why do people act like there's this gaping chasm between Keynesians and Sumner?

The answer is not that Keynesians don't care about NGDP, nor is it that they don't agree that the Fed should have a more expansionary stance. They are 100% on board with both of those things and as far as I can tell they have been from the beginning.
The real difference is that for Scott Sumner monetary policy is sufficient, and for Keynesians it's necessary, and there's a real chance it's not sufficient at a time like this. Sumner has a habit of strategically framing his position on this. He calls it "NGDP level targeting". Well what could be wrong with that? If the level is targeted then practically by definition we're doing OK, right? What Keynesians try to point out is that a level that is successfully targeted is a policy goal, not a policy. The policy is an expansionary monetary stance. The policy goal is targeting the level of NGDP. And the economic environment, agents' responses, etc. are what determine whether the policy is sufficient for achieving that policy goal.

Let's start with one objection by Keynesians (from Brad) that has been getting the most press: "The divide that worries me isn't sensible Democrat-sensible Republican. The divide that worries me is between economists who understand that the nominal money stock is not a sufficient statistic for the flow of nominal demand--especially when interest rates on short-term Treasuries are very low--and those who don't know enough to have figured that out, and who think that the nominal money stock is a sufficient statistic for the flow of nominal demand--even when interest rates on short-term Treasuries are very low."

This is the liquidity trap point. No, the interest rate mechanism is not everything. Yes, the interest rate mechanism is extremely important particularly when demand expectations are weaking MEC assessments anyway. Scott obviously scoffs at this. We say it's a good reason to not put all our eggs in one expansionary monetary stance basket in the effort to get total spending (again - note - we've gotta distinguish between the policy and the policy goal). That link up there is a good one because he also discusses other reasons to consider fiscal policy, including the expectations problem.

This is the Krugman point. Since the 1990s, Krugman has pointing to the possible new relevance of liquidity traps and has noted that a credible commitment to being irresponsible, monetary policy will still work in a liquidity trap. This is not new to the market monetarists, guys. It's all in the Japan paper. The problem is, a credible commitment to being irresponsible is hard for the monetary authority to do for the simple reason that all their pre-Fed work has been dedicated to convincing people that their policy responses are not going to be sustained indefinitely. There are Brad DeLong links above - if you want lots of links from me on why Krugman has always been on board with this (and why the distinction for Krugman too is really between policy and policy goals - two things that Sumner often confuses) see this post of mine.

So liquidity trap cuts out the interest rate mechanism. Anchored expetations and the Fed's track record (not to mention stubborn FOMC members) raise serious doubts about credible commitments to acting irresponsible can act on nominal spending through expectations.

That leaves fiscal policy.

And that's another difference between Keynesians and market monetarists. Sumner says that if the Fed is doing its job fiscal stimulus is pointless.

This is a reasonable statement to make.

But again, let's not confuse policies with policy goals. Has the Fed been doing its job? Has it succeeded in the expectations game that Krugman outlined many years ago? No.


Policies and policy goals are two different things. As far as I've been able to tell Brad DeLong has always been more or less on board with Scott's policy ideas. He has always been more or less on board with his policy goals. He has surpassed Scott in recognizing that those are two different things, and in recognizing that those are two different things he has been open to the idea of fiscal policy and not as sure about the efficacy of Scott's policy ideas (good ideas though they are).

So let's be careful when we talk about Keynesians and nominal spending.

Here's a good parting thought, also from an old post of Brad's:

"The government [he is referring to the Fed in this post] can buy more than Treasuries. After Treasuries, you buy GSE debt. And if that doesn't work, you buy bank and corporate debt. And if that doesn't work, you lend JPMorganChase $30 billion on the security of Jamie Dimon's dog. And if that doesn't work, you buy equities. And if that doesn't work, you buy the services of construction workers--by which time you are explicitly doing money printing-financed fiscal policy.

The thing that scares me is that I am not at all sure what or how much the Fed would have to buy. If you had asked me back at the start of 2008 how much the Fed would have to buy in order to keep nominal GDP on its pre-2008 growth trend, I would have said that it was almost certain that the Fed could do it by expanding its balance sheet from $1 trillion to $1.5 trillion. And if you had asked me in the middle of 2008, I would have said that it was almost certain that the Fed could do it by expanding its balance sheet to $2 trillion. And if you has asked me at the end of 209, I would have said that it was almost certain that the Fed could do it by expanding its balance sheet to $3 trillion. Yet here we are with a Fed with a $3 trillion balance sheet..."

Policies and policy goals are two very different things. An NGDP level target is a policy goal, not a policy.


  1. Setting a goal to solve a problem, and having a method to solve a problem are different things. Isn't having a method to prevent the problem also different ?

    Would monetary policies be more effective (sufficient?) in preventing a liquidity trap than in resolving a liquidity trap ?

  2. Just to clarify, Daniel, what I thought was "BS" was DeLong claiming that in early 2009, he thought the Fed was going to announce that it wanted to target NGDP. I think the post I quoted from him, back then, is as much of a smoking gun as one could ask for. To say interest rate cuts don't work at ZLB, and that all you're left with then is untested "quantitative easing" (his quotation marks to show he thinks it's newfangled), is about the furthest thing from a Sumnerian view imaginable. Indeed it is precisely the orthodox view that Sumner had to beat back over about 18 months.

    I will admit you have some links in this post that cast doubt on my thesis. I actually think the Caplan one helps me: Note that Caplan thinks it is a *good response* by him to say, "Well sure, if what we care about is NGDP, then maybe that would work..." So that post doesn't do what you think it does. You think you've got Bryan Caplan admitting, at the time, that DeLong cares about NGDP growth, when really what Caplan is doing is saying, "DeLong you obviously didn't make a good argument there, because your suggestion would just influence NGDP, and we economists all know that nobody cares about NGDP." I.e., that's exactly what I said economists besides Sumner and a few others were thinking, back then, including Fed officials.

    Your other link, where DeLong talks about the equation of exchange, is admittedly closer to what you need for your case. I still think I'm right, and that DeLong's discussion of Becker (the one I quoted in my post) is way more conclusive evidence than the one you dug up, but I can understand why you think I'm being too flippant with DeLong. It's not as open and shut as I thought when I first made the post, I will grant you.

    1. Bob -
      You would agree he cared about NGDP being off trend back then, right? I think I've shown he did. You would agree he advocated and thought the Fed would pursue large bond purchases to get NGDP back on track, right? I think I've shown that too (that is QE after all - which you agree here he was in support of). To me, that is the same thing as his recent assertion that:

      "as long as nominal GDP was below its pre-2008 trend, the Federal Reserve was going to buy bonds for cash–and keep buying bonds for cash until forecasts of nominal GDP were back on track."

      If you can see a difference between the two, you need to explain it to me because I don't understand it.

      It's true, Keynesians haven't always called it "NGDP level targeting". But that's why I highlight here that that's a policy goal, not a policy - and the policy they advocate (expanding the money supply when we're off NGDP trend) is the same policy Sumner advocates.

      The liquidity trap point is a genuine difference, as I've said on several occasions. But I've never known a Keynesian (maybe there are some out there - I can't think of any) to say that monetary policy should not be loose because of the liquidity trap. Indeed because of the expectations trap they say it should be very loose. The liquidity trap simply suggests its worth talking about fiscal policy now too.

  3. Speaking of monetary policy and reactions to expectations, it seems that monetary economists have a poor theoretical understanding of how to deal with uncertainty of agents. I've found one paper from a Google search that deals with monetary policy and decision-making under uncertainty, and I think economics could benefit from research coming from decision theory.

  4. Thanks for doing this. It's very hard for me to understand why Murphy seeks to perpetrate the misrepresentations that he does, on things that are so easy to check...


    Brad DeLong

    1. I sometimes think it's useful to collect material and arguments that are often left implicit.

      I also think a lot of this is Scott Sumner's "fault" (if you want to use the word "fault"). He is a master image-manager, and the image he seems to like is the voice-in-the-wilderness. And he's convinced a lot of people (including Bob) that he is such a voice. I found this interesting in Bob's comment above:

      "Just to clarify, Daniel, what I thought was "BS" was DeLong claiming that in early 2009, he thought the Fed was going to announce that it wanted to target NGDP."

      I couldn't find you claiming that anywhere in the statement that Bob posted. And this is why it's so critical to separate goals from policies. You clearly had an interest in returning nominal GDP to trend for years now. That has always been your goal which is all you said in the post. And to achieve your goal you've always seemed to say that the Fed should buy bonds, as you said in the post.

      So this is different from Scott - who combines his goal and his policy into NGDP targeting. You now support that, of course - but you never said you were promoting NGDP targeting as a policy back then. You only said you supported bond buying.

      Anyway, it is a subtle distinction, I think. But it is also something which - as you say - you've left a huge record on.

    2. The other frustrating thing about Scott, of course, is that he takes any hesitations (i.e. - any mention of either the liquidity trap or the expectations trap) as being an argument against doing monetary policy, when in actuality it is an argument for fiscal policy.


All anonymous comments will be deleted. Consistent pseudonyms are fine.