Thursday, November 14, 2013

John Cochrane and Keynesian Economics

John Cochrane has been writing stuff recently which is very helpful if you want to understand the difference between Old and New Keynesian economics, and the mysteries of the mechanics of New Keynesian models at the zero lower bound. One example is this working paper, in which he argues that the various "paradoxical" results associated with New Keynesian liquidity traps are artifacts of equilibrium selection, and that the paradoxes go away if we focus our attention on an equilibrium that New Keynesians typically ignore. If we want to take these models seriously, we should at least feel confident about what the models actually have to say.

Another related issue Cochrane raises is in this blog post. The basic point that Cochrane wants to make is that an Old-Keynesian model (i.e. IS/LM AS/AD) and a New Keynesian model are very different animals. Indeed, New Keynesian models would be much more recognizable to Ed Prescott than to John Maynard Keynes, if we chose to resurrect the latter. One way in which these models differ is in how fiscal policy works. Old Keynesians typically assumed that the economy was non-Ricardian, which helped to feed the multiplier process. However, in a New Keynesian model, there is a complete set of contingent claims markets, and Ricardian equivalence holds. Fiscal stimulus works by increasing anticipated inflation, and lowering real interest rates, causing consumers to substitute intertemporally. Current consumption must go up, as it is assumed that consumption reverts to trend in the future.

Cochrane is willing to take the New Keynesian story seriously, but he would like it if the people who are promoting fiscal stimulus were to defend their policy recommendations in terms of that model. Unfortunately the fiscal stimulus folks are either confused, or are deliberately obfuscating. Some people want to argue as if New Keynesian models and Old Keynesian models are indeed the same animal. Policy issues are explained in Old Keynesian language - e.g. stimulus works by way of the multiplier process - and then people appeal to New Keynesian models as if that somehow ratifies these old results, which is false.

Now, predictably, this kind of talk gets Paul Krugman upset, as of course what Cochrane is describing is the way Krugman operates. What gets Krugman particularly upset is this (from Cochrane):
But people love the story. Policy makers love the story. Most of Washington loves the story. Most of Washington policy analysis uses Keynesian models or Keynesian thinking. This is really curious. Our whole policy establishment uses a model that cannot be published in a peer-reviewed journal. Imagine if the climate scientists were telling us to spend a trillion dollars on carbon dioxide mitigation -- but they had not been able to publish any of their models in peer-reviewed journals for 35 years.
The problem here is the the story that people "love" is the Old Keynesian multiplier/paradox-of-thrift story. Somehow the average person finds it appealing, and so the story gets used all the time, in spite of the fact that it does not hold water - in the sense of being grounded in sound economic theory.

Krugman's reply to this is to argue that it's not Washington (or him for that matter) that is screwed up - it's the economics journals.
So consider two hypotheses. One — which Cochrane appears to believe — is that being inside the Beltway has rotted Janet’s and Olivier’s brains, not to mention that of all their researchers, causing them to revert to primitive concepts that “everyone” knows are false. The other — which is what I hear from young economists — is that there is an equilibrium business cycle claque in academic macroeconomics that has in effect blockaded the journals to anyone trying to publish models and evidence that stress the demand side.

Obviously you know which story I believe. The main point, though, is that trying to argue from authority is even sillier here than in most situations. There’s a huge difference between “nobody believes that” and “none of my friends will let that get published in the journals they control”.
Of course, most of those two paragraphs is nonsense. Janet Yellen and Oliver Blanchard are well out in the right tail of the quality distribution of Washington policymakers. To my taste, both are a little too wedded to Old Keynesian economics, but they listen, and I don't think those are the people that Cochrane is concerned with.

On other matters, what is an "equilibrium business cycle claque" anyway? Who exactly would those people be? First, if Krugman were to say the word "disequilibrium" to me, I would know what he meant, but I would have to correct him by pointing out that all the macro models I have seen in my life are equilibrium models. Old Keynesian, New Keynesian, stochastic growth, whatever - all of those models have some equilibrium concept. For example, you can fix the prices and let the quantities adjust - it's still an equilibrium model. I think what Krugman wants to say is that there is some "equilibrium business cycle claque" which is a collection of powerful people who control some important institutions, and reserve the right to filter out people and research they don't like. Here's a story. Long ago there was a group of outsiders, who either could not find jobs on the inside - in Cambridge Massachusetts, for example - or were literally kicked out of the Ivy League. Those people didn't spend their time whining about how badly they were treated, as far as I know. They went to work at out-of-the-way institutions like Carnegie Mellon University and the University of Minnesota, and ultimately remade macroeconomics - for the better, I think. They managed to get their work published, they helped start journals and professional societies, and built research institutions from the ground up.

Krugman's academic life was nothing like that. He was a Cambridge insider from the get-go, and anointed at an early age - not much struggle there I think. If these "young economists" who can't publish their "demand side" work even exist, the papers they can't publish must be real stinkers, as I don't see severe limits to the market for such stuff. The NBER - particularly the Fluctuations group and the "Monetary Economics" group - is very friendly to Keynesian economics. The Society for Economic Dynamics, started by what Krugman might think of as an "equilibrium claque," typically has plenty of New Keynesian economics on the program at its annual meeting. The QJE, where Eggertsson and Krugman just published their paper, is of course notable as an outlet for Keynesian work. The Journal of Monetary Economics, has been edited for years by Bob King, who certainly has Keynesian sympathies. That journal will soon be run by Ricardo Reis and Urban Jermann. I think it's fair to say that the former is a Keynesian. The editor of AEJ Macro (a newer and prominent macro journal) is John Leahy - a prominent Keynesian. In the group of co-editors at the AER, the only macroeonomist appears to be Marty Eichenbaum, who as we all know is now a hardcore Keynesian.

What could Krugman be thinking? We've found better ways to do macroeconomics than twiddling IS-LM models. Get over it.


  1. Steve,
    nice post. But, as I and other commentators pointed out to Cochrane, once you add liquidity constraints to New-Keynesian models the environment becomes non-Ricardian, and an increase in government outlays does affect consumption in a manner similar to that in Old Keynesian models. Cochrane's response was that he was talking about the simple new-Keynesian model. But it is not clear why people in Washington should have the strip down version in mind, and not the more complex version for which the old Keynesian model seems a reasonable approximation. In fact, that is the main point, I think, of Krugman-Eggertsson, that once you add bells and whistles, the New-Keynesian model behaves similarly to the old Keynesian one. This is not to say that economists should be using the IS-LM model when we conduct research, or even when we teach our students. But for the purpose of communicating policy to people who may not have taken a single economics class in their lives, I am not sure that using the outdated but simpler model is wrong IF the mechanism is similar to that in the more complex state-of-the-art model that the communicator has in mind (and I am sure this is not the simplest New-Keynesian model out there).

    Having said that, the fact that Krugman-Eggertsson was published in what is considered one of the top journals in economics shows that no bias exists against Keynesian ideas. If I was refereeing it, I would probably have rejected it as I have issues with several of its assumptions and predictions, which the authors do not bother to substantiate.

    1. So, we could add liquidity constraints to the basic NK model, and then argue about whether that makes any sense, fits features of the data, etc. Whether the results somehow "look like" what you get out of an IS/LM model seems irrelevant. I don't have to use IS/LM language to sell the ideas to someone who doesn't know much economics. If the ideas in my NK liquidity constraint model are good ones, I should be able to articulate that in a non-technical way. I think the reason people want to fall back on IS-LM language is that some of the people they are talking to have a smattering of macroeconomics, and that's IS-LM. The question then is: why are people still teaching that?

    2. Hmm, I see your point. Of course, I think that not everyone is as capable in articulating the ideas in more complex models, particularly dynamic ones, but yes, we should at least try.

  2. I think Nick Rowe has been doing an even better job than Cochrane. He might self-describe as math-deficient, but the quantity of his blogging imparts a quality of his own, and his greater engagement with commenters helps clarify things.

    1. I agree with you on two things: (i) the quantity of Nick's blogging; (ii) the deficiency.

    2. "the quantity of his blogging imparts a quality of his own."

      I have absolutely no idea what this word salad means. It indicates to me a person incapable of coherent thought.


  3. Cochrane is a whore of the financial industry, in particular a certain hedge funds, so it is obvious that he is not doing proper economics. Naturally somebody with vested interest in capital income could not care less about demand-sided policies which reduce unemployment and lead to gains for labour.

    Always interesting to see to skim over this blog and realize how many intelligent economists are right-wing scum that doesn't care one iota about 99% of the population while being able to pretend that it is all just about technical stuff and being able to completely ignore empirics. Europe has just done a giant test on the efficiency of fiscal policy during liquidity trap recessions and the results are clear.

    1. Sorry, I don't really care about what Cochrane does in his spare time, or how he votes. In the stuff I'm citing, he's trying to make some sense of economic theory and policy advice, with some success I think.

      "...right-wing scum that doesn't care one iota about 99% of the population..."

      That should be "...right-wing scum who do not care one iota about 99% of the population..."

    2. "somebody with vested interest in capital income could not care less about demand-sided policies which reduce unemployment and lead to gains for labour"

      That's right, the stock and real estate markets have not benefited from the Fed's demand-sided policies (QE, etc.)...if you live in Mars. And, clearly, the return to capital does not rise with employment...if you never took an econ course in your life.

      But I suppose this is the benefit of engaging in ad hominem attacks, you can be completely ignorant of the issue at hand and still get away with it; well, almost.

    3. Right, the Koch brothers care about high employment during recessions as an increase of employment increases the marginal efficiency of capital and thus their incomes.
      Getting a decent education in classical economics is one thing, being unable to use your brain afterwards is another.

  4. That was fun, but why not take this one on, too:

    Should be worth a laugh or two.

    1. I don't know whether to laugh or cry, Dave. In some places, undergrad macro looks exactly the same as when we took it. Actually, though, I think what I learned was a little more advanced than that. Sometime in the 1970s I heard mention of the investment accelerator, but not by anyone who took it seriously.

    2. It seems the vast majority (not some places, most places) of undergraduate macro theory courses include IS/LM/AS/AD. It features in many of the bestselling textbooks including those written by Mankiw, Bernanke, Blanchard and many others. Your arrogance ('laugh or cry') is thus quite astounding. Many of these courses also mention Ricardian equivalence and intertemporal optimization to introduce students to different modeling philosophies.

  5. Ouch!

    If you want to increase the natural rate of interest, to escape the ZLB, the NK model says you should *reduce* Gdot.

    If you add some hand-to-mouth agents to the NK model, it now says you should reduce Gdot and/or *increase* Tdot.

    (If you spend a minute thinking about the consumption-Euler equation you will understand why that is true. You need to increase Cdot for the Ricardian agents to increase the natural rate of interest, and you increase their Cdot by cutting Gdot and/or increasing Tdot.)

    Those policy recommendations are very different from ISLM.

    1. Nick,

      If you think that the zero lower bound is the problem, and your model is NK, there is a straightforward fiscal solution to the problem, which is an appropriate tax policy. See:

      I'm not sure what all this fuss about G is about.

    2. Steve: The New Keynesian model is even more different from the Old Keynesian ISLM model than John Cochrane thinks it is.

      It's not just a difference in the mechanism by which fiscal policy works. They make very different recommendations for fiscal policy.

      To escape the ZLB:

      The ISLM model says "increase G".

      The New Keynesian model says "reduce Gdot".

      That's what the fuss about G is about.

      (NK economists have missed this difference because they usually work in discrete time, where a temporary increase in G(t) implies a decrease in Gdot(t) going forward.)

    3. I still think that's an odd way to think about fixing a problem of distorted prices.