Tuesday, October 5, 2010

Fringe Economics

[Warning: Krugman content]
What do Post-Keynesian Economics, Austrian Economics, Institutional Economics, New Institutional Economics, and Marxist Economics have in common? People working in these areas sometimes call themselves Heterodox economists, and they work on the fringe of the profession. On rare occasions I have dipped into what these people do to find out what it is about. Typically, what I find is a grumpy bunch who dislike mathematics, and whose aims are mainly destructive - they want to discredit what mainstream economists are doing. To the extent that they have interesting and constructive ideas, you can usually find examples of mainstream economists who are tackling the same issues and doing a better job.

What makes a fringe economist? Apparently some are just born that way. My colleague Doug North, who lives in the office next to mine, is like this. Doug may indeed be the most successful fringe economist in the world (though "success" and "fringe" seem to be contradictory), and he has a Nobel prize to prove it. He has a disdain for mathematics, and seemed pretty suspicious of the invasion of his department by various macroeconomists and theorists, beginning six years ago. Other people grow into fringe economics, maybe out of necessity. One learns how to do economics in a particular way, a technological innovation in research methods comes along that destroys your human capital, and you are faced with two alternatives. Either you bear the costs of adopting the new technology, or you form your own support group and fight the adoption of the technology by others. Old-time Luddites and Saboteurs would understand the second approach well.

Mathematics is of course here to stay in economics. Samuelson's Foundations of Economic Analysis was an early handbook for the application of mathematics to economic problems, and once Arrow, Debreu, and McKenzie (for you Rochester loyalists) had made their contributions, there was no turning back. The fringe, working outside the mainstream, has had issues with essentially all mainstream fields in economics, but internal conflict within fields has also lead to fringe movements of disenfranchised types. Internal squabbling has been far more prevalent among macroeconomists than among microeconomists (though I'm sure people can give me plenty of examples of groups of economists who hate each other). Macroeconomists had a hard time absorbing the new methodology that was first introduced in the early 1970s that included dynamic optimization tools from mathematics, and various elements of received micreconomic theory (general equilibrium theory and game theory primarily). One Luddite was James Tobin who, after making some nice contributions to our understanding of asset pricing and financial intermediation, chose to take the destructive approach with regard to the ideas of Lucas and friends.

Now, this brings us to you-know-who. Krugman wants to discuss two "fringe groups." The first is the Austrians, who of course are genuinely fringe. Who are the other fringies? Krugman says:
More interesting, in a way, are the economics professors who are totally convinced that people like Brad DeLong and yours truly are ignorant, unwashed types who don’t know anything about modern macroeconomics.
I have no evidence on DeLong - no idea how washed the man is. With Krugman, however, we have good evidence of cleanliness from here, where we get this:
“Twenty-five cities in forty days,” he says. “The mechanics of washing up in hotel sinks because you’re not in any hotel long enough to use their laundry.”

It’s not so much the washing as the drying that presents a problem. Years of experiments have failed to yield a satisfactory solution. Krugman has discovered that it is slow and quite risky to use a hair dryer with any item that involves elastic. Long ago, in Tel Aviv, his roommates found him attempting to dry his underwear in a frying pan.

“The trick with underwear is to wring it out and then press down—”

Anyone who would rather dry his underwear in a frying pan than wear soiled undergarments can't be all bad.

Now, how ignorant is Krugman? The question is: "relative to what?" Surely he knows more than a ferret, or an albatross (my wife taught The Rime of the Ancient Mariner yesterday, and we had a discussion about albatrosses over dinner, so I had this on my mind). On economic issues, he would certainly give our graduate students a run for their money, but in terms of modeling skills, I think our first-year graduate students would win hands-down. Krugman is apparently quite proud of this 1998 paper. This is something our first-year graduate students could do in 20 minutes on an exam. Nothing to be ashamed of, of course. Krugman wasn't shooting for publication with this little ditty.

Does Krugman know more about macroeconomics than, say, Bob Lucas? I don't think so. Bob, at 73 years old, is still writing papers, giving academic seminars, and going to academic conferences. He is very much up on what is going on in the profession, he's receptive to new ideas, and he encourages younger members of the profession. Bob seems quite uninterested in earning a big income. He just loves economics. Krugman checked out of academic activities long ago, opting for highly-paid speaking engagements and NYT column-writing. But that's fine. We don't really expect him to be at the cutting edge of academic thought.

The strange thing is that Krugman wants to characterize the people who think of him as less-than-up-to-speed on macroeconomic thought as "fringe." I haven't taken a poll, but my informal reading of opinion among New Keynesians, New Monetarists, Prescott followers, or whatever, who are also active researchers (have published a paper in a respectable economics journal in the last 3 years), is that Krugman's "fringe sect" is actually the majority of mainstream macroeconomists.

Krugman goes on to say this:
For the record, I tend to think things through in terms of New Keynesian models, as in my old Japan paper, but often translate the results into IS-LM for simplicity. If that’s a crude, primitive approach, somebody should tell Mike Woodford.
What Krugman does not understand is that Woodford's framework is not IS-LM. One of Woodford's brilliant marketing triumphs was to dress up New Keynesianism in the language of Old Keynesianism. There is an IS curve and a Phillips curve, and you drop the LM curve in favor of a Taylor rule. However, the underlying model is basically a Prescott real business cycle model with monopolistic competition. The sticky prices give you relative price distortions that make the model amenable to doing standard welfare economics. It's really not IS-LM at all, but if people like Krugman think it is, that's helpful for Woodford.

Krugman goes on to discuss how some people, including John Cochrane and Narayana Kocherlakota, are hopelessly confused about basic economics. Then he finishes with this:
What’s going on here? I believe that what we’re looking at is people who know their math, but don’t know what it means: they can grind through the equations of their models, but don’t have any feel for what the equations really imply. Confronted with informal discussion that’s grounded in models but not explicitly stated in terms of math, they’re totally baffled. And so they lash out.

Sad, really.
Now it's Krugman that is sounding like a grumpy fringie. The message here is a Luddite message. Beware of economists who use mathematics. They have spent too much time with their math books and haven't a clue about economics. The rhetoric is typical. You are supposed to picture sad (and calm) Krugman in his arm chair, musing over his frustrated critics who are busy frothing at the mouth in their hopeless confusion.

There is nothing new about this, of course. We have lived with Luddites for a long time in the economics profession, and they are not going away. Coming from a Nobel-Prize-winner with a large audience, the message can of course be quite damaging. University administrators read the New York Times, and so do undergraduate students. Lazy scientists don't need encouragement. Students love it when they are told they really don't need to know anything technical. However, belief in that falsehood will not do them, or society, any good.

46 comments:

  1. I read that Krugman post and have been anxiously awaiting your ripost since. Thank you. Also, would you be willing to recommend a sort of "Must-do Math Classes for Macroeconomists" (especially for classes beyond Rudin Analysis) type of list for those of us currently being groomed for the failure that Krugman was describing?

    ReplyDelete
  2. I think the math thing is a little bit of a distraction. There are other fringes - guys who do agent-based modeling, econophysicists, the networks people etc. And then, there are the physicists, statisticians and mathematicians working as quants in finance who sometimes provide similar criticisms. There are also lots of microeconomists who worry about macro, though they may not voice it.

    That said, I very much agree math has a very important role in macroeconomics. But what math? Or, is it just about how seriously one takes the model - or maybe, just that we need to think about the trade-off between time spent learning high technique and institutional detail? I'm not sure.

    ReplyDelete
  3. I prefer the old Krugman:
    http://web.mit.edu/krugman/www/ricardo.htm

    How about just one Krugman post per week? Consolidated refutation of Krugman posts is ok.

    ReplyDelete
  4. Hamed,

    In depends in part what your math department is like. Beyond real analysis, students sometimes find topology useful, or stochastic processes. You have to be careful about how you use your time though. The problem is that typically a course taught in the math department will have only a small fraction of content that will be of immediate use in economics (of course, you never know when you can use some tool). Sometimes you can get a lot more out of an economics graduate course where you have an economist going through Lucas, Stokey, and Prescott (Recursive Methods ...) in detail. As with anything, you can get too much. You don't want to spend all your time learning math, or you lose contact with the economics. This is what Krugman claims to be worried about, but I think people like Kocherlakota actually have a good grip on the issues as well as the math.

    First anonymous: Yes, good point. This is another kind of grief that we have to deal with.

    Wonks: Yes, or maybe biweekly. I might even consider gradual Krugman withdrawal.

    ReplyDelete
  5. you literally have no idea what heterodox economics is. first i would say many heterodox economists are right to be suspicious of mainstream economics use of math (most statisticians, physicists,mathematicians etc are horrifies when they take a real deep look into mainstream macro). second, even taking that point there is plenty of heterodox thinkers doing great statistical work and dynamic modeling (steve keen, marc lavoie, wynne godley RIP, william mitchell, anwar shaikh, lance taylor etc). In fact, in my opinion the greatest criticism of mainstream economics is it's criticism of equilibrium and the best counter to mainstream formal modeling is stock flow consistent dynamic modeling. given all this i will give you that some heterodox economists are math phobic but in the same way that you wouldn't want me to judge mainstream economics based on paul krugman i would prefer you actually knew what you were talking about about heterodox economics as a whole before you talked about it. i would say that you rarely find mainstream economists debunking mainstream economics (because then it wouldn't be mainstream economics at all). The idea that heterodox economics isn't worthwhile because it debunks mainstream economics's
    absurdity is one of the stupidest arguments I've heard a mainstream economist make (and that's saying something).

    ReplyDelete
  6. http://www.businessspectator.com.au/bs.nsf/article/was-mathematics-at-fault-for-the-gfc-pd20091130-y9uku?opendocument&src=blb&is=financial%20markets&blog=keensian%20economics

    ReplyDelete
  7. I can't tell who the audience is for these Krugman posts. Economists can't find anything for them as the posts are quite free of content. Interested laypeople would be looking for evidence that your position has better predictive ability than his, but you never attempt to make that case. Certainly they would have no reason to be impressed with your assertion that his math is less sophisticated than yours unless you can show that your work provides better results. This is especially so when you can't even provide evidence of lack of mathematical sophistication on his part. No, the quote you provided does not support your position; it only shows he considers economics to consist of more than math. So it's not clear who you're trying to convince.

    ReplyDelete
  8. Nathan and last anonymous:

    I rest my case.

    ReplyDelete
  9. yes, when i think my point is obvious and doesn't need defending, all i do is assume that my detractors understand the fallacies in their arguments and "rest my case". Ironically, this seems to be the mainstreams response in theory too. "if i assume that your criticism is invalid, it is invalid."

    ReplyDelete
  10. It's true that Krugman's piece is simple mathematically. But that's no measure of its contribution. Solow's model is very simple, yet I assume you would concede he deserves his Nobel. For that matter, special relativity uses only high school math, yet a genius was needed to come up with it. Nothing is more facile than to say after something is invented, "Oh, that's simple. One of my students could have done it." But they didn't, and almost certainly wouldn't have. The only real measure of a piece of research is its influence, and Krugman's 1998 piece on deflation has been extraordinarily influential. He is simply much, much better than most at coming up with ideas that influence other researchers.

    ReplyDelete
  11. I ve been reading you for a while and I'd like to point something out about the relation between Krugman and you/the mainstream.

    It was not very long ago (less than one decade) that someone said something like:

    "Paul Krugman has been doing a lot of very effective writing attacking non-economists writing about economic matters. Paul is speaking for the whole profession in a very effective way and addressing the most important questions in social science"

    Guess whom? It's part of the interview to Bob Lucas in Snowdon&Vane's book on the state of macro.

    If you have followed krugman's work, both as a scholar and as a public pundit, you can see that he has been saying essentially the same things since he started attacking the supply siders and the strategic traders in the nineties.

    So, what has changed? I think it is mainly a matter of presentation, of emphasis. I really think you make a caricature of Krugman's -as I read his post, he is just saying that he finds it useful to translate results into an ISLM framework, just as Blanchard, Mankiw or Bernanke do in their intermediate books (do you think they are wrong as well?). On the other hand I think that Krugman has completely forgotten that his main asset was giving balanced view -he has become a troll.

    What I am trying to say is that both you and krugman are largely overemphasizing your differences.

    ReplyDelete
  12. "Nothing is more facile than to say after something is invented, "Oh, that's simple. One of my students could have done it.""

    Yes, I agree that simplicity is a virtue. Of course, Krugman did not invent the cash-in-advance model in his liquidity trap paper. Indeed, the model is in the toolbox of most any graduate macro student. I'm not exaggerating when I call it a homework problem. The point is that that Krugman likes to use this 12-year-old piece of work as an example of how up on modern macro he is.

    citoyen:

    Lucas is a generous guy who is entitled to his opinions. I don't always agree with him. One example is when he gets into quantity-theory mode.

    "he is just saying that he finds it useful to translate results into an ISLM framework, just as Blanchard, Mankiw or Bernanke do in their intermediate books (do you think they are wrong as well?)."

    IS-LM is ubiquitous in undergrad macro books. Indeed, you'll find it in edtions 1 through 3 of my book. In the fourth edition I do a much simpler kind of New Keynesian Woodford sticky price model that is much simpler and an easier extension of the other things I do. I think it is important for students to know this, if only because the ideas are influential and it is important to understand how everyone in the profession is thinking. I think it is impossible to successfully address current policy problems with an IS/LM model, which is part of Krugman's problem. In the old days, it was impossible in central banks to talk to management without using IS/LM AD/AS language. That was what those people knew, and you could not avoid it. There were some exceptions, though. For example, Al Broaddus, former Richmond Fed President, was receptive to learning about modern macro, and seemed quite capable of carrying on policy discussions in that context. These days, the management people know a lot more, and IS/LM essentially never enters the discussion, at least among people I know. They think in terms of models and ideas that have little to do with IS/LM. Indeed, it is essentially impossible to distill current policy ideas in a useful way and articulate them within some IS/LM framework. Krugman claims he's worked it all out in the background, but I don't believe him.

    ReplyDelete
  13. I cannot understand how you could make Doug North and all the (new) institutional economics belong to this fringe stuff. What is your criteria ? Mathematical knowledge and sophistication ? I think you'll agree if I say that this is not a very convincing one when what is at stakes is to evaluate what is the contribution of someone's work on how the economy works.

    North, (Oliver) Williamson, Thomas Schelling, Elinor Ostrom and so on... are all Nobel prized and their work uncontroversially illuminate our understanding of economic and institutional phenomena. The same is true of Daron Acemoglu's work, though it is heavily mathematized. Maths don't matter here. However, by contrast, I don't see clearly what modern macroeconomics has teach us during the last 20 or 30 years, things we don't already know with hydraulic-macro-style à la IS/LM or AS/AD. I'm not a macroeconomist, so I will really appreciate if someone could illuminates me on this matter...

    ReplyDelete
  14. Anonymous @ 7:54PM

    I feel sorry for you and your need to hide bravely behind your cloak of anonymity when making such remarks. Please spare us your amateur psychology.

    ReplyDelete
  15. "I don't see clearly what modern macroeconomics has teach us during the last 20 or 30 years, things we don't already know with hydraulic-macro-style à la IS/LM or AS/AD. I'm not a macroeconomist, so I will really appreciate if someone could illuminates me on this matter..."

    I guess you don't know what you don't know. Start by reading my intermediate macro textbook. The publisher charges an exorbitant price for it, but you can of course get it used, or borrow it:

    http://www.pearsonhighered.com/educator/product/Macroeconomics/9780131368736.page

    Another thing you could do is to compare journal articles in macro in 1970 to what you see today. You would not know it is the same subject. The strange thing is that textbooks (particularly at the intro level) and some of the public discussion (e.g. Krugman and a lot of what you see in the blogosphere) is back in 1970.

    ReplyDelete
  16. "The strange thing is that textbooks (particularly at the intro level) and some of the public discussion (e.g. Krugman and a lot of what you see in the blogosphere) is back in 1970"

    Your right and that's a disturbing fact, at least regarding public discussion (for the textbook matter, David Colander has argued for some time now that textbooks are typically well behind the state of arts - this is true in macro as well as in micro). How can we explain that the current public discussion of the financial crises and the economic recession is quasi-entirely held in the AS/AD framework? Where are the modern macroeconomists with their DSGE stuff (well not you but others)? I leave people like Cochrane or Fama since their recent claims seem (at least to me) to rely on completely unreliable assumptions (total Ricardian equivalence or total crowding out effect)...

    ReplyDelete
  17. Keynes has some great quotes. I was looking for the one where he discusses the long time it takes for adoption of new ideas, but this one is good:

    “The difficulty lies not so much in developing new ideas as in escaping from old ones.”

    Yes, some strange ideas persist in undergrad micro too. Does anyone know how to make sense out of short-run and long-run cost curves? The publishers are partly responsible for this. My publisher is pretty good in this respect, but sometimes they want to dictate content, and are afraid the book won't sell if it deviates too much from the existing middle-of-the-road.

    "Where are the modern macroeconomists with their DSGE stuff (well not you but others)?"

    In central banks, the level of the discussion has gone up a lot in recent years. However, most academics are busy earning a living. For most people, chattering in public about policy does not pay the bills.

    ReplyDelete
  18. First of all, thanks a lot for the answer.

    Let me clarify my thought. I started studying macroecoonomics in an IS-LM context (with Mankiw's book) and then, I learned to specify the specific functions (investment, money and consumption demands) in a more concise way with dynamic, stochastic micro-foundations, just as the progress from Blanchard undergraduate book to Romer's or Hejdra's would tell you.

    I can see that the approach is different when you use DSGE's and I believe you when you say that most people in the profession do not think in terms of IS-LM. Still, what is really hard for me to see is to what extent the new framework is incompatible with the old one. Most -I would say all, but maybe I am wrong- of the results can be translated in a IS-LM-AS language. I understand that this exercise might be seen as superficial for scholars, but I don't get:

    1)Why do you seem to see those exercises as illegitimate, misleading or old fashioned when, I think, they actually help -or at least some people find it helpful- to present ideas in a unified and simple framework which is more accessible than the intertemporal optimization stuff.
    2) You deny that they actually help to capture the intuition behind some complex ideas. For example, it was for me a very long way since I first got the idea that market economies suffer from lack of effective demand as shown in a simple keynesian cross framework, I was thereafter expose to Lucas supply functions, neutrality theorems and the benchamark RBC model and finally went back to the initial point when I read about search models and coordination externalities (Diamond), credit market imperfections (credit rationing and financial accelerator) and sunspot equilibria (Farmer etc,...).

    I mean, I don't think I am alone in this idea. Blanchard, which I hardly think can be qualified as a fringe economist, has written (those articles about the state of macro and what do we know that wicksell didnt know) that he thinks that simple old fashioned models are quite understimated.

    Finally, I'd like to ask you a question. In a set of article (for example that horrible one about the state of macro) Krugman has been pushing the idea that macroeconomics is quite divided, in terms of communities reading and writing for each other. Despite the fact I disagree with him, I see some apparent plausibility in that idea. Many of the statements about what people thought actually was the state of macro are extremely divergent; they seem to be living in different worlds. I am not talking of the blogosphere, I can try to provide examples if you wish. I mean, what I can observe is a world of people suggesting that they are the orthodoxy and the others are fringe economists.

    ReplyDelete
  19. Stephen Williamson writes:

    "One Luddite was James Tobin who, after making some nice contributions to our understanding of asset pricing and financial intermediation, chose to take the destructive approach with regard to the ideas of Lucas and friends"

    Well let me point to one paper that James Tobin published in 1975:

    "Keynesian Models of Recession and Depression," American Economic Review, American Economic Association, vol. 65(2), pages 195-202, May. 1975.

    Here he actually follows "Lucas and friends" setting up a macroeconomic model with an expectations-augmented Philips curve (something, I would think, "Lucas and friends" would agree with)

    In this paper he reaches the conclusion that Keynes tried to set up in chapter 19 of "The General Theory". Keynes asked whether flexible prices and money wages would move an economy out of recession.
    Tobin showed that the "Keynes effect" could be swamped by either a Mundell-Tobin effect or a debt deflation effect (mentioned by Fisher and Keynes). In conclusion he showed that more flexible prices and money wages could(would) actually lead to the an even more prolonged recession/depression.
    His model's basis is actually very "Lucas and friends". But he reaches a conclusion that, I believe, "Lucas and friends" would never arrive at.
    And this model actually sticked with him through the years since he emphasizes the same conclusion in 1992:

    "An Old Keynesian Counterattacks" Eastern Economic Journal, Eastern Economic Association, vol. 18(4), pages 387-400, Fall. 1992

    With regards your disagreement with the IS-LM model, I believe, you are "preaching to the pastor". As far as I can recall, the late Sir John Hicks, the originator of the model actually apologized later in his life for the model's lack of coherence with Keynes' real view of time and uncertainty as well as with the fact that stocks and flows weren't correctly "taken care of":

    "IS-LM: An explanation", Journal of Post Keynesian Economics, Vol. 3 (2), p.139-54. 1980.

    With respect to the New Keynesian approach with it's use of microfoundations in the form of (a) representative agent, I must admit, I thought the "Sonnenschein-Mantel- Debreu theorem" had proven its problems. That one cannot generally aggregate (different) individuals into one representative agent whose choice/action would result in the same choice as the collective.


    But since I'm not that clever in the ways of modern macro, I may be mistaken.

    ReplyDelete
  20. Citoyen:

    "1)Why do you seem to see those exercises as illegitimate, misleading or old fashioned when, I think, they actually help -or at least some people find it helpful- to present ideas in a unified and simple framework which is more accessible than the intertemporal optimization stuff."

    It's not difficult to translate "intertemporal optimization stuff" into simple language. Good ideas are simple ones, and it is just as easy to do that as to describe what is going on in an IS-LM model. Indeed, people who were educated in the 1970s and who later became microeconomists might tell you they found their undergraduate (and sometimes graduate) macro education impenetrable or nonsensical, but modern microfounded economics makes sense to them. Macroeconomists can now communicate with other economists - they use the same language.

    "they seem to be living in different worlds."

    Part of what Krugman's New York Times Magazine piece revealed was that he knew little about the state of current macro - what the schools of thought are, who is talking to who, etc. There are plenty of disagreements, but those are not typically about basic methodology. For example, Mike Woodford is a lot closer to Ed Prescott than he is to John Maynard Keynes.

    Reese2101:

    The James Tobin I was talking about was the guy who tried to keep the Yale economics department from being tainted by modern macro (which set them back for years), and who wrote "Asset Accumulation and Economic Activity."

    "That one cannot generally aggregate (different) individuals into one representative agent whose choice/action would result in the same choice as the collective."

    Yes, aggregation requires some very special assumptions. But all models are wrong; that's what makes them models. The fact that aggregation does not work does not mean that you can't learn something from a representative agent model. Of course there will be many issues such a model cannot say anything at all about. That is why much of modern macro research is conducted in heterogeneous-agent models. Woodford has some of those too.

    ReplyDelete
  21. Prof W: if your contention in this post is that heterodox economists are anti-math, you are mistaken. For instance both Duncan Foley and Steve Keen are very much anti-neoclassic, but both appear to be competent mathematically. For example, see Foley's paper on social security in an overlapping generations model with capitalists and workers.

    pe

    ReplyDelete
  22. Stephen,

    Mathematical models help clarify our thinking, but they should not be confused with the truth. You denigrate Krugman's modeling skills, but I give him credit for being able to state the basic assumptions of the models he employs in simple terms that add to understanding without resorting to mathematical mystification. I challenge you to do the same.
    Economics is a behavioral science with a few accounting identities thrown in. If you can't explain your model in words, don't expect that I'm going to be impressed that your grad students can solve differential equations. Any science worth its salt spends most of its time trying to find the shortcomings of present models when compared to observation. How does your model - whatever it may be - compare with present trends and how does it work when considering the shock of the recent financial crisis. Please tell us.

    ReplyDelete
  23. "If you can't explain your model in words, don't expect that I'm going to be impressed that your grad students can solve differential equations."

    Yes, I agree completely. If you like what Krugman does, that's fine too. But Krugman has another agenda, which is trying to convince people that modern macro is nonsense. You can see why I might have a problem with that.

    "How does your model - whatever it may be - compare with present trends and how does it work when considering the shock of the recent financial crisis. Please tell us."

    Read this and see if you like it.

    http://www.artsci.wustl.edu/~swilliam/papers/intermed09-10.pdf

    It has equations and words too.

    ReplyDelete
  24. Steve: "However, the underlying model is basically a Prescott real business cycle model with monopolistic competition. The sticky prices give you relative price distortions that make the model amenable to doing standard welfare economics."

    Sorry, but you are totally wrong here. It's certainly not an RBC model with monopolistic competition. The fact that both have an Euler equation doesn't make them the same. Bad monetary policy can cause aggregate fluctuations that are inefficient in the New Keynesian model. And this inefficiency has nothing to do with the relative price distortions that arise from staggered price setting, which are a secondary inefficiency; you get the same recession when all prices are set at the same time, so there are no relative price distortions.

    I'm trying to figure out why you think it's right, when it's so clearly (to me) wrong.

    And the best theory I can come up with is...it's the math wot dunnit! You see an Euler equation in both, and figure they must be basically the same (monop comp and relative price distortions aside).

    And I blame the New Keynesians too, for not making money explicit in the model, so it's not clear that the model only makes sense in a monetary exchange economy, where barter is too costly to do.

    ReplyDelete
  25. Nick,

    "Sorry, but you are totally wrong here. It's certainly not an RBC model with monopolistic competition."

    Have you read "Interest and Prices?"

    ReplyDelete
  26. Yes. Not thoroughly. Are you saying Woodford says it *is* an RBC model? Even more fun. I get to argue with you *and* Woodford! It's lovely when people don't understand their own models!

    Don't try to bulldoze me in a literature review pissing contest though. I'm trying to check that you really believe that, and it wasn't just an off-the-cuff comment you hadn't really given much thought to. Or maybe you meant something quite different than I thought you did. If you want me to explain more why I think you are wrong I will do so. I would like you to explain more why you think you are right. Because I really don't understand why you would believe that about NK models.

    ReplyDelete
  27. Nick:

    Look at Chapter 3, section 1, page 143, where he lays out "A Basic Sticky Price Model." Or you can read this early paper by Rotemberg and Woodford:

    http://ideas.repec.org/p/nbr/nberte/0233.html

    In Cooley's 1995 book Frontiers in Business Cycle Research, which summarizes the RBC literature (which is pretty much done by that time), there is a paper by Rotemberg and Woodford on imperfect competition in dynamic models. Shortly thereafter, Rotemberg, Woodford, and others are putting sticky prices into those models and linearizing to come up with 3-equation New Keynesian models ("IS curve," "Phillps curve," Taylor rule). By the time Clarida, Gali, and Gertler write their survey, they just write down the 3 equations, state that it is derived from an underlying optimizing model, and reference the derivations in other papers. The Genesis is clear - the thing evolves directly out of the RBC literature.

    ReplyDelete
  28. Steve: if you are just talking about the historical *origins* of the so-called New Keynesian IS curve, that's not something I would argue about. (Though if you simply replace the old Keynesian consumption function with Friedman's Permanent Income Hypothesis you get much the same thing). It sounded like you were saying something much more controversial.

    Because if you take an RBC model, then add sticky nominal prices, the result (usually) is a model that is very much not an RBC model. I read you as contradicting that.

    But what makes it a non-RBC model has very little to do with "The sticky prices give you relative price distortions that make the model amenable to doing standard welfare economics." It's not the *relative* price distortions that make it a Keynesian model. (Just have bunched rather than staggered price setting and you get rid of those distortions). It's the stickiness of the *general level of nominal prices* that distorts aggregate output (unless monetary policy is exactly right).

    ReplyDelete
  29. "Because if you take an RBC model, then add sticky nominal prices, the result (usually) is a model that is very much not an RBC model."

    A cow with a hat is still a cow. Typically, optimal monetary policy in a New Keynesian model reproduces the RBC equilibrium allocation - the right policy gives you the cow with no hat.

    "It's the stickiness of the *general level of nominal prices* that distorts aggregate output (unless monetary policy is exactly right)."

    Exactly wrong. You need to read "Interest and Prices." In Woodford's cashless economy with Calvo pricing, inflation is costly because of relative price distortions. Monetary policy can fix the problem by giving price stability (so long as the zero lower bound on the nominal interest rate is not violated). In the cashless model, the price level is irrelevant - only relative prices matter.

    ReplyDelete
  30. Steve: Ah! We do disagree (on some of this).

    It's not news to say that the job of monetary policy is to try (if possible) to keep the economy at the natural rate, which means (roughly) where it would be if prices were perfectly flexible. This idea pre-dates RBC and New Keynesianism. Old Keynesians would have described it slightly differently: something like using monetary policy to get a Keynesian economy to behave like a "Classical" (flexible price) one.

    But that doesn't make the economy Classical. It makes it an economy where good monetary policy matters for real output.

    If that's just a cow with a hat, I say it's some big hat. And it sure affects how the cow behaves.

    "In Woodford's cashless economy with Calvo pricing, inflation is costly because of relative price distortions. Monetary policy can fix the problem by giving price stability (so long as the zero lower bound on the nominal interest rate is not violated)."

    Agreed. And that result generally carries over into an economy with cash too, except for the shoe-leather cost of people economising on cash.

    "In the cashless model, the price level is irrelevant - only relative prices matter."

    Now the price-path *level* is irrelevant in a cashless economy. Precisely because it has no nominal anchor, like a fixed stock of cash. And the price-path level would be indeterminate, were it not for sticky prices and the central bank following some sort of Taylor Principle rule to make it determinate.

    But there are more ways for monetary policy to be bad than just choosing a non-zero target rate of inflation. Suppose the natural rate of interest falls, and the central bank does not observe it, and so does not drop the nominal rate, and so sets the nominal interest rate too high. The result is a recession. That recession is costly not just because relative prices are distorted, and so relative output across firms is distorted. It is costly because aggregate output falls.

    Or suppose the central bank suddenly decides to target a lower price level. The nominal price level cannot immediately fall to the new target. The nominal price level is too high relative to target. Again, the resulting recession is costly over and above any relative output distortions caused by relative price distortions due to non-synchronised price setting.

    And those results are just like the Old Keynesian ISLM model's.

    ReplyDelete
  31. In your first example, the central bank gets it wrong, but it comes down to a problem of the relative prices being wrong. In the second case, they make an error and - guess what - they get the relative prices wrong. It's always the relative prices. You can like IS/LM or you can like Woodford. Woodford deliberately led people to believe he was doing IS/LM, but he's not.

    ReplyDelete
  32. What's "wrong" in both examples is the real interest rate. Now, it's true that the real interest rate is a relative price -- the relative price of present vs future consumption -- but that's not the same as the relative price of one flavour of ice cream to another flavour of ice cream (which is the relative price I thought you were talking about).

    And again, it shares that feature with the ISLM. Assume the central bank sets a nominal interest rate, with the money supply perfectly elastic, and you get a horizontal LM curve. And if it sets too high a rate, relative to the natural rate (which is where the IS curve cuts "full employment", and the real interest rate is too high, you get a recession.

    Funny thing is, I actually agree with you that the Euler-equation "IS" is more different from the traditional IS than most New Keynesians think it is. But my reasons are very different from yours.

    ReplyDelete
  33. Nick,

    You could price any assets you want in that framework. When you are away from the optimum, all those asset prices will be wrong. There's nothing special about the real rate. At the base of it though is that the relative prices of goods are out of whack. I'm not sure what we get out of thinking that some model we use is "like" an IS/LM model. I sometimes find partial equilibrium supply/demand stories useful for giving students intuition. Robert Hall, for example, is very good at this. However, if we all woke up tomorrow having forgotten that IS/LM existed, I don't think we would have lost anything.

    ReplyDelete
  34. We really do disagree.

    Let me try the intuition this way:

    Let's drop the (silly) Calvo pricing assumption, and replace it with the (equally silly) assumption that firms have quadratic costs of adjusting their nominal prices. Let there be n firms each producing one flavour of ice cream. Each firm is run by a worker-owner, who gets utility from leisure, and who is also a consumer, with Dixit-Stiglitz preferences, so there's a taste for variety of flavours. Perfect symmetry, so all firms have identical production costs and demand curves, and price adjustment costs. So all firms always set the same price, whether or not the price adjustment costs are binding. Standard intertemporal preferences, so we get the standard consumption-Euler equation.

    A central bank sets a nominal interest rate, as a function of whatever.

    This model is identical to the standard Neo-Wicksellian/Woodfordian/New-Keynesian model, except: I've sawn off the labour market for simplicity; and I've replaced Calvo price adjustment with quadratic costs.

    So I have rigged the model so that relative prices are always equal to one, and are never wrong.

    Can this model have a recession, where output and employment falls below the natural rate, because of bad monetary policy, even though all relative prices are always "right"?

    I take it you would answer "no" (except for the labour wasted in adjusting prices)?

    I would answer "yes". Suppose the central bank sets the nominal interest rate too high for one period (and, for simplicity, assume it is common knowledge the bank will get it right next period, so the economy returns to the natural rate). Current consumption will fall relative to future consumption (even more so under rational expectations, since the recession will cause expected deflation and raise the real rate still further).

    Maybe I should blame Woodford. It's too easy to mistake the assumption of a "cashless" economy with a barter economy, rather than a pure credit monetary exchange economy. But I really think you have misunderstood his model. Maybe actually reading Interest and Prices carefully had negative value added.

    In a barter economy, of course, Say's Law applies. The underemployed ice cream producers simply resort to swapping each others' flavours.

    ReplyDelete
  35. But in a barter economy, of course, the idea that there's a central bank setting a nominal interest rate makes no sense anyhow.

    ReplyDelete
  36. Yes indeed. You should talk to Mike Woodford about that.

    ReplyDelete
  37. Nick,

    Actually Steve is right. Essentially all New Keynesian models are built on an RBC substructure. The main differences are (as Steve said) (1) monopolistic competition and (2) imperfect price adjustment. The sluggish price adjustment can take many forms (Taylor, Calvo, Rotemberg quadratic costs, 1-period ahead price setting, etc.) and in the end the basic implications are qualitatively the same.

    The pertinent “relative price” in the basic model is actually the real wage. After a demand shock (or a monetary expansion) firms seek to hire more labor at the given nominal wage. This bids up the nominal wage but the aggregate price doesn’t keep pace and thus the real wage rises. The higher real wage in turn elicits more labor supply from the households. This is how the baseline model works and is why so many New Keynesian researchers cast the problem in terms of a countercyclical markup. The markup is the ratio of the price to nominal marginal cost which in the standard model with linear production is just the nominal wage. If the markup is countercyclical then the New Keynesian framework has some hope since it is exactly this relative price which is distorted by demand/monetary shocks. If the markup is procyclical then either the model has to be abandoned or modified (e.g., with wage rigidity instead of price rigidity).

    Either way, the relative price movements (and all of the incentive effects which go with it) are entirely neoclassical. This substructure (the RBC model) may not be an attractive feature of the model but it is what the New Keynesian models are all built on.

    ReplyDelete
  38. Anonymous: You are giving a very different interpretation than Steve. But I disagree with you too.

    The "relative prices" that Steve is talking about are the relative prices of different firms' outputs. If I interpret Steve correctly (and he has not said I am mistaken in this), he is saying that the welfare losses due to bad monetary policy are only due to those relative price distortions and the relative output distortions they cause. And those relative output distortions disappear if you drop the assumption of staggered (Calvo) price setting.

    Your interpretation of the New Keynesian models is very different from Steve's, because you are allowing that bad monetary policy can cause aggregate fluctuations in output and employment, not just relative output fluctuations. And your results would not be (materially) affected if we dropped the Calvo staggered price setting, and assumed synchronised price setting instead.

    Your interpretation is more standard, and much more reasonable. But I still think you are wrong. In fact, you are wrong. Let me try to explain why.

    First, we can eliminate the labour market altogether, by assuming that each firm has one worker-owner (this is Larry Ball's old "yeoman farmer" metaphor), and the fluctuations in output and employment are exactly the same. Essentially, it makes no difference whether the workers are self-employed, or are hired by the firm in a competitive labour market.

    Second, try this thought experiment. Start in equilibrium at the natural rate. Now assume that the real wage is fixed by law (fully indexed minimum wage set at the current equilibrium wage). Now suppose bad monetary policy causes aggregate demand to fall. And nominal output prices are sticky, so firms' sales and output fall, and employment falls too, since it is linked to output via the production function.

    So we have a fall in aggregate output and employment even though the real wage stays fixed at the "right" (long-run equilibrium) level.

    And if the minimum wage law is suddenly abolished, so the nominal wage and real wage suddenly drops to clear the labour market, absolutely nothing happens to output and employment (except, over times, as firms cut prices more quickly now that their marginal costs have suddenly fallen). We have merely converted involuntary unemployment into voluntary unemployment, by dropping the wage down onto the labour supply curve.

    If you want the theoretical underpinnings for this result, you have to go back to disequilibrium macro of the late 1960's and early 1970's, where they distinguished between the "notional" labour demand curve (which assumes that firms are not sales-constrained, and is the marginal revenue product of labour curve), and the "constrained" labour demand curve (which takes into account the fact that firms are sales-constrained, and so only hire that quantity of labour they need to produce the goods they are able to sell).

    Look, Anonymous, your intuition is not unreasonable. You are not hopelessly wrong; just wrong. Many economists would think of the New Keynesian model the way you think of it. But it's wrong.

    The way this comment thread has gone, unfortunately, has convinced me that Paul Krugman is right. There is a Dark Age. Economists have forgotten what we used to know. And they don't understand their own maths.

    I'm a burned out ex-university administrator. Steve is a *much* more accomplished researcher in money/macro than me. You very possibly are too, Anonymous; your comment shows this. But you guys just don't have a clue. You don't understand what's going on in this very mainstream model. And I'm talking about New Keynesians nearly as much as "New Monetarists".

    ReplyDelete
  39. Or, put it this way, Anonymous: sure, the real wage is "wrong" when monetary policy is "bad", and causes aggregate fluctuations. But that's just a *symptom* of the problem, not a cause of the problem. The problem is a deficiency of aggregate demand in a monetary exchange economy, where direct barter of labour for goods is ruled out by (implicit) assumption. Even if you fix the real wage at the "right" level, but law, it won't fix the problem, and won't cause output to stay at the natural rate when monetary policy is bad.

    ReplyDelete
  40. Nick,

    You can't disagree with anonymous. He/she is just describing how these models work. You can't keep insisting the sky is green when we can all see it is blue. This gives you an important dimension on which a New Keynesian framework is very different from an Old Keynesian static sticky-wage AD/AS model. In the old models, a monetary expansion increases the price level, the real wage falls, and firms hire more workers (moving down the labor demand curve). In a New Keynesian model, a reduction in the nominal interest rate target is coupled with an increase in the real wage, which is required to induce workers to supply more labor.

    ReplyDelete
  41. Nick,

    It sounds like you are simply dissatisfied with the New Keynesian model itself, not Steve’s interpretation. The demand externalities that you have in mind (paradox of thrift effects for instance) are just not an important part of the model. Perhaps this is a shortcoming but as to whether the effect is present, it’s simply not there.

    The “yeoman farmer” interpretation doesn’t change the model. The incentive of the producers the same – when demand rises, prices of the other goods are unchanged and the yeoman farmer works more because he or she can purchase more goods at lower prices (his or her effective real wage has gone up).

    A fixed real wage is not typically included in the model. Fixed real or nominal wages can be added but still won’t undo the underlying structure which is fundamentally neoclassical (one difference though is that with fixed real or nominal wages, imperfect competition must be introduced in the labor supply curve).

    The “disequilibrium” analysis is also not typically part of the model. The New Keynesian model is set up and solved with conventional techniques. I imagine that you could add disequilibrium stuff though I have to admit I have no training or familiarity with such techniques so I can’t really say how that would change the framework.

    The mathematical analysis and interpretation of the model is fairly straightforward. One way to see what Steve is saying would be to simply list the equilibrium conditions for the standard RBC model and the New Keynesian model side by side. The conditions are all the same with two exceptions: the New Keynesian model has a price setting condition that is not present in the RBC model (this is the New Keynesian Phillips Curve or NKPC), and the non-stochastic steady state occurs at a point where price exceeds marginal cost (the steady state desired markup). These two mathematical conditions are exactly what Steve pointed to originally: imperfect price adjustment and monopolistic competition. All of the other conditions are the same.

    ReplyDelete
  42. Steve: there are two versions of the Old Keynesian model: the sticky wage version and the sticky price version. The sticky price version has exactly the same procyclical real wage as in the New Keynesian model. New Keynesians (typically) chose to go with the sticky price version.

    But the point I was making (in response to Anonymous) is that if prices are sticky it doesn't (much) matter if wages are sticky or not. You get the same output and employment fluctuations in response to bad monetary policy either way.

    ReplyDelete
  43. Anonymous: (thanks to both you and Steve for keeping this going, by the way).

    "It sounds like you are simply dissatisfied with the New Keynesian model itself, not Steve’s interpretation. The demand externalities that you have in mind (paradox of thrift effects for instance) are just not an important part of the model. Perhaps this is a shortcoming but as to whether the effect is present, it’s simply not there. "

    I am dissatisfied with the model, but I'm more dissatisfied with Steve's interpretation of it.

    There is a paradox of thrift in this model.

    Start in full equilibrium. Ignore investment for simplicity. Now assume there's a drop in time preference (people are more patient, so that parameter Beta changes). The natural rate of interest falls. Suppose the central bank fails to see that the natural rate has fallen, and keeps the nominal rate of interest the same for one or more periods. The Consumption-Euler equation tells us that there must be a fall in current consumption, and so a recession. And the Calvo (or whatever) Phillips Curve tells us that will cause disinflation. This is exactly the same result (qualitatively) you would get in an ISLM+Phillips Curve model.

    "The “yeoman farmer” interpretation doesn’t change the model."

    Agreed. It was just my way of showing that the underlying cause of recessions was not some sort of failure in the labour market, or distortion in real wages. Any change in real wages is a consequence of aggregate fluctuations, not a cause.

    (Comment too long. Cutting here.)

    ReplyDelete
  44. "The “disequilibrium” analysis is also not typically part of the model."

    Yes and no. It's there, just implicitly. Suppose we asked "what is the labour demand curve in this model?" In the long-run, when prices are flexible, the labour demand curve is the marginal revenue product curve. But in the short-run, when firms cannot adjust prices and so sales are exogenous wrt the firm, the labour demand curve is the inverse of the production function, provided real wages are less than or equal to marginal revenue product. This is exactly the same as the old Barro-Grossman 1971 disequilibrium model, except we replace "value marginal product" with "marginal revenue product", because of imperfect competition. In other words, when you solve for employment from short-run output, by using the production function, you are implicitly using a constrained demand curve. It's just not explicit. But if you ever asked the question "what would happen in the short-run if wages too were sticky", or some similar question, you would be forced to make this explicit.

    "The conditions are all the same with two exceptions: the New Keynesian model has a price setting condition that is not present in the RBC model (this is the New Keynesian Phillips Curve or NKPC), and the non-stochastic steady state occurs at a point where price exceeds marginal cost (the steady state desired markup)."

    The markup due to imperfect competition is not an issue here. An RBC model with imperfect competition is (usually, unless there's multiple equilibrium or something) is still an RBC model. Output is suboptimal, but money is (at least approximately) irrelevant, and any fluctuations caused by taste or technology shocks may be (at least roughly) efficient.

    It's the sticky prices that are the biggie.

    For starters, if prices were perfectly flexible, you could not have a central bank setting the nominal interest rate as a policy instrument. There would be no nominal anchor. Start in one equilibrium time-path, then double all prices along that time-path, and you are still in equilibrium. The price level would implode or explode if the central bank made the tiniest mistake.

    Furthermore, if prices were perfectly flexible, the real side of the economy would determine the real interest rate, independently of monetary policy, rather than the central bank setting the real interest rate, as it can do in the short run given sticky prices. This means that even though the equations describing the equilibrium conditions would look very similar, the chain of causation would be totally different. (This is getting back to Paul Krugman, and the kerfuffle over what Narayana Kocherlakota said.)

    I want to make one more fundamental point, about equilibrium conditions in a barter vs a monetary exchange economy. Right up Steve's street (I think). But I gotta teach my class. Later.

    ReplyDelete
  45. Briefly. Start with a RBC mode. There are two ways to introduce monetary exchange:

    1. The proper way. Explicitly introduce trading frictions. (Steve's way).

    2. The quick and dirty way. Assume all exchanges are infinitely costly, except exchanges of goods for money, which have zero transactions costs. With n goods, there are potentially n(n-1)/2 markets in a barter economy. You just assume there are only n-1 markets, in each of which money is traded against one other good.

    There's nothing wrong with the Q&D way, but you must be explicit about it. The New Keynesians aren't explicit. And it matters. For example, take a recession in an NK model. every firm wants to sell more goods, but buyers don't want to buy more. Why don't firms simply barter goods with each other? It's ruled out by assumption, but that assumption is never stated explicitly.

    If prices are perfectly flexible, it doesn't matter if those barter markets are "missing". So we don't need to see the equilibrium conditions in those missing markets. But if prices are sticky, it does matter. If there's an excess supply of apples for money, and an excess supply of bananas for money, apple growers would want to do direct barter trades with banana growers.

    If we allowed barter for example, then Steve's original interpretation of NK models (there's a relative price distortion and a relative output distortion, but nothing more), would be precisely correct.

    And the NKs are to blame for this, because they never make explicitly clear that they are ruling out non-monetary trade. Some of them even miss the fact that this model only makes sense with money.

    For example, suppose we had a barter version of the same economy, and had the government peg the real interest rate too high by pro-usury laws. That would not cause a recession. It would simply cause an excess supply of loans.

    ReplyDelete
  46. Fringie? You sure are arrogant and rude.

    ReplyDelete