Monday, February 10, 2014

Kartik Athreya and the mysterious allure of macro theory



I last heard of Richmond Fed macroeconomist Kartik Athreya about four years ago, when he wrote a screed attacking econ bloggers and rebuking people who read them (see reactions by Tyler Cowen, Scott Sumner, Matt Yglesias, Ryan Avent, and Brad DeLong). Macro is hard, Athreya said, and lay people should stop trying to understand it and leave things to the experts.

Now I see that Athreya, reprising his role as evangelist, has bowed to the inevitable reality that normal people want to understand something about macro. In an effort to oblige them, Athreya has written a book called Big Ideas in Macroeconomics, in which he attempts to explain modern macro theory to the lay public. I predict that very few people, even among the econ blog audience, will actually read the book. On the upside, that fact will probably annoy Athreya himself (hehehe, revenge, hehe). But on the downside, it sounds like a really interesting book, and I think  if people read it, they would have a much better understanding of what macroeconomists do these days.

I myself have not yet read the book, but David Glasner and John Quiggin have, so I'll be lazy and free-ride off of their effort. Apparently, Athreya discusses general equilibrium theory and a little game theory, and then goes into some of the modern macro models that incorporate these things. Interestingly, he apparently ignores models that bear on policy questions. Glasner writes:
The index [of the book] contains not a single entry on the price level, inflation, deflation, money, interest, total output, employment or unemployment. Which is not to say that none of those concepts are ever mentioned or discussed, just that they are not treated, as they are in traditional macroeconomics books, as the principal objects of macroeconomic inquiry. The conduct of monetary or fiscal policy to achieve some explicit macroeconomic objective is never discussed. In contrast, there are repeated references to Walrasian equilibrium, the Arrow-Debreu-McKenzie model, the Radner model, Nash-equilibria, Pareto optimality, the first and second Welfare theorems.
So what is Athreya doing here? I see two possibilities, which are not mutually exclusive.

Possibility 1: Athreya is trying to balance out the public discussion of macro. He knows that a lot of lay people and bloggers already talk about Keynes and Friedman, monetary policy and fiscal policy. His intent was to write a book about all the other macro stuff that the public doesn't know about - general equilibrium and game theory, incomplete markets and search frictions, and so on.

Possibility 2: Athreya just doesn't care that much wasn't that interested in writing a book about the stuff he leaves out. He's not interested in telling people about the history of economic hypotheses, only the history of economic methodology. He cares vaguely about policy and about the economy itself, but only insofar as it's an interesting application of his beloved methodology. (Update: The original language that I used here was mean unfair. What follows is closer to what I meant to say.) He is passionate about his beloved methodology, and wants to evangelize it to the public. The ideas he thinks are "big" are ideas about how to make macro models, not about what assumptions go in the models or what conclusions come out of the models.

I don't know Athreya, but my intuition is that whether Possibility 1 is true or not, Possibility 2 is true - Athreya seems to be in love with modern macro methodology. If he were just trying to balance out the public discussion, he would probably have been more explicit about that goal. Also, the fact that he doesn't seem to talk much about evidence is telling. "Big ideas" apparently doesn't mean ideas that are successful in explaining the data, it means ideas that are neato and cool.

Actually I think you see a decent amount of this attitude in the macro field - although some macroeconomists just use whatever methods they have to in order to explore their favorite ideas about the world, a large chunk seem genuinely in love with the DSGE modeling methodology itself (which we really shouldn't even be calling "DSGE" now that it includes game theory too).

I have always wondered why they love it so much.

Paul Krugman once wrote that macroeconomists are entranced by the "beauty" of their methods. But modern macro models aren't "elegant" in the physics sense. In physics, "elegant" means that you reduced something complicated to something really simple-looking. Modern macro models are more like Rube Goldberg machines full of delicately arranged moving parts. You usually have to solve them numerically. So maybe "beauty" is about complexity rather than elegance? After all, Rube Goldberg machines are beautiful, in their own way. Maybe the appeal of modern macro models is the amazement of having all of these moving parts and yet actually getting something halfway intelligible out of it all?

Or perhaps macroeconomists love the modern models because they're mentally exhilarating? Athreya's 2010 essay seemed to take great pride in how "hard" macro theory is. And I guess it is a little bit hard. It's harder than writing a blog, for sure. It's harder than the macro of days gone by. But as modern theoretical economics goes, it doesn't seem that hard. I'm pretty sure it's not as hard as game theory, decision theory, theoretical econometrics, or quantitative finance theory. If there are people who think that making DSGE models makes them badass math jocks, who are they comparing themselves to?

Or is it the fact that so many people criticize macro? Maybe the torrent of criticism - from the recession-hit public, from annoying bloggers, from certain other economists, even from some within the profession - creates a "circle the wagons" effect, like a culture of patriotism in a country under perpetual siege. Perhaps the simple fact that it's "our methodology" is enough to make it the object of love. It's true that macro methodology is unlike any other theoretical paradigm I've ever seen, so maybe its uniqueness interacts with the patriotism to produce a sense of pride.

Or maybe it's the subject matter that's intoxicating? A game theorist gets to say "Wow, I just figured out how people would interact in an infinitely repeated Prisoners' Dilemma!" Neat, but maybe not as neat as the macroeconomist, who gets to say "Hey, I just modeled the whole economy!"

Anyway, it's a mystery. If any macroeconomists can help explain the intoxicating allure of their modeling paradigm, I'd appreciate it.


Updates:

An email exchange with Steve Williamson made me realize that I said something that was much meaner (and more of a stretch) than what I meant to say. I issued an update above.

Just to be clear, this post is not criticizing Athreya's book. It sounds interesting, I'm going to read it, and I think you should definitely read it if you're interested in macroeconomics and don't already know all the stuff.

Other interesting hypotheses I've heard advanced for why people love macro methods:

1) Modern macro feels very new and "cutting edge" or "up-and-coming" to some people. (This makes sense to me, because macro had a very recent paradigm shift, back in the 80s.)

2) DSGE-type models may not be mathematically hard in an absolute sense, but they give people who come from non-math backgrounds a chance to do something harder and more technical and more imaginative than they ever did before, and thus feels empowering and liberating. (This also makes sense; I've definitely seen grad students gain mathematical confidence from doing DSGE models for the first time.)

Also, definitely read Costas Alexandrakis' thoughts on the culture of macro, in the comments. Also see Kurt Mitman, a young DSGE devotee, on why he loves it.

Here is a review of Athreya's book by Herb Gintis.

Steve Williamson is mad that I talked about Athreya's book without reading it, calling it "disgusting" that I would try to psychoanalyze Athreya without even reading what he wrote. But Steve needs to chill out. My main claim was that Athreya really loves modern macro methods. I don't see Steve contradicting that claim. But even if I was wrong about Athreya, my main observation - that there are a lot of people who looove modern macro methods a lot - doesn't really need Athreya as an example. Steve himself is a great example. The Athreya reviews by Glasner and Quiggin just got me thinking about macro-method-love. And I certainly was not criticizing Athreya's book; in fact, I recommended that everyone go read it. So calm down, Steve.

Steve also suggests that people actually read Athreya's research, and this I have done - well, a couple of his working papers, anyway. The stuff I've read is actually not very "macro-y" - it's pretty simple and intuitive theory about how credit markets work. In fact it seems more like the kind of thing Ricardo Caballero suggested macroeconomists work on, back in 2010.

Robert Vienneau read much of the book, and confirms that it's about history of methodology rather than history of ideas.

Chris House wonders which methods I'm talking about specifically, but then pretty much answers his own question:
Perhaps it is the conjunction of so many common elements that he associates with DSGE models. For instance, there is a good deal of “boilerplate” which shows up in DSGE models (the representative agent, the production function, the capital accumulation equation, and so on).
The method macro people seem to love most is general equilibrium itself. Almost anything with general equilibrium in it can probably now be labeled a "macro model" without anyone complaining. "Micro theory" people (often simply called "theory" people or "pure theory" people) occasionally use GE, but seem to have mostly lost enthusiasm for it. Interestingly, GE itself might be losing its total dominance in macro, as game theory is creeping in via wage bargaining in labor search models. But anyway, in general, Chris is right - DSGE "boilerplate" is really a conjunction of common "boilerplate" elements that people seem very attached to.

77 comments:

  1. well obviously I can't help you much on this one. I think there is a lot to the circle the wagons hypothesis, but that can't be the whole story (I mean people who really circled wagons drove those wagons into hostile territory for a reason).

    ReplyDelete
  2. DFLamont6:11 PM

    There is always room and a need for expert language, but a lot of this is needless complication that is common to many disciplines, which is the possession of "secret" or hermetic knowledge for the purpose of exclusion.

    People use jargon for the purpose of exclusion and status. It is one of the functions of language.

    It is easy to make a big goddamn mess of everything. It's hard to make things clear and simple. But there are many academics who make the mistake of thinking that because they make themselves impossible to understand, that everyone else is stupid.

    I found this to be the case with post-structuralism, which often suffers not only from being poorly translated from the original French, but is itself premised on a series of often terrible arguments cursed with bad thinking.

    Elegant theories can be problematic because they may function like the gears of a machine and be perfectly internally consistent without corresponding to the outside world. Complex theories can have a closer fit to data but may make it harder to find out glaring errors.

    ReplyDelete
    Replies
    1. Poststructuralism no good p

      Delete
    2. Quantum mechanics does explain much in the way of physcis, but scientists, engineers, and policymakers were able to employ good old Newtonian physics to put a man on the moon.

      I would analogize post-structualism with quantum mechanics when it come to language (words have meaning in reference to other words which they are not, etc.) but there was another theory (I forget the name, it was so long ago when I muddled through this) where words "do" what was intended given their shared social meaning which is sort of like Newtonian physics in that they "work" as intended.

      Delete
  3. Quiggin on Kartik Athreya http://johnquiggin.com/2014/02/10/macroeconomics-made-easy/

    ReplyDelete
    Replies
    1. Did I not link to that above? :-)

      Delete
    2. My apologies. I have a rare genetic condition: blue-black colour blindness :)

      Delete
  4. There are artists and musicians and there are people who, not being able to create anything themselves, indulge in art or music "appreciation."

    Athreya seems to want to indulge in economics appreciation - wallowing in the esthetic rather than trying to actually accomplish some real world good.

    Practical economic policy has been a disaster for the last fifteen years. It is long past time that a howling mob of angry citizens figuratively pulled people like Athreya from any position of influence on policy.

    ReplyDelete
  5. It must be the mathematical one-upping. There's no other way to compare people except looking at who's better at math. Risking confrontation with data is dangerous (actually the results are well known). As a bonus, the method offers new and exciting ways of curve fitting data. The more complexity, the more pleasure with noise is to be had.

    ReplyDelete
  6. "If any macroeconomists can help explain the intoxicating allure of their modeling paradigm, I'd appreciate it."

    Noah, I think that's what Kartik was trying to get across in the book that you didn't read.

    ReplyDelete
    Replies
    1. Well, I will read it! Gawrsh...

      Delete
    2. Yes, there's a movie I've heard about. Let me tell you about it before I go see it.

      Delete
    3. ...What's it called?

      Delete
    4. Anonymous11:13 PM

      It's called Noah Smith is a douchebag with no publications. Perhaps you've heard of it. I know I have.

      Delete
  7. The modern macro models are 85% signaling and 15% insight. But once you engage in signaling too often you can no longer tell the difference between the signal and the substance. Add in grad school and the psychological effects of sunk costs. We like what we understand. We don't like that which we don't (and some stuff that plain deserves it too). That's more or less what DFLamont is saying above, just a matter of degree.

    The other reason is quite simply that it's a coordination equilibrium. If you want to do macro, and everyone is doing this kind of macro, and this is the kind of macro that gets published, then you got to do it. Even if at first you recognize that it's only 15% insight or whatever, after awhile you get the outcome that all the journal editors expect it, all the referees expect it and even the audience of the journals expect it. Individually, maybe - even a majority - would like to switch, introduce more heterogeneity in the approaches. But who's dumb enough to be the first? So for sake of peace of mind your DSGE doing self might as well convince your not-too-crazy-about-DSGE self that everything is as it should be.

    ReplyDelete
    Replies
    1. Talking about signaling have you noticed that people proud of how modern their macro skill really like to name-drop and pull ranks?

      Also notice how they are most threatened by accomplished economists that break ranks than random cranks that have the potential to cause much more damage to the economy.

      Delete
    2. And how getting the facts wrong doesn't seem to bother them, nor far too man of their peers?

      Delete
  8. Academic macro isn't what lay people consider macroeconomics, which is the thing they remember from their Bachelor-level education. It deals with GDP, inflation, unemployment and productivity in a straightforward, policy-oriented way.

    So, "Macro theory" is basically a recent impostor, stealing the accomplishments of a field the public is familiar to claim authority. DSGE people should just call their field DSGE and try to prove on their own the value of their research.

    ReplyDelete
  9. Anonymous11:27 PM

    "If he were just trying to balance out the *pubic* discussion, he would probably have been more explicit about that goal."

    Probably the most amusing typo I've seen in an economics post lately. (Feel free to delete this comment when the typo is fixed.)

    ReplyDelete
  10. Anonymous4:16 AM

    It's about identity.

    Isn't it nice to have something that you can understand but people outside the profession cannot?

    ReplyDelete
  11. Anonymous4:24 AM

    DSGE model is an effective means of creating a barrier to entry to a large number of people from outside the profession. It sets the rules for academic engagement on issues of unemployment and poverty reduction which should be open to people with a broader set of skills and accumulated knowledge.

    It has nothing to do with finding the best ways of dealing with serious real world economic problems.

    ReplyDelete
  12. Herb Gintis has an provocative take on the book. A snip:

    "The fact is that I consider this an exemplary exposition of modern macroeconomics, and I think his defense of the theory is as good as one can find anywhere, the theory is in fact so weak that nothing can save it. People continue doing it not because it is good theory, but because it is the only game in town. I believe a complete revolution in macroeconomic theory is in the process of being born, although it will take some years to take over as the mainstream theory."

    The whole thing is here:
    http://www.amazon.com/gp/cdp/member-reviews/A2U0XHQB7MMH0E

    ReplyDelete
  13. Phil Koop8:52 AM

    Ah, the "intoxicating allure of macroeconomics". When did the flame burn so high and get so hot?

    I suspect you are right about possibility 2, in which case one wonders whether Athreya has found his best home at the Richmond Fed. But I think you have overlooked some others:

    Possibility 3: Athreya wants to skate clear of policy issues so that his personal opinions will not be conflated with those of his employer.

    Possibility 4: Athreya reckons that if people who have no actual responsibility for implementing policy feel free to pontificate on the subject, then turnabout is fair play.

    ReplyDelete
  14. This comment has been removed by the author.

    ReplyDelete
  15. Anonymous9:01 AM

    I f you think macroeconomic models that do not address the Lucus critique with micro-foundations are unscientific, you get an explanation of what the current research program is about and why they don't even bother addressing the Old Keynesians. Hey, they'd say, Ptolemaic models of the solar system outperformed Copernican models until Kepler came along. While the models may not be perfect, we're on the right path because our research program is scientific. The Old Keynesian research program is like adding epicycles to a model that has the sun orbiting the earth. So I don't think the DSGE guys believe they have a methodology fetish. They think they are on the right path. Unfortunately or fortunately, it requires a lot of technique to do macro right, but that's not the point. They blow off other technical criticisms as secondary to the thrust of the program. I'm just trying to profile the movement, not endorse it. Does this explanation seem reasonable?

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
    2. Anon: I think that's why they do it, but it doesn't explain why some love it. Actually there are a number of macro people who do it because it's the only game in town, but who don't seem to love it.

      Delete
    3. Anonymous10:57 AM

      Agreed. Even is the program is on the right path, the current models are awful at explaining past activity and doubly awful at forecasting. You'd expect these guys to be very humble. In 2011, Steve Williamson predicted inflation would rise to 5-10%. In 2012, he again predicted increased inflation to 5%.

      Delete
    4. Anonymous11:05 AM

      Noah -- If you think you're on the cutting edge of an advancing scientific research program, then of course you love it. You might bitch a bit about the difficult techniques (or revel in them), you might be concerned that the models are too primitive to explain past data, but you love the fact that you're on the bleeding edge of advancing your discipline and that your solving problems and deficiencies of the old, failed program. Once again, does this explanation sound reasonable?

      Delete
    5. Once again, does this explanation sound reasonable?

      I guess, but wouldn't this be equally true in any field, then?

      Delete
    6. Anonymous2:44 PM

      I think that people, especially in the earlier stages of an ascending research program, are a pretty happy group. The new "Young Turks" as it were replacing the old "Young Turks." As the research program matures, barriers are met, the optimism gets tempered, and so does the love. But there's always folks who love the technique and think they are in a path-breaking program. I think you see this in the natural sciences as well but I'm very interested in your opinion given your physics background. Recall that the early work in the RE, RBC, and the whole DSGE methodology garnered a bunch of Nobels. Pretty exciting stuff to be in on the ground floor of that (if you buy into the research program). If you don't love what you're doing then, you're in the wrong field no matter how misguided or brilliant the research is ultimately judged. My two cents.

      Delete
    7. I've seen a similar love of the subject among physics grad students I've met (and I've met a lot). Physics is something that's very easy to fall in love with, for a lot of reasons. Most of the PhD students I knew (and know now) had absolutely no intention of pursuing a physics career after graduation - they did it for the love. I can understand that...I majored in it in undergrad for the same reason, and I would have gone on to grad school just for fun if other events had not interceded (but let's not go there)...

      But I don't often see the same passion among game theorists, decision theorists, theoretical econometricians, or quant finance people, that I do in macro. Actually, there was very little passion for macro methods at Michigan, even among our top macro job candidates. It was people from other schools who had the enthusiasm. Also - and this is something I didn't mention in the post - the macro enthusiasm seems much more concentrated among foreigners than Americans.

      I think the "cutting edge" hypothesis is interesting, but don't you think the crisis threw macro onto the back foot in that regard? Now everyone is doing financial-friction macro, but it seems more like a reaction to the crisis - something people would have preferred not to have to do - than a really exciting new technique. What do you think?

      Delete
    8. Anonymous3:24 PM

      I think the DSGE program is maturing so I'd expect less enthusiasm than at its beginning. (It's been a while since the ground-breaking work.) But its new enough that you don't see it in undrgrad texts. I think the political nature of macro (as say compared to theoretical econometrics) might also contribute to the love (but now I'm really speculating, i.e. making stuff up). And the financial crisis could have had two effects on the "love". One group can see that the program isn't working well and that all it's trying to do patch things up/over. Not much love there. The other, the ones who feel the love, see it as a giant intellectual challenge within the program and are energized to find the answer. I wonder if there was less love at Michigan than at Minn or Chicago? Not a big Kuhn fan, but those involved in the revolution feel more love than those doing normal science, i.e. making the new framework rather than working within the new framework once it's established. Are we getting anywhere? Ask Steve. I get the impression he loves it, but that might be because he still feels the glow from back in the day when this stuff was getting started.

      Delete
    9. Steve definitely seems to love it more than a lot of the younger people, and Michigan seems to love it less than elsewhere (though they are good at working with it). Interesting thought about the politics...maybe to some, DSGE is a big anti-government crusade. Prescott certainly seemed to think along those lines.

      Delete
    10. Anonymous6:39 PM

      We gotta get Colander on this. Maybe he could figure it out in his next survey of graduate students. By the way, what were Mich econ grad students excited about?

      Delete
    11. Getting out of grad school, mainly. ;-)

      People I knew were pretty excited about structural estimation and decision theory.

      Delete
  16. So you're judging an entire book based on what someone else wrote about the index? FAIL! I've edited several technical books the past few years. Producing the index is a hard, tedious process and is often considered an afterthought.

    The most likely explanations for those terms not being in the index are (1) pure oversight or (2) being so broad they don't warrant inclusion in an index.

    Maybe you should read books before you criticize them.

    ReplyDelete
    Replies
    1. I don't think I criticized the book, did I? It sounds good.

      Delete
  17. Anonymous10:34 AM

    Economics is of course art not science. DSGE macro is like 12-tone music - nobody listens to it but it is intellectually challenging to compose it. Krugman is Mozart, composing stuff people understand and like.

    ReplyDelete
  18. Anonymous10:52 AM

    Macro is not hard, is made hard to give the impression that we are producing something of value. If the job of Athreya were left to the rules of his models (free and rational markets), he would be homeless.

    ReplyDelete
  19. Anonymous10:54 AM

    Terry Tao has a post on the "post-rigorous" stage of mathematics:
    http://terrytao.wordpress.com/career-advice/there%E2%80%99s-more-to-mathematics-than-rigour-and-proofs/

    The basic idea is that mathematicians reach a stage where they can discuss mathematical concepts in fairly general terms. But in order to reach this stage, you have to spend sufficient time messing around with epsilons and deltas in proofs until you're able to talk about the higher-level concepts without making basic mistakes.

    In macro the analogous convention to a mathematical proof is a model. Thus macroeconomists can and do discuss general concepts, but they do so with a known set of models in the background.

    The problem is that lots of people are talking about the same things, but without a model in the background. Now some of these people are actually making sense, but there's a cultural gap -- they can't write out a model with epsilons and deltas. It's like a mathematician criticizing a physicist for talking about observed regularities as though they are certain mathematical results.

    So Athreya's book is rather like if a lot of journalists and commentators were talking about the Riemann hypothesis in vague general terms, and a mathematician wrote a book called "Great ideas in analysis" that explained how you prove things, and the journalists made fun of him because "Riemann hypothesis" didn't show up in the index.

    ReplyDelete
    Replies
    1. Anonymous10:57 AM

      One more thing: There are plenty of economists who seem not to have moved into the post-rigorous phase, and don't seem able to talk about economics in general terms outside a particular model. Unfortunately, it's often these types that criticize the general public for lack of knowledge of "real" economics.

      Delete
  20. Noah, I want to ask you a question (or two actually) that I asked under John Quiggin's post also.

    I will try to set this up: I never studied macro mathematically because I formed the suspicion at the beginning of my studies about 40 years ago that mathematics would never be able to precisely predict complex systems in any subject area, including economics and ecology. So I studied math up to intro calculus, and then jumped over to Gödel’s late ruminations on foundations, because I already suspected that our real-world problems will turn into policy crises that are going to need some other standard of agreement besides mathematical certitude. And my suspicion seems to have been borne out about prediction of complex systems, aside from use of accounting identities, or morphologies described by power laws like Zipf’s law. I think it comes out of a very simple problem, a foundational type of problem: Newton’s N-body computation problem repeats itself, or has an analogy, with the number of compartments in your model.

    So here is my question: Do you think that the need of the New Classicals to adhere to the “efficient markets hypothesis”, and to “rational expectations”, is partly an unconscious preference to avoid the intractability that would result from introducing an additional compartment, a “finance” compartment, into the models?

    And, to ask a related question of New Old Keynesians, do you think that IS-LM adequately covers the reality of the financial system, for macro policy purposes?

    ReplyDelete
  21. I won't be reading it at thirty five bucks..

    ReplyDelete
    Replies
    1. Apparently Athreya's macro knowledge trumped his micro in this regard.

      Delete
    2. Apparently Athreya's macro knowledge trumped his micro in this regard.

      OUCH, BURN

      Delete
  22. Noah,

    I think the most accurate view of how science progresses is the one articulated by Lakatos. In summary, and I am afraid I won't do justice to his view, science evolves as the result of competition among different research programs. In economics, the dominant program (or paradigm if you like Kuhn's terminology) has been winning comfortably. The question I believe you are asking is why. Here are some possible explanations:

    1. The professional incentives (e.g. the selection criteria set by professional journals) are biased. I certainly think this is true, as probably is the case in most sciences. However, as Steve Williamson has pointed out, this didn't prevent a research program that was considered "out there" at the time, and found refuge in non-ivy universities (e.g. Carnegie Mellon and Minnesota) to steal followers from the status quo and eventually dominate.

    2. The dominant program has been extremely able to deflect criticism by modifying what Lakatos calls "auxiliary hypotheses". It has done so, for example, by adding to its models frictions, costly computation, etc. while retaining its core hypotheses (e.g. optimization and equilibrium).

    3. As bad as the dominant paradigm is, it is still better than most other alternatives. This is also true to a large extent, as most critics of the current paradigm seem to advocate a return to Marxian economics, Austrian economics, IS-LM, etc. I think there are many good reasons why these programs lost followers: they describe a world that is no longer with us, they cannot make quantitative predictions, they were unable to formulate internally consistent theories, etc. On this, you should read the review of Kartik's book by Gintis.

    4. Young economists are more poorly educated about the world. This is an observation of mine. Economists of the older generation are usually much more interesting to talk to. Many have held non-academic jobs, often blue collar ones, before pursuing their graduate studies, have lived in different parts of the world, and have read extensively, They especially know their history well. I am afraid that the preoccupation of younger economists with Bellman equations is often a sign of their inability to talk intelligently about worldly affairs. Having said that, there are economists who can do both well, an example that comes to mind being Daron Acemoglu.

    4. Economists are very risk-averse. One research program that I feel could pose a serious challenge to the dominant program is that formulated by Nelson and Winter (http://www.amazon.com/Evolutionary-Theory-Economic-Change-Belknap/dp/0674272285), which involves agent-based evolutionary models, and which acknowledges that while people act with purpose to improve their circumstances they do no necessarily optimize. The problem is that because of its quantitative rigor it requires a significant investment in skills, just like the dominant program, but with more uncertain results. Therefore, many "apostates" are content to just throw stones at the leading paradigm without doing anything about it. On this, let me close with a story. A few years back I had a chance to interact with Peter Diamond at a conference of the Eastern Economic Association. Diamond asserted that the DSGE research program is a dead-end. When I asked him what should replace it, he said that young economists should figure this out. But considering that the research that earned Diamond his Nobel Prize was pretty much within the dominant program, I couldn't help but wonder: Did he realize this late in his career, or did he knew it all along but was waiting for someone else to take the risk of finding a replacement? And if a well-known economist at MIT is not willing to take the risk, then who should?

    ReplyDelete
    Replies
    1. I was more wondering why people personally love it, not why it's dominant. I think you're right about the reasons it's dominant.

      Delete
    2. Anonymous3:03 PM

      CA -- I think the Lakatos framework is a great way to look at competition between theories or research programs and it works pretty well for macro. The profession accepted that the Lucus critique and decided micro-foundations was the answer. While there are lots of problems with DSGE (see Waldman, Noah and a bunch of others), the profession doesn't see a research program that solves the primary concern of the Lucas critique and addresses the Waldman/Noah criticisms. And without an alternative, DSGE will persevere as the dominate methodology.

      I just wish those DSGE guys would admit that their program isn't developed enough to give any legitimate policy prescriptions. The IS-LM framework is plainly kicking its ass. There should be no embarrassment for that. In lots of areas, rules of thumb or what Krugman has called "middlebrow theories" are much better forecasters.

      Delete
    3. "I just wish those DSGE guys would admit that their program isn't developed enough to give any legitimate policy prescriptions."

      I agree, and many do.

      "The IS-LM framework is plainly kicking its ass"

      On this we will have to agree to disagree. Example: Traditional IS-LM has no built-in expectations. Most current discussions about policy heavily involve expectations, and rightly so.

      Delete
    4. Anonymous3:41 PM

      CA -- I should have been more clear -- The IS-LM framework is plainly kicking its ass AS A FORECASTING TOOL during the Great Recession. The IS-LM liquidity trap analysis predicated no inflation and there hasn't been any. A lot of the MN/CHI RBC and DSGE guys have been incorrectly forecasting much higher inflation for years.

      Delete
    5. But IS-LM model forecasts deflation, which hasn't happened either.

      As far as forecasts go, don't judge by what people say. Judge by what the models say. RBC models have no money, so they are agnostic as far as inflation is concerned. Chicago is not known for RBC. Lucas' model does predict lower inflation when output and employment is lower. However, in my opinion this hardly makes it a desirable tool for analysis, as it does not contain a financial sector, so the most important part is missing. Same is true for New-Keynesian models (which, by the way, also predict deflation).

      Delete
    6. Anonymous5:44 PM

      Sloppy of me. Interpreting these models (and what is considered a standard version) isn't always easy. Krugman says he slaps on downward wage rigidity to prevent deflation. I think the discussion among Noah, Krugman, Delong, Rowe, Williamson, Andolfatto, Sumner, Smith, Murphy, Yates, Wang, Avent and other about how to interpret Williamson's formal models and their validity is informative. http://noahpinionblog.blogspot.com/2013/12/does-qe-cause-deflation.html and the many links therein.

      Delete
    7. Noah: "I was more wondering why people personally love it, not why it's dominant. I think you're right about the reasons it's dominant. "

      Noah, as you know, there's a lot of sub-fields to go into you you don't want to do DSGE macro. Many are more challenging, presumably many are fun, and many are more lucrative.

      Delete
    8. CA Thanks for pointing to the Nelson & Winter book. Looks interesting. But it was published 30 years ago. If this paradigm was going to take over, you would think it would have some momentum by now.

      Delete
  23. RE: CA's comments on Diamond. Actually, I wouldn't say that Diamond's best stuff is part of the dominant paradigm. If you look at the original search papers, he says quite explicitly that he wants models that break completely with Walrasian GE. The DSGE people who want to just add search go to Pissarides and Mortensen, not Diamond, where they find unique equilibria, constant returns to scale matching functions, constrained Pareto-efficiency and, you know, THE BEST OF ALL POSSIBLE WORLDS MORE OR LESS.

    ReplyDelete
    Replies
    1. kevin,

      here is how Diamond (1984) describes his model in the first page of "Money in search equilibrium".

      "This paper studies a simple general equilibrium search model where money is
      used for all transactions (no barter or credit). The model considers only steady
      state rational expectations equilibria."

      Delete
    2. Yeah but that's because he's saying "I can keep all the assumptions of the dominant paradigm and only break one (the Walrasian auctioneer) and get completely different results. In this case, multiple equilibria, multiple natural rates of unemployment, multiple y*s that would normally go into the DSGE Euler equations. And all that with flexible prices too.

      Delete
    3. Fine, but why even bother if you think the dominant paradigm is a dead-end? Why try to reform it rather than start something completely new? I mean, it's his right, but when he comes out after he has made a career as working within the dominant paradigm and says something like this, well, what am I to think? To use a related story, Chris Pissarides ended a talk recently by saying, I was born a Keynesian and will die a Keynesian...but in the meantime I earned a Nobel Prize. And people laughed. Am I the only one who does not find such cynicism amusing?

      Delete
  24. Full disclaimer: I know Kartik and like him and am very sympathetic to his research agenda.

    A couple of things:
    1. I think it's tricky when you paint all of modern macro in one broad swath. The label DSGE - Dynamic Stochastic General Equilbrium, could be applied to say Kydland and Prescott, since it is a dynamic, stochastic GE model. But generally (at least from where I'm sitting) DSGE refers to models that use the representative-agent paradigm (possibly with added margins/frictions/etc).

    I see that as facing off against heterogenous agent macro - starting with the Bewley-Hugget-Ayagari models for steady-state environments, and then under a Krusell-Smith type set up to accommodate aggregate shocks.

    Then, for macro-labor the standard paradigm now is Mortensen-Pissarides, which depending on the flavor can be somewhere in between representative agent and heterogenous agent macro.

    The point I'm trying to make, however, is that different types of models have different advantages for the question that's specifically trying to be answered. But, at the end of the day, I think all of us macro economists realize (or should realize) that, as Prescott put it, all models are wrong. They're just abstractions that we use in the hopes of learning something about the economy. Clearly not every margin or contingency can be modeled, but we write down the best model that we can to try to answer the question of interest, hopefully capturing all of the first-order features.

    2. Why do I (and I think others) love the modeling paradigm that we've chosen?

    I'm particularly interested in policy-oriented research. The allure of GE models, in my opinion, is that we make the assumption that people are rational and optimize. Then, we bring our models to the data and estimate parameters such that the micro and macro behavior of people looks correct. Then, we can change something in the economy, say, the tax rate or unemployment benefit duration, or the bankruptcy code and see what happens to the macro economy based on the re-optimization of all of the individual households. We can see what happens at the micro and macro level, and evaluate the distributional effects of policies. To me that's the allure, is that it's very clear what the assumption that's being made is (rationality/optimization), and then it's clear to interpret the results and draw conclusions from that assumption. Will we always be correct? Of course not. But this is science, and all models are wrong, and we know that.

    3. I've done some work with agent-based modeling. When I was a graduate student at Oxford in physics I did work on what would be considered ABM. I found it to be interesting and wrote a paper on it looking at traders in an artificial market. I think that that kind of modeling is more useful for forecasting than policy analysis. I will confess I'm not fully up on that literature, but my question is what is the discipline put on people's actions that we could then use them to generate counterfactuals?

    4. Finally, and I'm not the first person to say this - I forget who first said it - but if say, I could write down a model that would perfectly predict/forecast the macro-economy, I wouldn't be sitting in my graduate student office right now, I'd be living a life more like: http://richkidsofinstagram.tumblr.com/
    Some of the people that worked in my research group at Oxford didn't publish the full details of their research in their theses because they went to hedge funds upon graduation because they thought they could make money (this was written about in the Economist at the time).

    ReplyDelete
    Replies
    1. All models are wrong is not a justification of models that are always wrong. There has v to be some useful information about the world coming from the models. If you can't point to robust evidence for explanatory offer of a class of models then as a class they are useless. It is not science when you are from evidence add far as you can.

      Delete
    2. Yes to Krzys point above. Six years into a depression, when the hyperinflation didn't arrive and fiscal policy has been shown not to be neutral, you kind of lose the right to be going "all models are wrong, you know ..."

      Delete
    3. vimothy11:12 AM

      All models are wrong, but some are useful

      It might be wise to remember how that quote ends:

      "All models are wrong, but some are useful; the practical question is how wrong do they have to be to not be useful."

      -- George Box, Robustness in the Strategy of Scientific Model Building

      Delete
    4. Adding on to Krzys - Kurt, you're raising a strawman in your comment. The issue isn't that macro can't perfectly predict the economy; the issue is that large chunks are flat-out wrong, and many in the field are just fine with that.

      Delete
  25. I'm partly just guessing, but it seems to me guys like Athreya love their newfangled macro methods largely because they're somewhat more removed from the political debates.

    If you want to argue policy, there's no need for complex methodology. Everyone knows that the vast majority of policy-interested economists goal-seek their economics conclusions to fit their political leanings. Older methods work fine for that and are also much more convincing when hastily explained in blogs. Think of Krugman and his hastily drawn ISLM diagrams, sometimes ruing the cost of war, other times ruing the counter-stimulus of fiscal drag and not mentioning that most of it's defense cuts. For that kind of work, the complexity of recent macro theory would be a huge detriment. What one wants for such work is theory that's simple enough to be convincing to laypeople with similar political leanings and flexible enough to allow one to take opposite positions depending on circumstances. The modern public figure economist is really a rhetorician who sometimes resorts to equations and diagrams to argue his case.

    So it's understandable that someone like Athreya would pride himself in believing he's above all that and engaged in science not rhetoric.

    ReplyDelete
    Replies
    1. Anonymous1:51 PM

      "Think of Krugman . . . sometimes ruing the cost of war, other times ruing the counter-stimulus of fiscal drag and not mentioning that most of it's defense cuts."

      I guess you have to be a macro-geek, rather than a layperson, to see those statements as inconsistent. I think Krugman would say, convincingly, that you can rue the cost of war (be it Iraq or WWII), while at the same time acknowledging the fiscal stimulus it provides. You could then argue that non-military spending would have equal (or sometimes greater) stimulative effect, and wish that our spending could be allocated differently between guns and butter.

      He could similarly point out that his criticisms of deficit spending on Iraq and Afghanistan arose when the economy was basically at full employment, and his concern about fiscal drag came in the wake of the Great Recession. You know, the old "when the facts change . . . what do you do,sir?" & etc.

      The one observation I do feel qualified to make as a layman is the length to which people will go to try to accuse Krugman of being inconsistent.

      Delete
    2. " Think of Krugman and his hastily drawn ISLM diagrams, sometimes ruing the cost of war, other times ruing the counter-stimulus of fiscal drag and not mentioning that most of it's defense cuts. "

      Two things - first, he's been very clear why he says that certain times things are good or bad. That alone is better than 90% of the economists on the right. Second, he's actually broken the counter-stimulus down a few times.
      Third, military spending is a very inefficient form of stimulus; an necessary war is obviously worse.

      Delete
    3. I'm not saying that Krugman is the only inconsistent economist. I'm saying he's a typically inconsistent economist. His only consistency is that he will always reach the economics conclusions that suit his political leanings. And that's absolutely typical of economists. The right wingers do it just as much.

      Your point on the timing of his positions is fair, but I don't believe that was important. He criticized war spending because it was Republicans doing it and decried budget cuts without mentioning that they were mostly defense cuts because it was Republicans doing it. If you choose to believe otherwise I don't think you're being fully honest with yourself.

      Another case was his recent piece explaining why lower employment due to Obamacare was okay because it was voluntary. Sure the decision not to be employed is voluntary, but the reduced demand affects many other people. Either reduced aggregate demand is bad or it's not. He will argue either way, because his economics is in service to his rhetoric, not the other way around. And that is the rule not the exception among economists who write on public policy.

      Delete
  26. Anonymous4:14 AM

    "Everyone knows that the vast majority of policy-interested economists goal-seek their economics conclusions to fit their political leanings."

    (1) All economists should be interested in policy.

    (2) There is no such thing as a model in economics that is politics - neutral. Your micro-foundations for a start will immediately start off with an assumption about human behaviour and social interaction which is absolutely political.

    ReplyDelete
    Replies
    1. Yes I agree. A model behaves how its economist creator tells it to behave and produces the conclusions it is pre-programmed to produce.

      But it is at least theoretically possible than a macroeconomist could base a model on empirical observations and let the conclusions fall where they may. And I'm saying the vast majority don't do that: they devise models in such a way so as to reach a preferred conclusion.

      Delete
  27. Anonymous7:38 AM

    Well, I have read some parts of the book. It´s a huge book, considering its main task as written down in the early pages. The parts I have just skipped, I don't think I will miss many relevant ideas. Well, is there one "major idea" about macro in the entire book? I don't think there is.
    For example, the defense of Rational Expectations is just old talk ... in old dressing. There is nothing absolutely new there, not an argument, not even the language. It looks like we are reading a text written down 15, 20 or 30 years ago.

    Well, If someone comes up to me and say that what they do, or what they know, is really, really very very difficult, such that it should be left entirely to the trade specialists, I will be suspicious.
    The point is that there is a great difference between knowing really something, and playing around with things just to have a job. There is a funny story (actually true) about Richard Feynman (been considered as one the most brilliant physicists ever), also famous for being able to explain the most sophisticated results in physics to the layman using simple words. Once, in a talk at Berkeley (I think), he was explaining a fundamental point in modern physics, about which he had been one of the greatest contributors. Someone in the audience was perplexed by his explanation, and questioned the logic of Feynman's argument. Feynman's replied that there was no problem; he would explain the correctness of his argument by the means of simple words, such that all the audience would understand quite well, he knew the issue so well, so ...
    Funny enough, Feynman were not able to do that. Then he replied: "maybe, we still do not as much about this issue as we thought".

    There you have it: (1) one of the greatest scientists of the last 100 years had a tremendous pleasure in turning sophisticated physics into common language; (2) an economist arguing that modern macro is really a very difficult thing, "leave it to the professionals, and read my book if you want to catch up a little bit, even though I do not talk about inflation, unemployment, monetary policy, and so on".

    I am left with the impression that Athreya displays some limitations about his understanding of modern macro (I mean models, results, variations on the theme). Is it so difficult to explain the logic and the main results of e.g. the baseline New Keynesian Model (kind of Clarida, Gertler and Gali model) to a broad audience? I do not think so. It may be difficult to DEMONSTRATE the steps to a particular result (e.g., if you do not known dynamic processes, stochastic processes), but not the whole logic and what the main results mean in terms of economic policy.

    Athreya's book is, in the words of Feynman, and intimidating book. Big book, too pretentious: a claim for knowledge, where there is little.

    ReplyDelete
  28. This is a masterpiece:

    "When I was in undergraduate Engineering school, before my operation research days and the involvement with forecasting and finance, it was not a secret that many students who could not cope with reality, math, laboratories and hard work dropped out of Engineering school and enrolled in Marketing and Economics. We considered those disciplines the “dropout hideouts”, places where people just talked about meaningless things and did no real work. Yet, those disciplines today control the fate of the world. Why? The answer is because they are the majority. Everyone thinks he or she can do economics or can market a product. Just watch TV for just one day. You will see mostly economists parading and ads for products but you will rarely see an engineer or physicist explaining how reality works. The hot topic is always “the economy, stupid”. This is because everyone can talk about it and has the misconception that he understands what is being talked about."

    Too much true in this piece. The rest is also interesting: http://t.co/i2gKthIRD0

    ReplyDelete
  29. When I retired as a computer scientist, I thought it might be interesting to study macroeconomics. I was surprised and dismayed to discover the field is totally dominated by a sort of pseudo-math. Equilibrium models seem pretty quaint from an outside perspective. The damage to the field by this is much larger than you economists have realized. You are missing out on so many of the advances in information theory, mathematics and computing that ALL the other sciences are benefiting from, while you all sit around and gaze navels with each other. No one can enter your field unless they are brought in young enough to be bamboozled. The sooner you fix this and move on, the better.

    ReplyDelete
    Replies
    1. Anonymous11:19 PM

      You are an amazingly arrogant and poorly informed twat.

      Delete
  30. Anonymous11:22 PM

    You know what I find funny? The arrogant posturing in the comments by a bunch of people so poorly informed about the field they're criticizing that they think Noah knows something. You people are truly pathetic, and I'm going to laugh at you while lying in my piles of money.

    ReplyDelete
  31. Anonymous9:08 PM

    Telling Williamson to calm down is like telling the sun to cease being hot. He's got as much misplaced confidence as the rest of the profession combined.

    He seems to think the whole world is subject to some Williamsonian version of the Dunning-Kruger Effect, that not a single person has enough intelligence to even contemplate how much more intelligent he is than the rest of us.

    I've found that it's rare someone with so much confidence will ultimately have anything useful to say. Arrested development and all that. It's a precarious thing to be certain in an uncertain world. May he live long enough that everyone sees how misplaced it all was.

    ReplyDelete
  32. Anonymous3:54 PM

    "DSGE-type models may not be mathematically hard in an absolute sense, but they give people who come from non-math backgrounds a chance to do something harder and more technical and more imaginative than they ever did before, and thus feels empowering and liberating."

    An extraordinary remark. If anything macro needs "liberating" from DSGE.

    I suspect this will not happen though, even if we have had three decades of it monopolising the field, crushing any critical reasoning, and being of very marginal use to policy makers. The reason is the more dominated it is by things like DSGE and Game theory, the more it will attract people who like such things. And the more it will be dominated by DGSE/Game theory.

    Meanwhile the discipline will get narrower and unable to benefit from different perspectives, skills, methods and accumulated knowledge. It will skip merrily along, as policy makers continue to address every new crisis with three quarter century old knowledge.

    ReplyDelete