Wednesday, March 11, 2015

Lucas and His Critics

Thanks to Brad DeLong and Noah Smith for resurrecting a 2011 panel discussion on the roots of modern macroeconomics, which includes Michael Lovell, Robert Lucas, the late Dale Mortensen, Robert Shiller, and Neil Wallace. If you're interested in the history of macro thought, this is fascinating.

What's of interest to DeLong and Smith is something else altogether, though. As Noah lets on at the outset, it's basically macro-dissing. Noah tells us about two kinds of macro-dissers:
Attacker Group 1: "Old Keynesian" economists who want to use aggregate-only models.

Attacker Group 2: Decision theorists and other micro theorists who want to make macro use more realistic models of agent behavior.
DeLong and Smith are more-or-less in the first camp and the second, respectively, though they might be better-slotted in a third category - journalists who no longer practice economics (much), but have a lot to say about what economists do or should do. The focus of these two macro-dissers is some comments in the panel discussion by Bob Lucas. What we get is Lucas's comments, taken out of context, with parts edited out, and filtered by DeLong's and Smith's notions of what modern macroeconomists do. If you're thinking this won't be entirely grounded in reality, you're on the right track.

Somewhere in the panel discussion, Shiller makes some comments about behavioral economics, which sets Lucas off. He has very different views, as you might already know. Here are Lucas's comments (not edited):
One thing economics tries to do is to make predictions about the way large groups of people, say, 280 million people are going to respond if you change something in the tax structure, something in the inflation rate, or whatever. Now, human beings are hugely interesting creatures; so neurophysiology is exciting, cognitive psychology is interesting – I’m still into Freudian psychology – there are lots of different ways to look at individual people and lots of aspects of individual people
that are going to be subject to scientific study. Kahnemann and Tversky haven’t even gotten to two people; they can’t even tell us anything interesting about how a couple that’s been married for ten years splits or makes decisions about what city to live in – let alone 250 million. This is like saying that we ought to build it up from knowledge of molecules or – no, that won’t do either, because there are a lot of subatomic particles – . . . we’re not going to build up useful economics in the sense of things that help us think about the policy issues that we should be thinking about starting from individuals and, somehow, building it up from there. Behavioral economics should be on the reading list. I agree with Shiller about that. A well trained economist or a well educated person should know something about different ways of looking at human beings. If you are going to go back and look at Herb Simon today, go back and read Models of Man. But to think of it as an alternative to what macroeconomics or public finance people are doing or trying to do . . . there’s a lot of stuff that we’d like to improve – it’s not going to come from behavioral economics. . . at least in my lifetime. {laughter}
That seems pretty interesting. Lucas is open to thinking about alternative views of human behavior, but he's making an assessment about what the productive avenues for research in macroeconomics are. He says behavioral economics does not appear to be one of them. And I think he's right - for now, at least.

This is interesting, as I had never thought about the macroeconomic problem in the way Lucas has laid it out. Modern macro is built up from "conventional" economic theory - competitive analysis, information economics, game theory, for example. In a macro model - with heterogeneous agents, a representative stand-in, or whatever - economic agents are rational, in that they are maximizing some objective function given available information and constraints. But the basic economic theory we work with is almost never used to make sense of the behavior of the individual. The one study I know about is John Rust's 1987 Econometrica paper. That's a dynamic programming model of bus engine replacement, fit to the observed behavior of Harold Zurcher. But I don't think the standard consumer behavior we teach in econ 101 would do a very good job of predicting what I do when I go out shopping. And I'm actually trained to think like an econ 101 consumer, so econ 101 certainly won't explain my neighbor's behavior. But we teach the things we do in econ 101 for good reasons - we think that this basic theory does a decent job of explaining how reasonably large groups of people behave. Not the 280 million that Lucas mentions - I think our usual notion of "large" in this respect is much smaller.

So, when we say that macroeconomic theory has "microfoundations," what we mean is not that it is built up from theory that explains the behavior of individuals. For a lot of economic behavior, we're not going to do very well in explaining the behavior of an individual. And, as Lucas notes, behavioral economists can't do it either. Rather, what "microfoundations" is about is finding the model elements - optimizing behavior, constraints, information - that explain the behavior of large (that's large in the "larger than one but much smaller than 280 million" sense) groups of economic agents from first principles. Then we can make predictions about the effects of policy on the behavior of really large (i.e. 280 million for example) groups of people. So, until we come up with something better, that's the state of the art.

Finally, I thought Noah's description of "attacker group 2" was interesting. This was "decision theorists and other micro theorists who want to make macro use more realistic models of agent behavior." Two reactions to this:

1. It seems a lost cause to try to "make" some other researcher do something that you would like them to do. These are quite independently-minded types. Best to do it yourself.
2. As Lucas points out, we don't care about someone else's notion of what might be "realistic." We go with what's useful.


Also, it's not as if macroeconomists never get outside the box - in the behavioral sense. In contrast to what Noah seems to think, we're not straight-jacketed into some narrow class of models. Some examples:

Krusell/Smith 2003
Recent survey by Driscoll/Holden at BoG
Greg Mankiw blog post

So people have experimented with behavioral economics in the macro literature, they have held conferences, they have written papers. But it's not widely used. And Lucas goes some way in explaining why.

67 comments:

  1. When has Smith ever practiced economics ? His comparative advantage is journalism/blogging, which he does well.

    ReplyDelete
  2. Stephen, George Stigler made similar criticisms to Lucas about the lack of the ability of behavioural economics to generate simple, powerful wide ranging predictions back in 1963 using his vast knowledge of the history of economic thought.

    His quote is at http://utopiayouarestandinginit.com/2014/10/09/george-stiglers-1963-critique-of-behavioural-economics/

    ReplyDelete
  3. "So, when we say that macroeconomic theory has "microfoundations," what we mean is not that it is built up from theory that explains the behavior of individuals."

    Surely that's nonsense anyway? When we are talking about an economy we are talking about a social system. Nobody says the key to understanding an organised social system is understanding the behaviour of individuals. It is about the dynamic of a whole group and its institutional make-up. I guess this might be related to the so-called Two Cambridges dispute of the 1960s and the so-called aggregation problem. Clearly we need some empirically well-founded insight from sociology and history here.

    Great post, and interesting debate. For a different view I was also interested to see what Lars Syll says about this subject on the veracity of the record on the use of rational-expecations economics:

    "One could perhaps accept macroeconomic models building on rational expectations-microfoundations if they had produced lots of verified predictions and good explanations. But they have done nothing of the kind. Therefore the burden of proof is on those macroeconomists who still want to use models built on these particular unreal assumptions."

    https://larspsyll.wordpress.com/2015/03/11/brad-delong-on-silly-robert-lucas/

    ReplyDelete
    Replies
    1. "It is about the dynamic of a whole group and its institutional make-up."

      Sure, but how do you put that in a model?

      Delete
    2. "Sure, but how do you put that in a model?"

      Isn't that what we taxpayers are paying you to do? To figure that out???

      Delete
    3. "Sure, but how do you put that in a model?"

      Emergent properties. Other disciplines do it (more outside social sciences). But I suspect it will require some radical, fundamental rethinking by (Anglo-American) economists about the nature of "theory" and "economy."

      Will keep this cryptic for now. You'll just have to wait for me to finish my book first...

      Delete
    4. "But I suspect..."

      Suspicions aren't very helpful. If you're proposing an alternative you think is superior, we need more than that.

      Delete
  4. There's a lot of neat decision theory from the last couple decades that is making its way into macro. Lots of work on ambiguity, for example, where agents don't know the probabilities. And even some work on hyperbolic discounting, as you note. Here are a couple conferences on the former:

    http://www.cvstarrnyu.org/2nd-conference-on-ambiguity-and-robustness/
    http://www.carnegie-rochester.rochester.edu/crApril14.htm

    ReplyDelete
  5. "Lots of work on ambiguity, for example, where agents don't know the probabilities."

    Does this overcome the aggregation problem and is it useful for understanding the behaviour of a complex social system, such as a national economy?

    ReplyDelete
    Replies
    1. Put another way, does it simply replace one 'representative agent' with another?

      Delete
    2. This is basically what Noah was talking about. Using results from decision theory in macro. You would have to read the papers, but indeed, I know some of it is a representative agent with unusual preferences. However, you have to get beyond words to something practical we can use, if you want to propose an alternative. We all know the world economy is a "complex social system," but where does that take you? What's the model that idea generates?

      Delete
    3. I think if one is going to go down the behavioral road the best way to aggregate is with agent-based computational models. However, it is unclear at this stage whether the added realism but also complexity adds much to our understanding of how economies behave. At the same time, I think more resources should be devoted to such research programs.

      Delete
    4. I have seen some of this work, but I didn't learn anything from it. Some individuals follow some rules of the thumb in making decisions. Maybe they update their rules of thumb based on how the rules perform. Maybe they somehow learn from each other. The results are very sensitive to what the nature of the suboptimal behavior is. I tend to think - until someone convinces me otherwise - that suboptimal behavior is probably a minor issue for aggregate phenomena. We have enough to think about with constraints, information frictions, commitment problems, etc.

      Delete
  6. I think the broader point both Smith and DeLong are making is that if micro-foundations are impossible to build "from the bottom up" then why do Lucas et al. criticize so heavily the approach taken by old fashion Keynesians? Especially since the main thrust of the criticism is that old-fashion Keynesians use models that are not grounded in micro-foundations?

    On another note-- your ad hominem attack, describing them as more journalist than economist, seems a bit unfair? I mean, aren't you a blogger too?

    ReplyDelete
    Replies
    1. "...aren't you a blogger too?"

      Sure, that's pretty obvious. But I do all the regular academic stuff too.

      Delete
    2. This is what makes this blog so invaluable - it is written by someone who does serious macroeconomic research. Who else does this - Andolfatto, Cochrane , maybe James Hamilton.

      Delete
    3. Looks like this "serious research" did not prevent Williamson from making bad predictions about inflation whereas the predictions of his archnemesis Krugman, made after the financial crisis began, were spot on: low inflation, long recessions, no rising rates for public debt.

      Delete
    4. Sometimes I make bad ones, sometimes I make good ones. Check this out (last paragraph):

      http://newmonetarism.blogspot.com/2013/12/fomc-statement.html

      As for Krugman, he's the deflationary trap guy, right? That's the phenomenon we have never seen, but which Krugman likes to predict.

      Delete
    5. No idea what a "deflationary trap" is supposed to mean. If you refer to the fact that low inflation / deflation is bad during a balance sheet recession, well, this is ancient stuff and self-proclaimed Fisherites like yourself should really know it. You know, Fisher wrote more than just definition of the real interest rate. :D

      Delete
    6. About predictions, after the financial crisis broke out Krugman predicted that inflation will stay low, that T bill rates will not soar and that the recessions will not be a short one. So in terms of empirical validity his model performs better than yours (whatever you use). I roll with the theory that actually matches the facts while you can defend loonie like Lucas until kingdom come.

      Delete
    7. "No idea what a "deflationary trap" is supposed to mean."

      I'm sympathetic, but if you go to Krugman's blog and do a search on that, you'll find a lot of stuff.

      Anonymous 3:42: Sure, that's what Krugman says, of course. His model is IS-LM, which actually doesn't "perform" at all.

      Delete
    8. I am a different anonymous then Anonymous 3:42. Maybe Krugman got effects right, but his claim to causality may not be warranted ...

      Delete
    9. Well, I can say that I have never learned anything from Krugman about macroeconomics. And I don't know any practicing macroeconomists - academics and people who actually make policy decisions - who take him seriously either. I can see how it would sound convincing for a lay person, or someone with some undergraduate economics training - he's very good at writing and persuasion.

      Delete
    10. Mark Thoma and Brad de Long take Krugman seriously, since Thoma's blog is mostly rehashes of Krugman's articles, with Thoma's value added being "I agree with Krugman.

      Delete
    11. Yes, I know. That's the primary cheering section.

      Delete
    12. Neither DeLong nor Thoma is a practicing macroeconomist. DeLong hasn't done anything in years except low-tech history, and Thoma has never done anything.

      Delete
  7. The case for an activist monetary policy is about whether you can manage the leads and lags on monetary policy. If you can't do that, Keynesian macroeconomics can be as true as it wants but will be merely of academic interest to academics.

    A key part of any response to Friedman rests on the ability of forecasters to do their jobs with tolerable accuracy. After reading the annual reports of the Fed, Milton Friedman noticed the following pattern:

    "In the years of prosperity, monetary policy is a potent weapon, the skilful handling of which deserves the credit for the favourable course of events; in years of adversity, other forces are the important sources of economic change, monetary policy had little leeway, and only the skilful handling of the exceedingly limited powers available prevented conditions from being even worse."

    Central banks pay due to the implications of the leads and lags on monetary policy only as an ex-post facto rationalisation for disappointment.

    ReplyDelete
  8. Hi Stephen,

    I think you should pay closer attention to the difference between an explanation and a prediction.

    ReplyDelete
    Replies
    1. I think you should pay closer attention to being specific, so I can figure out what you're trying to get across.

      Delete
    2. How does a macro model with a representative agent that maximizes “some objective function given available information and constraints” explain “how reasonably large groups of people behave,” apart from predicting economic outcomes?

      Milton Friedman argued that a model should be judged by the accuracy of its predictions, not by the realism of its assumptions. If we’re interested in explanation rather than prediction, should we still ignore a model’s assumptions?

      Reading your blog, my impression is that you're taking for granted positions in debates about the nature of science that are much less widely shared than they once were. If you're interested in the current state of this debate, you might take a look at Hilary Purtnam and Vivian Walsh, eds., The End of Value-Free Economics (2012), which includes essays by Amartya Sen and Partha Dasgupta.

      Delete
    3. "How does a macro model with a representative agent..."

      Seems you're taking for granted what is in macro models. The nature of these things is not what it once was. You can still find some representative agent models in macro - sometimes that's useful. But more often not. Philosophizing about what should go in a macro model is apparently very different from putting one together and making it work. Assumptions of course don't come out of nowhere. But you have to be creative about what to leave out. That's the whole trick.

      Delete
    4. I don’t know what it means to, "philosophiz[e] about what should go in a macro model . . .” Philosophy does not compete with science; it has nothing scientific to offer economists.

      But when you describe the real job of the macroeconomist as that of “putting [a macro model] together and making it work,” I want to ask, “what counts as ‘making it work’”?

      Delete
    5. Philosophizing about macro models is what you do, and what Noah Smith does, for example. I can do the same when I think about what a brain surgeon does. I can sit in my armchair and muse about how the brain surgeon goes about his/her business. But I have never been in an operating room, except to have things done to me. Making a macro model work is playing around with it until you think it captures well the phenomenon you're interested in. Then, suppose you're going to use it for policy. You start using it, and see what happens. If it seems to be producing bad outcomes, or it no longer seems to be explaining the phenomena of interest, then you modify it, or abandon it and move on to something else.

      Delete
    6. “Philosophizing about macro models is what you do, and Noah Smith does, for example. I can do the same when I think about what a brain surgeon does.”

      This is a poor analogy because the criteria for judging the results of brain surgery are widely shared within the medical community, whereas the criteria for judging macro models are not so widely shared. There are lots of model-builders, and there’s a lot of disagreement among them. The recent debate between you and Nick Rowe over the relationship between interest rates and inflation, and the discussion between Roger Farmer and David Andolfatto on the concept of involuntary unemployment are good examples.

      One last question “from the armchair,” Should theories should be judged solely on the basis of their predictive success as Milton Friedman urged, or should we also “look under the hood” as Dan Hausman recommends?

      Delete
    7. "This is a poor analogy because the criteria for judging the results of brain surgery are widely shared within the medical community, whereas the criteria for judging macro models are not so widely shared."

      I'm almost certain that's incorrect. My guess is that brain surgeons have their own intra-professional controversies, and disputes over what procedures to use, for example. And some of the science may be bad, if you look at how they do it. So I think the analogy is fine. I'm just trying to help you understand that I can explain some things about how economics is done in lay terms for you, but there are things that you won't entirely understand until you have done them yourself. We're all specialized enough that there will be some deep parts of professional practice that only the practitioners will get.

      On your "last question:" This is roughly how I explain this to students, and borrows a bit from Lucas, actually:

      1. Start with a question we want to answer.
      2. Build the model specifically to answer the question at hand. If the model happens to be useful for other stuff, all the better, but our goal here is simplification. We want to get by with the least amount of detail we can, otherwise we can't understand what is going on. The model is an abstraction, and to be illuminating it can't include all the detail that exists in reality, or we can't work with it.
      3. Now try this model out on the phenomena we know something about. Does it explain phenomena that are closely related to the question we want to answer? Does the model somehow fit the facts well? If it doesn't, we might want to step back and modify the model, or start over.
      4. Now start to use the model to answer the question we started with at step 1. What does the model tell us? Why does it tell us that? If we have done a good job in the modeling process we can understand all the bits and pieces in the model that went into giving us that result. Maybe the result is sensitive to some parameter, and we don't have good information on what that parameter should be. Then maybe we have a problem. That's why building it up from preferences, endowments, technology, and information structure works well - we have a bunch of components and can interchange them to see what happens.

      Step 4 is part of "looking under the hood," at least the way I think about it. For example, if "predictive success" was based on some elasticity being 0.2, but other studies estimate that elasticity to be 2.2, and 2.2 means the model doesn't predict well, we might want to think twice about it.

      Delete
    8. Wow this sounds nearly identical to "construct validity" and "convergence" in social psychology. And just as vacuous.

      Delete
    9. You'll have to explain that. Why is it vacuous? I'm just describing to you a scientific method. In spirit it's not all that different from what is done in the natural sciences, though there are some key things that make economics different.

      Delete
    10. SW, thanks for the thoughtful reply. I may get back to you.

      Anonymous,

      I'm not sure what you mean by "'construct validity' and 'convergence' in social psychology," but the two phrases sound a bit like the distinction between the "coherence theory of truth," which takes a "broad view" of the relationship between thought and reality (or between models and statistics), and the "correspondence theory of truth," which contemplates a closer relationship between our ideas and a concept-independent reality. This particular distinction may well be "vacuous," but the chain of reasoning that leads to this conclusion isn't necessarily vacuous. One chain begins by simply noticing that the very idea of a theory of truth is beyond comprehension, but maybe this distinction is quite different from the social psychology version.

      Delete
    11. Jargon is a signal that, indeed, there is nothing going on. Anonymous says that what I said is vacuous, just like "construct validity" and "convergence," which is some jargon, the meaning of which is obscure to us. You'll note that I didn't use any jargon in 1-4. That's all plain English, I think.

      Delete
  9. Steve I follow your blog and find it interesting. I think the general justification of a theory is , following Friedman, is in its power to make sense of the surroundings..in fact , if you follow the wonderful tribute to Gary Becker (in the Chicago website) , you will find that the uniqueness of Chicago school was in its ability to explain / rationalize the surroundings. Its Price Theory program and even macro used to be fairly empirically oriented. In a broad sense, under the leadership of Lucas , Chicago macro has moved surprisingly away from that hallowed tradition. I recollect in Jim Cassidy profile of Chicago economists following the crisis, Gary Becker recounting that after a phd dissertation(obviously on rational expectations and microfoundation orientated) interview attended by both Friedman and Becker, Friedman turned to Becker and said "aren't they going too far.." or some such word to similar effect. The issue is , has it?

    ReplyDelete
    Replies
    1. Well, this is all bygones now. Becker is no longer with us, and Lucas is 78 and no longer actively engaged in the Chicago economics department, though he may do a bit of teaching. But Lucas is/was in fact very much empirically-oriented. Sure, his key contributions were in macro theory, but he has plenty of empirical work, and the theory certainly had the data in mind. For example, "Expectations and the Neutrality of Money" (JET 1972) is about explaining Phillips curve correlations, in part. So, there's a sense in which Lucas was indeed in the old Chicago tradition. He loved Friedman, though my impression is that that love was not mutual. Old Chicago, e.g. circa 1965 was very low-tech. Lucas is high-tech, and has a lot of respect for theory. Definitely not in the Friedman mold.

      Delete
  10. For the record, Anonymous is referring to a Heckman interview, not Becker.

    http://www.newyorker.com/news/john-cassidy/interview-with-james-heckman

    ReplyDelete
    Replies
    1. Note, in that interview, there is this:

      "Heckman then criticized behavioral economists, such as Berkeley’s George Akerlor and Yale’s Robert Shiller, for suggesting that the roots of the crisis lay in irrational behavior: overconfidence, animal spirits, and so on. For the most part, individuals responded to market incentives and reacted rationally, he insisted."

      Here's another:

      Interviewer: What about Robert Lucas? He came up with a lot of these theories. Does he bear responsibility?


      Heckman: Well, Lucas is a very subtle person, and he is mainly concerned with theory. He doesn’t make a lot of empirical statements. I don’t think Bob got carried away, but some of his disciples did. It often happens. The further down the food chain you go, the more the zealots take over.

      Delete
  11. The 'micro vs macro' dichotomy seems misplaced to me. Ditto the 'micro' in microfoundations. Prof Williamson has indicated as much (we're not starting with a model of individuals). The real difference is between modeling founded on what might be called informed, forward-looking optimization (RBC etc.) by the herd versus one dependent on aggregate behavioral norms that don't necessarily accord with utility maximization. Is that fair?

    ReplyDelete
    Replies
    1. Roughly, yes. Though on the "versus one dependent on aggregate behavioral norms that don't necessarily accord with utility maximization," I could also say that some of the alternatives are not even internally consistent.

      Delete
    2. Where do norms come from? They come from what people like. Why do you suppose you know that terms belong in my utility function?

      Delete
  12. Isn't Lucas the source for 'no such thing as involuntary unemployment'? Pretty offensive and totally absurd. The construct that produces such thinking invites attack and criticism - deservedly.

    ReplyDelete
    Replies
    1. Can find this quote in any Lucas paper? What is offensive and absurd is you!

      Delete
    2. Though I am not sure that you deserve a more thoughtful response, Lucas' point was that involuntary unemployment is an ill defined concept and not very useful in terms of explaining what is going on in the economy. Hence, business-cycle models should focus on the fluctuations of employment hours instead. The only way to interpret this statement as a claim that there is not involuntary unemployment is if you don't know how to read.

      Delete
    3. An employment contract, like marriage, requires that two people agree. If one would-be party to the contract does not agree, then it doesn't happen. We could talk about people being involuntarily single, I suppose, but the relevant behavior we are interested in, I think, is the search behavior. An unemployed person is engaged in active search for employment. Old-fashioned ways of thinking about that - in competitive equilibrium environments - didn't get very far, as those models are not equipped to think about search. Search models allow one to think about unemployment in a useful way. Those models are more enlightening about the determinants of unemployment, and how governments might help labor markets work more efficiently. But people in those models are always making choices. The unemployed are people who choose to search for work because they think there is light at the end of the tunnel - potentially they will find a job that will make them better off. So voluntary or involuntary doesn't enter into the discussion. But that doesn't mean that the outcomes are efficient, and there can certainly be a role for government. So, anonymous is just confused. He doesn't possess some moral high ground relative to Bob Lucas. He's just engaged in misdirected vitriol.

      Delete
    4. This quote has been put to another blogger and history of economic thought specialist for a different perspective who has replied with a post:

      https://larspsyll.wordpress.com/2015/03/16/stephen-williamson-on-involuntary-unemployment-and-search-models/

      In some ways it seems this becomes a circular argument about the usefulness of models when dealing with social systems that involve complex interrelations between components. On this some social sciences went one way, economics another. Macro-economics would seem to have particular issues in that there are some practicalities that require estimates - for example we need multiplier estimates when drawing up a national budget and the impact on other economic variables of a decision regarding an appropriate operating interest rate target. And this surely requires a model. However, it is up to the historian to go through documents of cabinet meetings and central bank meetings and look at to what degree these models are taken into consideration when it comes to the actual call, and how much is actually subjective judgement that takes into account a lot of unquantifiable and unknowable information that is not contained in them. Then perhaps we can start to put the usefulness of these models into perspective.

      Nanikore.

      Delete
    5. "However, it is up to the historian to go through documents of cabinet meetings and central bank meetings..."

      Even better, sometimes you get to be there while these decisions are being made. But when you say "the historian," who is that? For example, if I read the FOMC minutes, does that make me a historian?

      Delete
    6. "if I read the FOMC minutes, does that make me a historian?" Of course it does. You are reading primary evidence. That is the best thing you can possibly have to understand the actual behaviour of agents and the real world target system. All macro-economists should be historians.

      This is from cabinet discussions, 1976 - the time of the Sterling Crisis, perhaps the biggest crisis in Britain since the Great Depression and surely the moment of calling in the career of a macro-economist (Britain, who little more than half a century earlier had the world's reserve currency, now found itself in the humiliating position of having to go to the IMF):

      " On the hand, those operating in the foreign exchange market
      already appreciated correctly the importance of the public
      sector borrowing requirement. They wanted to know that the
      Government were concerned about Its size and recognised
      the need to reduce it. There could be political advantages in
      early action on next year's borrowing requirement before any
      possible recourse to the IMF In the autumn.
      c. Although overseas opinion had taken some account of It,
      there had not yet been a full appreciation at home of the
      significance of the recently announced increase in the United
      Kingdom's recoverable reserves of gas and oil. The
      Department of Energy were endeavouring to produce more
      telling monthly statistics which would point up the value of
      these assets, upon which emphasis could also be laid at the
      Energy Conference later in the month. However, it was
      observed that foreign opinion had already taken full account of
      the value of the oil and gas reserves, and it was only for tills
      reason that the country had obtained help on the scale which
      had recently been arranged. The additional resource growth
      from North Sea oil and gas would be relatively small.
      THE PRIME MINISTER, summing up the discussion, said that..."

      See:
      http://filestore.nationalarchives.gov.uk/pdfs/small/cab-128-59-cm-76-8.pdf

      I think when Bernanke had to make the call, he relied on a vast knowledge of historical case studies, not on rational expectations models with sticky prices. It also does not surprise me that Yellen wants "more diversity in the economics PHD". That means people who are able to handle conflicting and complex normative information. (I guess that was what Lars Syll was getting at in that admittedly rather blunt cartoon re Levine). Carney has said that "no theorist is able to explain the complexities of central banking".

      Delete
    7. OK, but how do I put some structure on what I'm reading in the FOMC minutes, to help me understand it? Don't I need theory, or some alternative theories?

      Delete
    8. " Carney has said that "no theorist is able to explain the complexities of central banking"."

      Who cares? Why should I be interested in the thoughts this guy. As well say "Bill Bixby thinks macroeconomics is on the wrong track, wants more studies of green people."

      Delete
    9. It most certainly does matter. Understanding the behavioural thoughts of individuals in macro is questionable. Society as a whole is another matter in macro-economics. The exceptions to being interested in an individual would be the behaviour of a central bank governor (believe me this can affect a central bank reaction function!) or a HOG, or a finance minister, or if you are in a oligopolistic capitalistic system, the behaviour of key players in finance, industry, or perhaps labour unions. Historians most certainly follow transcripts of what central bank governors say very carefully to build up an understanding of their behaviour and how they were understanding and responding to events, together with that of their institutions, and this will eventually help us understand MP more generally.

      Delete
    10. "OK, but how do I put some structure on what I'm reading in the FOMC minutes, to help me understand it? Don't I need theory, or some alternative theories?"

      Not immediately, no. In fact you should put these aside for a while, and work like an investigator. You put all the evidence together, even if presents you with contradictory information. (If you do not do this you will repeat the sins of Sargent 1981.) You then synthesise the information - pointing out the contradictions. You do not let the model synthesise the information for you. Once you have done this you consult models, preferably a few, to see how your initial findings of your independently conducted analysis has conformed or differed from them. Your objective is to find out the truth, or as much of it as possible, not to conform your analysis to a model. (It is the role of theology to make the theory of evolution to conform to that of Genesis, it is not the business of scientists or social scientists.) It is quite possible your investigation has conformed what we already know or what theory tells us - great, it is nice to be reassured. But the chances are it does not. Once you have consolidated your thoughts as coherently as you can with a view to a policy recommendation, you are now ready to draw up a model. Along the lines of the the way you suggested re Piketty's global wealth tax I would imagine is the right way to go.

      I think we should watch out for gadgets as explanations. They are not. Although you have to make allowances to Lars Syll, as he is not a native English speaker (even if he is a brilliant linguist) I agree with his point;

      "Being able to model a “gadget world” — a world that somehow could be considered real or similar to the real world — is not the same as investigating the real world. Even though all theories are false, since they simplify, they may still possibly serve our pursuit of truth. But then they cannot be unrealistic or false in any way. The falsehood or unrealisticness has to be qualified."

      https://larspsyll.wordpress.com/

      I would go on to say that the qualification (call it the "other frictions) is probably the most important part of the analysis, more important than the basic model. In fact analysis should actually start there.

      The Keynes quote is also important in the link above. Steady states and returns to equilibrium are basically ahistorical concepts.

      ISLM (or worse) are gadgets. They do not really give us an understanding of the cause of problems like liquidity gluts, and therefore do not give us the solution. I have been looking at interwar period liquidity gluts, what was causing this was structural changes happening in the financial sector and industry - oligopolisation, risk averse behaviour by financial institutions and a lack of ability of small firms and households to access bank credit, and a lack of a need of large firms to access them (among a number of other causal factors). Once you know this, you are able to target your policy accurately. And to get there, you need to get out and look at the historical documentation - letters to banks, transcripts of central bank meetings, household balance sheets.....

      Gadgets are not the way someone can handle 2008, the Sterling Crisis or 1929. And ultimately this what we are aiming to do. Chris Sims said in his Nobel Prize Acceptance Speech that models failed in 2008 because of "exceptional events". How silly. It is precisely these events we are trying to in the first place avoid and then if we have to, deal with.

      In the example above, we get a fascinating insight into expectations formation. It also begs other questions - how was it that the financial sector managed to have such a huge influence over the government and its decisions regarding the PSBR? What sort of capitalist system is that and when did this begin to form (ref - Angus Maddison and what was beginnning to happen re capital market liberalisation in the latter half of the Bretton Woods Period.)

      Delete
    11. John D, your nonsense will not make any more sense no matter how many times you say it.

      Delete
  13. It seems like the gist is: building a macro model up from the behavior of individuals sounds great, but it's hard to do in practice. We do the best we can with the computational power, mathematical theorems, and economic intuition we have available. But, Smith and Delong make a good point about how this impacts Lucas' criticism of "old-Keynesian" models. If I write down a micro founded macro model and estimate my parameters and say my model passes the Lucas critique because my parameters are policy-invariant, but I'm missing some underlying mechanism such that, during one policy regime my parameters are estimated to be theta and during another policy regime my parameters are theta^\prime, then my parameters are not really policy-invariant. Perhaps my model presents a nice story, but my micro foundations don't necessarily improve my model's ability to make conditional forecasts about the macroeconomy. In some sense, I am still estimating aggregate relationships and assuming they will hold in the future.

    Since we cannot fully "micro found" a macro model as Lucas suggests (by aggregating up from subatomic particles), indeed we cannot even get close, how can Lucas be so critical of models that he judges to be less micro founded than his? Indeed his criticism is often on the grounds that they are less micro founded, not on the grounds that they don't fit macro data as well.

    ReplyDelete
    Replies
    1. "Since we cannot fully "micro found" a macro model as Lucas suggests (by aggregating up from subatomic particles), indeed we cannot even get close..."

      That's your assumption, and it's not correct. One can follow the Lucas program, do everything by the book, and still have a crappy model that is not structural, just because you have done the job badly. That doesn't mean you can't do it right, and it doesn't mean you should go back to IS-LM, or some such, which is not going to help under any circumstances.

      Delete
    2. But how do you know you have done it right? If two models fit the macro data equally well, then on what grounds do you argue that one model is better micro founded and less subject to the Lucas critique than the other?

      You misinterpreted my point. I am not arguing against the "Lucas program". I am simply saying that Lucas sometimes seems very critical of people essentially making the same point that he is making in the paragraph you quoted.

      Delete
    3. "If two models fit the macro data equally well, then on what grounds do you argue that one model is better micro founded and less subject to the Lucas critique than the other?"

      That's one of the key problems in macroeconomics, if not in economics in general. I've got an example of this that I like to show students, which involves showing how the data will look the same in two worlds: (i) frictionless world; (ii) world with frictions in which the policymaker is correcting the frictions in the right way. Sometimes two theories could be observationally equivalent in terms of the aggregate data, but there is some micro data that will allow you to distinguish between the two theories. But maybe the data is bad - which it often is in economics. It's a tough business.

      On Lucas, I don't think we should assume we know what is in his mind. Like anyone, his ideas change over time. He's not even a hardliner on his own critique. The money demand function has the same kind of problems vis a vis the Lucas critique as the Phillips curve does, but Lucas doesn't have a problem with estimating money demand functions.

      Delete
  14. " But the basic economic theory we work with is almost never used to make sense of the behavior of the individual. The one study I know about is John Rust's 1987 Econometrica paper."

    This may be the sense that this is the only model I can think of that is literally estimated on data from one individual. But there are a whole bunch of structural papers in labor and marketing that look at individual decisions and estimate models of rational behavior on this scale but using panel data. Some of them are very successful (and many of them are precisely about predicting consumer shopping behavior). A good example might be Misra and Nair: http://www.simon.rochester.edu/fac/misra/mkt_salesforce.pdf

    ReplyDelete
    Replies
    1. Yes, but what those people are doing is modeling average behavior. Labor economists do not take an individual, Joe Schmoe say, and try to explain what he is doing. They look at the behavior of many individuals - the more the better - and try to uncover regularities in average behavior, conditional on what they can observe.

      Delete
  15. Further to Anonymous 4.25 above

    The difference between a science and modern macro-economic theory is that a physicist has to work on the unknowable (the origins of the universe) which requires abstract theoretical musing. But if this contradicts anything that is knowable, it is immediately abandoned. Modern macro-carries on in spite of the evidence and despite a huge amount of knowable information which is available which cries out for primary investigation.

    Why? Well I think the Keynes quote here gives a convincing, if miserable explanation:

    "The strength of the self-adjusting school depends on its having behind it almost the whole body of organized economic thinking and doctrine of the last hundred years. This is a formidable power. It is the product of acute minds and has persuaded and convinced the great majority of the intelligent and disinterested persons who have studied it. It has vast prestige and a more far-reaching influence than is obvious. For it lies behind the education and the habitual modes of thought, not only of economists but of bankers and business men and civil servants and politicians of all parties …"

    https://larspsyll.wordpress.com/2015/03/29/did-krugman-get-it-right-this-time-read-my-lips-he-didnt/

    On another subject, as pointed out in a comment in the above link, Krugman's persistent flagging and defence of ISLM is because it continually gives better predictions than the alternatives - and this is the leg he stands on. But surely,that is not enough. Unless it gives an explanation of how we get liquidity traps, what caused them in the first place. it is absolutely useless as a device to get the right policy response. So what does Lucas mean when he says "go with something useful?" Useful for policy?

    ReplyDelete