Tuesday, July 27, 2010

Solow in Grumpy-Old-Man Mode

Here is a report on hearings for the House Committee on Science and Technology. For mainstream economists who despair about the state of the profession and the public's misunderstanding of what we do, this will not help your mood.

The witness panel is basically 4 quacks and Chari. The witnesses appear to have been asked to give testimony on the "DSGE model" which the committee seems to view as representative of what a typical macroeconomist is up to. Robert Solow's testimony is representative of what is in the other testimony, though Page mainly tries to sell his ideas on complexity theory and does not do much criticizing. Solow has plenty of bad things to say about macroeconomics, but comes off as totally cluless. Here is how he starts out:
Here we are, still near the bottom of a deep and prolonged recession, with the immediate future uncertain, desperately short of jobs, and the approach to macroeconomics that dominates serious thinking, certainly in our elite universities and in many central banks and other influential policy circles, seems to have absolutely nothing to say about the problem. Not only does it offer no guidance or insight, it really seems to have nothing useful to say.
Well unfortunately, it's Solow who has nothing to say. Here is his main problem:
I do not think that the currently popular DSGE models pass the smell test. They take it for granted that the whole economy can be thought about as if it were a single, consistent person or dynasty carrying out a rationally designed, long-term plan, occasionally disturbed by unexpected shocks, but adapting to them in a rational, consistent way. I do not think that this picture passes the smell test. The protagonists of this idea make a claim to respectability by asserting that it is founded on what we know about microeconomic behavior, but I think that this claim is generally phony. The advocates no doubt believe what they say, but they seem to have stopped sniffing or to have lost their sense of smell altogether.
I should get Solow to sniff my models before I write up the papers. First, as Chari points out in his testimony, the majority of work in macro these days is done in heterogeneous agent models. Representative-agent is a useful baseline model in what we teach, but most serious research departs from that. Clearly Solow hasn't been paying attention. Next we have this:
An obvious example is that the DSGE story has no real room for unemployment of the kind we see most of the time, and especially now: unemployment that is pure waste. There are competent workers, willing to work at the prevailing wage or even a bit less, but the potential job is stymied by a market failure. The economy is unable to organize a win-win situation that is apparently there for the taking. This sort of outcome is incompatible with the notion that the economy is in rational pursuit of an intelligible goal. The only way that DSGE and related models can cope with unemployment is to make it somehow voluntary, a choice of current leisure or a desire to retain some kind of flexibility for the future or something like that. But this is exactly the sort of explanation that does not pass the smell test.
Now we know that Solow is really out to lunch. For more than thirty years, economists have been working on general equilibrium models of search and matching (including the Mortensen-Pissarides framework that Chari mentions) that have been used extensively to help us understand how labor markets work, and the role for policy in intervening in those markets. Then:
The point I am making is that the DSGE model has nothing useful to say about anti-recession policy because it has built into its essentially implausible assumptions the “conclusion” that there is nothing for macroeconomic policy to do.
Solow should take Chari's cue and attend the SED meetings. He would see a lot of papers where models are constructed in which there are financial and search frictions, and where there are important things that policy can do.

Question: Who chose these witnesses? Why can't the House Committee choose witnesses who are actually serious macroeconomists engaged in research and policymaking? The committee could have talked to, for example: Kiyotaki, Gertler, Woodford, Lucas, or Sargent, and actually learned something useful.


  1. I should get Solow to sniff my models before I write up the papers.


  2. I think you're being a bit hard on Solow, but more importantly rather disingenuous about "the DSGE model." Sure there are hetero-agent models, but how many of them are in Gali's textbook (0), or Woodford's, or Walsh's? How many papers prior to Blanchard-Gali (wp version '06) combined the "workhorse NK model" with Mortensen-Pissarides? Christiano (and co-author(s) I can't remember offhand) have a DSGE model with "real-life" unemployment, but how much work was being done in that area prior to the current recession? Ditto financial frictions (apart from Bernanke-Gertler-Gilchrist)?

    My prior is that when you say "workhorse DSGE model," the vast majority of the macro profession has in mind exactly the sort of model that Solow's criticizing.


  3. 1. No, I'm not being hard enough on the guy.
    2. DSGE is "dynamic stochastic general equilibrium," which covers virtually everything we do in macro. I think what Solow is criticizing (it's hard to tell; he doesn't say) is Kydland-Prescott (1982). New Keynesians appropriated the name DSGE to apply to a class of sticky-wage and sticky-price models.
    3. DSGE as in "dynamic stochastic general equilibrium" models with "real-life" unemployment, i.e. people in the model who would answer the appropriate BLS question with a "yes":

    Andolfatto (AER, 1995)
    Shimer (AER, 2005)

    and there are many more.

    3. Financial frictions: See my 1987 JPE paper. We have been doing this stuff for a long time.

    Get with the program, Pete!

  4. It seems to be clear that you can say nothing about involuntary unemployment if you assume that such a thing dose not exist, and it seems equally obvious that government intervention is not required if it is assumed that without government intervention everything is just fine.

    Those talking about heterogeneity in DSGE models seem not to have grasped the message of the Sonnenschein-Mantel-Debreu theorem: With enough heterogeneity, anything can happen, and individual rationality does not carry over to aggregates even if you have only two households (Hicks 1956).

    Williamson in clueless young-man-mode.

  5. "It seems to be clear that you can say nothing about involuntary unemployment if you assume that such a thing dose not exist"

    What is involuntary unemployment? We observe someone spending time searching for work. Why do they search? Because they prefer working to not working. That is exactly what is going on in a search model.

    "With enough heterogeneity, anything can happen, and individual rationality does not carry over to aggregates even if you have only two households (Hicks 1956)."

    So, what's your point? All models are useless?

    "Williamson in clueless young-man-mode."

    Thanks for the complement. At 55, the census thinks I'm a senior citizen.

  6. So you kept young mentally. Good to know.

    Involuntary unemployment: There are several definitions found in the literature. None precludes search, and none refers to mental states. Have a look at Keynes' definitions. (I erred about your age because you seemed unaware of that.)

    There are other (different) definitions, such as "involuntary unemployment" referring to a state where workers don't find a job at the going wage rate, that is, the wage paid to similar employed workers. Definitions depend on the purpose, of course, but most seem to relate to uncleared markets, in some sense. Uncleared markets do not preclude search. (Let me add that the concept of uncleared markets does relate to different equilibrium concepts. A Walrasian equilibrium may differ from a Marshallian equilibrium or a Swedish (ex ante-ex post) equilibrium, or a game-theoretic equilibrium, and so this is a difficult concept. I cannot go into this, but I see no awareness among DSGE people either.)

    Regarding your question about heterogeneity: I guess I said that many claims made on behalf of DSGE models are plainly misleading. This does not imply that the models are useless. Some people obviously enjoy dealing with them, even if they are unable to defend them. The TINA argument (that there is no better alternative) looks like defending astrology because there is no better way to foretell the future.

    Saying that DSGE models don't convince me does not mean that all macro models are worse. The largely neglected endogenous business cycle models to be found in older books (which you may have not read, due to your age) may offer quite a number of insights...

    Anyway, keep young. And, given your real age, you may consider not calling people like Robert Solow, Sidney Winter, David Colander and Scott Page "quacks."

  7. 1. I think the view of most mainstream economists these days is that the concept of "involuntary unemployment" is not useful. Our measure of unemployment is the number of people who say yes to a particular question asked in the BLS household survey. These are people who are not employed, and who are actively searching for work. In an Econ 101 sticky-wage model, if the wage is above the market-clearing level, what would the people who can't find work at the going wage tell the BLS. In what sense are they "actively seeking" work? They certainly are not spending any time doing it.
    2. In Minnesota they like to say "it takes a model to beat a model." What's yours?
    3. Quacks are quacks, old or young.

  8. Steve,

    1) if by "the program," you mean "DSGE modeling," I think I'm already pretty "with it", thanks. I've even taught your book ;-)

    2) Solow may well have in mind the original RBC-type model as his main target of criticism, but I think much of what he says applies equally well to the sticky-price NK version too.

    My main argument was that issues like hetero agents, unemployment in the BLS sense, and credit/financial frictions have not been major areas in DSGE modeling (in the NK sense) until recently. I certainly didn't mean that nobody's working on them.

    Here's some data from the indexes of the books by Gali, Woodford, and Walsh (3rd ed.):
    The phrase "heterogeneous agent(s)" appears exactly 0 times.

    The word "unemployment" does not appear in Woodford's index. The references in Walsh lead to discussions of a vertical long-run PC, the idea that policy may respond to unemployment, and a discussion of inflation bias with persistent unemployment. Nothing with BLS-style unemployment.

    Gali mentions unemployment in his "extensions," and has this to say (p. 188): "The framework analyzed in this book does not incorporate unemployment explicitly, and, hence, it is silent about the the determinants of its level and fluctuations, or its potential role in the design of monetary policy."

    "financial frictions" or "credit frictions" are also treated in Gali's extensions. Walsh has added some material to his chapter 11 since 2e.

    "Unemployment" similarly appears nowhere in the papers by Smets & Wouters (either JEEA or AER), Christiano et al JPE '05, ACEL (version of 12/18/04), or Woodford's AEJ: Macro "New synthesis" paper.

    Back to hetero agents, I searched for "heterogeneous agent" on ECONLit, limiting results to those after 2000 (arbitrarily). I got 576 hits, only 200 of which I could search explicitly. Of those, 95 appeared in what I'd call "leading" macro or general journals (18 journals, listed below). That works out to just over half an article per journal per year, which doesn't strike me as a lot. Sixteen appeared in 3 special issues. Further, all 95 were published in 2006 or later.

    To be fair, I haven't tried to count how many representative agent, no-BLS-unemployment, frictionless finance papers have appeared over a similar time frame, but my prior is that the rate would be substantially higher.

    3) To clarify, I think there's a great deal of useful information to be gained from DSGE models, even of the RBC, representative agent, no-BLS-unemployment, etc. variety. I also think it's important to acknowledge that criticisms such as Solow's do in fact apply, to a greater or lesser extent, to a wide range of them.


    PS -- here's my list of journals: AER, AEJ (including one in Micro), JME, JMCB, JPE, EJ, JEDC, IER, Econometrica, Macro Dynamics, RED, RESTud, QJE, JET, Berkeley Macro, EER, JIE, JEconometrics. I made no attempt to filter the hits more finely, eg by separating "micro" from "macro" from "finance" papers.

  9. Pete,

    I think we are more or less on the same page now. You clearly understand that "DSGE" these days usually refers to an NK model. However there is nothing in Solow's statement to indicate that he understands that.

    "My main argument was that issues like hetero agents, unemployment in the BLS sense, and credit/financial frictions have not been major areas in DSGE modeling (in the NK sense) until recently."

    Yes, exactly. A legitimate criticism of the profession was that a lot of people, inside and outside central banks, focused their attention on NK models. The baseline NK model focuses exclusively on the sticky price friction, at the expense of everything else. Some larger NK models (Smets-Wouters, Christiano-Eichenbaum-Evans) throw in everything but the kitchen sink, and ultimately seem to me no more than data description. It's certainly not my fault that people neglected monetary and financial frictions for years - my work and the work of people I talk to is all about monetary and credit frictions, banking, etc.

  10. I have one question (just to avoid a type of previous confusion)

    Do DSGE models assume "loanable funds"?

  11. I assume by "loanable funds" you mean some kind of role for credit, whether through an explicit banking sector or otherwise.

    1. If by DSGE you mean "New Keynesian model," (as some New Keynesians have in mind), then the answer is "to a limited degree." Typical New Keynesian sticky price models don't have a role for credit, but there are recent attempts to include this - Curdia and Woodford's recent work, for example. I think this is somewhat crude, but that's what they have to offer.

    2. More broadly, there is a wide array of dynamic stochastic general equilibrium models that include credit arrangements in interesting ways. My 1980s work, work by Bernanke/Gertler/Gilchrist, Kiyotaki and Moore have credit and financial intermediaries. My survey work on New Monetarist Economics with Randy Wright shows how to put banking arrangements into monetary models of search and matching.

  12. No, I mean mainstream loanable funds theory where market clears via interest rates and government is part of it.

  13. I have read some about the "Rational Being" of macro models, but, in my modest opinion, may be this lack of the "social networking" associated to the "real" human beings

    I follow those that consider the human being more a "social being" than a rational one, and this has a big impact in the way the economy works

    I am not a professional economist, I am an engineer, I have been woking for some multinational companies, and there is certainly some let's say "connections" between them regarding to decisions that, is some cases, affect the market and prices, because "it is very difficult to sustain huge investments, research and employ a lot of people with too much uncertainties"
    This is only one part of the kind of realtionship that people build-up to take decisions and "control and predict their lives"

    A "rational being" that do not build social networks of mutual interest may be is a rational being, but not a human one

    It this included in antway in the DSGE models?

  14. Sorry, I have some more questions:

    In the real work most of the economist are paid by institutions or groups that are in connections with financial institutions and governments

    In connection with the fail of professional economists, as a group (not talking about exceptions), to detect the huge world economic crisis. Could be the "proximity" to the financial, economic and political power the ultimate reason because they fail to detect this dramatic crisis even using so much advancel tools like the DSGE models?
    (The money is confidence and everybody knows that economist are not here to make people lose money)

    How a theory could be proctect itself to the "wishful thinking" of the people that use it? Does this proof that economy is not a "real" science?

  15. DFC and Ekkehart are serious kooks.

    Steve, I can't believe you put up with these people.

  16. The real economy is dead. Take a look at the recent small business optimsim index which declined again in July. Small businesses cannot offset weakness in the US with international sales. Get ready for QE 2.

  17. "The real economy is dead" and mainstream economists are day-dreaming about their out-of-this-world models, while being well "paid by institutions or groups that are in connections with financial institutions and governments".

    No need to pay any attention to what you say, guys.