Wednesday, July 25, 2012

Structure, Lucas Critique critiquing, etc.

There has been some blog conversation about microfoundations, the Lucas critique, etc., recently, so I thought I would add my two cents. Noah Smith wonders how you are supposed to tell when your model satisfies the Lucas critique, and when it doesn't. Good question. Here's a quote from an economist who has been a strong supporter of the Lucas Critique:
The sentence “My model is structural” as now used seems to me to be equivalent to “My model is good” or “My model is better than your model.”
We're trying to construct models that are useful for analyzing the effects of policies. But if that's our intention, then those models should give us the right answers. What's structural? That depends on the model, and the policy, and we know that the model can't be perfect. Our ability to estimate - or calibrate - parameters is limited, and we have to leave some things out, otherwise we have something that is too complicated to understand. So we have to accept that any economic model is going to be imperfect in its predictions about policy. This is not going to be like using Physics to put people on the moon.

But what's the big deal? Noah seems endlessly perturbed that economics is not like the natural sciences. There are no litmus tests that allow us to throw out bad theories so we can be done with them. But that makes economics fun. We have to be creative about using the available empirical evidence to reinforce our arguments. We have to be much more creative on the theoretical side than is the case in the natural sciences. We get to have interesting fights in public. Who could ask for more?

Here's another issue. Noah seems to think that the Lucas critique is a convenient bludgeon, designed for use by the right wing:
This is just one more judgment call in macro. Which means one more place where personal and political bias can creep in. In the comments on my earlier post, someone wrote: "I think of the Lucas Critique as a gun that only fires left." What that means is that in practice, the Lucas Critique is generally brought up as an objection to models in which the central bank can stabilize output. Or in other words, consensus has a well-known hard-money bias.
On the up side, this is certainly milder than some things I hear. Some people out there think that the entire post-1970 research program in macroeconomics is a right-wing bludgeon. As I point out in this post, the Lucas critique was as much a problem for the money demand function as for the Phillips curve, and the quantity theorists certainly were (are) not lefties. As well, I'm sure that Paul Krugman would be all in favor of structuralism if there existed a watertight theory of nominal wage and price rigidity. Then we would be spared stuff like this:
But what if you have an observed fact about the world — say, downward wage rigidity — that you can’t easily derive from first principles, but seems to be robust in practice? You might think that the right response is to operate on the provisional assumption that this relationship will continue to hold, rather than simply assume it away because it isn’t properly microfounded — and you’d be right, in my view.
Krugman understands that if he said that at a serious economics conference, or wrote it in a paper submitted to a serious economics journal, that he would probably be given a hard time - and that would be right, in my view. This is part of what we do to sift ideas. The theory has to be convincing.

But the perception that Lucas critique arguments are more often used against arguments favoring government intervention than against those which do not is probably correct. Economic arguments that justify no government intervention are often easy to construct and understand. Justifying intervention is hard. Try to actually define what an externality is, or to construct a Keynesian model with all the proper working parts. Those things are difficult.

35 comments:

  1. "This is not going to be like using Physics to put people on the moon"

    Having studied theoretical physics I have serious doubts here. Perhaps you misunderstand what physics is about, what laws in physics actually mean and how they all fit together.

    "Justifying intervention is hard"

    On economic grounds. Because it is all politics.

    ReplyDelete
  2. "This is not going to be like using Physics to put people on the moon"

    Let me try in another way. Physics tries to do science in the sense of Popper, while economics - in the sense of Kuhn. While I generally side with Kuhn rather than Popper, the consequence of such approach is that scientific progress in economics can go to any arbitrary end-point. In other words in the sense of Kuhn mainstream economics advocates for the scientific push-progress. That is if you push a cart your direction is generally random. Physics, on the other hand, starts with a pull progress. It goes from one theory to another and never looks back. But it also always assumes that current dominant theory can be just plain wrong. Nothing like this seems to me in mainstream or any other branch economics.

    ReplyDelete
  3. Not sure what your serious doubts are. If I'm putting someone on the moon, that's mostly Newtonian physics. I know the gravitational forces exerted by the moon and the earth on the space vehicle, and I work out how much force I have to apply to the vehicle and in what direction, to get it out of the earth's orbit, into the moon's orbit, and to land the thing on the moon. Someone tells me that they want to use monetary policy so that 10 million more people are employed in the United States. How do I figure out what they should do? How do I know this is even feasible? Forget the politics. This is a difficult job for a scientist.

    ReplyDelete
    Replies
    1. Hm. Mostly? Surely if you want to build an internal combustion engine it is mostly classical physics. But why do you chose putting a man to a Moon as a comparable task in physics to employing 10m people via monetary policy in economics? That is why I have doubts that you understand what physics is about. Putting man on the moon is like micro-physics assuming that microeconomics has everything right.

      And why do you, as a SCIENTIST, a priori limit yourself to just monetary policy? Isn't it a purely political position? Are you really talking science?

      Delete
    2. "Are you really talking science?"

      Yes.

      Delete
    3. I'm using the person on the moon because it's an exercise in applied science where we have demonstrated that we can be pretty accurate. It's hard to show you in a blog why this is the case, but I'm certain we'll never do that with economic policy. That's not the fault of the economists, it's the nature of what we have to work with. Maybe, as you point out, the politicians get in the way, but those people can also be funny about climatology and evolutionary theory.

      Delete
    4. "Yes."

      Really? So you say you never look back? Why do think that monetary policy is the only tool? Your position as a scientist is less than short-sighted. I do not think you are doing science but rather multiplying your beliefs. Thank you but I have enough of that stuff on tv every day in and out.

      Delete
    5. Also, look at the technology they needed to put men on the moon. I don't know if you've been to Johnson Space Center in Houston, TX, but mission control from the late 60s was pretty low tech. The maths they needed to put men on the moon were built on stable relationships, while economics, even if we have accurate models (haha), requires that we be able to demonstrate their accuracy and access and process all of the relevant data /right now/, because certain relationships (MPC, fiscal policy environment, tech level, etc.) can change before the policy can be put in place. You wouldn't, for instance, talk about a "gravity shock" in physics the same way you'd talk about tech bubbles and shocks in economics.

      Delete
  4. "It goes from one theory to another and never looks back."

    "But it also always assumes that current dominant theory can be just plain wrong."

    Actually, we do a little better than that in economics. We know that the theory is wrong, not that just that it CAN be wrong. Nevertheless, theories that are wrong, in the sense that they can be rejected if you collect enough data, can be useful. As a result, we never throw anything out. We're pack rats. We can go backward and find a germ of a useful idea in stuff written 100 or 200 years ago.

    ReplyDelete
    Replies
    1. "We can go backward and find a germ of a useful idea in stuff written 100 or 200 years ago."

      Good. But can you please tell me what has been rejected once and forever as just plain wrong?

      Delete
    2. Everything. It's all plain wrong. That's the point. Being plain wrong does not mean it's not useful. The whole idea is to figure out what is useful and what isn't. On some things economists more or less agree. On others we don't. That doesn't make us idiots.

      Delete
    3. Idiots - I did not say it. Coming from natural sciences I do not understand it. Maybe it is really some special type of fetish... It is just a disaster that we all have to suffer by-side effects from this fetish of a "closed circles of limited people" as our saying goes.

      Delete
  5. "The reason that we can criticize modern macroeconomic models is that they are explicit about their assumptions about fundamentals such as preferences, technologies and resource constraints.

    Of course, being precise means making simplifying assumptions that are often questionable – there are many elements of modern macro models that I question.

    However, replacing them with models based on vague intuitive ideas is not an improvement."

    -Valerie Ramey, from her comments on DeLong/Summers

    ReplyDelete
    Replies
    1. Here's the whole discussion, if others are interested:

      http://weber.ucsd.edu/~vramey/research/Discussion_of_DeLong_Summers.pdf

      Delete
  6. But the perception that Lucas critique arguments are more often used against arguments favoring government intervention than against those which do not is probably correct.

    Yes, this is all I was saying. I do not think the Lucas Critique was designed for use as a right-wing bludgeon, I just think that's the way it turned out in practice. That's why I said "in practice" in my post...

    But even if the Lucas Critique had been designed as a right-wing bludgeon, that does not in any way diminish its correctness. I believe in assessing the strength of ideas on their own merits (and I occasionally even manage to live up to this belief)...I don't think anyone can deny that the Lucas Critique is right.

    ReplyDelete
    Replies
    1. I don't think anyone has argued that it's not correct. Lucas lays it out and gives you an example. People have argued about how much it matters, though. Sims in particular I think, who wanted to argue that you could make policy inferences from VARs.

      Delete
  7. Dear Professor Williamson,

    I am writing frm Taiwan. Nice post. Over the last few months, I have found that your positions on macro are becoming more and more reasonable to me.
    Still, Lucas's own models are not immune to Lucas Critique. Rational Expectations and RBC are not convincing at all. Axiomatic consistency does not imply empirical consistency. Even the whole idea of utility is problematic. Realistic microfoundations require a deeper understanding of aggregation of human decision-making (e.g, neuroeconomics, institutions, social psychology etc). Macro and micro, as they currently stand, are not science as Noah Smith has argued, and sadly will not be science decades from now.

    ReplyDelete
    Replies
    1. Try coming up with a useful model instead of pronouncing something is or is not "science" (a truly pointless exercise if ever there were one), and we might pay attention to your ideas. As it stands, your thoughts merit less attention than my dog's.

      Delete
  8. This "debate" on the Lucas critique is weird. Read the famous Sargent-Lucas 78 'after keynesian economics', and it's pretty clear that whether the Lucas critique matters is an empirical question. That is, do we see actually parameters shifting following a policy change? As you say, this is a question that depends on the model and the policy change. No model is perfectly structural. Or if you prefer, no structural model is perfect. But there were a series of papers in the 1980s (incl an AER by Sims) that tried to test for the importance of the Lucas critique. It wasn't terribly conclusive as I remember.

    ReplyDelete
  9. If the natural sciences were as "fun" as the social sciences, it wouldn't be very fun for humanity.

    "Still, Lucas's own models are not immune to Lucas Critique. Rational Expectations and RBC are not convincing at all"
    The Lucas Critique is not about whether something is "convincing" but whether it's invariant to policy changes. Rational expectations isn't a model, but a component that goes into models. RBC (more creditable to Prescott than Lucas) was one of the first attempts to satisfy the Lucas critique, but merely satisfying that doesn't necessarily make it a correct (sorry, useful) model of reality.

    ReplyDelete
  10. Further, RBC was not set up as a model to think about policy, though of course you can extend it (as people have) to do the job.

    ReplyDelete
  11. Actually, I have repeatedly disagreed with the view that Popperian falsification is what physics does. Even while Newton was alive, observations had uncovered discrepancies between the orbits of planets and the theory's prediction. But the successes of the theory in other aspects were so many, that people were willing to ignore its falures, at least until a new theory with more explanatory power came along. As Kuhn writes:
    "No theory ever solves all the puzzles with which it is confronted at a given time; nor are the solutions already achieved often perfect. On the contrary, it is just the incompleteness and imperfection of the existing data-theory fit that, at any given time, define many of the puzzles that characterize normal science. If any and every failure to fit were ground for theory rejection, all theories ought to be rejected at all times."

    And hysists Sokal and Bricmont seem to agree with this view when they write:
    "When a theory successfully withstands an attempt at falsification, a scientist will, quite naturally, consider the theory to be partially confirmed and will accord it a greater likelihood or a higher subjective probability. ... the history of science teaches us that scientific theories come to be accepted above all because of their successes."

    I know that Noah does not like judgments calls, neither do I. But if they are unavoidable in physics, it is reasonable to expect that they are even less avoidable in economics.

    ReplyDelete
  12. CA, I do not think it is choice between Popper and Kuhn. I consider the problem of economics as pretending doing popperian pull-progress while in reality doing kuhnian push-progress. This is a contradiction which physics and other natural sciences clearly do not have. And it is pretty much this contradiction which is challenged consciously or unconsciously by outsiders.

    ReplyDelete
  13. I agree that there is no choice. Kuhn was writing not about economics, but about the natural sciences as a historian (how things have been working), and not a philosopher (how things should work). He showed that natural sciences "progress" not according to the principle of falsification, but in fact in a much more subjective fashion where argumentation is important. And physisists Sokal and Bricmont agree. So it seems to me that no science progresses in Popperian fashion, but at least economists are honest about it.

    ReplyDelete
  14. Your last statement I would never ever ascribe to economists. The fact that there exist at the same time many competing and sometimes mutually exclusive branches of economics tells me that economists are too arrogant to be honest about the type of progress they "practice". To an outsider all these debates look like all and each of them already found an ultimate truth which others are just too stupid to get. Alternatively lets stop calling it science and then the topic is closed. But they do not do it either.

    ReplyDelete
  15. So how many versions of string theory (not to mention all the other theories) are there in physics? And how does that lead to a conclusion of arrogance? Look, I understand that a camel can only see the other one's hump and not its own. But seriously, you have to do be able to do better than that.

    ReplyDelete
  16. String theory as well as quantum theory of gravity are just contenders in their attempts to explain everything. They openly admit it. I.e. that they are just attempts.

    But then they are purely theoretical exercises. Whatever they dream about, the outcomes of that dreaming for all practical purposes are irrelevant for our world. That is not something one can say about economists.

    ReplyDelete
  17. Actually you are wrong! Many economists would argue the same about their research. What motivates them is the desire to solve a puzzle, even if they hope that the solution may end up having practical relevance after all. But I am sure string theorists feel the same way (or else why should we keep funding their research?) The sample of economists you base your assertions on is biased!

    ReplyDelete
    Replies
    1. My problem is not with economists. It is about economics. It is ok to do nuclear physics. It is NOT ok to allow germans to crucify greeks. And the list is virtually endless. My sample is biased but only because the whole world seems to be falling apart on existing economic ideology of unlimited needs. If just 5% of population is enough to feed everybody, what should we really do with the other 95%? We do not really need them. That is my problem.

      Delete
    2. Well, I am Greek, and I certainly do not blame economics for what is happening in Greece. In fact, I blame the lack of economic education among the Greeks, and to a lesser extent among the Germans.

      Delete
    3. And human wants go beyond feeding. You really should take some economics classes before you start criticizing it.

      Delete
    4. "If just 5% of population is enough to feed everybody, what should we really do with the other 95%? We do not really need them. That is my problem."

      That says more about your failure to understand economics than it does about actual economists. Heed CA's advice before you make yourself appear even more of a fool than you have already managed.

      Delete
  18. Finally, because I have better things to do, an indicative sample of the many contributions of economic theory that have made a difference in people's lives can be found in the book "Better Living Through Economics" by John Siegfried. Summary here:
    http://www.hup.harvard.edu/catalog.php?isbn=9780674036185
    Unfortunately, some economic background is necessary to make the most from the book.

    ReplyDelete
  19. Why does Krugman thinks arbitrary introducing downward nominal wage rigidity help anyway? Even assuming wage rigidity, if firms can vary the hours-worked of existing employees and / or hire new employees, with no existing contractual arrangement, at any given wage rate then all the employment fluctuations should be taking place on the intensive margin. So what exactly is Krugman arguing? Is he arguing that hours are somehow sticky? Does he believe that there is a technological restriction on firms hiring new employees? Does he believe that there is downward nominal wage rigidity for new employees, even though no wage as been set? Does he believe the reservation wages of the currently unemployed are so high that there is no wage an employer could offer that they would accept? None of those seem particularly appealing assumptions so, no, arbitrarily introducing downward wage rigidity is not helpful. If you think downward wage rigidity is important, you are going to have to put in the effort to show why and how it matters - and that can only be done with micro foundations.

    Also, mere observation of something doesn't mean it matters much. That's what a model is for. Obviously menu costs quite literally exist (like reprinting menus!), but without a model it's hard to know if it matters. There are plenty of ways you could have menu costs and have them not matter much, like in the Golosov and Lucas paper from a few years back.

    ReplyDelete
    Replies
    1. I like this paper:

      http://ideas.repec.org/p/nbr/nberwo/17520.html

      It's a model with price dispersion. What "clears markets" is the whole distribution of prices. So in a dynamic setting, you can have shocks that move the distribution of prices around, but from period to period only a few producers actually have to change their prices. So prices can appear to be sticky, but in fact money is neutral in the model. Thus, what you think you are seeing actually tells you nothing about the results of the policy experiment.

      Delete