Site Meter

Monday, April 16, 2012

Pulled back from comments again

First anonymous is Justin Dylan. I was totally unfair of me to complain that he posted as anonymous as that is the default given my blogger settings. Before, I admitted the problem is my ilhtmliteracy.

Second, I have no idea why I had a ragegasm. I do fairly often. I use the blog to blow off steam. It makes me less unpleasant to be around. It is better to type out rage than to speak (or more likely) shout it out. No one has to walk away to not read what I type, while people do walk away to not hear what I say (and sometimes I follow them).

I am fairly calm now. I can briefly restate my problems with micro foundations. Then I will reply in detail.

1. I agree that the rational expectations assumption doesn't have strong implications. I think (and have frequently written and believe I have rigorously proven) that it has no observable implications at all. That's why I try to avoid the phrase "rational expectations hypothesis." I think the word "hypothesis" should be reserved for testable hypotheses, and I don't think there is any testable hypothesis there.

2. Some micro founded work has improved economic theory. But I think it has just undone some of the damage done by other micro founded work. Given my first claim, I think it is possible to get any micro founded result one wants. If people think that economic theory tells us something, they are wrong, that is, any implication of any effort in economic theory can only be something which might be true and we have to look at the evidence to see that if it is true. I don't think it can lead us. But I am sure that it has lead many people astray. Proving case by case that the core assumptions of economic theory don't imply this that or the other thing is therefore useful. But the same useful goal could be achieved (if I am right) by general acceptance of the idea that the core assumptions of economic theory have no implications at all.

3. I suspect the sincerity of people who provide micro foundations for Keynesian assertions (more generally for assertions that some people foolishly think are inconsistent with the core assumptions of economic theory). I suspect that I am not the only one who starts with the conclusions and works back to the assumptions required to reconcile those conclusions with rational utility maximization. In particular, I have fairly strong suspicions about Stiglitz and the late R Dornbush. I fear that it is very costly to admit that one does this (I sometimes wish I hadn't publicly admitted it here on this blog). I don't see the use of starting with the conclusion then finding the right assumptions (this helps explain my extremely low research output).

4. The Lucas critique of Friedman's methodology of positive economics. There is a monstrous equivocation at the base of the effort to find micro foundations. On the one hand, it is argued that it would be good to understand the true psychology, sociology, technology etc which causes economies to behave as they do. The advantages of knowing this are argued at length, for example by Lucas in "Econometric Policy Evaluation: A Critique" but no one has ever disdained understanding the causes of things or knowing the truth.

The problem is that economists also argue that it doesn't matter that the micro assumptions in micro founded models are false. When acutal micro foundations are considered, suddenly the exactly opposite argument is made: that models may be useful even if they aren't true, that the economy might act as if the false assumptions about psychology were true etc.

Each argument is valid as far as it goes, but the research program relies on both of them at once. Some clearly false assumptions must be made, because if they were true and ignored models which fit the historical data would not be useful for policy analysis.

The first (Lucas critique) is valid in that, other things equal, it would be better to have approximately accurate micro foundations. Other things are not equal and micro founded models rejected by the data are preferred to ad hoc models which aren't rejected by the data because uh well I don't know why.

The second is valid because "might" makes right. The approach of making false assumptions and finding their implications might lead us to useful policy conclusions, because anything at all might work.

OK now responding



Robert,
It certainly wasn't my intention to upset you, let alone enrage you, so I apologize for that. I merely meant to point out that dismissing the microfoundations literature, as I thought you were doing, seems an unreasonable position to me.

That is what I was doing. Also don't apologize. I get enraged regularly for no good reason.

I'm beginning to see that people understand "microfoundations" in different ways, and it appears that the New Classical economists may have given the idea a bad name.

I was not writing only about New Classicals. My disdain was aimed at new Keynesians. Also at my own work.

But I agree strongly with Stiglitz (http://onlinelibrary.wiley.com/doi/10.1111/j.1542-4774.2011.01030.x/pdf) that it is "important that macroeconomics be based on the right microeconomic assumptions, those consistent with actual behavior, taking into account information asymmetries and market imperfections".

But Stiglitz considers only the deviations from economics 101 (first semester) that he likes best. What about irrationality ? A model with rational agents information asymmetries and market imperfections is absolutely rejected by all available data. Assumptions that are wrong in fewer ways than new Classicals are is not the same "the right microeconomic assumptions". Stiglitz asserts that the world really is in Nash equilibrium. He is way too smart to believe this, but he sees only the points where he disagrees with the new Classicals and ignores the points where he chooses not to argue with them.

The use of representative agent models with perfect competition, complete markets, perfect information, instantaneous market clearing etc etc, as has become standard, makes it more or less impossible to address any worthwhile macroeconomic questions. But it is that particular, and somewhat perverse, choice of microfoundations which is at fault and not the idea of microfoundations in general.

That's not what I was talking about. We agree that RBC models are silly, but I was criticizing modern economic theory much more generally.

I also agree with Stiglitz that the rational expectations hypothesis (REH) is not necessarily always the problem. Once you move away from the New Classical (or most DSGE) models, REH no longer has the same strong implications that it has in these contexts.

I think it obviously has no implications and is not, therefore, properly speaking, a hypothesis.

Macroeconomics has probably lost 20 or 30 years, but interesting work is now being done, by Roger Guesnerie, Roman Frydman and others, both in reevaluating REH and assessing its implications (see for instance, http://www.econ.nyu.edu/user/frydmanr/RFandESPIntroductoryEssay_WhichWayForward_Revised3_16_2012final.pdf ). Hopefully much more will follow.

I will have to look at that to comment.

Finally, I would quibble with you on what is probably a semantic point: I don't think the issue is to consider "the possibility that we are not Bayesian" as you put it, but rather to recognize that Bayesian arithmetic only in applies in Savage's "small worlds" in which we can know what all possible states of the world are (or every conceivable possible course of future events), and assign probabilities to them. In a "small world" new knowledge never leads us to reconsider the basic model we use in determining our beliefs, or feel the need to alter our model of the world after being surprised by a chain of events whose implications had not previously been considered (all of this paraphrases Binmore).

This just shows that Savage's work has been abused by people who ignore the qualifications he made. I didn't know this (although I should have guessed).


The New Classical model describes just such a "small world" in which we all agree on the relevant model of the economy and in which true "surprises" can never occur. But neither can most of the macroeconomic phenomena we are interested in, which I guess is why Robert Lucas finds it all so surprising.

well I agree with that.

Thanks very much for a stimulating discussion.

Respectfully,
Justin Dylan

Thanks for the thoughtful comment from which I learned.

No comments: