Friday, June 10, 2011

Sometimes when non-Austrians cite Hayek I think Hayekians can get the wrong impression of what they're saying...

...A good case in point is Ricardo Caballero. Now I've been critical of him insofar as I think he's employed this classic technique of dumbing down the people he's criticizing to make his criticism seem more relevant. But as far as what he's actually claiming we should be doing, I think I'm pretty much in agreement with him. And I'm not sure a lot of Hayekians would be even though they were excited he talked about Hayek in his recent Journal of Economic Perspectives paper.

He was recently interviewed by the Minneapolis Fed. Here he is on the proper use of modeling:

"I’m about to teach a course in which I will, in the introduction, talk briefly about this methodological issue. But I still need to teach the basic models. That won’t change. In fact, I think it is very important to clarify that I am not antimodel. On the contrary, the economy is so complex that there is little hope of understanding much without models. I just don’t want these models to acquire a life that is independent from the purpose they are ultimately designed to serve, which is to understand the functioning of real economies.

The critique part of the paper you refer to argued that the current core of macroeconomics has become so mesmerized with its own internal logic that it begins to confuse the precision it has achieved about its own world with the precision it has about the real one.

There is absolutely nothing wrong with building stylized structures as just one more tool to understand a piece of the complex problem. My problems with this start when these structures take on a life on their own, and researchers choose to “take the model seriously”—a statement that signals the time to leave a seminar, for it is always followed by a sequence of naïve and surreal claims."

A lot of people criticize economic models for being an imperfect representation of the economy - for not being "true", and they act condescendingly towards the simpletons that would employ models. The fact is this is a fundamental misunderstanding of why we model, which I think Caballero expresses brilliantly here. We don't model to reproduce reality - that is impossible. Reality is too complex. We model to get a sense of the dynamics of specific processes or mechanisms that we think are important. You can't just talk your way through the implications of most of these processes. It's too complex. Math helps you coordinate a lot of moving parts that prose could never coordinate. But you always have to keep in mind that you're just trying to understand the processes at work in the consideration of multiple interacting variables, and this is a tool for doing that. You're not reproducing reality. Science isn't about uncovering "truth" with a capital T - it's about uncovering useful knowledge.

He goes through uncertainty issues too (he calls it "Knightian uncertianty"). He provides an interesting discussion of complexity and presents and good case for why complexity makes inference from microfoundations to macrodynamics hard:

"The economy is an incredibly complex object—and I mean “complex” in the sense of very hard to understand. This complexity is not something we can just get rid of in the process of writing simple models, for it is central to economic behavior during crises. My work with Alp is an attempt to capture a small part of this complexity problem and its role during financial crises.

The basic idea is that the economy is a very complicated network of connections, but most of the time economic agents can go about their daily activities without worrying about those complications. To succeed, you—or financial institutions in our model—just need to be good at understanding your local environment.

However, as crises cross a certain threshold, all of a sudden it is no longer enough to understand the local environment. You begin to worry about indirect hits through the network. So what was a relatively simple optimization problem quickly becomes an immensely complex one. At that point, we have moved from a world of more or less well-defined risks to one of (Knightian) uncertainty and, as we discussed earlier, decision makers then become ultraconservative. And the most attractive individual decision is simply to withdraw

He makes the skepticism about microfoundations more explicit here:

"The quantitative implications of this core approach, which are built on supposedly “micro-founded” calibrations of key parameters, are definitely on the surreal side. Take, for example, the preferred “micro-foundation” of the supply of capital in the workhorse models of the core approach. A key parameter to calibrate in these models is the intertemporal substitution elasticity of a representative agent, which is to be estimated from micro-data. A whole literature develops around this estimation, which narrows the parameter to certain values, which are then to be used and honored by anyone wanting to say something about “modern” macroeconomics.
This parameter may be a reasonable estimate for an individual agent facing a specific micro decision, but what does it have to do with the aggregate

Here I think he's being a little unfair to the discipline (although making an excellent point). If you know the literature on intertemporal substitution elasticities, then you know that this question about the difference between micro and macro elasticities is absolutley central to the discussion. The (modern) literature turns on the question of comparing micro and macro estimates and explaining why macro elasticities are so much larger than micro elasticities. The same goes with the wage cyclicality literature (actually, the two literatures are related). Macroeconomists don't just blindly plug in micro-elasticities. They understand this issue. It would be nice if as a result they would pay less attention to microfoundations. Instead they've generally tried to bridge the gap. That's a good effort too, I suppose. The point is, they're not naive, and the point is Caballero seems to have healthy skepticism of "microfoundation" approaches which I like.


The point is this - if you read some self-styled Hayekians (I have Russ Roberts in mind) you get a few takeaways from them:

- Microfoundations are essential - in fact, macroeconomics is nothing but microeconomics
- Modeling is highly misleading - prose is better
- Look at Caballero! Look! He's quoting Hayek and agrees with us!

I think it should be clear that (1.) appreciating Hayek does not require you to devolve into this Luddite approach to macroeconomic modeling, (2.) Caballero has a very pro-model view, but he views models much the way I've presented them on F&OST - as a way of understanding the market better not reproducing "reality", and (3.) he's skeptical of microfoundations which is an excellent prejudice for a macroeconomist.

UPDATE: On facebook, a former public policy professor of mine writes of this interview, "When I received my graduate training, macroeconomics was still part of the required content of the field of public finance. I like to quip that I know (or don't know) enough macro to be dangerous. This part of the interview with the current chair of the econ department at MIT makes some interesting points generally about the proper use of models in economic analysis and thinking. My own experience is that students -- esp. public policy students -- actually become more receptive to the use of models as analytic tools when models are presented as useful ways of tackling compex questions that provide insights, if not literal answers." [emphasis mine]


  1. As you know, I'm a fan of the virtues of (relevant) simplification. One of my favourite parables is Jorge Louis Borges' "On Exactitude in Science":

    In that Empire, the Art of Cartography attained such Perfection that the map of a single Province occupied the entirety of a City, and the map of the Empire, the entirety of a Province. In time, those Unconscionable Maps no longer satisfied, and the Cartographers Guilds struck a Map of the Empire whose size was that of the Empire, and which coincided point for point with it. The following Generations, who were not so fond of the Study of Cartography as their Forebears had been, saw that that vast Map was Useless, and not without some Pitilessness was it, that they delivered it up to the Inclemencies of Sun and Winters. In the Deserts of the West, still today, there are Tattered Ruins of that Map, inhabited by Animals and Beggars; in all the Land there is no other Relic of the Disciplines of Geography.

  2. You know, the Flat Earth Society has a model too.

  3. After reading some of your posts I can't help but think that living in the blogosphere world gives an overly rosy - even naive - view of how the majority of researchers publishing in the top journals use the models they develop. I say this because it seems that those who blog tend to be genuinely more interested in thinking about a lot of the issues that are completely ignored by the majority of professors at top schools whose sole desire is to knock off those first few top-journal papers.

    re: "The fact is this is a fundamental misunderstanding of why we model, which I think Caballero expresses brilliantly here. We don't model to reproduce reality - that is impossible."

    Caballero seems like a thoughtful guy, and I think there's a lot of them, but I just don't think they comprise the majority of those publishing in the top journals.

    Here's what I've been told, *explicitly*. The idea to publishing in the top journals is that you first *pick some specific aspect of reality that you want to recreate* - eg the education gap across countries. Then, using one of accepted methodologies - infinitely lived, OLG, etc - construct a set of plausible assumptions that *replicate* that aspect. Make sure you can derive some analytical results, and most importantly, derive the corresponding policy implications. If you find some interesting aspect, and your assumptions are reasonably, and cleverly, constructed, then you've got a chance at a decent journal.

    And this is advice from tenured professors who graduated from MIT, Princeton, UM, Rochester, etc. The reality is that the majority of professors at top schools have one thing on their mind - publish and get tenure. Most have never seriously thought about what the proper role of a given model is. What's important is that they find a niche by doing research in relatively narrow-minded way.

  4. eardniw -
    That's interesting - I would have thought exactly the opposite is the case. Talk is ridiculously cheap on the blogosphere. That has some advantages - people without a deep education like me can speculate and give some ideas a shot. But it also means that people can make broad, sweeping, unfair criticisms like "these guys are naive and don't understand how complex the world is". Getting these sorts of papers in journals does certainly require immersing yourself in that world - but I think the critics are wrong to assume that the people who know these models best don't know the limits.

    Another good example of this is the Tom Sargent interview that I write about here:

    Where it becomes clear that the rational expectations and New Classical economists aren't nearly as naive about their work as many bloggers want to make us believe.

  5. I would guess the counter-point is that no matter how many people say, yeah, we realize the model isn't reality, we know enough about the ways humans interact to realize that it is treated as reality. Good thinking is provisional; but knowledge is rarely actually thought of that way. This is explained by all sorts of behaviors; group think being an example.

    There are ways to combat that, but it is a constant issue.

    I also do not think that the peer review process does much to help at all.

    I've read the work of sociologists who study hard sciences like physics, etc.; I am sure there is a similar sub-field for fields for like economics, etc,.

  6. "he's skeptical of microfoundations which is an excellent prejudice for a macroeconomist."



All anonymous comments will be deleted. Consistent pseudonyms are fine.