Sunday, July 27, 2008

The Map Is Not The Territory

Suppose an orthodox economist hands you a map and says, "This is a map of New York City." You look at it and say, "It is not. It is a map of the London tube system."

Or suppose an orthodox economist hands you a map. And you look at it and say, "This cannot be right. Here are two interesecting contour lines supposedly of different elevations."

Suppose the orthodox economist responds, "Assumptions do not need to be realistic."

I have encountered several economists who distract from those pointing out the logical inconsistencies and factual errors in their theories. They make dismissive non sequiturs about methodology, as illustrated in the parables above.

13 comments:

Gabriel said...

Well, no. Also suppose that what the "orthodox economist" is showing you can only happen in a known subset of circumstances. In which case, their error is not logical invalidity but claiming a level of generality not supported by the theory.

Anonymous said...

what if that known subset of circumstances can never occur in the real world?

Anonymous said...

Well, no. Also suppose that what the "orthodox economist" is showing you can only happen in a known subset of circumstances.

Well, it's easy enough to avoid accusations of committing a logical fallacy, often with the magic words "one-good model." The circumstances you refer to are entirely mathematical though, of the form if a then b, if a' then b', etc. That these conclusions have any bearing on an economic 'reality' is a matter of faith.

-h.e.

Anonymous said...

"Well, no. Also suppose that what the "orthodox economist" is showing you can only happen in a known subset of circumstances."

Yes, well, if those circumstances are unlikely to ever occur then the model is pretty useless. Sometimes those assumptions would never actually occur, ever, which makes the model even more surreal.

"In which case, their error is not logical invalidity but claiming a level of generality not supported by the theory."

And that raises the question of why design models and make policy decisions based on them when reality if the "level of generality not supported by the theory"?

Iain
An Anarchist FAQ

Robert Vienneau said...

It seems to me that if one is claiming an unsupported level of generality, one must accept a statement of the form: [for all x in S, if p(x) then q(x)] and [there exists a in S, such that p(a) and not q(x)]. That's a logical contradiction. (S is the domain of the theory, p() is the assumptions, and q() is the conclusions.)

Suppose we are talking about results that exclude capital-reversing. In the multigood case, the subset of circumstances is not known.

I guess the much lauded Paul Romer would be an example of an orthodox economist not to be trusted in map-making.

Gabriel said...

Robert,

No disagreement from me, but you sometimes make it sound as if all "orthodox economists" do is work involving some capital index, which I don't think to be the case. Maybe in macroeconomics, but that's by definition an applied field, often guilty of much worse crimes.

H.E.,

Unlike the long run equilibria studied by Sraffa and the classics, which Robert recaps here, which happen in the Real World(tm) all the time, right?

Anonymous said...

Gabriel,

I'm not a Post Keynesian economist. I'm just re-iterating McCloskey's point, which is that economic theory by itself is little more than an exercise in rather uninteresting mathematics. In most branches of economics, there is also at least a casual interest in also observing the economic world.

-h.e.

Anonymous said...

To push the McCloskey thing further, it makes sense to think of economic theory as an exercise in applied storytelling. In good economics, the math is used to tell a story and check for internal contradictions. It's just a language, nothing more, nothing less. Models serve the same purpose in sociology, psychology, etc. I'm reminded of this old gem from Krugman:

http://web.mit.edu/krugman/www/hotdog.html

No model, ever, is sufficiently general to be correct, but many are useful. This goes for all kinds of theorizing, not just mainstream economics. We'll never have a model that's 100% true, but I don't see the alternative as complete nihilism.

I'd file the ergodicity issue under "Controversies over things which Samuelson said in the 1960s that turned out not to matter much." Defining your state space properly should take care of things; then the usual math goes through. Hell, even those dreaded RBC models and the associated VAR specifications are based on nonergodicity, and these models still have well-behaved solutions. What's the problem with nonergodicity again?

-Chris

Anonymous said...

In good economics, the math is used to tell a story and check for internal contradictions. It's just a language, nothing more, nothing less. Models serve the same purpose in sociology, psychology, etc.

But orthodox economic theory is not just a language. Frankly I think you are confusing mathematics, which is a general study of relationships between various kinds of abstract objects, and theories within other disciplines, which seek to discover relationships between variables of interest. Orthodox theory prescribes a particular vision of how variables are related (i.e. methodological individualism, equilibria). This is not simply a language through which competing theories can be expressed.

-h.e.

Robert Vienneau said...

I thought Canadians were supposed to be polite. Chris is not invited to comment here.

Robert Vienneau said...

Gabriel doesn't seem to be addressing anything I say.

My Manchester School paper explicitly notes it is not about aggregation problems, i.e., a capital index.

I think of an example of capital-reversing as an a in that logical contradiction I noted in my prior comment. As such, it is irrelevant if an example exists in the "Real World". Furthermore, such examples seem to have implications for non steady-states. Burmeister and Hahn have both said so, too.

Sraffa provides both an internal critique of the logic of neoclassical theory and a foundation for the construction of an alternative. Those who promote the alternative claim that their theory has something to say about tendencies existing in the "Real World". But they do not necessarily claim that "long run equilibria" ever happen in the "Real World".

YouNotSneaky! said...

"I thought Canadians were supposed to be polite. Chris is not invited to comment here."

WOW! Can anyone, not just Robert, but ANYONE, anyone in the world point out to me where exactly Chris is being impolite?

Am I missing something or is this just one of the
"different model = illogical",

"counterexample = proof of the negative universal",

"disagreeing with Robert = dishonesty, impoliteness and harassment"

things?

WOW.

peter said...

In the real-world, the vast majority of actions are not reversible -- capital, for example, once invested in physical plant and labour cannot readily be uninvested. Thus, at least in the real world, paths are usually 1-way not 2-way streets, and the intermediate states of the world achieved along each path are likely to differ greatly. Along one path, there will be houses built by the workers employed by the business into which the invester sank his capital, while these houses will not exist along a path where that invester kept his money under his mattress.

Given this fact, it would seem to me obvious that any real economy exhibits path dependence, since path-independence would imply that all those different intermediate states somehow end up at the same final state. This is highly unlikely, to express myself politely.