Thursday, October 24, 2013

On the fourfold root of the principle of insufficient reason

Economist James Crotty has a guest post on the blog Triple Crisis titled “The Man Who Won a Nobel Prize for Helping Create a Financial Crisis.” (The man in question is Eugene Fama, he of “efficient markets hypothesis” fame.) Crotty accuses much of modern-day economics of being an intellectual con. He's hardly alone in that claim, but I think he makes a better case than usual.

Critiques of the foundations of economics are a dime a dozen. The normal procedure is to take issue with the field’s premises. Rational actors with perfect information, market equilibria, exogenous shocks – what crazy, unrealistic assumptions and simplifications! Who would ever think you could derive useful insights about human behavior from models based on such constructs? And so we declare economics a ridiculous pseudoscience, feel very smug and holier-than-thou, and get on with our lives.

Crotty, however, offers an augmented analysis. The bait-and-switch, as he describes it, has not one step, but four:

Step one: Formulate a theory based on assumptions that are “grossly unrealistic” and “patently false.” (Crotty’s characterizations)
Step two: When those assumptions are criticized, assert that their realism or lack thereof doesn’t matter, as long as they yield good predictions. (methodological positivism)
Step three: Test your theory, but do so using methods that can’t possibly falsify it.
Step four: Declare your theory validated. Use it to implement your ideologically preferred policies, and use its “scientific” status to trump any and all objections.  

To me, steps one and two are not necessarily objectionable. That’s not what Crotty argues – he vehemently criticizes all four steps. For him, unrealistic assumptions are bad assumptions, ipso facto. The thing is, though, “unrealistic” assumptions have often served science quite well. Consider physics. Reducing a thrown stick or a falling piano to a point mass is “grossly unrealistic” and “patently false,” but it’s a great way to calculate a trajectory using Newtonian mechanics – which happens to be a “patently false” approximation to relativity theory. Unrealistic ontology, but excellent predictive power.

Moreover, it’s not necessarily clear before the fact whether a given theoretical construct will turn out to be realistic or not. Lamarkism looked plausible for awhile, as did the luminiferous ether. We’re accustomed to think of atoms and molecules as “real” because of what we’ve been taught in school, but imagine what atomic theory would sound like to an Amazonian tribesman or someone from 1500. “You mean to tell me solid objects are made out of infinitesimal particles that are mostly empty space? That’s realistic?!?" Well, yes it is, but we needed roughly a century of highly sophisticated physics research to know why.

So I’m OK with steps one and two. But I am absolutely not OK with step three. Scientific theories are supposed to be testable, dammit. According to Crotty, however, testing economic theories with econometrics is the ultimate grading on a curve – the theory always passes:  
[E]conometric tests can at best provide suggestive, not conclusive evidence in support of the empirical validity of predictions generated by economic theories. With today’s computing power, it is possible to run literally millions of regressions to test a theoretical proposition. … As a result, virtually any hypothesis can be shown to be statistically significant if enough different regressions are run. This is why both sides of every important debate in economics can provide econometric evidence in support of their positions.

Falsifiable in principle, unfalsifiable in practice. For an ideologue, the perfect science! Notice that Crotty doesn't accuse economists of ignoring empirical data – the standard complaint. They're checking their work, all right, he says – just not in a way that would ever be decisive. 

And so, ever since 2008, we’ve had step four ad nauseum. Our model works; if reality isn't cooperating, it must  be your problem.

Sounds like economists really need to read their Popper.

P.S.: For what it’s worth, Noah Smith makes Crotty’s point with less ire and more sympathy in section No. 10 of his long post on judgment calls and the philosophy of science:

[A]lthough I am not a strict “falsificationist”, I think that rejecting models is almost certain to be an essential feature of using evidence to select the best set of models to use in practice.
 And what I suspect is that macroeconomics went so long without any hope of matching any data that it developed bad habits. Internal consistency and the collective intuition of macroeconomists were overemphasized, and what little data there was was often ignored. Theoretical tolerance became the norm, and models that were essentially never useful remained prominent in the toolkits of economists and policymakers alike. And the large reliance on judgment seems (unsurprisingly) to have allowed some political bias to seep into the profession.

No comments:

Post a Comment