Emanuel Saez wins the Clark Medal

This is excellent news. One important aspect is that Saez is very reality based, that is empirical. As you have noted from time to time, the economics profession has had a very high opinion of theory in which results are rigorously derived from clearly false assumptions. Saez is, as far as I can tell from his web page, entirely focused on the data and not on theory. I think this might be the first Clark medal awarded to someone for analysing data with minimal and informal theorizing.

Oddly his often co-author Thomas Picketty was a theorists theorist when he was young (and I mean really really young like 20 or so) and shifted to totally empirical because he didn’t believe in the theory — an amazing example of intellectual integrity.

via Brad DeLong and Matt Yglesias. Check out Ken’s comment at DeLong’s place.

Debate from comments which I found interesting after the jump (and also in comments of course).

I am not absolutely opposed to theory. However, theory should be based on evidence. Existing economic theory is mostly based on pure imagination and then on dismissing the rejection of hypotheses based on data either because auxiliary hypotheses are always needed or because models are false by definition and false models might be useful.

Give me an example of anything useful ever accomplished in the history of humanity by an approach similar to that of economic theory in which theoretical speculation comes before not after striking empirical regularities are observed.

Anything. Ever.
More ▼Like this comment? [yes] [no] (Score: 0 Da 0)Community assigned karma score: 0 Da 0 Rispondi:Cancellaeditcontrolla
2slugbaits replies:Today, 18.59.51“Black holes? Gravitational constants? Bending of light? Aren’t hese examples of natural phenomena whose existence was predicted by theory and later confirmed by empirical tests?


I reply

Your examples ignore my qualifier “striking empirical regularities are observed”. The prediction of the bending of light came from general relativity — the striking empirical regularities were than the location of planets can be predicted given the model that they orbit in ellipses with the sun at a focus, that their angular momentum is constant, that the period of their inverse to the average distance to the sun to the 1.5. These striking empirical regularities were described by Kepler. Based on these striking empirical regularities, Newton proposed a theory of gravity which was used to, among other things, predict that there was a planet around where Neptune was found (and not by coincidence).

In developing his theory of gravity, Einstein matched the same striking empirical regularities used by Newton, plus the innumerable data which fit Newton’s model. Also he fit the fact that the perihelion of mercury precesses 90 seconds of an arc per mercurian year.

The theory which fit all of those facts successfully predicted how much light was bent by gravity.

So we have very striking stylized facts based on observation. We have a theory which fits a huge amount of data. We have a prediction — then we have confirmation.

I didn’t mean to suggest that the theory simply restated the empirical regularity which was already observed. What I wrote and what I meant was that a theory which doesn’t fit a whole whole lot of facts really well has not yet amounted to anything.

In physics a good shortcut to a theory which fits a whole lot of facts is one that gives the old theory as a special case under the conditions where the old theory fit an amazing number of facts. Thus the later theorist (Einstein) doesn’t have to deal directly with the data fit by the old theory (locations of planets) because he knows the stylized fact that, except for the precession of the perihelion of Mercury, the old theory is consistent with the data (and that anomaly depends on the assumption that the Sun is a rigid sphere and that no forces but gravity are relevanat and that there is not another planet even closer to the Sun where it would be almost impossible to see).ù

In contrasst, the prediction that there are black holes has not been confirmed empirically. It is known that there are extremely massive and dense bodies such that *according to theory* they should be black holes. However, what is actually observed is the accreation disk outside of the event horizon. The claim that there is a black hole as opposed to an extremely massive and dense object which is not a black hole because the theory is wrong is based on pure extrapolation based on the assumption that the theory is true.

No empirical observation confirms the theoretical prediction that there should be black holes. The data show that there are objects massive and dense enough so that *if* the theory is true they are black holes.

My view is that theorizing is worth while once one has achieved something with atheoretic empiricism. A theory which fits the facts is worth investigating. Before we have such a theory, we can try to think of one, but we shouldn’t claim we know things are true because they are implied by the theory until it has surprising success in fitting the facts, that is, does much much better than reduced form empirical models with the same number of free parameters.

We can also follow the approach which I think is more likely to be fruitful (because unlike the start with a theory approach it has had some success in the past) and try to develope stylized facts with a descriptive approach.

I am not criticizing Newton or Einstein. I am arguing one shouldn’t try to be a Newton until there has been a Kepler. No example has yet been presented of a useful Newton without a preceding Kepler (I hope I have made clear what I mean). I do not believe that there is one.