Monday, October 10, 2011

Krugman Gets It Correct on the Two New Nobel Prize Winners

Krugman writes (My emphasis):


This is a statistical techniques prize — both men worked on methods for
extracting insight from the data history provides us, which generally don’t
offer anything like a controlled experiment. As the detailed explanation of the
prize (pdf) explains, before Sargent and Sims came along, econometrics consisted largely of estimating models you had no good reason to believe based on identifying assumptions (if you don’t already know, you don’t want to) that lacked credibility. S and S played a key role in developing methods that let the data
speak instead.
Well, almost all correct. He is correct that it is a "statistical techniques prize" and that before before Sargent and Sims came along "econometrics consisted largely of estimating models you had no good reason to believe." But, I believe there is still no good reason to believe in econometric models, while Krugman apparently believes that Sargent and Sims some how sprinkled wonder dust upon, what he seems to admit, was before that a fraudulent science.

As I indicated, in my first post on Sargent and Sims, Hayek does an impressive job of destroying the econometrics way of thinking.

When all is said and done, when one studies these econometricians for any length of time, it becomes clear that their mumbo jumbo equations are good for nothing other than justifying interventions by governments in our lives and soothing their inferiority complexes.

Ludwig von Mises had them pegged decades ago:



It is only the passionate pro-socialist zeal of mathematical pseudo-economists that transforms a purely analytical tool of logical economics into an utopian image of the good and most desirable state of affairs.

Here's Murray Rothbard cutting apart the mathematical model building approach:



Not only measurement but the use of mathematics in general in the social sciences and philosophy today, is an illegitimate transfer from physics. In the first place, a mathematical equation implies the existence of quantities that can be equated, which in turn implies a unit of measurement for these quantities. Second, mathematical relations are functional; that is, variables are interdependent, and identifying the causal variable depends on which is held as given and which is changed. This methodology is appropriate in physics, where entities do not themselves provide the causes for their actions, but instead are determined by discoverable quantitative laws of their nature and the nature of the interacting entities. But in human action, the free-will choice of the human consciousness is the cause, and this cause generates certain effects. The mathematical concept of an interdetermining "function" is therefore inappropriate.

Indeed, the very concept of "variable" used so frequently in econometrics is illegitimate, for physics is able to arrive at laws only by discovering constants. The concept of "variable," only makes sense if there are some things that are not variable, but constant. Yet in human action, free will precludes any quantitative constants (including constant units of measurement). All attempts to discover such constants (such as the strict quantity theory of money or the Keynesian "consumption function") were inherently doomed to failure.

Finally such staples of mathematical economics as the calculus are completely inappropriate for human action because they assume infinitely small continuity; while such concepts may legitimately describe the completely determined path of a physical particle, they are seriously misleading in describing the willed action of a human being. Such willed action can occur only in discrete, non-infinitely-small steps, steps large enough to be perceivable by a human consciousness. Hence the continuity assumptions of calculus are inappropriate for the study of man.

Other metaphors bodily and misleadingly transplanted from physics include: "equilibrium," "elasticity," "statics and dynamics," "velocity of circulation," and "friction." "Equilibrium" in physics is a state in which an entity remains; but in economics or politics there is never really such an equilibrium state existing; there is but a tendency in that direction. Moreover, the term equilibrium" has emotional connotations, and so it was only a brief step to the further mischief of holding up equilibrium as not only possible, but as the ideal by which to gauge all existing institutions. But since man, by his very nature, must keep acting, he cannot be in equilibrium while he lives, and therefore the ideal, being impossible, is also inappropriate.

The concept of "friction" is used in a similar way. Some economists, for example, have assumed that men have "perfect knowledge," that the factors of production have "perfect mobility," and so on, and then have airily dismissed all difficulties in applying these absurdities to the real world as simple problems of "friction," just as the physical sciences bring in friction to add to their "perfect" framework. These assumptions in fact make omniscience the standard or ideal, and this cannot exist by the nature of man.
Bottom line: There are simply no constants in the field of human action, like there are in the physical sciences, water freezes at 32 degrees in Washington D.C., Minnesota and New Hampshire. The price of any good may stay constant for awhile, but that doesn't mean it will do so for ever. Just ask Long Term Capital, which blew itself up by using equations which assumed that bond prices would stay in a certain range. Further, actions of groups of individuals are not constant. Just ask the banks and ratings agencies, which assumed that defaults on subprime mortgages would stay below a certain limit, in their equations, blew up the subprime sector.

It's a good bet that water will freeze at 32 degrees, but there is no equation in the world of human action, whether written by Krugman, Sims or Sargent that can contain such a constant. Thus, all their equations are susceptible to blowing up at any time.

Krugman had it right at the start of his post:
...econometrics consisted largely of estimating models you had no good reason to believe
It's the same today.

8 comments:

  1. You are correct, sir. I am in such a business and human behavior have few "constants", but the entire system is non-linear, non-stationary with significant feedback loops that form a difficult "Gordian Knot" of causality. "Causal" (not really) relationships reverse often and the focus of the human element shifts constantly. Our technologies use a walk-forward adaptive process in which there are no constants, that change with not time, but relationships seen. More advanced than those methods used by the Nobel winners, but my prize comes from the compensation from my customers for the value delivered. Can it be modeled, yes, to some degree, for a while, long enough to be useful, but the system must adapt constantly, but use few if any constants.

    ReplyDelete
  2. Raise your hand if you love Rothbard......no wonder statists loathe and despise him.

    ReplyDelete
  3. What I can't understand is why most academics simply can't grasp what you've written here Bob. What is so difficult to understand about it?

    ReplyDelete
  4. Side note: Variations in the pressure can change water's freezing temperature. But the point still stands even that is a known property and has a proven equation to calculate the change.

    ReplyDelete
  5. Economic interaction is a sort of self-organizing neural network where the individual neurons are trying to maximize their self-interest. Each neuron is also complex with different weighting functions and a vast array of inputs. A standard mathematical modeling approach has no chance of predicting the outcome of inputs to the network.

    ReplyDelete
  6. Great post, wenzel. Our current state of the economy and an even bleaker future is a perfect example of the failure of this econometric nonsense.

    ReplyDelete
  7. Oszag said health care would move away from double blind clinical studies to clinical modeling, like econometrics. Oddly, he recommended economic studies go the other way, toward rigorous experimental studies.

    Peter Orszag said the government would through stuff against the wall to see what sticks, which you heard in person.

    The feds are slinging away.

    http://peureport.blogspot.com/2011/10/cheating-to-garner-prize-clueless-uncle.html

    ReplyDelete
  8. You may be aware of this already, but I just learned this... "The Prize in Economics, as it is referred to by the Nobel Foundation, was established and endowed by Sveriges Riksbank" ... sveriges riksbank being the swedish central bank. Would they really award the prize to anyone who pointed out how horrible having a central bank is?

    ReplyDelete