Bollettino

As readers of my previous post can tell, LI is in a bit of misery right now. Free fall, hysteria, calls to my brother, walks over bridges with an eye to trajectory, fall, unconsciousness, drowning.

But let’s get away from the personal, shall we? And take up the subject of models. Economic models.

In the Summer of 2001, the Journal of Social Research published a special issue on numbers and economics. This turns out to have been a timely topic, for at the moment, we are seeing economic numbers bifurcate in an unusual manner. On the one hand, we are in the midst of a strong business recovery – on the other hand, we are in the midst of a credit bubble, a wage meltdown, and a growth in unemployment and partial employment that is effecting us all – LI’s desperation, which see.

One aspect of this disturbance in the global economy is the three year collapse of economic forecasts. Although the Bush tax cut model was really not about sustaining us in a recession, the forecasts that have emanated from the White House are not just mendacious. They are underpinned by orthodox economic models. If the economy was recovering from a post-World War II recession along regular lines (given the absence of anything like an oil shock), we shouldn’t be seeing the sluggishness in the job market, or the slowdown in income increases, that we are seeing. Everybody, I think, agrees about that. In the Outlook section of the Washington Post today, there is an amusing article that uses Neal Stephenson’s Snow Crash as a template for understanding the current anxiety about outsourcing. In Stephenson’s dystopic America, the only things that Americans produce competitively, any more, are micro-coding, t shirt slogans, and pizza delivery. As the author, an editor at U.S. News remarks, we might not be producing micro-coding competitively any more.

Read WP's article along side this thumb-sucker from the NYT. The pizza deliveryman future is no joke. The article cites Bill Gates recent campaign to get more students in IT classes:

[Gates cites] recent Bureau of Labor Statistics projections for 2002 to 2012 indicating a 57 percent increase in the number of jobs (up by 106,000) for network systems and data communications analysts and a 46 percent rise (up by 179,000) in positions for software engineers in applications.

But some economists point to those same federal forecasts to poke holes in the argument that the key to job creation is more sophisticated education and knowledge. Yes, the greatest increase is expected to be for registered nurses (an increase of 623,000 jobs) and college and university teachers (an increase of 603,000).

But according to forecasts issued last month by the Bureau of Labor Statistics, 7 of the 10 occupations with the greatest growth through 2012 will be in low-wage, service fields requiring little education: retail salesperson, customer service representative, food-service worker, cashier, janitor, waiter and nursing aide and hospital orderly. Many of these jobs pay less than $18,000 a year. Forecasting an increase of 21 million jobs from 2002 to 2012, the bureau predicted 596,000 more retail sales jobs, 454,000 more food-service jobs and 454,000 more cashier positions."


LI, when not hitting the walls, is curious about the deep structure of the theoretical problem here. What are economic models, anyway?

In Measure for Measure: How Economists Model the World into Numbers, by a Dutch economist, Marcel Boumans, the answer is: models are instruments for seeing.

Here’s how he states his argument.

“This paper will argue that in economics, models function as such instruments of observation--more specifically, as measuring instruments. In measurement theory, measurement is the mapping of a property of the empirical world into a set of numbers. This paper's view is that economic modeling is a specific kind of mapping to which the standard account on how models are obtained and assessed does not apply. Models are not easily or simply derived from theories and subsequently tested against empirical data. Instruments are constructed by integrating theoretical and empirical ideas and requirements in such a way that their performance meets a previously chosen standard.”

Boumans’ first section starts with an intriguing quote:
“…Morrison and Morgan (1999) have shown that models in economics still function as if they were physical instruments. They can function as such because they involve some form of representation. This representative power enables us to learn something about the thing it represents. But,
we do not learn much from looking at a model--we learn more from building
the model and manipulating it. Just as one needs to use or observe the use
of a hammer in order to really understand its function, similarly, models
have to be used before they will give up their secrets. In this sense, they
have the quality of a technology--the power of the model only becomes
apparent in the context of its use (Morrison and Morgan, 1999: 12).”


This is a deconstructive moment. Boumans has presented the model, in his abstract, as a way of seeing – but what he seems to be tending towards is a way of reading. The conflation of reading and seeing is at the heart of the Derridian renewal of ecriture – a renewal that seems to have been forgotten, even among Derridians. That it is forgotten is generated by its structure – the difference between seeing and reading functions, in the Derridean version of Western metaphysics, as a self-erasing concept – it emerges only to vanish. The problem with the deconstruction is that it seems impervious to historical contingency. Actually, we think that deconstruction’s a-historical structures can be usefully historicized, and that Derrida’s attention to privileged metaphors and examples reflects the way the structure is historicized. Hence, that Heideggerian hammer that turns up, unexpectedly, in Morrison and Morgan’s quote.

But let’s not go that route. Instead, let’s go to sections 4 and 5 of Boumanss paper, on calibration. Here, again, there is a dominant instrument – a familiar one – the clock. The clock is an example of what Herbert Simon, and Boumans, calls the artifact:
“To clarify this definition of an artifact, Simon uses the example of a clock. The purpose of a clock is to measure time. The inner environment of the clock is its internal construction. Simon emphasizes that whether a clock will in fact tell time is also dependent on where it is placed. The artifact is molded by the environment: a sundial performs as a clock in sunny climates, but to devise a clock that would tell time on a rolling and pitching ship it has to be endowed with "many delicate properties, some of them largely or totally irrelevant to the performance of a landlubber's clock" (6).
The designer insulates the inner system from the environment, so that an
invariant relation is maintained between inner system and goal, independent
of variations over a wide range in most parameters that characterize the
outer environment (9).
In contrast to physics, in which one is able to create stable environments for measurements, in economics one has often to take measurements in a constantly changing environment.”

If the environment keeps changing, the example of the clock suggests putting things into models that don’t change – invariants. Boumans provides a very interesting discussion of what those invariants consist of, and how they were formulated in the post War era. He touches on one set of invariants that proved very popular: Kaldor’s “stylized facts” of growth. Kaldor developed this typical pattern of growth, supposedly from comparing the paths of development in capitalist economies, and inscribed it in a template. This template then became a regulator – the parametric invariants to which models of business cyclic behavior would refer.

The problem, as Boumans acknowledges, is the deleterious effect of stylization on ‘fact.”

“Although we have seen that equilibrium business-cycle modelers aim to model from invariants, the choice to take these stylized facts as empirical facts of growth is dubious. Solow already remarked that "there is no doubt that they are stylized, though it is possible to question whether they are facts" (1970: 2). The danger is that stylized facts may turn out to be more stylized than factual. Hacche provided an account of the British-American evidence relating to Kaldor's six stylized facts and showed inconsistencies between economic history and Kaldor's stylized facts:
the data for the United Kingdom provide little support for the hypothesis
that there is some "steady trend" or "normal" growth rate of capital or
output or both running through economic history--which is what Kaldor's
stylised facts suggest--unless the interpretation of the hypothesis is so
liberal as to bear little meaning (1979: 278). “

It seems to us that we have run into a problem in the last three years – and really, a problem stretching back into the nineties. It is that the stylized facts of growth, derived from the Depression through the Reagan years, no longer give us accurate readings.

Boumans ends his paper on an excessively modest note: “To come back to the title of this paper, now put as a question--"How Do Economists Model the World into Numbers?"--my answer is that economists, after a century of mathematical modeling, now prefer very simple mechanisms with the faith that they will be calibrated in the future.”

Ourselves, we think that the essence of the problems now facing us come down to the problem of composition. That is, given the equilibrium models economists use to forecast the effect of policy, we have ignored, for too long, as an effect that can be theoretically cancelled out, shifts in the composition of an economy. These shifts, both in the relative wealth that defines the class structure and in the economy’s mechanisms for production and consumption, have been considered by economists to be secondary correlates, infinitely permeable, that derive from flows of capital that happily obey the equilibrium models economists set up.

Well, we are now seeing the revenge of composition on the models of the economists. A rare and terrible moment.