Thursday, May 7, 2015

Ain't Misbehaving?: further thoughts on the Kahneman-Tversky experiments

I've been reading through some previous Sandwichman posts with the aim of extending the essay outlined in yesterday's Decisions... Decisions. When I was writing that, I resisted the temptation to call attention to the seemingly irrelevant detail that the hypothetical disease was described as "Asian." What is that about? Does originating in Asia give the disease some exotic elan? Or does the superfluous detail lend it an air of realism? Could this faux verisimilitude distract from the very unreal condition of having allegedly definitely-known probabilities of the results of various treatments?

I am wondering if perhaps many of the subjects in the experiment disregarded the literal but incongruous descriptions of the outcome probabilities and substituted more intuitive -- but also more ambiguous -- interpretations. Under these re-interpretations, the two sets of choices would not be equivalent. I am thinking here of the kind of "perceptual reorganization" investigated by Jerome Bruner and Leo Postman in their classic 1949 study, "On the perception of incongruity: a paradigm." Their assumption -- that "most people come to depend upon a certain constancy in their environment and, save under special conditions, attempt to ward off variations from this state of affairs" -- would seem to have salience in the Kahneman-Tversky experiments.

Is it not the case that the behavioral economists are looking at the case in terms of individuals failing to attain some presumed standard of rationality attributed to a so-called rational agent. This concedes too much "observation, fixed by reason" to so-called rational choice scheme. What if we view these anomalous outcomes from the perspective that the normative model is a magical pseudo-scientific one,"hedged round by observances, mysteries and taboos"?

Reprising material from an earlier post, rather than merely linking to it: In Magic, Science and Religion, anthropologist BronisÅ‚aw Malinowski discussed the interplay between the systematic rational knowledge and the magical pseudo-science of the Trobriand Islanders, observing that "even with all their systematic knowledge, methodically applied, they are still at the mercy of powerful and incalculable tides, sudden gales during the monsoon season and unknown reefs."

It is in dealing with these formidable uncertainties that magic comes into play. "Science," Malinowski explained, "is founded on the conviction that experience, effort, and reason are valid; magic on the belief that hope cannot fail nor desire deceive." In contrast to the reliance of science on "observation, fixed by reason," the domain of magical pseudo-science is "hedged round by observances, mysteries and taboos."

The "mainstream/heterodox" distinction in economics is otiose (and odious). The demarcation that matters is between observation of economic regularities, which is limited, and the proliferation and persistence of economic pseudo-science in the face of "powerful and incalculable tides" and "sudden gales." "Theorists have a natural urge toward precise and determinate theorems or laws," John Maurice Clark wrote 65 years ago. "But..." he continued:
    "...the facts of economic life show little consideration for this urge, and remain, to a large extent, perversely and persistently indeterminate. This is the skeleton in the closet of economic theory. What is a proper attitude for a would-be science, forced to deal with such refractory material? One thing economists do is to construct hypothetical simplified 'models.' These can be used in two ways: as an approach to reality or as an escape from it. My problem is how to promote the first kind of use and set up safeguards against the second." 
    Would Clark's attitude toward this "skeleton in the closet of economic theory" make him "heterodox"? How has the bureaucratically-imposed conventional cost-benefit analysis and the Kaldor-Hicks criterion that justifies it achieved its canonical status? How about the notion of shirking in New Keynesian models of sticky wages? The ritual invocation of the lump-of-labor fallacy claim? Ceteris paribus? General equilibrium?

    The urge for formulas in economic analysis is strong, especially from official "deciders" who yearn for guidelines, criteria or rules-of-thumb that will immunize their decisions from criticism for favoritism, arbitrariness or bias (all the more convenient if favoritism and bias are non-transparently built-in to the formula!). In an article also published in 1950, Paul Samuelson wrote:
    "Improved measurement of national income has been one of the outstanding features of recent progress in economics. But the theoretical interpretation of such aggregate data has been sadly neglected, so that we hardly know how to define real income even in simple cases where statistical data are perfect and where problems of capital formation and government expenditure do not arise."
    In his article, Samuelson warned that "the last word on the subject will not be uttered for a long time." Not that anyone would still be listening when that proverbial "last word" (or even the next word) was uttered. Hedged in by observances of bureaucratic standards and procedures, mysteries of discounted net present value and taboos on interpersonal comparisons of utilities, the aggregate data of national income came to ritually stand in for its own interpretation.

    Usage and custom have shifted the burden of proof from the believers in economic magic to the skeptics. Disproving the magic is impossible. As Malinowski explained:
    First of all, magic is surrounded by strict conditions: exact remembrance of a spell, unimpeachable performance of the rite, unswerving adhesion to the taboos and observances which shackle the magician. If any one of these is neglected, failure of magic follows. And then, even if magic be done in the most perfect manner, its effects can be equally well undone: for against every magic there can be also counter-magic [ceteris paribus]. 

    No comments:

    Post a Comment