the execrable fallacy of predictive markets, or: do I really still have to debunk this shit?
as much as i’d like to believe this is true…
Traders on the University of Iowa’s real money Iowa Electronic Markets believe control of the U.S. House of Representatives is a toss-up after this fall’s mid-term Congressional elections.
…it vexes me that my university, and a lot of other people who ought to damn well know better by now, are still tacitly promoting the view that financial markets constitute some sort of transcendent superhuman hive mind that can actually see into the future — rather than dumb machines that are occasionally and incidentally useful for purposes of getting shit done, but whose primary function has never been anything more or less than to make certain assholes richer than certain other assholes, usually at the expense of the rest of us.
of course, the Iowa Electornic Markets’ promoters will say, i’m misrepresenting their claims — their futures market is merely an interesting social experiment. they don’t claim to predict the future, only to gauge other people’s predictive instincts in a more reliable way than conventional polling does, by tying the outcomes to participants’ rational self-interest. that’s not out-and-out stupid, i’ll grant. and it’s true that certain kinds of questions — if formulated in the right way, and subject to carefully calibrated laboratory controls — are conducive to being answered in a meaningful way with an approach like this.
but that kind of nuance is secondary to the message that IEM’s “research” is designed to pump out year after year, and generally lost on the readers of newspaper articles like the above, which dutifully publish every copy-pasted press release they get from the UI Tippie College of Business, right down to strategically vague and unsubstantiated claims like the following:
Begun in 1988, the IEM is a research and teaching tool that has achieved an impressive prediction record, substantially superior to alternative mechanisms such as opinion polls. Such markets have been significantly more accurate than traditional tools in predicting outcomes ranging from political election results to movie box office receipts.
the point that’s flogged again and again, in 22 years of publicity literature out of the IEM, isn’t anything so subtle or academic as the observation that human decision-making is a complex phenomenon that’s further complicated by economic considerations. no, the take-away, in case you missed it, is that market behavior — the most potent distillation of human reason, purified of wishful thinking and ideological cant, incentivized with the prospect of real financial reward and punishment, and massively aggregated by high-speed computer networks — is “significantly more accurate than traditional tools in predicting outcomes.” you’re not being asked to consider the truth of this statement within the qualified and elaborately engineered context of the experiments that are actually being carried out, you’re just supposed to internalize the austrian-school dogma that Markets are smarter than mere humans, and way, way smarter than human political and social institutions. the Market, in its most ideal (read: unregulated, unimpeded, unaccountable) form, is an algorithm so powerful that it can actually predict the future.
this may have been a tenable position as recently as early 2008, but anybody still seriously espousing it ought to be publicly ridiculed, then tarred and feathered and ridden out of town on a rail. so the IEM has successfully predicted “movie box office receipts” — good for them! where were they on lehman brothers? on madoff? did the omniscient Market see the BP oil spill coming? if they did, it could only have been because investors knew the company was taking unconscionable risks in order to maximize short-term gains — and the Market has never balked at wreaking massive environmental (or social, or political, or even economic) damage when short-term gains were on the line.
it’s long past time to put this discredited, obsolete premise to bed. the Iowa Electronic Markets, like most attempts to assert economics as a straightforward empirical discipline encompassing a comprehensive and reliable theory of human behavior, is junk science. or rather, at best, it can’t tell us anything useful if we can’t figure out how to ask it the right questions. and as long as the people asking the questions are only interested insofar as they stand to make or lose a buck in the short run, we never will.
