As I commented last fall, the accuracy of Efficient Market Theory (or lack thereof) bears on how much effort it's worth investing in prediction markets, where they can be applied, how they should be structured and overseen, who ought to participate, and how we should interpret their results. In short: everything.
What the ongoing debate over Efficient Market Theory does not do however, is invalidate the idea of using markets for information aggregation. It merely raises some common-sense caution flags. I offer this post by blogger Barry Ritholtz (on inertia in Apple's share price) as a succinct summary of this caution: "markets do usually get there - not always, but most of the time, and not right away, but eventually." Hat tip: Bob Weber.
Such tongue-in-cheek cynicism aside, several things seem especially relevant to those interested in using markets for prediction:
1) Markets need time. As I discussed earlier this week, there may be an optimal run time and event horizon for prediction markets - something that early experiments are only beginning to grapple with. Unlike securities markets, the definitive date for final evaluation of prediction market contract makes their trading characteristics 'tighter' than those of regular securities (which must reflect discounted future cash flows out into a rather long future.)
2) Not all markets are created equal. Some applications of prediction markets are, frankly... stupid. Complex, pie-in-the-sky contract bundles, vague contract definitions, wide spreads, tiny floats, and 'build-it-and-they-will come' participant recruitment are signs of faith overcoming reason in some early experiments.
3) It's all about interpretation. Some would invest prediction markets with God-like powers of prediction and rely on them exclusively for big-ticket decisions. That irrational and short-sighted. They hold great promise, but let's be clear: they are supplemental to existing decision-making processes. In very few cases do they offer a plausible substitute for them.
4) Consider the alternatives. I touched on this last October, but I'll re-frame it here: Groups and organizations are more frequently dysfunctional than not when it comes to taking full advantage of their collective wisdom and making rational, timely decisions. One need only look at the amount of money ($billions), spent on 'knowledge management' systems over the past decade or so and compare that to their demonstrable ROI. (What ROI?, you may ask. Exactly.) The alternatives to prediction markets are not perfection (as some would have it), but often an inconsistent, inexplicit, closed, hierarchical, and politically-constrained decision making process. In that light even an Efficient Market Theory qualified with a dozen asterisks can begin to sound pretty good.