F.A. Hayek, for a long period at the University of Chicago, emphasized the ability of markets to aggregate dispersed knowledge, ensuring that prices incorporate new information. Prediction markets, aggregating dispersed knowledge, have been flourishing on the Internet, and many of us have been impressed by their accuracy. For example, the Hollywood Stock Exchange has done well in forecasting Oscar winners and box office receipts. But many people are skeptical; they believe that as these markets become better known and more manipulable, they will often be proved wrong.
We are about to have a new test of prediction markets in the political domain. As of this writing, the markets say that the Senate will stay in Republican hands and that Democrats will capture the House. Some markets offer more specific forecasts. For example, the Washington Stock Exchange now predicts that Jim Talent will defeat Claire McCaskill in Missouri; that Bob Casey will defeat Rick Santorum in Pennsylvania; that Robert Menendez will beat Tom Kean in New Jersey; that Jim Webb will beat George Allen in Virginia (a surprising prediction, at least to me); that Joe Lieberman will defeat Ned Lamont in Connecticut; and that Bob Corker will defeat Harold Ford in Tennessee. It will be interesting to see if there is significant movement, in any of these predictions, over the next days.
It should be possible to learn a great deal once we know about the accuracy of such predictions. The Washington Stock Exchange does not involve real money; overseas markets, some of which allow a number of relevant bets, do. Will real money be shown to matter? Will the markets do well in predicting not only who wins the Senate and the House, but also who wins particular races? Might there be some major surprises? We'll have the answers soon.