For a while, it seemed like economists had the answer: create a futures market for political candidates. A decade ago, the University of Iowa received a special waiver to allow gambling on U.S. elections. And so the Iowa Electronic Markets were born to test an economic theory: the efficient markets hypothesis.

The theory is simple. It states that the current price of any commodity in an exchange market reflects the total amount of information available at the time; in other words, it's the "correct" price or the best possible prediction.

In terms of politics and predicting elections, the implication is that free markets are smarter than polls or pundits. However, it seems that for the time being at least, economists will have to go back to the drawing board with the rest of us, for the futures markets performed as badly, if not worse, than astrology, sports superstitions and every other method for handicapping the presidential race.

There are actually two presidential markets in Iowa. The more popular version is called the winner-take-all market, in which investors buy futures contracts that pay $1 each on Nov. 10 if your chosen candidate wins the popular vote as reported by the New York Times that morning.

The other market is called the vote-share market; it pays out $1 multiplied by the proportion of the popular vote received by that candidate on the same day. So if you bought Bush vote share futures at 40 cents apiece, and then he garnered 48 percent of the popular vote, you'll walk off with 8 cents for each contract you owned.

This year, both methods failed. As Bush and Gore quibble over who will claim the presidency, the Iowa Election Markets seem to mimic the confusion, delivering two contradictory results.

Iowa's winner-take-all market pegged Bush as the clear winner, with confidence in a Bush win soaring in the days before the election. But on Election Day, the market swung wildly and finally flipped to favor Gore. The ambivalence held: Even as late as Nov. 10, the day the market closed, Gore futures were selling for 96.9 cents, not the full dollar they should have cost, given the payout. The vote-share market was more stable, but closed at an anomalous pricing of 49.1 cents for Bush and 48.1 cents for Gore.

At first blush, the efficient markets theory seems a reasonable way to predict elections. Think about it: If a pollster calls you up and asks you who you plan to vote for or who you think will win, you have little incentive to carefully consider your answer. Maybe you say you'll vote for Bush or Gore, but when Election Day comes around, you never make it to the voting booth. Maybe you aren't a very representative voter. These are the intangibles that have driven generations of political scientists crazy.

But if you, the respondent, has to put your money where your mouth is, it's a different story altogether.

The Free Software Project
Read Andrew Leonard's book-in-progress on Linux and open source -- and post your comments.