The Mystery

UPDATE:  I didn't read the IEM winner-takes-all description carefully enough.  It only pays out if the candidate for the party (they don't switch to the actual candidate names until after the nominations) takes more than a plurality of the popular vote.  So, it is not necessarily predicting an outright victory.  The market is only saying that there is  63% chance that Obama gets more than McCain in the popular vote.  The Intrade market, however, does pay out on victory (i.e., Electoral College victory) by the actual candidate.

One of the mysteries of the current campaign is the disconnect between the polls, which are fairly consistent in showing a statistical tie between McCain and Obama, and the various presidential "futures markets," which predict a 63% chance of an Obama victory.  My business partner puts more faith in the futures markets than the polls, believing that the format of the polls makes them unreliable.  

There is something to that position.  With the futures markets, the transaction is totally anonymous and there is no interaction with a pollster to muddy the waters.  With polls, you can get unintended effects from poorly phrased questions and you risk the possibility that the respondent is telling the pollster what they think he or she wants to hear.  This is probably the root of the supposed "Bradley effect," which posits that respondents will say they want to vote for the minority candidate because they don't want to sound racist, but they vote differently once they get in the polling booth.  If you believe that theory, however, then Obama would actually be behind McCain, which makes the disconnect between the polling and the futures markets even more pronounced.

 

One possibility for the disconnect is that the national polls don't reflect the reality of the American Electoral College, which is, of course, the way we actually pick the President.  Most of the Electoral College tracking maps that I see have Obama with a fragile lead in electors, which could be why the futures markets give him better than coin toss odds.  But the move of one or two swing states (e.g., Ohio, with 20 electoral votes) from Obama to McCain reverses the results and McCain has recently been showing some gains in those states.  

In the Iowa Electronic Markets, this theory about the disconnect appears to be consistent with the two different contracts that trade: the vote share contract and the winner-takes-all contract.  The first is a measure of the popular vote and will pay out in cents on the dollar, the percentage of the popular vote cast for the candidate, while the second only pays based on who wins.  The vote share contract seems to reflect the current polling data, with only 4 cents separating the contracts (note: the link updates with current prices, so it may be a different price quote when you click through.)  The winner-takes-all contract, however, has the 64/36 split, indicating a market perception that McCain will pick up a lot of votes in solidly Republican states that will add to his popular count, but he will lose key battleground states and the Electoral College.

Intrade, the other major futures market, only features a winner-takes-all contract as far as I can see and the results are consistent with Iowa.  They do, however, feature a VP choice contract, which is interesting to watch.  My favorite pick for McCain, Alaska governor Sarah Palin, is trading at 10.2 on the bid.

The one result I absolutely dread is that Obama wins the popular vote by a large margin due to big turnout in solid blue states like New York and California, but loses in the electoral college.  If you think we are divided now, just imagine what it will be like if people think the electoral college "stole" the election from Obama.  Oy vey!