19 October 2004

Predict the Election Outcome

We cannot know the future. But we seem to be intensely curious about it. During election seasons, we are bombarded by the results of polls indicating the preferences of voters at the moment. Right now (20 October 2004) most polls show a "statistical dead heat" between President Bush and Senator Kerry, among "likely voters". This suggests that the two candidates would receive nearly the same number of votes ("if the election were held today").

As discussed in a previous post, polls are statistical efforts. They use the responses of a limited number of people to estimate the attitudes of a larger group, in this case the entire electorate. But there are other ways of predicting the future.

Markets aggregate the opinions of large numbers of independent actors who are willing to risk their money based on their belief that they, individually, have a better idea about the future than the rest of the participants in the market. Considerable research indicates that the opinions of large numbers of people, informed or not, aggregated in this way, are likely to be more accurate than the opinion of even the best-informed and most prescient individual. Historically, markets have proven as good as polls in predicting the outcome of elections, or better.

For example, take a look at the Iowa Electronic Market ( http://www.biz.uiowa.edu/iem/ ). This market currently (as of 18 October) indicates that President Bush will receive 50.6% of the combined Bush-Kerry vote, and Senator Kerry will receive 49.5%. Note that this is not a statistical estimate. These are the current market prices. Market participants are willing to spend $0.495 to receive $1.00 if Kerry receives the majority of the Bush-Kerry vote, and nothing if he doesn't, and $0.506 to get $1.00 if Bush gets more of those votes, but nothing if Kerry gets the majority.

Note that in a market this valuation is not based on the political preferences of the participants. It is based on their collective cold, hard calculation of the likely outcome.

But the really interesting market is the "winner-take-all" market. Because of the operation of the Electoral College, the majority of the votes does not always open the door to the White House. In the Iowa Markets, participants are collectively willing to pay $0.606 to receive $1.00 if Bush is elected, and only $0.396 to get $1.00 if Kerry wins (and nothing if he loses). Effectively, this means the market thinks Kerry has only a 40% chance of winning, while Bush has a 61% chance. ( see )

If you had $100 to invest in one of these markets, what would you do?

No comments: