Monday, January 07, 2008

The (lack of) wisdom of crowds

Do not fear to be eccentric in opinion, for every opinion now accepted was once eccentric.
- Bertrand Russell

For those of you unfamiliar with the Iowa Electronic Markets, they are worth a look. The idea behind them is that markets aggregate differences in information and opinion in an efficient way and hence provide a better sense of what is likely to happen than typical polling results. The current economist most associated with this idea is Charlie Plott. (In that linked paper, the authors write that the idea goes back at least to Hayek.) But for the markets aimed at predicting which presidential candidates will be selected in the Democratic Convention, either there was some huge news just before the Iowa Caucuses that caused the prices of the various candidates to change dramatically or, my preferred interpretation, the market just got it wrong before the Iowa Caucuses. Apart from the actual returns and the exit survey results of that day, there really wasn’t any big news leading up to the Caucuses. But there was a certain interpretation of the information that was readily available. That interpretation prevailed. But it was quite wrong. Let’s take a look.

The assets that are traded on this market pay off $1 if the particular candidate gets selected by the party to run for president. It pays nothing otherwise. There are four assets, one each for Hilary Clinton, John Edwards, and Barack Obama, and one for the rest of the field. Here are the results for December. Focus on the column called LastPrice. Evidently, Clinton was the clear favorite then. Her price never falls below $.55 and for the much of the time it is priced about $.60. Obama’s price never reaches $.40 and for much of the time it is below $.30. These results indicate that Clinton was the clear favorite in December, at least according to the markets.

Consider how this information contrasts with these CNN Poll results reported on January 1, where the conclusion is it’s a dead heat between Clinton and Obama. Now take a look at the IEM results for January. On New Year’s Day, things were just like December. The following day Obama’s price started to rise and Clinton’s started to fall, with a fairly dramatic change on the 3rd and Obama overtaking Clinton on the 5th (the day of the Caucuses). By the 6th, Obama became the clear favorite, according to the markets.

The question is – did many people just make up their mind in Iowa at the last moment, with the great bulk of these voters going for Obama? That would be one possible explanation. Or did they already have their minds made up in December, but the IEM certainly and the polls to a lesser degree, didn’t measure this well at all. That’s my interpretation. The high turnout suggests to me that people are upset with the status quo and want to do something about it. And I believe that’s been going on for quite a while. Sure, people are dismayed about the price of oil and it reached $100/barrel only after New Year’s, as did the news that unemployment spiked last month. But these things seem to me more of a trend, not something fundamentally new.

There is something new going on. But right through December we’ve been told “same old, same old.” So there really wasn’t any way for most of us to know how the bulk of the voters felt. This is a kind of tyranny of the status quo.

In some of his remarks during the Saturday debate, Obama talked about listening to the voters and letting their voice speak. If he does ultimately win the Presidency and his positions become the status quo how, exactly, will he enable the voter’s voice and what will he do to hear it? That’s a question without an easy answer.

No comments: