Sunday, May 16, 2004
Can we take the Iowa Electronic Market seriously?
For some time, the Iowa Electronic Markets have allowed people to trade futures based on election outcomes. For the 2004 presidential election, IEM offers a vote share market in which one can buy contracts for either Bush or Kerry that pay off at the vote share of that candidate times one dollar. The current price of a Kerry contract is about 48 cents; Bush is about 52 cents.
There are some big caveats. Participants in the market are limited to $500 dollars investment. So, if I think the VS market is off by a few cents (and I do), I could invest up to $500, and, by a quick bit a arithmatic, stand to profit a whopping $15 if I am right. The overall vote share market currently consists of a mere $11,377, a very shallow market indeed.
Markets are a great mechanism for aggregating information, but it seems to me that the strictures placed on IEM make "market" something of a misnomer. I realize the IEM would love to operate without these restrictions, so I'm not trying to criticize their methods. But they talk tough about how useful they are as a predicting mechanism, and I just don't see it.
IEM data may still be useful for a variety of research purposes (for an incomplete bibliography of work relying on IEM data, click here). It's a neat laboratory to study investor behavior, but with some big caveats. The chief among them is surely that with such small sums involved, the goals of investors need not be maximizing returns from futures, but shifting the price of the market itself. They may do this to change impressions of market participants (or, indeed, the public at large, since the IEM gets a fair bit of attention). They may do it as self-expression. In most markets, such people will be quickly parted of their money, and the price will govern as usual. But with a tiny market, tiny shares, and participants self-selected for their interest in politics, and I can imagine many scenarios in which investor preferences, beliefs, or strategies may shift the market price. Discipline, of course, will always settle in on payday (e.g., day before IEM data are about as good as polls). But that's not enough to make this a reliable predictive tool, or justify giving IEM attention months away from a close election.
I'll stick to survey data, thanks.
There are some big caveats. Participants in the market are limited to $500 dollars investment. So, if I think the VS market is off by a few cents (and I do), I could invest up to $500, and, by a quick bit a arithmatic, stand to profit a whopping $15 if I am right. The overall vote share market currently consists of a mere $11,377, a very shallow market indeed.
Markets are a great mechanism for aggregating information, but it seems to me that the strictures placed on IEM make "market" something of a misnomer. I realize the IEM would love to operate without these restrictions, so I'm not trying to criticize their methods. But they talk tough about how useful they are as a predicting mechanism, and I just don't see it.
IEM data may still be useful for a variety of research purposes (for an incomplete bibliography of work relying on IEM data, click here). It's a neat laboratory to study investor behavior, but with some big caveats. The chief among them is surely that with such small sums involved, the goals of investors need not be maximizing returns from futures, but shifting the price of the market itself. They may do this to change impressions of market participants (or, indeed, the public at large, since the IEM gets a fair bit of attention). They may do it as self-expression. In most markets, such people will be quickly parted of their money, and the price will govern as usual. But with a tiny market, tiny shares, and participants self-selected for their interest in politics, and I can imagine many scenarios in which investor preferences, beliefs, or strategies may shift the market price. Discipline, of course, will always settle in on payday (e.g., day before IEM data are about as good as polls). But that's not enough to make this a reliable predictive tool, or justify giving IEM attention months away from a close election.
I'll stick to survey data, thanks.
