WHO WILL WIN ON TUESDAY? LOOK TO THE BETTING HOUSES

Saturday, October 30, 2004

If the up-and-down swings in the presidential polls have left you dizzy, maybe you should look at the political betting shops instead. They're volatile, too, but they tend to be more accurate than most polls. People answering pollsters' questions have no particular reason to be truthful, even if they do know their own minds, while people who have real money on the line have good reason to make thoughtful and informed decisions about what is likely to happen.

Since 1988, the Tippie College of Business at the University of Iowa has operated the Iowa Electronic Market, which among other issues tracks the presidential race. As of Friday, contracts for a Bush win were trading about 6 cents above contracts for a Kerry win. The spread was wider for a while in September, but has consistently favored Bush since the end of August.

There's a separate contract for share of the popular vote, and it is saying that the vote is likely to be agonizingly close. But we knew that.

The Iowa market is tiny, though, because traders are limited to an investment of $500 under federal rules on commodities trading. It's a research tool, not a gambling venue. So a lot of the betting action has moved to an Irish site, tradesports.com, which does trade sports but will allow people to bet on pretty much anything they want to, including politics.

Two researchers, Justin Wolfers then of the Wharton School of Business at the University of Pennsylvania and Eric Zitzewitz of the Stanford University Graduate School of Business, asked tradesports to create contracts that allowed people to bet on a combination of events, for instance that Bush would win and that Osama bin Laden would be captured before Oct. 31. The difference between the price of those contracts and the one for a Bush win alone shows that the market believed that bin Laden's capture would lift Bush's re-election chances to 91 percent. At the time, the standard Bush contracts were trading at $66.60.

As of Friday, bettors at tradesports were estimating Bush's chances of winning at 51 percent, and his share of the popular vote at 51.5 percent.

Betting on elections is nothing new, as Paul Rhode and Koleman Strumpf, of the economics department at the University of North Carolina at Chapel Hill, have written.

In fact, before scientific polling became common, newspapers followed the odds as they covered the election.

The authors note that Andrew Carnegie, returning from a visit to Scotland late in October 1904, was quoted in The New York Times: "From what I see of the betting . . . I do not think that Mr. Roosevelt will need my vote. I am sure of his election . . ."

Huge amounts were at stake in some election years. In 1916, the betting volume in the New York City markets (converted to 2002 dollars) reached $165 million.

Concern that people would try to influence elections by manipulating the markets eventually led to the end of the formal political market "and once polling was invented, journalists forgot it ever existed" Wolfers said Thursday.

But the evidence suggests that the markets are surprisingly resistant to manipulation, he says. Traders who are "buying publicity," as he calls it, trying to move prices artificially so market-watchers will think that a candidate is gaining or losing support, can do so but only for a short time.

Bush contracts on tradesports have apparently been the subject of several such attacks. Don Luskin of the weblog poorandstupid.com wrote on Oct. 16, "Yesterday the price of the futures were sold down from about 55 to 10 with a single 10,000-lot order entered by a single trader . . . Within moments after the order was completed, the price recovered back to the low-mid-50s."

Wolfers told me that the first time something like this happened, in September, people in a tradesports forum were talking about how much money they made. One wrote, "Is this part of the new Kerry attack strategy -- flood the markets? Let's take their money."

Though it's not publicly known who the trader might be, another forum participant observed that it might be worthwhile for the Democrats if the media began reporting that Bush and Kerry were close to a tie. At the time, Bush contracts were around 62 to 64, and the price quickly recovered to that level. "Sites like ours are more accurate than any poll possibly could be," someone wrote in a forum post.

Markets collectively have much more information than any of the individuals in them, which is why market economies work better than planned ones. Markets also require participants, if they want to be successful, to separate what they hope for from what is probably true, and to make realistic judgments about what other people are likely to do. If you have to "put your money where your mouth is," you will find out whether you really believe what you say.