Maybe we can just get rid of the pollsters
In California Race, an Online Invitation to Profit
By Guy Gugliotta
Washington Post Staff Writer
Sunday, September 7, 2003; Page A06
Put your money where your mouth is. Will California voters keep Davis? Will they get rid of him and pick Schwarzenegger? Or will it be Bustamante? And if so, by how much?
At the Iowa Electronic Markets,http://www.biz.uiowa.edu/iem/, you can spend $5 or $500 online buying and selling futures for the California recall election, with 12 different outcomes open for trading and an opportunity to treble your investment if you don't let sentiment get in the way of greed.
"We ask you to think about not what you want to happen, but what you think is going to happen," said Joyce Berg, a member of IEM's board of directors and an accounting professor at the University of Iowa's Henry B. Tippie College of Business -- inventor and custodian of the electronic market. "You make money here by buying low and selling high."
IEM traders are predicting that about 54 percent of California voters will choose to recall Gov. Gray Davis (D), that actor Arnold Schwarzenegger (R) will get around 39 percent of the gubernatorial vote, and that Lt. Gov. Cruz M. Bustamante (D) will get 32 percent.
But the market has been running only since Aug. 27, and the differences are still wide between the buying prices bid on various outcomes ("contracts") and the selling prices asked for them. With more "traders" and more transactions, the differences between "bid" and "ask" will shrink.
Berg noted, for instance, that bidders late Friday were willing to buy a Schwarzenegger contract for 34 cents (corresponding to his predicted percentage of the vote), but sellers wanted 45 cents: "If somebody asked me today, 'Could you pick a winner?' -- to me, it says no," Berg said. "The more traders you have, the better the data and the smaller the spreads. This is far from over."
Sound confusing? That was the intention, at least in part.
The IEM was the brainchild of a group of University of Iowa professors dismayed at the quality of polling during the 1988 presidential election. Could markets do a better job?
There were several advantages. "Markets are good at aggregating information," Berg said. A political futures market would reflect not only polling data, but also knowledge gleaned from newspaper stories, television appearances, stump speeches, debates, and assessments of the effectiveness of spin and strategy -- all of this crystallized into a single money number rating the likelihood of a given outcome.
This approach has two advantages over polls, Berg said. Traders do not cross their fingers, lie to the pollster and hope for the best -- they're risking money, albeit modest sums. Second, the market does not worry about "likely voters," "Latinos," "union members" or "elderly voters."
"Our traders aren't a random sample," Berg said. "We want informed people who are interested." Traders will have seen the poll numbers for various ethnic, age, regional and occupational groupings and will use the information in their investment decisions. If they don't do it right, disaster awaits.
"We have found this to be an excellent way to teach students about markets," Berg added. "In general, students don't appreciate the rule that if you don't know anything, you probably don't want to be trading."
Since 1988, the IEM has traded more than 50 markets in 13 countries. In elections for which the IEM had polling data, the market outperformed the polls by a bit more than four-tenths of 1 percent. Polls averaged 1.93 percent deviation from actual outcomes, whereas the market averaged 1.53 percent.
IEM was significantly more accurate than the polls in the 1988 and 1992 U.S. presidential elections, significantly less so in 1996, and ended in the middle of the pack, as muddled as anyone, in 2000.
IEM was clearly the inspiration for the Defense Department's effort to develop an online futures market to predict terrorist events, an initiative that was abandoned this summer. Berg called this an "interesting idea" and noted that regardless of the morality of such an enterprise, it would probably be difficult for a terrorist to bid up the possibility of an attack in one place while setting up another target.
"In the political markets, you can try to push the price up for one candidate, but it will move right back down," Berg said. "Other traders will say 'this is stupid,' and the bubble will burst. And besides, efforts to game the system are very detectable."
The California recall election presents unusual challenges for IEM because the balloting is double-barreled, and the second-round results, in the event that Davis averts being recalled, would not be published.
To get around this dilemma, IEM put together two packages -- one on Davis's prospects and the other on the second ballot question: Who should replace Davis if he is recalled? As in all IEM markets, traders could predict voting percentages or simple win-lose outcomes.
The resulting array of possibilities can be bewildering not only to rookies but also to seasoned political junkies. Former Clinton White House official Mark Kleiman, a public policy expert at the University of California at Los Angeles, used IEM last week to predict that Bustamante had a 40 percent chance of being governor "when the dust clears," with Davis and Schwarzenegger at 30 percent apiece.
"Wrong," he wrote on his Web site a couple of days later, after watching IEM. Recalculating, he gave Schwarzenegger 37 percent, with Bustamante at 33 percent and Davis holding at 30 percent.
But in a telephone interview the next day, he reconsidered yet again.
"Maybe large numbers of Democrats will only vote if they see Schwarzenegger move ahead," Kleiman said. "Maybe that's why the traders go for Bustamante."
washingtonpost.com |
