For years, prediction markets have forecast elections with eerie accuracy. Now, they’re coming soon to a policy near you.
Want to know who the next U.S. president will be? Forget the daily barrage of polls: For 20 years, an online project run out of the University of Iowa has predicted the winner of presidential elections more accurately than opinion surveys.
It's the stuff of pundits dreams, but the Iowa Electronic Markets (IEM) is the only legal market of its type in the United States. Antigambling laws prevent other such markets from taking online bets with real money, a sanction the IEM avoids because it is primarily used for research. But soon, the IEM might have competition. This summer, corporations and economists lobbied the U.S. Commodity Futures Trading Commission (CFTC) on one of the regulatory body's upcoming decisions: Should more prediction markets like the IEM be permitted on U.S. soil?
Proponents say such a move would be far more than a good gamble. "The net result [of legalizing the markets] will be to provide more accurate public assessments of political risks around the world," the Eurasia Group, a risk-assessment consulting firm, argued in a letter to the CFTC this summer. "This is an unmitigated public good."
News about the ruling could come before 2009, says Bruce Fekrat, special counsel at the CFTC. But whatever the commission decides in the short term, one forecast is clear: With or without real money, political prediction markets are coming to America.
Many in the policy world are already convinced about prediction markets reliability. Google uses them to forecast the impact and success of business decisions; public-health experts use markets to track influenza strains. They are not a crystal ball, but based on the academic research, prediction markets are better [forecasters] than anything else available, boasts John Delaney, CEO of prediction market Intrade. Indeed, a recent study published in the International Journal of Forecasting found that the IEM came closer to predicting election outcomes than opinion polls 74 percent of the time.
The exchanges work much like a stock market. Instead of companies, a trader buys "yes" or "no" contracts for a given event: a Republican will win the White House, or more than 350 people will catch avian flu this year. The price of the stock, which fluctuates as it is bought and sold, translates into the percentage chance that the event will happen. For instance, a price of 62 on a contract betting that the Guantnamo Bay prison will be shuttered by December 2009 means traders collectively think the closing has a 62 percent probability. Where money is not allowed, companies use fake currencies (such as the Gooble of Google) or prizes as an incentive to trade. Other markets avoid U.S. restrictions altogether by basing their servers abroad, as does Intrade, operating from Ireland.
Economists like the markets because they are fast, fluid, and efficient. "You look at the price of Obama stock [trading at 87.6 on Intrade as of Oct. 27] vs. the McCain stock [trading at 12.6]," offers David Perry, president of prediction markets consulting firm Consensus Point," and it is so easy to interpret." There is also a strong incentive for knowledgeable traders to put their money where their mouth is: If they're right, theres a payoff when the contract comes due.
Predictions of specific political events have also met with some success. In the lead-up to Saddam Hussein's capture in late 2003, an Intrade contract predicted the event two days before information about the ousted dictators imminent capture became publicly available. "There was an incentive for somebody who has private information to put it in the public domain," Delaney explains. Traders on Intrade also correctly forecast the resignation of U.S. Defense Secretary Donald Rumsfeld.
A twist on the predictions model might be consulting firms setting up political futures markets. If the CFTC rule change goes through, client companies could buy contracts to hedge against unforeseen events that could damage their business, be it the collapse of the Bolivian government or a renewed outbreak of war in Georgia. "Political risks could mean tens of thousands of dollars to me whether I like it or not, and there will be some firms interested in offering those [predictions as] products," explains Justin Wolfers, an economist at the University of Pennsylvania's Wharton School.
If prediction markets are so great, why not legalize? The question may be just as much a moral as a regulatory one. Take the case of the Pentagons Defense Advanced Research Projects Agency (DARPA), which in 2003 tried to start a prediction market that included 10,000 traders. The idea was to forecast political events in the Middle East -- such as Palestinian leader Yasir Arafat's assassination -- allowing the United States to tailor its security and diplomatic policies accordingly.
But betting on the death of a leader or another country's demise raised qualms that Congress could not ignore. One senator who helped spike the program called it "ridiculous" and "grotesque." Economists worried too. Nobel laureate Joseph Stiglitz feared the market would create an incentive for bad behavior: Terrorist groups such as Hamas could get a payoff by betting that a terrorist attack would occur, and then carrying it out. Manipulation was another concern. Investors might try to influence real-world events, as an investor recently did by pushing the price of John McCains Intrade "stock" up 10 points, perhaps in the hopes of boosting the Republican presidential nominee's chances on Nov. 4.
The markets' supporters claim such criticisms usually dont stand up. If terrorists, for example, placed bets on their upcoming missions, the trades would inform policymakers of the impending doom -- the very motive for creating the markets in the first place. "It's like leaving their business card at the crime scene," says Delaney, the Intrade CEO. Manipulation, too, is easy to track and costly for the manipulator. An investigation quickly pinpointed the inflationary McCain trader, who surely lost thousands of dollars.
Since DARPA's 2003 debacle, just one other government entity has dabbled in the markets. The Joint Planning and Development Office, a body set up to coordinate between government agencies, hopes to use prediction markets in designing NextGen, a revamped air-traffic control system for the United States set to launch in 2025. Traders from the airline industry, government bodies, and technical firms will be able to bet on when interim deadlines can be met, giving managers a more accurate timeline of progress. As a bonus, the markets will help the bureaucracy take into account factors such as changes of administration and shifting congressional priorities. "Politics is very hard to talk about for a civil servant. So the market allows you to avoid difficult discussions," says Yuri Gawdiak, the NASA engineer charged with setting up the system.
"Prediction markets can also help officials judge whether certain policies will work," says Robin Hanson, chief scientist at Consensus Point and an economist at George Mason University who worked on the Pentagon's futures project. If you are imagining a major initiative in African health, or if you're trying to stabilize a trade agreement, you can ask the markets whether or not you can expect success.
Cleaning up Wall Street, Delaney argues, might be another good use for prediction markets. In lieu of traditional ratings agencies such as Standard & Poor's or Moody's, for instance, the markets could be used to evaluate corporate debt or predict bank failures: As of Oct. 22, Intrade gave a 94 percent chance that more than 15 U.S. banks would collapse by mid-December, and a 57 percent chance that the number would exceed 20.
Hanson, however, worries the financial crisis might prevent the markets from going forward. "A whole bunch of things lately have been blamed on speculators. That doesn't sound like a good environment for authorizing a wider range of [market tools]." But regardless of what U.S. regulators decide, the policy world is jumping in. And while there is no way of knowing the outcome of the CFTCs ruling, there is one tried-and-true way to predict it. How about a market?
Iowa Electronic Markets