Thursday, May 10, 2007

Philadelphia's Future of Political Markets

I've been participating in a political futures market on the Iowa Electronic Markets, which, according to Slate's Melonyce McAfee, is the only market in the United States that legally allows people to bet real money on the outcomes of political races:
Why is the IEM allowed to accept cash? The government has said it won't prosecute. The Iowa markets were started in 1988 as a teaching aid at the University of Iowa's Tippie business school to give students experience in a real futures marketplace. In 1993, the not-for-profit entity received a "no-action letter" from the Commodity Futures Trading Commission, which regulates certain markets associated with financial, commercial, or economic interests. CFTC staff said it would leave the IEM alone as long as it operates under certain stipulations—that it exists for academic purposes, doesn't engage in advertising, and administrators don't profit from it. No other political prediction market in the United States has gotten the same deal.
Link. The markets are designed to predict the winner of elections by utilizing "the wisdom of crowds," the idea being that a crowd of people, aggregating their information and judgment, will end up at the optimal outcome. Futures markets, some say, are as good or better at predicting events than are individual experts or even polls. The general idea is that you buy a share in a particular event (say, Barry Goldwater being elected president) and your return is based on whether that event comes to pass. In a winner-takes-all (WTA) market, you would earn, say, a dollar if Goldwater was elected and zero if he was not. In a vote-share (VS) market, you would earn as many cents on the dollar as Goldwater earned percentage points of the vote--if Goldwater got 43 percent of the vote, you would earn back 43 cents, whether or not Goldwater won.

The market I've been trading in is meant to predict the outcome of the Philadelphia mayoral primary, which is five days away. Since Philadelphia is a heavily Democratic city, the winner of the Democratic primary is more than likely going to be the winner of November's general election and the next mayor. There are five major candidates in the primary, and almost all of them have lead in the polls or otherwise been the presumptive winner at some point in the campaign. Right now polls show former city council member Mike Nutter surging ahead of businessman Tom Knox, who had been leading for the past couple months. Before that, congressmen Bob Brady and Chaka Fattah had each been considered frontrunners.

I joined the market and started trading with a measly five dollars because I wanted to help in the prediction, because I'm a little bit of a wonk and had been following the race more than your average citizen, and because the Philadelphia Inquirer and University of Pennsylvania put up my five-dollar stake. So has the market done a good job at efficiently predicting the winner?

Not really. Of course, the election hasn't happened yet, so I can't say who the winner will be or how right the market was. But what I can say with some certainty is, probably due to low participation and skewed results by a few idiot traders, the market has failed.

The WTA market is pretty straightforward and there is some hope for it. In that market, you should simply buy shares in the candidate you think is going to win the election. The price you are willing to pay should be based on the chances you perceive the candidate having. In that market, Nutter is currently selling at the highest price, 70 cents, because as an aggregate the participants think Nutter has a 70 percent chance of winning. Knox is behind at 39 cents. The order makes sense, but the prices should tell you something is a little bit wrong: there's no way those two candidates could have those chances of winning, because the total chances for all candidates must add up to 100 percent.

It's the VS market that is completely haywire. The rules there make for different incentives: rather than just buying the candidate you think is going to win and bidding up as high as you think his chances of winning are, you should bid on any candidate where you think the price of the share, in cents, is lower than the share of the vote the candidate is going to get. In a perfectly equal five-man race, that would mean each candidate would get 20 percent of the vote, and each share would pay 20 cents. The Philadelphia race isn't perfectly equal, but it is fairly close. The current frontrunner is polling at 31 percent and the runner-up at 21 percent, with another 21 percent undecided. It's pretty unlikely that anybody, and certainly not more than one person, will get more than 40 percent of the vote. If you accept that premise, it would be irrational to pay more than 40 cents for any single candidate. Yet the asking prices for shares in three of the candidates in the Philly VS market are higher than 40 cents, and have stayed there for days: Knox is at 42 cents, Nutter at 47, and Fattah at a ridiculous 93.5 cents. Again, the order here, though a little surprising, isn't that unbelievable: Fattah could win the election with Nutter and then Knox runners up. But there hasn't been much trading in the VS market for weeks, and those asking prices are to blame. It would be totally irrational to pay 93.5 cents for Fattah unless you think he will win more than 93.5 percent of the vote. No one would ever think that possible—this is Philadelphia, after all, not the Soviet Union. Yet at least one person does: the last price at which Fattah sold was 92.5 cents. That trader will lose money on that share unless Fattah's immediate family are the only voters next Tuesday. Good luck with that one, buddy.

My guess is that the Philadelphia market has flaws because it has too few participants and because at least a few of those participants it does have (like whoever paid 92.5 cents for a share of Fattah in the VS market) are either completely irrational or don't understand the rules of the game. That's the problem with relying on the wisdom of crowds: you need a crowd, and you need it to have some wisdom.

Indexed by tags , , , , , , , , , .


Post a Comment

<< Home