Forest Nelson and several academics from the University of Iowa were sitting down to lunch in March of 1988, just after the Michigan Democratic caucus. Michael Dukakis had been predicted to beat Jesse Jackson by a landslide. Instead, Jackson easily pulled off a startling upset.
“One of the other two guys said, ‘Boy, if the futures markets in Chicago did as bad a job of predicting the November price of corn as those opinion polls did of predicting the very next day’s election, then those markets wouldn’t exist,’” recalled Nelson, an economics professor still at the university. “Of course, that threw the whole question on the table.”
The three started scribbling on every napkin at hand, and then every other napkin they could borrow. The lunch went on for hours.
Right there, the group created what came to be known now, more than two decades later, as the Iowa Electronic Markets, a real-money political prediction market among a relatively small group of people based in Iowa that has routinely been more accurate than national public opinion polls at predicting the outcome of presidential elections.
For years, the concept of trading on future political events as if they were bushels of corn has been a nonprofit academic exercise (as well as, more recently, fodder for the Ireland-based market Intrade).
But even with a real dollar or two involved, those were essentially paper trades. Just before the holidays, the North American Derivatives Exchange proposed trading political event contracts on a regulated market in the U.S.
If this sounds like an intriguing but discomforting idea — do we really want people betting on our democratic process? — the Commodity Futures Trading Commission agreed. After the New Year, it asked Nadex to put the plans on hold while it seeks public comment and mulls whether or not it would be a good idea to dramatically expand a concept that has worked as a limited university research project. Iowa’s contracts are worth a dollar. Nadex wants to set them at $100.
• • • • • • • • • • • • • • •
Even the Iowa market has spent considerable time under the scrutiny of the Commodity Futures Trading Commission, or CFTC.
“We faced several legal issues early on, some of which we weren’t even aware of,” Nelson said. “We thought it was such a good idea as a research project that we blindly went ahead without really thinking about what we were doing — until later, someone said, ‘Boy, is this legal?’”
CEO of Nadex: Safeguards in Place
You correctly identified a very important concern regarding market manipulation. We spent considerable time thinking about this issue when we designed the contracts and we believe the safeguards we have in place should deter such activity in our market:
• The North American Derivatives Exchange is a retail market. We propose to limit the position an individual or entity can have to 2,500 lots (except market makers, which have higher limits). These relatively low limits combined with enough liquidity should minimize the ability and incentive for anyone to even attempt to manipulate the market.
• Like many other prediction markets, our market is a fully collateralized market; that is, the market participant has to fund, in full, the maximum potential loss on the contract at the time of the trade. As a result, market participants cannot rely on leverage to increase their potential impact on the market.
• Studies with which I am sure you are familiar (such as by Paul W. Rhode and Koleman S. Strumpf) have shown that attempts to manipulate such markets don’t work. Even if prices can be moved temporarily, they quickly revert to market levels. We think our markets, with the regulatory structures that we have in place, should be even more resistant to manipulative efforts.
• Nadex has extensive real-time market surveillance systems which will alert us to any unusual price movements and allow us to take prompt action in the event that an attempted manipulation does occur.
I believe that it is time for Professor Nelson’s theoretical work at the IEM to (as he says) “turn out in actual applications.”
— Yossi Beinart, CEO of Nadex
In its first election year, the Iowa market operated with the blessing of the state’s attorney general under a small-stakes gambling statute that allowed for, say, Kiwanis clubs to have weekly poker night. But the market was limited to the “club” of the University of Iowa. Only students, faculty, and staff were permitted as traders. When the researchers wanted to expand their trading base — and, theoretically, the market’s accuracy — in 1992, they went to the CFTC for permission.
In response, they got a “no action” letter. Technically, what they were doing was illegal. But the CFTC concluded that the market didn’t violate the public’s interest, and so it was permitted to continue under a series of restrictions, and under the understanding that all of this was meant to be an academic undertaking. The researchers argued that they had to have some money on the table, or the market wouldn’t work (and even a dollar seems to be enough to get people’s attention). But no single trader could invest more than $500 in the market, and the market itself had to be limited to 1,000 traders. The researchers also weren’t allowed to advertise or charge commission.
The Iowa market has been operating under that no-action letter ever since (with periodic inquiries to ensure everyone is in line). It’s hard to imagine that Nadex would want to run under similar limitations.
Today, the Iowa market still consists mostly of traders in Iowa, with some others from around the country and globe. It offers two kinds of $1 contracts: one on the vote share a presidential candidate is expected to win, and the other on the overall likelihood of his or her winning. On a winner-take-all contract, if you believe Mitt Romney has a 60 percent chance of winning the presidency this November, you ought to be willing to pay 60 cents for a contract predicting that outcome. If you think he’ll win with 52 percent of the vote, then you should pay that much in pennies for a vote-share contract.
Under this system – and even with all these Iowans – the whole market has been surprisingly accurate.
“All the pollsters come to us and say, ‘Your markets can’t work right because they don’t have representative samples, your traders don’t fit the national electorate,’” Nelson said. “We’d compute the numbers and say, ‘You’re exactly right. Our traders are too white, too rich, too educated, there are too many of them in the academic community.’ In almost any way you can think, they’re biased in some way compared to the national electorate.”
Yet the system still works. Nelson explains the logic this way: if you had a car problem, you wouldn’t go poll 100 random people on the street about what to do about it. You’d go to expert mechanics. And the traders in the Iowa market have made themselves experts in a similar way. The market doesn’t ask them who they plan to vote for, but who they think will win. By virtue of their participation in the market — and that they have money on the line — traders have a serious incentive to become as knowledgeable about the topic as possible, just as stock traders and hedge funds do for equities or more traditional commodities. The process is similar to surveying experts; the survey is just done on a free market.
• • • • • • • • • • • • • • •
Through years of tweaking this system, Nelson and his colleagues have addressed a lot of their original research questions: Can a market device predict a non-market outcome? Can a market predict an election better than polls can? Do the traders have to represent the full population? How much money do you need on the line to keep them interested? And exactly how do you design the market to tap into that potential for expertise among traders to yield an accurate result?
In a way, by answering many of these questions, the Iowa market has cleared the way for a “real” market like Nadex to enter the field. Nelson feels torn about this.
“From the very beginning, we thought this was such an obvious, such a natural thing, why haven’t the futures markets in Chicago been doing this thing all this time?” he asked. “We gave them five to10 years before they would catch on, and it would be offered by those ‘real exchanges in Chicago.’ The fact they haven’t is in one sense a surprise, and a bit of a disappointment. I have been really disappointed by that. On one hand, we’re theorists, but we’re also interested in seeing our theoretical work turn out in actual applications.”
On the other hand, though, Nelson says the academic in him wants to point out that there are still a lot of unanswered questions. Yes, the Iowa market has been running for 24 years. But that’s really just six presidential election cycles. That’s not enough data to really prove over time how accurate these markets are, and what ensures their accuracy.
Nadex, for its part, has made two primary arguments for why the market would be a good idea. It argues that political contracts would help retail investors control risk by hedging against their current positions (if, say, the oil company you invest in isn’t likely to benefit by Obama’s re-election, at least you’d make some money off that event if it happened).
The other argument — and Nadex cited the evidence of the Iowa Electronic Market in its announcement — is that such a market would provide the general public with more accurate information about an issue of enormous public importance than we’re currently getting from opinion polls.
Nelson is more swayed by the second argument. But the CFTC has also come back to the same question it originally asked the Iowa researchers: Is this in the public interest? More specifically, in Nadex’s case, would such a market create incentives to try to manipulate elections (above and beyond all the Super PAC and Citizen’s United-style juggling that already occurs)?
The ability to guard against such manipulation largely lies in the design of the market. But everything Iowa has learned all these years doesn’t necessarily offer a perfect blueprint for how to scale up the concept to what Nadex has in mind.
“It’s relevant for the CFTC to ask precisely that question,” Nelson said. “Is this something that could be used to manipulate election outcomes? That’s an important question. I don’t have an answer for it.”