In July 2003, press reports began to surface of a project within the Defense Advanced Research Projects Agency (DARPA), a research think tank within the Department of Defense, to establish a Policy Analysis Market that would allow trading in various forms of geopolitical risk. Proposed contracts were based on indices of economic health, civil stability, military disposition, conflict indicators and potentially even specific events. For example, contracts might have been based on questions like “How fast will the non-oil output of Egypt grow next year?” or “Will the U.S. military withdraw from country A in two years or less?” Moreover, the exchange would have offered combinations of contracts, perhaps combining an economic event and a political event. The concept was to discover whether trading in such contracts could help to predict future events and how connections between events were perceived. However, a political uproar followed. Critics savaged DARPA for proposing “terrorism futures,” and rather than spend political capital defending a tiny program, the proposal was dropped.This is the first paragraph of an article on Prediction Markets (pdf) by Justin Wolfers and Eric Zitzewitz in the Journal of Economic Perspectives (Volume 18, Number 2. Spring 2004. Pages 107–126). The uproar over the DARPA plan was the first most people would have heard of prediction markets.
Prediction markets, which are also called “information market” or “event futures” are markets where participants trade in contracts whose payoff depends on unknown future events. If prediction markets are truly efficient then the market price will be the best predictor of the event being traded on, and no combination of available polls or other information can be used to improve on the market-generated forecasts.
The best known of the prediction markets is most likely the Iowa Electronic Market, run by the University of Iowa. The original Iowa market was setup in 1988 to trade in a contract that would paid 2.5 cents for each percentage point of the vote in the presidential election won by Bush, Dukakis or others. The Iowa Electronic Market has explained to run markets based on the 2003 California gubernatorial election, the 2004 presidential election, the 2004 Democratic presidential nomination, 2008 presidential nominations for both the Democratic and Republican parties, the 2008 U.S. presidential election and how the Federal Reserve will alter the federal funds interest rate. Event markets have now been set up by universities in other countries trading on the outcomes of their own elections. Examples are the Austrian Electronic Market run by the Vienna University of Technology or the University of British Columbia Election Stock Market that focuses on Canadian elections.
Prediction markets work by harnessing the “wisdom of crowds”. Inside every trader's head are beliefs and information about the future. Traders use their information and beliefs to buy and sell stocks in prediction markets. The price that emerges from this trading represents the sum total of the crowd's wisdom. Prediction markets work well. In every head-to-head comparison of prediction markets against rival forecasting methods in the last twenty years, prediction markets have trumped the competition. For example, since 1988,Iowa Electronic Market has successfully predicted every US presidential race. The accuracy of Iowa's prediction markets were compared with 964 polls taken in the five presidential races that occurred in that time, and came out on top three out of four times.
Prediction markets work so well for two basic reasons. First, they ask the right question. Pollsters ask who you will vote for. Prediction markets ask who is going to win. Second, prediction market traders put money where their keyboards are, so it pays to do a little homework.
Wolfers and Zitzewitz detail the three basic contracts that can be traded in prediction markets.
First, in a “winner-takeall” contract, the contract costs some amount $p and pays off, say, $1 if and only if a specific event occurs, like a particular candidate winning an election. The price on a winner-take-all market represents the market’s expectation of the probability that an event will occur (assuming risk neutrality).For example, a contract that pays $1 if there is a National Prime Minister after the next election and zero otherwise. If contracts trade at, say, 80 cents then there is a 80% chance that there will be a National Prime Minister heading the government after the next election.
Second, in an “index” contract, the amount that the contract pays varies in a continuous way based on a number that rises or falls, like the percentage of the vote received by a candidate. The price for such a contract represents the mean value that the market assigns to the outcome.So, for example, a contract may pay $1 for each percentage point of the vote that Labour gets in the next election. That is, it would pay $55 if Labour gets 55% of the vote.
Finally, in “spread” betting, traders differentiate themselves by bidding on the cutoff that determines whether an event occurs, like whether a candidate receives more than a certain percentage of the popular vote. Another example of spread betting is point-spread betting in football, where the bet is either that one team will win by at least a certain number of points or not. In spread betting, the price of the bet is fixed, but the size of the spread can adjust. When spread betting is combined with an even-money bet (that is, winners double their money while losers receive zero), the outcome can yield the market’s expectation of the median outcome, because this is only a fair bet if a payoff is as likely to occur as not.For example, a contract pays even money if the Greens win more than y*% of the vote in the next election. Such a contract costs $1, pays $2 if y > y*. Pays $0 otherwise. You bid according to the value of y*.
Thus the three basic forms of contracts will reveal the market's expectation of a specific parameter: a probability, mean or median, respectively.
The good news is that prediction markets have now been set up in New Zealand. iPredict is a company owned by Victoria Link Ltd, itself wholly owned by Victoria University of Wellington, and the Institute for the Study of Competition and Regulation (ISCR), whose primary purpose is academic research. They have developed a series of markets based on New Zealand's political and economic future.
Currently there are markets on:
- Who will be the next Prime Minister?
- Party Vote Markets: What share of the party vote will each party get?
- Other Politics: Will Winston Peters resign or be sacked as Minister of Foreign Affairs
- Will Peters be re-elected to Parliament?
- Will New Zealand to go into recession in 2008?
- What will third quarter inflation be?
Update: Kiwiblog posts on iPredict
Update 2: A not very useful puff piece on iPredit from the TVNZ website. Video from the TVNZ business program available here. TV One news item here.
Update 3: The Visible Hand in Economics comments on Putting your money where your mouth is
Update 4: The National Business Review notes that New predictions market kicks off tomorrow and New online prediction market launches today.
Update 5: TV3 news says Prediction website puts chances of Peters resignation at 84%