Jump = to the=20 Navigation Bar

Can Markets Predict the Future?

"By election day, the markets with an average absolute = error of=20 around 1.5 percentage points, were considerably more accurate than = the=20 Gallup poll projections, which erred by 2.1 percent."

Prediction markets -- also known as information markets or = events=20 futures -- first drew widespread attention in July 2003 when it = was=20 revealed that the Pentagon's Defense Advanced Research Projects = Agency=20 (DARPA) was establishing a Policy Analysis Market to allow trading = in=20 various forms of geopolitical risk, including economic and = military=20 scenarios. The objective was to discover whether trading in such = contracts=20 could help predict future events. Bowing to a storm of criticism = that it=20 was proposing "terrorism futures," DARPA dropped the program. But = other=20 prediction markets, dealing with everything from sports and = entertainment=20 to elections and finances, have emerged and gained growing = interest and=20 participation.

In Prediction Markets (NBER Working Paper No. 10504), authors = Justin=20 Wolfers and Eric Zitzewitz describe the types of contracts that = might be=20 traded in prediction markets and then survey several applications, = with=20 special attention to market design issues. Finally, they assess = the=20 predictive value of such markets.

Wolfers and Zitzewitz begin by noting that much of the = enthusiasm for=20 prediction markets derives from the efficient markets hypothesis. = In a=20 truly efficient prediction market, the market price will be the = best=20 predictor of the event, and no combination of polls or other = information=20 can be used to improve on the market-generated forecasts. Wolfers = and=20 Zitzewitz do not insist that prediction markets are literally = perfectly=20 (or fully) efficient; however, they acknowledge that a number of = successes=20 in these markets, both within firms and with regard to public = events such=20 as presidential elections, have generated substantial interest = among both=20 political and financial economists.

In a prediction market, the researchers note, payoffs are tied = to=20 unknown future events, and how the design of how the payoff is = linked to=20 those events can elicit the market's expectations of many things. = A in a=20 "winner-takes-all" contract, for example, the contract costs a = specific=20 amount and pays off a specific amount, and only pays if a specific = event=20 occurs, such as a particular candidate winning an election. The = price on a=20 winner-takes-all market represents the market's expectation of the = probability that an event will occur. By contrast, for in an = "index"=20 contract, the amount that the contract pays varies in a continuous = way=20 based on a number that rises or falls, like the percentage of the = vote=20 received by the candidate. Finally, in "spread" betting, traders = bid on=20 the cutoff that determines whether an event occurs, like whether a = candidate wins more than a certain percentage of the popular = vote.

The various types of contracts may reveal the market's = expectation of a=20 specific parameter: a probability, a mean, or median, = respectively. But=20 prediction markets also can be used to evaluate uncertainty about = these=20 expectations, for example a family of winner-takes-all contracts = that pays=20 off only if the candidate earns 48 percent of the vote, 49 = percent, 50=20 percent and so on. This family of winner-takes-all contracts then = will=20 reveal almost the entire probability distribution of the market's=20 expectations. A family of spread-betting contracts can yield = similar=20 insights.

With these factors in mind, Wolfers and Zitzewitz examine the = data=20 compiled from analyses of the University of Iowa's Iowa Electronic = Market,=20 which has offered trade on presidential election contracts since = 1988.=20 Charting the pricebids for the past four presidential elections, = the data=20 show that as election day drew nearer, the prediction markets' = projected=20 candidate vote shares grows more accurate. Prediction markets also = beat=20 appeared better calibrated than independent analysts on the = probability of=20 the ouster of Saddam Hussein. The Hollywood Stock Exchange = likewise has=20 proved highly accurate in predicting opening weekend box office = success=20 and Oscar winners.

Even some prediction markets with very small participation have = shown=20 striking results. An internal market at Hewlett-Packard produced = more=20 accurate forecasts of printer sales than did the firm's internal=20 processes, and at Siemens an internal market predicted the firm = would=20 definitely fail to deliver a software project on time, even as = traditional=20 planning tools said the deadline could be met. In each firm, the = traders=20 numbered only between 20 and 60 employees.

Wolfers and Zitzewitz maintain that the success of prediction = markets,=20 like all markets, depends on their design and implementation. Some = key=20 design issues include: how buyers are matched to selltraders; the=20 specification of the contract; whether real money is used (some = prediction=20 markets operate for entertainment purposes and use make-believe = currency);=20 and the kind of information available to provide a basis for = trading. When=20 such factors are weighed judiciously, Prediction markets are = better at=20 pricing some events than others. The markets, like many = individuals, are=20 not always well calibrated on small probability events. In = addition,=20 markets on complex events, or events where there is likely to be = inside=20 information, often fall to attract sufficient liquidity.

Wolfers and Zitzewitz conclude with cautious optimism. They = believe=20 that prediction markets are extremely useful for estimating the = market's=20 expectation of certain events. Simple market designs can elicit = expected=20 means or probabilities, while more complex markets can elicit = variances,=20 and contingency markets can be used to elicit the markets' = expectations of=20 covariances and correlations.

Prediction markets have their limitation, the researchers = caution, but=20 they may be useful as a supplement to more traditional means of=20 prediction, such as opinion surveys, expert panels, consultants, = and=20 committees.

-- Matt Nesvisky

=20