Recent empirical research shows that prediction markets yield high-quality predictions about future events.
A prediction market is a market in which participants bet on a future event. The outcome of the event determines winners and losers, and pre-event prices in the market reflect participant expectations.
For example, markets on the outcome of the 2012 World Series suggest that the Chicago Cubs have about a 1-in-1,000 chance of winning. Large prediction markets exist for political events, economic statistical releases and corporate events (such as company sales results), to name a few.
Economic reasoning alone is unclear as to the relationship between prices in these markets and the probabilities of the outcome of events they are supposed to predict, and unclear as to whether the markets will be successful in terms of sustaining participation. Ultimately, that’s an empirical question, not a matter of economic reasoning.
The efficient-prediction perspective says prediction market prices should be close to the probabilities or expectations of the featured events because market participants receive clear and direct monetary rewards for accurate predictions (winning their bets) and monetary penalties for inaccurate predictions (losing their bets). These penalties and rewards give market participants incentives to gather information and form the best prediction they can.
The efficient-prediction perspective does not say that markets are a flawless crystal ball. For example, the Intrade betting market was saying that the Supreme Court was only 25 percent likely to uphold the mandate portion of the new health-care law, which the court did uphold. That market never said that the ultimate outcome was impossible, just one-third as likely as the alternative.
However, the efficient-prediction perspective presumes that the market exists on a scale large enough to create significant rewards for those with accurate predictions. Another perspective, “no-trade,” says prediction market participation will be low, if not zero, because traders suspect that a person would take the opposite side of their trades only because he had superior knowledge about the outcome. A market cannot survive if its only participants are “insider traders.”
Paradoxically, a prediction market cannot succeed unless it multitasks – it must serve an additional purpose separate from predictions, so that the participants with information about the outcome have counterparts to take the other side of their trades.
In the case of sports and political markets, that additional purpose is entertainment – people enjoy engaging in the activity and are willing to participate even if their expected profits are zero or negative. These people are participating for various reasons other than prediction.
(It is sometimes said that Las Vegas-type sports-event betting markets, such as the one that produced the odds on the Chicago Cubs, are not pure prediction markets because some participants are bookmakers and others are casual bettors, so the odds do not reflect precisely how the money is bet. But that’s my point: successful prediction markets do not consist solely of insider traders).
The entertainment and other nonprediction functions of the market will also be reflected in the market prices, so they no longer solely represent probabilities or accurate expectations. For example, sports fans may enjoy cheering for the home team and in doing so might place more entertainment value on a home-team bet. If so, market prices may reflect not only the home team’s probability of winning but also the magnitude of the entertainment value.
The market for United States government bonds offers another example. The prices of those bonds, especially the gaps in prices between bonds with and without inflation-indexed payouts, reflect market expectations about future expectations. But the prices may also reflect changes in the value of other rewards to market participation, such as liquidity or the satisfaction of regulatory requirements.
The relationship between prediction market prices and event probabilities is therefore an empirical question, requiring many events in order to compare market predictions with alternatives. The economists Erik Snowberg, Justin Wolfers and Eric Zitzewitz have conducted a number of empirical studies and summarized them in a recent paper. They find that market prices more accurately predict events than do professional forecasters and polls that are often given authority in assessing the likelihood of various outcomes.
In many cases, poll results and professional forecasts offer no information about event outcomes beyond what is already reflected in prediction market prices.