Five Open Questions About Prediction Markets 
Justin Wolfers Eric Zitzewitz
(7882 words)

Abstract Interest in prediction markets has increased in the last decade, 
driven in part by the hope that these markets will prove to be valuable 
tools in forecasting, decisionmaking and risk management – in both the 
public and private sectors. This paper outlines five open questions in the 
literature, and we argue that resolving these questions is crucial to 
determining whether current optimism about prediction markets will be realized.

This Draft: January 21, 2005

Interest in new applications of prediction markets is focused in three 
domains: forecasting, decision making, and risk management. The success of 
corporate prediction markets in forecasting printer sales (Chen and Plott, 
2002) and project management (Ortner, 2002) has stimulated interest in their 
application to other business problems, and firms such as NewsFutures, Net 
Exchange and Incentive Markets have sprung up to meet this demand. Among the 
applications being discussed are “decision markets” in which securities are 
traded that pay off based on an outcome (e.g., revenue from a product) if 
and only if a particular decision is made (e.g., the product is launched). 
The idea behind these markets is to elicit knowledge from within an 
organization that might otherwise be lost somewhere in a poor organizational 
form.

Similar ideas have been discussed in the policy realm. DARPA’s Policy 
Analysis Market would have launched securities designed to capture the 
probability of specific events, along with contingent securities designed to 
capture the outcome of specific policies. During the 2004 primary season, 
the Iowa markets ran securities designed to predict the general election 
success of the Democratic contenders, while later in the year Tradesports 
ran similar contracts designed to capture the effects of geopolitical and 
economic events on the election (see Wolfers and Zitzewitz, 2004b for 
details). Abramowicz (2004) and Hahn and Tetlock (2004) envision using 
prediction markets to assist in policy making by extracting expert opinion 
in a credible and objective manner. Hanson (2003) goes even further, arguing 
that policy makers should simply define a GDP+” measure of social welfare, 
and that policy decisions should be based entirely on market-based 
predictions of which policies maximize this social welfare measure.

...

At the same time, contracts on more “serious” topics (economic numbers, 
financial market outcomes, contingent political contracts) attract volume 
that compares favorably with volumes on the successful Iowa vote share 
markets. These traders may have been attracted more by the substantial free 
media Tradesports has received, and uninformed trading in these contracts 
may more of a combination entertainment and overconfidence motives. As the 
wonkishness of the contract rises, however, volume and liquidity falls 
rapidly. The few Tradesports markets of the sort planned by the DARPA Policy 
Analysis Market (e.g., will there be a Palestinean state by 2005?) have not 
been very successful.

...

The consulting firms running corporate prediction markets have taken the 
same approach as experimental economists on campuses: they have subsidized 
participation, allowing everyone to leave a winner, albeit to varying 
degrees. The Policy Analysis Market was also supposed to have involved a 
subsidy for participation. If the information to be generated by the market 
is important enough to justify the subsidy and a willing financier can be 
enlisted, then this approach can work. It is important, however, that 
subsidies are not designed in such a way that they can be gamed, as one 
could imagine with per-trade subsidies or with some schemes for 
subsidization via a money losing market maker.