The Politics of Information Markets, and Information Markets in Politics Adam Meirowitz Joshua Tucker Princeton University Politics of Information Markets • The Legacy of PAM – Political Reaction – Media Reaction – Result for Program • Lessons from PAM for future political information markets The Legacy of PAM • Summer of 2003: Defense Advanced Research Projects Agency (DARPA) funds development of a Policy Analysis Market for futures on political events in the Middle East – Robin Hanson knows far more about the project. What was DARPA’s proposed informaton market called? “Unbelievably stupid” - Sen. Dorgan “A futures market in death” - Hillary Clinton “make believe markets trading impossibilities that turn the stomach” - Sen. Wyden “Dr. Strangelove caught in bizarre betting parlor” What happened? • July 28, Sens. Wyden and Dorgan (D) press conference • Morning of July 29, NYT and major press sources cover • Afternoon of July 29, the program is cancelled • Deputy Defense Secretary Wolfowitz: “I share your shock at this kind of program. We’ll find out about it but its being terminated.” Postscript • Admiral Poindexter eventually retires in embarrassment • Bipartisan criticism • All this over a market that never opened Number of Newspaper articles on PAM (July-August 2003) Lexis-Nexis search Opposed Neutral In favor Total News 7 39 2 48 Opinion 49 4 14 67 Total 56 43 16 115 Media issues • Majority of coverage was superficial dealing with “morality” • A few stories speculate about: – Inducement of terrorism-profit motive – Terrorists getting rich (External) – Manipulation of market by terrorist (Internal) • “Primacy of Iowa” as reference point Packaging • Problem: Profit as inherently dangerous concept politically • Solution: Frame markets comparatively Profit as inherently dangerous concept politically • “Immoral” to profit on “bad” events • Threat to manipulation when profit is at stake seems more consequential in a political context – Throwing a basketball game vs throwing an election. – Manipulating a stock price vs. a terrorist attack Packaging suggestions • Frame market as improvement over other aggregation institutions – Organizations already exist for these purposes, so question is not about topic – Question is: are markets better at aggregating information? • Do the actual welfare comparison so politicians see the trade offs • Provide this information as part of introduction so politicians, press can reference it in initial evaluations Information Markets in Politics • Incentives when information market is imbedded in a policy making setting • Information from market prices can effect beliefs about states which effect preferences about policies • But this may lead to an incentive problemManipulate beliefs to influence policy selection Simple informational environment • Two states x ε{a,b}, two policies p ε{a,b}. • Some agents (θ=1) want p=x some (θ=-1) want p≠x. • Agents know which preference type they are and learn a signal s that is informative about x. • Mechanism aggregates information and preferences and selects p • Question: Can we get the full information majority rule core policy? Communication and voting • Without transfers communication will not work. • Some type (θ=-1) is likely to be in minority • They want to influence beliefs about x • So lying about s is a desirable deviation State contingent transfers • With transfers that depend on message and state truthful implementation is possible. • Transfer to i needs to compensate for probability that i’s information moves posteriors about x in a meaningful manner. Message contingent transfers • With transfers that depend on just other messages truthful implementation is possible. • Incentive to be truthful as best predictor of other messages is your own message. Decentralizing transfer schemes • Give agents choice of Arrow Securities on state or on majority message. • Prices are selected to satisfy IC conditions and budget balance. How about endogenous prices • Markets as games (Shapely and Shubik, 1977; Jackson and Peck, 1999) • Endow traders with risk free asset and Arrow security that pays off if x=a. • Allow trader to decide if she wants to sell B units of the risk free (action 1) or sell Q units of the risky asset (action 0). • Set price of P=Bn1/Qn0 Positive result • We can fix values of B and Q to achieve the full information majority rule core. But…. • Letting participants determine how much they want to trade, can be problematic. • Finite populations –full aggregation fails (Dubey, Geanakopolis and Shubik 1987) • Failure of efficient markets hypothesis in these games without politics (Peck and Jackson) • Open question when imbedded in political problem.