Costly Actions, External Incentives and Prediction Markets
Consider a prediction market of multiple rounds with a security contingent on a certain event whose final outcome is decided by the agents who also trade in the market. One such prediction market is one in which two agents, Alice and Bob, are trading on the likelihood of a project both are working on complete. Prior research either only considers the expected rewards in the prediction market or if external incentives are present, then only a low number of rounds in the prediction market, to our knowledge at most 2. In addition, the existing literature assumes that when external incentives exist, there is no net difference between the cost of different actions agents may take outside of the prediction market. For example, it is the same cost for either Alice to work hard to complete the project as it is for her to ``loaf'' and not work hard. In this work we consider a 2-round round setting in which agents' cost of external actions differ. We show that when external action costs differ but are within a proper range, a prediction market is incentive compatible regardless of the initial market estimate, something that currently is not shown in the existing literature.
Cite this version of the work
Chen Di (2017). Costly Actions, External Incentives and Prediction Markets. UWSpace. http://hdl.handle.net/10012/12221