An Experimental Implementation of Cooperative Games

Ailin Leng | The University of Queensland

I propose a continuous time frame to experimentally implement cooperative games. The proposed procedure is not restricted by the set of players or characteristic function and can be applied to any general cooperative game with transferable utility. A classic buyer-seller cooperative game is implemented as an example. In Treatment 1, any of the two buyers can form coalitions with any of the three sellers. In Treatment 2, Buyer 1 and Seller 1 cannot trade, i.e., the coalition including only Buyer 1 and Seller 1 contains zero payoffs. The two treatments have an identical theoretical prediction according to the core. Each buyer claims one unit of the payoff and each seller gets nothing. Results show that buyers claim significantly higher amounts, but sellers still keep a portion of the surplus. Moreover, one seller is usually left out and gets zero payoffs. The core correctly predicts the relative payoff differences within and between treatments, while the Shapley value correctly predicts the average payoffs over periods.