Persistence of power: Repeated multilateral bargaining with endogenous agenda setting authority

We extend a simple repeated, multilateral bargaining model to allow successful agenda setters to hold on to power as long as they maintain the support of a majority of other committee members. Theoretically and experimentally, we compare this Endogenous Power environment with a standard Random Power environment in which agenda setters are appointed randomly each period. Although the theoretical analysis predicts that the two environments are outcome equivalent, the experimental analysis shows substantial differences in behavior and outcomes across the games. The Endogenous Power environment results in the formation of more stable coalitions, less-equitable budget allocations, the persistence of power across periods, and higher long-run inequality than the Random Power environment. We present evidence that the stationary equilibrium refinements traditionally used in the literature fail to predict behavior in either game.