Volunteering a Public Service: An Experimental Investigation

In some public goods environments it may be advantageous for heterogeneous groups to be coordinated by a single individual. This “volunteer” will bear private costs for acting as the leader while enabling each member of the group to achieve maximum potential gains. This environment is modeled as a War of Attrition game in which everyone can wait for someone else to volunteer. Since these games generally have multiple Nash equilibria but a unique subgameperfect equilibrium, we tested experimentally the predictive power of the subgame-perfection criterion. Our data contradict that subjects saw the subgame-perfect strategy combination as the obvious way to play the game. An alternative behavioral hypothesis – that subjects were unable to predict accurately how their opponents would play and tried to maximize their expected payoff – is proposed. This hypothesis fits the observed data generally well.