Collective action: Experimental evidence

We conducted a laboratory experiment to test the comparative statics predictions of a new approach to collective action games based on the method of stability sets. We find robust support for the main theoretical predictions. As we increase the payoff of a successful collective action (accruing to all players and only to those who contribute), the share of cooperators increases. The experiment also points to new avenues for refining the theory. We find that, as the payoff of a successful collective action increases, subjects tend to upgrade their prior beliefs as to the expected share of cooperators. Although this does not have a qualitative effect on comparative static predictions, using the reported distribution of beliefs rather than an ad hoc uniform distribution reduces the gap between theoretical predictions and observed outcomes. This finding also allows us to decompose the mechanism that leads to more cooperation into a “belief effect” and a “range of cooperation effect”.