In this paper we test the effectiveness of a compensation mechanism when a negative externality is produced. It allows agents suffering from the negative externality to compensate those who reduce its production. Transfers are implemented via a two-stage design which is 10 an adaptation of Varian’s mechanism. It has been previously tested in the lab with different types of games, and its effectiveness turns out to depend on the experiment, for unclear reasons which we try to decipher in this paper. Three possible explanations, choice overload, mere coordination and inequality, are proposed and studied. We show that, other things equal, the larger the size of the strategy space, the lower the mechanism’s efficiency (choice overload ef15 fect). Perhaps surprisingly, the data show that the appearance of additional equilibria does not jeopardize effectiveness (no mere coordination effect). Finally, inequality of outcomes plays a key role (fairness effect).