<p>Endogenous peer reward mechanisms have been shown to be effective in promoting contributions in public goods games (Yang et al. 2018). However, prior research has not shown whether such a mechanism itself is capable of being fully self-determined by decision-makers voluntarily. In a laboratory experiment, we endogenize the peer reward mechanism by allowing players to select their preferred fraction of total income in the public goods game to use for rewarding their peers’ public goods contributions. Players are then sorted into groups based on their declared fraction, with the group level fraction implemented as the minimum of a group’s individual preferences. We confirm that the reward mechanism successfully reverses the typical time decay in contributions found in public goods games, while maintaining high overall contribution levels, and in addition, implemented “tax” rates increase over time across all groups, demonstrating that players successfully learn the benefit of a higher fraction of income devoted to peer rewards. Contributions and reward vote-casting are both empirically increasing in the tax rate, and full contributions can be reached in practice at the 33% tax threshold. Our study shows that endogenous self-determination of sanctioning capacities and resource redistribution can be effective in sustaining high levels of cooperation.</p>