I investigate if, how, and why the effect of a non-deterrent contribution rule in a public goods game depends on how it is implemented: endogenously chosen or externally imposed. At the aggregate level, my experimental design allows me to disentangle to what extent the total difference in participants’ contribution behavior between the two endogenously chosen treatments, where the rule is either democratically chosen or rejected, is driven by the direct effect of the contribution rule, self-selection into institutions, information transmitted via democratic participation, and democracy per se. I find that contributions to the public good are higher if the non-deterrent contribution rule is democratically chosen than if it is democratically rejected. When the effects of self-selection and information transmitted by voting are taken into account, the total difference in contributions is not directly driven by democratic participation. At the individual level, heterogeneous treatment effects, which depend on whe her participants have been overruled by their group members in the decision making process, affect participants’ contribution behavior in ways that offset each other at the aggregate level. My findings suggest that participants are more willing to comply with a regulation that they themselves disapprove of if the majority of the group members has democratically decided in its favor.