The evidence on the effectiveness of participatory development approaches in low-income countries is ambiguous. We randomly vary governance modalities to study elite capture in Ethiopian forest user groups and explore implications for livelihoods of group members. Top-down monitoring and punishment increases consumption and income, and decreases inequality. In contrast, internal monitoring has no effect on livelihoods. Additional heterogeneity analysis, based on observational data, reveals that while top-down monitoring works in groups where forest benefits are unimportant, internal monitoring improves economic outcomes in those groups where forest benefits are an important component of rural livelihoods. This suggests that participatory approaches work if targeted participants have strong incentives to voluntarily contribute effort.