How to Select the Leader in a One-Shot Public Goods Game: Evidence from the Laboratory

We tested five treatments: a baseline control with simultaneous contributions and four leadership treatments—Random Leadership (RL), Voluntary Leadership (VL), Random Leadership from Voluntary Candidates (RL-VC), and Voluntary Leadership from a Random Candidate (VL-RC). The data revealed several notable findings. First, introducing a leadership role did not consistently boost overall contributions compared with the control, and the pure voluntary mechanism (VL) performed the worst.

Groups without any leader contributed significantly less than those with a leader. VL-RC was the most effective endogenous mechanism: Imposing a higher barrier to leadership, it reduced the diffusion of responsibility and free-riding among potential leaders. In contrast, mechanisms allowing multiple leaders (VL and RL-VC) tended to induce more selfish behavior, as evidenced by lower average contributions from leaders in these settings.

Additionally, leaders generally contributed more than followers. However, the strength of the leader-follower influence varied across treatments. There was a significant positive correlation between leader and follower contributions in VL but not in other treatments.

The findings suggest that the informativeness of signaling and reciprocity of leadership selection mechanisms is crucial for public goods provision. Policymakers and organizational leaders shall create barriers in selecting candidates and privately encourage the candidate who values the common good the most to lead by example voluntarily.