Does transparent leadership promote cooperative groups? We address this issue using a public goods experiment with exogenously selected leaders who are able to send non-binding contribution suggestions to the group. To investigate the effect of transparency in this setting we vary the ease with which a leader's actions are known by the group. We find leaders’ suggestions encourage cooperation in all treatments, but that both leaders and their group members are more likely to follow leaders’ recommendations when institutions are transparent so that non-leaders can easily see what the leader does. Consequently, transparency leads to significantly more cooperation, higher group earnings and reduced variation in contributions among group members.

Previous research offers compelling evidence that leaders suffice to effect efficiency-enhancements on cooperation, yet the source of this effect remains unclear. To investigate whether leadership effects can be attributed exclusively to the common information that leaders provide to a group, irrespective of the source of that information, we design a public goods game in which non-binding contribution suggestions originate with either a human or computer leader. We find that group members’ decisions are significantly influenced by human leaders’ non-binding contribution suggestions, both when the leader is elected as well as when the leader is randomly chosen. A leader's suggestion becomes an upper bound for group member's contributions. Identical suggestions do not impact the group members’ decisions when they originate with a computer, thus supporting to the view that information provided by human leaders is uniquely able to establish welfare-enhancing norms.

Previous research provides compelling evidence that human leaders are sufficient for efficiency-enhancing leadership effects on cooperation in groups, yet little is known regarding the contexts in which human leadership is necessary in this regard. For example, a simple explanation is that leaders provide a common signal to which group members naively respond. Because it is not necessary for a common signal to originate with a human leader, such an explanation could have critically different policy implications than are suggested by currently accepted accounts of a leader's value. We here report rigorous evidence on the extent to which leadership effects in a social dilemma require human leaders. We find that group members' decisions are significantly influenced by human leaders' non-binding contribution suggestions. On the other hand, those same suggestions do not impact group members' decisions when they do not originate with human leaders.

Does transparent leadership promote cooperative groups? We address this issue using a public goods experiment with exogenously selected leaders who are able to send non-binding contribution suggestions to the group. To investigate the effect of transparency in this setting we vary the ease with which a leader's actions are known by the group. We find leaders' suggestions encourage cooperation in all treatments, but that both leaders and their group members are more likely to follow leaders' recommendations when institutions are transparent so that non-leaders can easily see what the leader does. Consequently, transparency leads to significantly more cooperation, higher group earnings and reduced variation in contributions among group members.

Previous research offers compelling evidence that leaders suffice to effect efficiency-enhancements on cooperation, yet the source of this effect remains unclear. To investigate whether leadership effects can be attributed exclusively to the common information that leaders provide to a group, irrespective of the source of that information, we design a public goods game in which non-binding contribution suggestions originate with either a human or computer leader. We find that group members' decisions are significantly influenced by human leaders' non-binding contribution suggestions, both when the leader is elected as well as when the leader is randomly chosen. A leader's suggestion becomes an upper bound for group member's contributions. Identical suggestions do not impact the group members' decisions when they originate with a computer, thus supporting to the view that information provided by human leaders is uniquely able to establish welfare-enhancing norms.