Effects of Democratic Leader Selection in Inter-Group Games

• In Event: Experiments in Political Economy

Abstract

Major decisions about inter-group interactions are made by leaders, and groups vary significantly in how those leaders come to power. Our proposed research assesses how leader selection mechanisms affect inter-group contests, which are a natural analogue to conflictual interactions like war or lobbying for a preferred policy. We conduct our study in an experimental setting in order to precisely control the parameters of the interaction between groups, isolating the causal effect of variation in our experimental protocol. In our experiment, we set up inter-group contest games, vary characteristics of leader selection mechanisms, and observe changes in how leaders behave. In this study, we are specifically interested in the effect that democratic leader selection has on inter-group contests.

Existing experimental literature on inter-group competition focuses largely on how individuals contribute to inter-group conflict in the context of free-riding problems. Furthermore, few works examine the role of leadership selection, and the ones that do examine its effects on intra-group dynamics. There is a dearth of experimental work that examines how representatives behave in inter-group contests. As many inter-group interactions in the real world involve delegates making decisions on behalf of their group members, assessing the effect that various selection mechanisms have on the representatives’ behavior is important.

We articulate three channels through which we expect selection mechanism to affect leader behavior: 1) who self-selects into the pool of possible leaders (pre-selection), 2) who is selected by particular mechanisms (selection), and 3) how the selection mechanism itself affects the behavior of leaders (post-selection). Our experiments are designed to assess precisely how and through which channels the democratic selection mechanism affects leader behavior, which in turn affects inter-group contest outcomes.

Existing work on election in the lab almost universally finds socially positive effects of democratic elections in within-group games. Elected leaders are less corrupt, engage in less graft, provide more public goods for the group, etc. Contrary to this, we expect that being an elected representative need not have a universally positive effect in an inter-group setting. We expect that democratically selected representatives will exert excessive efforts in inter-group contests to the detriment of group performance outcomes and global welfare (i.e. the sum of welfare for the two groups).

Consistent with these expectations, we found large, negative effects of democracy on group welfare in our pilot project. Individuals exerted more, costly effort as democratically elected representatives compared to when those same people played the same game individually and compared to when they were asked to choose for their group, without an election. The post-selection mechanism generated the majority of this effect, as the process of being elected altered the behavior of individuals, rather than high-effort individuals self-selecting into running or being selected by group members in the election. This emphasizes the importance of the electoral mechanism, in and of itself, on inter-group behavior. Building on the in-person pilot experiment, we plan to conduct the interactive games online. This allows us to sample respondents who represent a broader segment of the general population than the usual undergraduate convenience sample.

Authors

• Stephen Chaudoin, Harvard University

• Sarah Hummel, University of Illinois at Urbana-Champaign

• Yon Soo Park, Harvard University