Are group members less inequality averse than individual decision makers?

Haoran He, Marie Claire Villeval

TL;DR

This paper investigates how inequality aversion differs between individuals making decisions alone and those making decisions in groups. Surprisingly, while initial group proposals show more inequality aversion, the final group decisions reveal a similar level of aversion as individual decisions, suggesting that group dynamics and the median member's preferences play a crucial role in shaping outcomes.

What Problem Does This Paper Address?

The existing literature on inequality aversion, a concept describing people's concern for the fairness of payoff distributions, has largely focused on individual decision-making. While some studies suggest groups tend to make more selfish decisions than individuals, the specific impact of group dynamics on inequality aversion, particularly how individual preferences aggregate into a group decision, remained less explored. This paper aims to bridge this gap by directly comparing inequality aversion in individual versus group settings, examining how group decision-making processes influence these preferences.

Methods

The researchers designed a laboratory experiment using two games: the Ultimatum Game (UG) and a Modified Dictator Game (MDG). These games were adapted to measure both disadvantageous inequality aversion (concern about receiving less than others) and advantageous inequality aversion (concern about receiving more than others).

The experiment employed a within-subject design, meaning participants experienced both individual and group decision-making conditions. Key aspects of the methodology include:

  • Strategy Method: Participants made decisions for all possible scenarios, allowing for a detailed understanding of their preferences.
  • Unanimity Rule: Group decisions required all members to agree, providing a clear mechanism for studying the aggregation process and the role of veto power.
  • Treatments: Three main treatments were used:
  • I-I (Individual-Individual): Participants made decisions individually in both UG and MDG.
  • I-AG (Individual-Anonymous Group): Participants made individual decisions and then group decisions anonymously.
  • I-NAG (Individual-NonAnonymous Group): Similar to I-AG, but group members were identified, allowing for the study of social image concerns.
  • Data Analysis: Statistical analyses included Wilcoxon signed-rank tests, Tobit regressions, and analysis of concession indices to quantify differences in decision-making and aggregation processes.

Datasets

Not discussed in the paper. The study relies on data generated from their experimental design.

Key Insights and Ideas

  • Initial Proposals vs. Final Decisions: The most striking finding is the divergence between initial group proposals and final group decisions. While initial proposals in groups reveal higher inequality aversion than individual decisions, the final group outcomes mirror individual decision-making levels.
  • Median Player Dominance: The aggregation process within groups appears to be primarily driven by the member with median preferences. This suggests that extreme preferences, while present, may not dictate the final group outcome.
  • Anonymity's Limited Role: Lifting anonymity within groups did not significantly impact initial proposals or final group decisions, suggesting that social image concerns, while present in individual decisions, have a less pronounced effect when proposals are aggregated.
  • Beliefs About Norms: The increased inequality aversion in initial group proposals is attributed more to beliefs about prevailing group norms or the change in decision environment rather than efficiency or social image concerns.

Key Results

  • Increased Aversion in Initial Proposals: In most treatments, individuals expressed significantly higher disadvantageous and advantageous inequality aversion when making their initial proposals in a group setting compared to their individual decisions. This effect was observed across various treatments, with p-values often below 0.001.
  • Similar Aversion in Final Group Decisions: In contrast to initial proposals, the final group decisions showed no significant difference in inequality aversion compared to individual decisions in most treatments. For example, in the I-AG and I-NAG treatments, p-values for comparisons between individual and group decisions were often above 0.10.
  • Aggregation Dynamics: The number of rounds needed to reach group consensus was influenced by the distance of preferences from the median member. Larger deviations from the median led to slower convergence.
  • Concessions: Individuals further from the median made larger concessions per round, indicating a willingness to compromise to reach an agreement.

Limitations and Future Work

  • Disentangling Explanations: The authors acknowledge that their design does not fully disentangle whether the increased inequality aversion in initial proposals is due to beliefs about group norms or simply the change in the decision environment. Future research could directly elicit beliefs to address this.
  • Order Effects: While some order effects were explored, the authors suggest further investigation into how starting with group decision-making versus individual decision-making might influence subsequent behavior.
  • Communication and Identification: The study used restricted communication and minimal physical identification. Future work could explore how face-to-face communication or stronger group identity manipulation might alter group decision-making and inequality aversion.
  • Cultural Context: The study was conducted in China. Comparing these findings with collectivist versus individualistic societies could provide valuable insights into the cultural influences on inequality aversion in groups.

Summary updated

BibTeX
@article{he2017are,
  author = {Haoran He and Marie Claire Villeval},
  title = {{Are group members less inequality averse than individual decision makers?}},
  journal = {Journal of Economic Behavior \& Organization},
  year = {2017},
  doi = {10.1016/j.jebo.2017.04.004},
}
Figure