Give Everybody a Voice! The Power of Voting in a Public Goods Experiment with Externalities

Christoph Engel, Bettina RockenbachSSRN 2014 · DOI

Introduction

In many social and political scenarios, the internal cooperation of a community creates "spillovers" that affect passive outsiders. These externalities can be positive, such as a city park enjoyed by tourists, or negative, such as industrial pollution affecting downstream residents. Standard economic theory predicts that "insiders"—those making the decisions—will typically ignore these effects on "outsiders."

This research investigates whether expanding the "polity" by giving everyone a voice through non-binding voting can mitigate these externalities. Using a "veil of ignorance" approach, the study examines if participants will vote for socially optimal rules before knowing whether they will be active contributors or passive bystanders.

Experimental Design

The researchers conducted a laboratory experiment with 154 participants at the University of Erfurt. Participants were divided into groups of seven, consisting of four "actors" (who could contribute to a public good) and three "bystanders" (who remained passive).

The experiment proceeded in two phases. Phase 1 established a behavioral benchmark without voting. In Phase 2, before roles were assigned, all seven members voted on a non-binding recommended contribution level. The study utilized four distinct treatments to isolate the effects of different externality types and the impact of voting itself.

Summary of experimental treatments and group composition

The treatments included:

  • NX (No Externality): Bystanders are unaffected by actor contributions.
  • NE (Negative Externality): Contributions reduce bystander payoffs.
  • PE (Positive Externality): Contributions increase bystander payoffs.
  • ReNX (Control): A "restart" treatment without voting to control for the natural boost in cooperation often seen when a game is reset.

Key Findings

The "Veil of Ignorance" and Moral Coordination

When participants voted without knowing their eventual roles, they tended to favor socially responsible outcomes. In the Negative Externality (NE) treatment, voting successfully lowered contributions, effectively protecting outsiders from harm. In the neutral (NX) treatment, voting acted as a powerful coordination signal, increasing contributions by 25.8% compared to the control group.

The Paradox of Positive Externalities

The most striking finding was the failure of voting in the Positive Externality (PE) treatment—the very scenario where cooperation is most socially desirable. While groups often voted for maximum contributions (20 tokens), actual compliance was poor once roles were assigned. Actors' contributions in the PE treatment were significantly lower than in the NX treatment.

Payoff Envy vs. Efficiency

The breakdown in the PE treatment was driven by "unfavorable payoff comparisons." While actors were willing to cooperate for mutual benefit, they were reluctant to do so if it resulted in passive bystanders receiving a "windfall profit." This "payoff envy" suggests that actors feel relatively poorer when bystanders benefit significantly without contributing, leading them to prioritize relative status over absolute social efficiency.

Conclusion

The study demonstrates that while non-binding voting can "internalize" negative externalities and boost neutral cooperation, it struggles to overcome distributional concerns in positive externality settings. When a policy requires actors to accept a perceived distributional loss relative to bystanders, non-binding recommendations may be insufficient. Future research should explore whether legally binding enforcement mechanisms or different group histories could mitigate this "payoff envy" and facilitate socially optimal outcomes in all externality conditions.

Summary updated

BibTeX
@misc{engel2014give,
  author = {Christoph Engel and Bettina Rockenbach},
  title = {{Give Everybody a Voice! The Power of Voting in a Public Goods Experiment with Externalities}},
  year = {2014},
  doi = {10.2139/ssrn.2519479},
  howpublished = {SSRN},
}

Loading related content...

Figure