Group Cooperation Diverges onto Durable Low versus High Paths: Public Goods Experiments in 134 Honduran Villages

We performed large, lab-in-the-field experiment (2,591 participants across 134 Honduran villages; ten rounds) and tracked how contribution behavior unfolds in fixed, anonymous groups of size five. Contribution separates early into two durable paths, one low and one high, with rare convergence thereafter. High-path players can be identified with strong accuracy early on. Groups that begin with an early majority of above-norm contributors (about 60%) are very likely finish high. The empirical finding of a bifurcation, consistent with the theory, shows that early, high contributions by socially central people steer groups onto, and help keep them on, a high-cooperation path.

💡 Research Summary

This paper presents the findings of a large-scale “lab-in-the-field” experiment investigating the dynamics of cooperative behavior in public goods provision. Conducted across 134 rural villages in Honduras with 2,591 participants, the study observed individuals in fixed, anonymous groups of five over ten rounds of a standard public goods game. Participants’ decisions were linked to pre-collected data on their social networks (friendship and adversarial ties) and individual attributes.

The core empirical finding is a stark bifurcation in contribution trajectories. Very early in the sequence of rounds (within the first few), individual and group behavior diverges onto one of two durable paths: a high-cooperation path or a low-cooperation path. Subsequent convergence between these paths is rare, indicating path dependence and the existence of multiple, stable equilibria. A critical threshold was identified at the group level: if approximately 60% or more of group members contributed above the average in the initial rounds, the group was very likely to sustain high cooperation until the end. Conversely, groups failing to reach this early critical mass tended to get locked into a low-contribution trap.

The study identifies key predictors of entering and remaining on the high-cooperation path. Social network structure, specifically the number of friendship ties an individual has (a measure of social embeddedness), is a strong and robust predictor. Individuals with more friends contributed more on average and were more likely to be high-path players. Personal attributes also mattered: men contributed more than women, higher education was associated with higher contributions, and non-religious individuals and Protestants contributed more than Catholics on average. Notably, baseline household wealth or food security did not robustly predict cooperation after controlling for social network position, suggesting social capital is a more crucial driver than material resources in this context.

The authors employ rigorous panel data methods to unpack the mechanisms. Using specifications with individual fixed effects and village-by-round fixed effects, they control for time-invariant individual heterogeneity and village-specific shocks. To identify peer effects, they leverage a leave-one-out measure of peers’ past contributions and implement instrumental variable strategies that exploit as-if-random variation in group composition and deeper lags as instruments. The results show that the instantaneous (same-round) peer effect is negligible. In contrast, the lagged peer effect is substantial and approximately one-for-one: a one-unit increase in the average contribution of one’s peers in the previous round leads to about a one-unit increase in one’s own current contribution. This pattern is consistent with a process of norm updating based on observed group behavior rather than within-round imitation.

Additional analyses reveal dynamics such as the asymmetry in path switching (falling from high to low is more common than rising from low to high), the buffering role of friendship ties against downward moves, and how higher education increases sensitivity to peer behavior. A variance decomposition indicates that most unexplained variation lies at the individual level.

In conclusion, the paper provides compelling field-experimental evidence that cooperative outcomes are highly path-dependent and bifurcate into durable high and low equilibria. The early behavior of a critical mass of individuals, particularly those who are socially embedded, can steer an entire group onto a sustained high-cooperation trajectory. These findings highlight the importance of social networks and early-stage interventions that seed cooperation through central community members for promoting successful collective action.