Testing evolutionary theories of human cooperation via meta-analysis of microfinance repayment

Explaining how cooperation evolves is a major research programme in the biological and social sciences. In this study, we tested evolutionary theories of human cooperation in a real-world social dilemma: joint liability microfinance, in which groups of borrowers must cooperate to successfully repay a shared loan. We used pre-registered Bayesian multilevel models to estimate meta-analytic associations between loan repayment and proxies of four evolutionary mechanisms proposed to support cooperation: relatedness, reciprocity, partner choice, and punishment. A systematic search of the microfinance literature yielded 73 effect estimates for 11 proxies of evolutionary mechanisms analysed in 11 separate meta-analyses. Punishment-based variables showed the strongest positive meta-analytic associations with loan repayment, with mixed results for other mechanisms. However, estimates varied widely in their certainty, with generally high levels of between-study heterogeneity. Our results provide some evidence for evolutionary mechanisms supporting cooperation in real-world contexts, but also indicate there are non-generalisable findings and/or reproducibility issues in the microfinance literature.