Self-reliance crowds out group cooperation and increases wealth inequality

The study examines how the option of self-reliance—solving a shared problem individually via private means—affects cooperation in the provision of public goods. Classic public goods problems feature non-excludability and free-riding, risking under-provision. Prior work has focused on mechanisms to curb free-riding (e.g., punishment, partner choice, repeated interaction), but has largely overlooked that some individuals can avoid dependence on public goods entirely by investing in private solutions. This creates a different social dilemma: whether to solve problems collectively (public good) or individually (private solution). The authors ask: Does the availability and affordability of private solutions reduce cooperation? How do within-group asymmetries in the ability to be self-reliant shape contributions, social welfare, and inequality? They propose that self-reliance may depress cooperation by separating fear of exploitation from greedy free-riding, and that asymmetries in self-reliance can erode communal provision and amplify wealth disparities.

The paper situates its question within research on public goods and social dilemmas showing vulnerability to free-riding and mechanisms to sustain cooperation (punishment, partner choice, repeated interactions). It highlights work indicating that local, excludable group goods can crowd out global public goods, but notes ambiguity about whether effects reflect increased free-riding or a switch to private/self-reliant solutions. Literature on heterogeneity in resources and productivity shows mixed effects of asymmetry on cooperation; however, how differences in ability to exit dependence on public goods (self-reliance) affect provision and distributional outcomes remained unknown. The study extends step-level public good and coordination frameworks to incorporate private solution options and asymmetries in endowments/self-reliance.

Design: Laboratory experiment with groups of 4 facing a stylized collective action problem each round. Participants allocate resource points (RP) between a private pool (individual solution) and a public pool (collective solution), keeping any RP not invested. Earnings from kept RP are realized only if the participant reaches either their private threshold (ci) or the public threshold (cp) is met by the group; otherwise, kept RP are lost. Parameters: Public solution threshold fixed at cp = 180 RP. Private solution threshold varied across counterbalanced blocks of 10 rounds: ci ∈ {∞, 75, 65, 55, 45}. When ci = ∞, private solutions are unattainable and the task reduces to a step-level public goods game. Endowments and conditions: Total group endowment per round = 360 RP. Symmetry condition (25 groups): each member endowed with e = 90 RP. Asymmetry condition (25 groups): two members endowed with e_low = 60 RP (more dependent), two with e_high = 120 RP (less dependent). Thus, private solutions are easier for high-endowment members when available. Within the asymmetry condition, individuals retained their endowment type across the experiment. Procedures: Participants made simultaneous allocation decisions each round and received feedback on all members’ allocations to private and public pools, which thresholds were met, and earnings. After the last round of each block, members voted on delegating the group’s decision for that round to a third party; if a majority supported delegation, the third party’s pre-specified decision replaced the last round’s group decisions. Voting outcomes and third-party decisions were only revealed at the end of the experiment to avoid strategic learning about others’ preferences. Third-party condition: An additional sample of n = 61 participants acted as third parties. For each ci level and resource distribution (symmetric vs. asymmetric), they specified how each group member should allocate resources between private and public pools. If a group voted to delegate, the computer randomly implemented one third-party decision corresponding to that condition. Measures: Main behavioral outcomes included success in creating the public good, frequency of private solutions, contributions by endowment type, failure rates (losing kept RP), wealth/earnings, and inequality (e.g., Gini). Social preferences were measured post-task via the incentivized Social Value Orientation (SVO) slider measure. Risk preferences and fairness evaluations were also collected. Sample and payment: Total N = 261. Main experiment: 200 participants assigned to 50 groups (25 symmetry, 25 asymmetry), 50 rounds per participant (5 blocks × 10 rounds). Third-party sample: 61 participants. Payment conversion: 100 RP = €0.50; main participants received a €6.50 show-up fee plus earnings from three randomly selected rounds per block and from SVO/risk tasks; average earnings ≈ €11.60. Third parties received €3.50 show-up fee. Ethical approval obtained from Leiden University Psychology Research Ethics Board; informed consent collected. Analysis: Mixed-effects regressions, non-parametric tests (Mann–Whitney U, Wilcoxon signed-rank), chi-square tests, t-tests, Spearman correlations, and mediation analyses with bootstrapped confidence intervals; analyses conducted in R 3.3.3.

• Decline in cooperation with self-reliance: When private solutions were unattainable (ci = ∞), 78% of groups created the public good (no difference by symmetry/asymmetry; cluster-adjusted chi-square P = 0.27). As private solutions became cheaper, cooperative success decreased (mixed-effects regression, P < 0.001).
• Extreme crowd-out in symmetric groups at low ci: When ci ≤ 45, 0% of symmetric groups reached a public solution; 93% of group members solved the problem privately.
• Partial resilience under asymmetry: At low ci, 26% of asymmetric groups still created a public good and only 55% of members solved privately (symmetry vs. asymmetry: Mann–Whitney U = 92, P < 0.001).
• Exploitation dynamics: Less dependent (e_high = 120) members often contributed nothing to the public good, shifting burden to more dependent (e_low = 60) members. With private solutions available, more dependent members allocated more to the public pool than others (within asymmetry: Wilcoxon W = 240, P = 0.04; vs. symmetric 90-RP members: Mann–Whitney U = 569, P < 0.001).
• Higher failure risk for the dependent: More dependent members most frequently lost their remaining RP when thresholds were unmet (within asymmetry: Wilcoxon W = 288, P < 0.001; vs. symmetric: Mann–Whitney U = 506, P < 0.001).
• Strategic self-reliance and coercion: Less dependent members withdrew support for the public solution when more dependent members contributed less, especially when private solutions were available only to the less dependent (ci ∈ {75, 65}), inducing higher public contributions by the more dependent (Supplementary Note 5).
• Wealth inequality: As private solutions became cheaper, earnings disparities widened between less and more dependent members (mixed-effects regression, P = 0.01). In symmetric groups, wealth distribution remained more uniform (mixed-effects regression, P < 0.001). Symmetric groups were wealthier overall than asymmetric groups (group-level t-test, t(41) = 2.06, P = 0.04). Learning to coordinate on public solutions over rounds occurred in symmetry for ci = ∞ and partly for ci ∈ {75, 65}, but not in asymmetry when private solutions were available.
• Social preferences as drivers: Less prosocial SVO among less dependent members predicted lower contributions to the public solution, which in turn predicted lower earnings for the more dependent (mediation model; standardized path coefficients approx. 0.43 from SVO to cooperation and 0.79 from cooperation to welfare; significance indicated in figure: * P = 0.03; ** P < 0.001).
• Delegation to third parties: Voting favored delegation more among more dependent members (51%) than less dependent (37%); paired t(24) = 2.20, P = 0.04. Higher individual earnings correlated with lower support for delegation (Spearman r = -0.41, P = 0.003). Only 30% of opportunities resulted in installing a third-party decision-maker. Third parties would have reduced the wealth gap (group-level t(67) = -4.48, P < 0.001), favored public over private solutions, and redistributed via the public good to reduce inequality.

The findings demonstrate that introducing viable private solutions creates a self-reliance dilemma that depresses collective action beyond classic free-riding dynamics. When self-reliance is feasible, individuals—especially those with higher endowments—shift toward private provision, reducing public good creation and exposing more dependent members to higher failure risk and losses. Asymmetries in self-reliance reshape intra-group incentives and bargaining power: less dependent members can strategically withhold public contributions (or opt out via private solutions), effectively coercing more dependent members to contribute more, which increases earnings inequality and undermines overall social welfare. Social preferences moderate these effects, making outcomes for the dependent contingent on the prosociality of the better-off. Third-party decision-makers favor communal solutions and redistribution, but groups rarely adopt delegation when advantaged members can maintain higher earnings under decentralized decisions. These results highlight how private options and inequality in self-reliance polarize preferences over governance and impede coordination on efficient public goods, informing debates on privatization, welfare provision, and institutional design.

This study introduces and empirically validates the self-reliance social dilemma: the availability of private solutions crowds out public goods provision, reduces social welfare, and amplifies wealth inequality—especially under asymmetric endowments. Groups coordinate effectively on public solutions when they must, but the option to exit via self-reliance undermines co-dependence and collective efficiency. Policy implications include the need for institutions that sustain contributions to communal solutions even when private options exist, and for curbing inequality in self-reliance to balance individual freedom with group dependence. Future research could test field implementations, explore institutional mechanisms (e.g., taxation, subsidies, regulation, compulsory contributions), examine dynamic wealth and preference evolution, and assess heterogeneity in social preferences and legitimacy of third-party governance across cultures and contexts.