Prosociality predicts labor market success around the world

The study examines whether the positive association between prosociality (altruism, reciprocity, trust) and labor market success, well-documented in WEIRD contexts, generalizes globally. Motivated by theory suggesting social skills reduce coordination costs and improve workplace efficiency, and by evidence linking non-cognitive skills to economic outcomes, the authors test the relationship across diverse societies. They use globally representative data to assess whether prosociality predicts higher income and lower underemployment/unemployment beyond demographics and cognitive ability, and whether this relation varies by country, continent, or development level.

Prior research shows prosocial elements such as trust facilitate economic exchange and reciprocity can enforce contracts, leading to more efficient market outcomes. A broad literature documents returns to non-cognitive and socio-emotional skills for life outcomes. Deming’s model suggests social skills reduce coordination costs, enabling specialization and cooperation. However, much evidence derives from WEIRD samples, raising concerns about external validity. Cross-societal differences may stem from market integration, community size, and religious institutions. The paper situates its contribution within this literature by testing generalizability at a global scale.

Data: Global Preference Survey (GPS) embedded in the Gallup World Poll 2012, covering 76 countries representing ~90% of the world population; ~80,000 individuals, median ~1,000 per country, one randomly selected respondent per household. Sampling weights used for representativeness. Measures of labor market success: (1) Log household-level annual income in international dollars (PPP). (2) Individual-level underemployment indicator (working less than desired). (3) Individual-level unemployment indicator (not employed but actively seeking work). (2) and (3) defined for labor force participants only (excludes full-time students, retired/disabled individuals, homemakers). Prosociality measurement: Combines three experimentally validated facets from GPS—altruism, positive reciprocity, and trust—each elicited via hypothetical choices and Likert items. The three standardized facet scores are aggregated using principal component analysis (PCA); the first principal component (eigenvalue 1.486) is used and standardized (mean 0, SD 1). Negative reciprocity is excluded based on theoretical and empirical grounds. Empirical strategy: OLS regressions with subnational region fixed effects; standard errors clustered at the country level; observations weighted by Gallup sampling weights. Main specification regresses outcomes on standardized prosociality. Controls added sequentially: gender, age, age^2; cognitive ability proxied by self-reported math skills. A sample restriction to individuals not in a partnership is used to assess robustness to intra-household labor division. Heterogeneity analyses include: (i) within-country nonlinearity via prosociality quartile dummies; (ii) separate regressions for facets (altruism, reciprocity, trust); (iii) country-specific income premia and cross-country analyses relating premia to continent, WEIRD vs non-WEIRD grouping, countries’ average prosociality, and log GDP per capita. Software: Stata 14.2.

• Income: A one standard deviation increase in prosociality predicts approximately 7.8–8.0% higher household income in pooled data without cognitive controls (coef ~0.079–0.078, SE 0.009). With cognitive ability controlled, coef ~0.060 (SE 0.009). Among non-partnered individuals, coef ~0.058 (SE 0.012).
• Underemployment: Prosociality is associated with a 0.013 (SE 0.004) lower probability of underemployment in baseline models; 0.011 (SE 0.004) with cognitive controls; 0.013 (SE 0.006) among non-partnered. Relative to the unconditional underemployment rate of 21.3%, the reduction is ~6.3% (~1.3 percentage points).
• Unemployment: Prosociality reduces unemployment by 0.009 (SE 0.003) in baseline, 0.008 (SE 0.003) with cognitive controls, and 0.012 (SE 0.006) among non-partnered; relative to the 11.7% unconditional unemployment rate, the reduction is ~7.8% (~0.9 percentage points).
• Nonlinearity: Higher prosociality quartiles are associated with greater labor market success; effects are stronger in the bottom and middle quartiles, indicating a concave relationship.
• Facets: All three facets individually predict income—altruism coef 0.057 (SE 0.008), positive reciprocity 0.074 (SE 0.007), trust 0.016 (SE 0.007).
• Cross-country heterogeneity: Income premia of prosociality are significantly positive for most countries; notable negative premia in Canada and Pakistan. No significant differences across continents (Wald test p = 0.520, N = 76) or between WEIRD (North America, Europe, Australia) and non-WEIRD regions (Asia, Africa, South America) (p = 0.711). No systematic relation between country premia and average prosociality (Spearman ρ = −0.091, p = 0.436) or log GDP per capita (ρ = −0.065, p = 0.579).

The findings address the research question by demonstrating that prosociality robustly predicts better labor market outcomes—higher household income and lower underemployment and unemployment—across a large, globally representative sample. The relationships persist after accounting for demographics and a proxy for cognitive ability and are not confined to any particular continent, development level, or average national prosociality, underscoring broad external validity. The concave pattern suggests the largest marginal gains from prosociality accrue at lower to middle levels. All three prosocial facets contribute, with positive reciprocity yielding the highest income premium and trust the lowest. While the study is correlational and cannot rule out reverse causality or omitted variables, the consistency of results across specifications and countries suggests that labor market models incorporating prosocial and social skills may have wide applicability. Policy relevance is highlighted by evidence that prosociality is malleable, suggesting potential benefits from interventions that foster prosocial behaviors and beliefs.

This paper provides global evidence that prosociality is positively associated with labor market success. Using validated measures across 76 countries, the study documents substantial income premia and lower risks of underemployment and unemployment linked to prosociality, with similar effects across continents and development levels. All main facets—altruism, reciprocity, and trust—contribute, and marginal returns appear concave. Future research should identify causal mechanisms, address potential reverse causality and unobserved confounding, and explain cross-country heterogeneity (e.g., institutional, cultural, or market factors). Evaluating interventions that enhance prosocial skills and assessing their long-run labor market impacts across diverse settings are promising directions.

• The analysis is cross-sectional and correlational; it cannot establish causality or rule out reverse causality and omitted variable bias.
• Underemployment is a subjective measure and may itself be influenced by prosociality.
• Cognitive ability control (self-reported math skills) may be a “bad control” if cognitive ability and prosociality co-develop, potentially biasing estimates.
• Earnings are not observed; the study uses household income (PPP) rather than individual earnings, which may introduce household-level confounding and intra-household allocation issues (addressed partly via a non-partnered subsample).
• Measurement relies on self-reports and hypothetical items (though experimentally validated); remaining measurement error may attenuate or bias estimates.