Effect of the contextual (community) level social trust on women's empowerment: an instrumental variable analysis of 26 nations

The study investigates whether higher contextual (community-level) social trust causally increases women’s empowerment. Women’s empowerment—measured as autonomy in household decision-making and ownership of assets—has been linked to improved household welfare, children’s health and nutrition, labor market participation, and use of health services. Prior work has emphasized social capital, often operationalized as networks and participation, but largely at the individual level. The authors identify a gap: the theoretically posited positive role of community-level social trust on women’s empowerment has not been empirically tested and confirmed across countries. This paper advances theory by focusing on community-level trust (a collective resource) rather than only individual trust or network participation, and empirically evaluates its effect while addressing endogeneity. The context is 26 post-communist countries where women’s empowerment weakened during transitions to market economies, and where comparable multinational data are available.

The paper situates women’s empowerment as a multifaceted construct involving resources (e.g., asset ownership) and agency (e.g., decision-making), drawing on Kishor (2000) and Kabeer (1999). Prior studies link social capital to women’s empowerment via networks and participation (e.g., Mayoux 2001; Alemu et al. 2018; Po and Hickey 2020; Nayak and Panigrahi 2020; Machio et al. 2022), highlighting knowledge diffusion and assertiveness through group membership. However, these studies predominantly examine individual-level social capital and network participation. The authors adopt Putnam’s conceptualization of social capital as a contextual, community-level resource, emphasizing social trust (a cognitive dimension per Coleman, 1988). They distinguish particularized vs. generalized trust (Newton and Zmerli, 2011) and note that community-level trust amalgamates bonding and bridging features. Mechanisms through which community trust may foster empowerment include lowering transaction costs, improving market reliability and institutions, and enhancing employment and entrepreneurship opportunities (e.g., Bjørnskov 2008; Kwon et al. 2013; Habibov and Afandi 2017; Collier 2002). The authors also review methodological challenges (reverse causality, unobserved heterogeneity, measurement error) and position their use of instrumental variables as an advance over descriptive cross-sectional analyses.

Design and data: Cross-sectional analysis using the 2016 Life in Transition Survey (LITS III) by EBRD/World Bank. Sample comprises 23,292 women across 26 post-communist countries in Eastern and Central Europe and Eurasia (including former Soviet republics and Mongolia). Approximately 1,500 individuals were interviewed per country with harmonized sampling and instruments for cross-country comparability. Outcomes (women’s empowerment): - Index of Women’s Asset Ownership: Sum of binary indicators for owning land, a bank account, and a dwelling (range 0–3; higher = more ownership). - Index of Women’s Autonomy in Household Decision-Making: Sum of three binary items assessing who decides on (a) daily expenditures/bills, (b) saving/investing/borrowing, (c) major household purchases. Coded 1 if the woman decides alone or jointly (with partner or family) and 0 if decisions are made by partner/others (range 0–3; higher = more autonomy). Predictor of interest: - Community-level social trust: Derived from LITS question “Generally speaking, would you say that most people can be trusted or that you can’t be too careful…” on a 0–10 scale; individual responses averaged at the community level to construct the contextual trust measure. Instruments (for IV estimation): - Main set A: Population density (persons/km²; US Census Bureau International Database, 2016). - Main set B: Willingness to take risk (country-level average from LITS, scale 1–10). - Alternative instruments for robustness: Absolute latitude (community’s distance from equator; from LITS) and Caloric Suitability Index (agricultural potential post-1500; Galor and Özak, 2016). Covariates: Individual and family-level characteristics—age groups, marital status (married, single, widowed, divorced, separated), education (1–8 scale from no education to master’s/PhD), employment (binary), income decile (1–10), rural residence, family composition (shares of children 0–7, older children 8–17, adults 18–64, elderly 65+), as well as country-level Human Development Index (UN, 2016) and annual GDP growth (World Bank, 2016). Statistical approach: Both single-stage OLS and two-stage least squares (IV) regressions are estimated. First stage regresses community trust on covariates plus one instrument; second stage regresses each empowerment index on the predicted trust and covariates. Diagnostic tests include first-stage robust F statistics for instrument relevance; Stock–Yogo weak instrument tests; Durbin and Wu–Hausman tests for endogeneity; and, when applicable (two instruments), Sargan and Basmann overidentification tests. Estimation conducted in Stata 15. Sensitivity analyses stratify by rural/urban and by age (younger vs older women), add individual-level trust as a control, and employ alternative instruments.

- OLS (descriptive): A one-unit increase in community trust is associated with +0.103 in the Asset Ownership Index (p<0.001); no significant association with Decision-Making Index. - IV main results (Set A: instrument = population density): - Women’s Asset Ownership: +1.215 (SE 0.072), p<0.001. - Women’s Decision-Making: +0.270 (SE 0.081), p<0.001. First-stage robust F statistics >518, exceeding Stock–Yogo critical values; Durbin/Wu–Hausman tests significant, supporting endogeneity and the use of IV. - IV main results (Set B: instrument = willingness to take risk): - Women’s Asset Ownership: +0.555 (SE 0.054), p<0.001. - Women’s Decision-Making: +0.540 (SE 0.078), p<0.001. First-stage robust F statistics >622; endogeneity tests significant. - Subgroup robustness (population density instrument): - Rural: Asset +1.871 (SE 0.219), p<0.001; Decision-Making +0.432 (SE 0.192), p<0.05. - Urban: Asset +0.944 (SE 0.071), p<0.001; Decision-Making +0.226 (SE 0.087), p<0.01. - Subgroup robustness (risk-taking instrument): - Rural: Asset +0.569 (SE 0.083), p<0.001; Decision-Making +0.617 (SE 0.120), p<0.001. - Urban: Asset +0.520 (SE 0.073), p<0.001; Decision-Making +0.513 (SE 0.107), p<0.001. - Age stratification (population density instrument): - Older: Asset +1.265 (SE 0.097), p<0.001; Decision-Making +0.211 (SE 0.107), p<0.05. - Younger: Asset +1.121 (SE 0.121), p<0.001; Decision-Making +0.313 (SE 0.138), p<0.05. - Age stratification (risk-taking instrument): - Older: Asset +0.429 (SE 0.073), p<0.001; Decision-Making +0.602 (SE 0.111), p<0.001. - Younger: Asset +0.675 (SE 0.092), p<0.001; Decision-Making +0.553 (SE 0.127), p<0.001. - Additional robustness: - Controlling for individual-level trust, community-level trust remains positive and significant (e.g., Set A: Asset +1.509, Decision-Making +0.320; Set B: Asset +0.728, Decision-Making +0.706; all p<0.01). Individual-level trust enters with negative coefficients, likely reflecting multicollinearity with contextual trust. - Alternative instruments (latitude and Caloric Suitability Index) yield positive significant effects (Asset +0.479; Decision-Making +0.183; both p≤0.01). Overidentification tests (Sargan/Basmann) are non-significant, supporting instrument validity. - Across all IV specifications, first-stage F values are high (typically >90 and often >300), weak instrument tests are passed, and endogeneity tests are significant, indicating IV estimates are preferred over OLS.

Findings confirm the theoretically posited causal link: higher community-level social trust increases women’s empowerment through greater asset ownership and enhanced participation in household financial decisions. The results are consistent across instruments, robust to extensive controls and to inclusion of individual-level trust, and hold in rural/urban and younger/older subgroups. The contrast between small or null OLS associations and large positive IV estimates underscores biases from reverse causality, omitted variables, and measurement error in single-stage models. The study reinforces the view of social trust as a contextual public good that lowers transaction costs, improves institutional effectiveness, and facilitates economic engagement, thereby creating conditions that expand women’s access to assets and decision-making authority. Practically, community-based initiatives that build trust—such as formalized women’s groups and supportive legal-regulatory environments—can be effective levers for gender equality policies in post-communist settings.

This study is the first multi-country IV analysis demonstrating that community-level social trust causally strengthens women’s empowerment, measured via asset ownership and household decision-making. By focusing on the contextual dimension of social capital and employing strong instruments, it advances both theory and empirical methodology, showing that OLS underestimates the true effects. Policy implications include prioritizing community-based interventions and formal support for women’s organizations to nurture trust and empowerment. Future research should: (1) broaden empowerment and trust measures (e.g., additional asset types and trust dimensions); (2) develop purpose-built surveys enabling richer indicators and country-by-country causal comparisons; (3) examine heterogeneity across socio-economic contexts and levels of existing empowerment; (4) further scrutinize instrument exogeneity and explore alternative, plausibly exogenous sources of variation; and (5) incorporate qualitative approaches to identify mechanisms and the most salient dimensions of empowerment and trust.

- Cross-sectional survey design limits temporal inference; despite IV, full causality cannot be definitively established. - Instruments, while theoretically motivated and statistically validated, may still have residual correlation with outcomes through unobserved channels; exclusion restrictions cannot be fully guaranteed. - Measures of trust and empowerment are attitudinal/subjective and may contain measurement error. - Empowerment indicators are limited to asset ownership (three items) and household financial decision-making (three items), potentially omitting other empowerment facets. - Country-by-country causal comparisons were constrained by available sample sizes; broader, purpose-built datasets could improve granularity and generalizability.