Offline social capital, online social capital, and fertility intentions: evidence from China

Global fertility has declined markedly over recent decades and is projected to fall below replacement in most countries by 2100. China, the world’s most populous country, has experienced sustained sub-replacement fertility and now faces ultra-low fertility alongside population aging. Despite policy shifts from the former restrictive family planning regime to encouragement of higher parity (separate two-child policy in 2013, universal two-child in 2016, and three-child policy in 2021), fertility remains low (TFR ≈ 1.3 in 2020). Chinese fertility intentions are shaped by territorial diversity, urbanization, and deep cultural norms valuing lineage and family continuity, alongside modernization and changing values that emphasize individual development and quality of life. Recent shocks such as COVID-19 further lowered intentions among women of childbearing age. Given the centrality of intentions as predictors of behavior, this study investigates how social capital—offline and online—shapes fertility intentions in China, intentionally abstracting from direct policy effects to focus on social mechanisms.

Social capital encompasses trust, norms, and networks that facilitate cooperation and resource access. It can be parsed into cognitive (e.g., social trust) and structural (e.g., social networks/participation) dimensions, and in the contemporary era also into offline and online forms. Prior work links social capital to numerous outcomes (education, health, employment) and suggests it may affect fertility intentions via support, information, and norms. Empirical evidence mostly from Europe indicates offline social trust and networks can raise fertility by enabling work-family balance and reducing perceived costs through support. In contrast, online social capital, proxied by social networking site use or Internet access, may reduce fertility intentions by spreading contraceptive information, increasing opportunity costs for women, reducing partner time and marital satisfaction, elevating consumption aspirations, and amplifying negative perceptions of childbearing. The study posits three hypotheses: (H1) Offline social capital is positively associated with fertility intentions. (H2) Online social capital is negatively associated with fertility intentions. (H3) Both offline and online social capital affect fertility intentions through perceptions of pensions (reliance on children), filial piety, and social justice.

Data and sample: The study uses the nationally representative Chinese General Social Survey (CGSS) 2017, covering 31 provinces/autonomous regions via multistage stratified PPS sampling. Respondents aged 18–60 were retained to reflect contemporary childbearing horizons and evolving reproductive technologies. After excluding missing/outliers, N = 7,518. Variables: Dependent variable is policy-free fertility intention measured by the response to “How many children do you want to have if there are no policy restrictions?” coded as 0, 1, 2, or 3 (for three or more). Core independent variables capture social capital: (1) Offline social capital includes social trust (agree that most people can be trusted: 1 = relatively/strongly agree; 0 = otherwise including neutral) and social network (frequency of socializing/visiting in free time: 1 = sometimes/often/always; 0 = never/rarely). (2) Online social capital is proxied by Internet use frequency over the past year: 1 = sometimes/often/always; 0 = never/rarely. Controls span demographics (gender, age, education, hukou, marital status, health, working hours—logged), social insurance (medical, pension), household (children, son ratio, annual family income—logged), and region (West baseline vs. Central, East, Northeast). Model: Because the dependent variable is ordered categorical, the main specification is an ordered probit (oprobit): Fertility_i = β0 + β1(offline trust) + β2(offline network) + β3(online) + β4(controls) + ε_i. Marginal effects are computed to interpret probability shifts across outcome categories. Endogeneity is assessed via instrumental variables using the conditional mixed process (CMP): first-stage probits for endogenous regressors and second-stage oprobit for fertility intentions. Instruments: self-rated social class (social level 1) for trust, self-rated socioeconomic status (social level 2) for network, and log(first online age) for online frequency. Robustness checks include ologit and OLS of log-transformed intended number of children.

- Descriptives: Mean fertility intention = 1.85 (below replacement 2.1). - Baseline associations (oprobit, N=7,518): - Trust: 0.075 (SE 0.028), p<0.01; Network: 0.057 (0.027), p<0.05. Combined offline measures remain positive and significant. - Online frequency: −0.079 (0.036), p<0.05; when included jointly with offline, online remains negative (−0.082, 0.036), p<0.05. - Key controls: Rural hukou positive (~0.298–0.314, p<0.01); married positive (~0.171, p<0.01); son ratio negative (~−0.204, p<0.01). Age positive; others mixed. - Marginal effects (Model 5): Per unit increase in social trust, P(Y=0) −0.004, P(Y=1) −0.018, P(Y=2) +0.009, P(Y≥3) +0.013. For online social capital, the probability of willingness to have 2 or ≥3 children decreases by 0.010 and 0.014, respectively, as online frequency increases. - Endogeneity (CMP IV): Instruments are strongly related to endogenous regressors; first-stage coefficients significant. Second-stage signs and significance persist (trust/network positive; online negative at 5%). atanhrho_12 indicates endogeneity; CMP estimates are preferred and align with baseline. - Robustness: Results hold under ologit and OLS (log intended children). For example, ologit: trust 0.137 (0.051), p<0.01; network 0.103 (0.049), p<0.05; online −0.144 (0.066), p<0.05. OLS: trust 0.021 (0.009), p<0.05; network 0.018–0.019 (0.008), p<0.05; online −0.021 to −0.022 (0.011), p<0.1. - Heterogeneity (subsamples): - Gender: Women—trust 0.100 (0.038), p<0.01; network 0.071 (0.038), p<0.1; online −0.147 (0.050), p<0.01. Men—coefficients smaller and generally not significant. - Region: East—trust 0.102 (0.038), p<0.01; network 0.087 (0.036), p<0.05. Midwest—offline effects weaker/negative in narrative; online more adverse for Midwest than East. - Age: 18–29—online −0.564 (0.214), p<0.01; 40–60—trust 0.074 (0.037), p<0.05. The 30–60 group is more influenced by offline social capital; online primarily reduces intentions among 18–29. - Mechanisms (Table 7): - Pension concept (relying on children): trust +0.138 (0.031), p<0.01; online −0.139 (0.040), p<0.01. - Filial piety score: trust +0.048 (0.018), p<0.01; online −0.057 (0.025), p<0.05. - Perceived social justice (1–5): trust +0.596 (0.027), p<0.01; online −0.098 (0.035), p<0.01. Overall: Offline social capital increases, and online social capital decreases, fertility intentions, with effects concentrated among women, Eastern residents (offline), Midwestern residents (online), and younger adults (online). Findings support H1–H3.

The study demonstrates that social capital shapes fertility intentions in China in opposite ways offline versus online. Offline trust and networks likely enhance intentions by increasing perceived support, lowering subjective costs, and facilitating work–family balance—effects especially salient for women and older reproductive ages with more stable social ties. In contrast, online engagement is associated with exposure to information and norms that raise opportunity costs, spread fertility anxiety, strengthen individualistic preferences, and contribute to perceived social unfairness—effects particularly pronounced among younger adults and women. Regional patterns reflect different social structures and Internet penetration: Eastern regions see gains from marginal offline accumulation, whereas Midwestern regions exhibit more entrenched traditional norms offline yet are vulnerable to online-induced declines in intentions. Mechanism analyses indicate that offline capital reinforces traditional norms of pension and filial piety and enhances perceived social justice, all of which increase intentions; online capital attenuates these norms and lowers perceived fairness, reducing intentions. These results suggest that policy strategies to raise fertility intentions should not only reduce direct costs but also cultivate social environments of trust and support, and address online information environments and fairness perceptions that influence reproductive decisions.

This study integrates offline and online social capital into a unified framework to explain fertility intentions in China. Using nationally representative data and robust methods, it finds that offline social capital (trust, networks) raises, while online social capital (Internet use) lowers, fertility intentions. Effects operate partly through pension reliance norms, filial piety, and perceived social justice. Policy implications include strengthening social trust and in-person social interaction; enhancing supportive community networks; improving the online information environment and media guidance to reduce fertility anxiety; and elevating perceived fairness via distributional, procedural, and interactive equity in public services. Future research should incorporate richer multidimensional measures of social capital, leverage longitudinal or quasi-experimental designs to identify causal pathways, and situate fertility within the broader agenda of sustainable population development.

- Measurement: Offline social capital is proxied by only two items (trust, socializing), and online by Internet frequency; broader, multidimensional measures are needed. - Policy context and inertia: The analysis abstracts from the historical and inertial effects of family planning policies due to data limitations; future work should compare cohorts and periods across policy regimes. - Design: Cross-sectional data limit causal inference; longitudinal or quasi-experimental evidence is needed. Additionally, future studies should also consider global sustainability dimensions alongside fertility promotion. - Generalizability of mechanisms: Pension, filial piety, and fairness perceptions are context-specific; external validity to other cultural settings requires caution.