Abandoned children in China: the son-preference culture and the gender-differentiated impacts of the one-child policy

Direct administrative records of child abandonment in China are not publicly available, leaving a gap in understanding the consequences of fertility control, especially the one-child policy (OCP). Prior research has relied on small surveys or indirect estimates and has focused more on related phenomena such as informal adoption and trafficking, with limited insight into how policy, culture, and economic conditions shape abandonment. Using large-scale self-reported cases from the BCBH platform, the study investigates how OCP strictness and patriarchal (clan) culture influence abandonment incidence across provinces and whether these effects differ by child gender. Conceptually, the OCP functions like an implicit tax on over-quota births; under son preference, stricter enforcement could both raise abandonment (children become more expensive) and increase prenatal sex selection (especially against females). Therefore, the net gender effect is ambiguous. The study also considers rapid socioeconomic changes since the late 1970s (education, urbanization, incomes) and the diffusion of ultrasound technology enabling prenatal sex selection. Hypotheses: (1) Stricter OCP is positively associated with the number of child abandonments. (2) In provinces with high son-preference culture, more boys are abandoned, though this may be offset by prenatal sex selection against girls.

Existing studies identify that abandonment and missing children in China have been shaped by both policy and culture. Research on welfare homes and adoption has shown a high share of abandoned girls and disabled children (Shang et al., 2005; Kay et al., 1998; Zhang, 2006). Demographic analyses document "missing girls" associated with the OCP and son preference (Ebenstein, 2010; Almond et al., 2018), with the diffusion of diagnostic ultrasound driving prenatal sex selection (Chen et al., 2013). Provincial-level analyses have linked son preference and policy to distorted sex ratios and regional disparities (Poston et al., 1997; Gu et al., 2007; Murphy et al., 2011). Developmental factors (education, urbanization, income growth) and state policies interact with patriarchal norms to influence fertility and gender outcomes (Chung & Das Gupta, 2007; Fan et al., 2011; Knight, 2014; Wang & Chi, 2017). Recent work has also examined informal adoption and trafficking networks using BCBH data (Wang et al., 2018; Ma et al., 2020) and son preference in trafficking markets (Xiong, 2023).

Data source: The authors web-crawled and text-mined the BCBH platform, compiling 107,679 self-reported cases of parent–child separation (1980–2020; crawled by March 8, 2022). They extracted and validated provincial address information. Four cases lacking gender data were removed. Cases where the child left parents at age >12 were removed to avoid including runaways. Using fuzzy text identification, they excluded non-voluntary separation (kidnapped, stolen, robbed, trafficked, coerced) following Ma et al. (2020). The resulting dataset was aggregated to province–year counts by gender and normalized by provincial 2000 census population to obtain abandonment density. Policy measures: The OCP “penalty rate” (minimum social maintenance fee requirement as a share of per-capita disposable income) was extended to 1980–2020 using provincial administrative documents (following Ebenstein, 2010). Provincial birth quotas were constructed weighting population bases across policy categories. The composite OCP strictness indicator Pol = penalty rate / birth quota (higher values imply stricter OCP). Cultural measures: (1) Genealogies-to-population ratio (from Shanghai Library’s genealogy collection) as the primary proxy for clan/son-preference culture; (2) Han population share (2000 census) used for robustness. Controls: Province-level birth rate (China Population and Employment Statistics Yearbook), average schooling (illiteracy rate lagged by 3 years), urbanization rate, disposable income per capita (log) and its square. Empirical design: Baseline OLS panel regressions with robust standard errors, province fixed effects, and year fixed effects: Y_it = α0 + α1 Pol_it + α2 (Pol_it × Cul_it) + X_it + η_t + ξ_i + ε_it, where Y_it is provincial abandonment density (all, boys, girls). They pooled 1980–2020 and also conducted subgroup regressions by OCP strictness zones (Strict: one-child; Medium: 1.5-child; Loose: two or three children) and by gender. To assess changes after policy relaxation, they used 2012 as a breakpoint (reflecting OCP relaxations beginning 2011 and extended in 2013 and 2015) and estimated models for pre-2012 and post-2012 periods. Robustness checks: (a) Replace genealogy proxy with Han share; (b) Exclude 2,486 cases with disease/disability; (c) Split analyses by data source (children-seeking-parents vs. parents-seeking-children) to assess reporting biases. Descriptive analysis documented temporal waves (peak around 1990) and spatial heterogeneity by province, with notable gender patterns and links to skewed sex ratios at birth.

- Descriptive trends: Child abandonment surged from the late 1970s, peaking in 1990, then declined; girls accounted for more than half of abandonments in most years. A smaller uptick appeared in 2015–2016 during the transition to the universal two-child policy. Provinces with strict OCP and strong clan culture (e.g., Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Hubei, Hunan, Guangdong, Sichuan) saw more girl abandonments in the 1990s, while Guizhou exhibited relatively more boy abandonments. Skewed provincial sex ratios at birth paralleled these patterns, especially in high clan culture provinces (e.g., Jiangxi, Fujian, Guangdong). - Baseline regressions (Table 2): For the full sample, stricter OCP (Pol) increased abandonment density (e.g., Pol coefficient 0.777 to 1.355, significant at p<0.01). For boys, Pol was positive and significant (0.601 to 0.560, p<0.01). For girls, the main Pol effect was not significant without interaction (0.183, n.s.), but became positive and significant when including interaction, while the interaction Pol×Cul was strongly negative (≈−4.11, t≈−6.01), indicating that in stronger clan culture provinces, OCP’s impact on girl abandonment is attenuated. - Culture interaction: For all children, Pol×Cul was negative and significant (≈−4.33, t≈−3.96), and for boys it was not significant (≈0.123). For girls it was negative and highly significant (≈−4.11), consistent with substitution toward prenatal sex selection (skewed SRB) rather than postnatal abandonment of girls in strong clan culture settings under stricter OCP. - Policy regime change (Table 3, pre- vs post-2012): Before 2012, Pol was positive and significant for all (≈1.463, t≈−4.19), boys (≈0.489, t≈−2.26), and girls (≈0.962, t≈−4.24). Pol×Cul was negative and significant for all (≈−2.076) and girls (≈−2.169), not significant for boys. After 2012, neither Pol nor Pol×Cul significantly affected abandonment for any group, indicating weakened or null effects following OCP relaxation. - Additional insights: The attenuation of OCP’s impact on girl abandonment in high clan culture provinces aligns with more abnormal SRBs during stringent OCP and ultrasound diffusion, suggesting prenatal substitution. Possible underreporting biases and differential mortality/adoption pathways may contribute to lower observed abandonment of girls. - Robustness: Results held when using Han share instead of genealogies, excluding disabled/sick cases, and splitting by data source; core findings remained largely unchanged.

The findings confirm that stricter OCP implementation was associated with higher child abandonment, supporting the policy-as-cost mechanism. Gender-differentiated effects highlight that while boys’ abandonment rose with stricter OCP, the anticipated rise for girls is mitigated where son-preference (clan) culture is strong, consistent with a substitution from postnatal abandonment to prenatal sex selection. The strong negative interaction between policy strictness and clan culture for girls, coupled with persistently skewed sex ratios at birth, underscores the role of ultrasound-enabled prenatal selection in shaping outcomes under binding fertility constraints. Temporal analyses show that after OCP relaxation (post-2012), these associations dissipate, suggesting that loosening birth restrictions reduced the incentives that previously drove abandonment and sex-selective behaviors. Spatial patterns indicate that regional policy regimes and cultural intensity jointly structure abandonment risks, and may interact with informal adoption and trafficking networks. The results refine understanding of how fertility policy interacts with entrenched patriarchal norms to generate complex gendered consequences, highlighting that policy-induced costs can redirect, rather than eliminate, discriminatory practices.

Using a large, self-reported dataset from the BCBH platform, this study documents the evolution of child abandonment in China and identifies how OCP strictness and clan (son-preference) culture jointly shaped abandonment, with marked gender differences. Stricter OCP increased abandonment overall; boys’ abandonment rose with policy strictness, while for girls the policy effect weakened in provinces with strong clan culture due to substitution toward prenatal sex selection, as reflected in skewed sex ratios at birth. After the 2012 relaxation of the OCP, policy effects on abandonment largely disappeared. Contributions include: assembling a 40-year, province-level dataset of abandonment from micro reports; constructing a composite OCP strictness measure; and revealing gender-differentiated and culture-moderated effects. Future research should incorporate direct abortion data when available, expand measurements of OCP penalties beyond monetary fines (e.g., employment sanctions), and account for concurrent structural changes (marketization, urbanization). Improved administrative data access and linkage with health and social service records would further clarify mechanisms and address reporting biases.

Key limitations include: (1) lack of direct abortion data limits the ability to conclusively identify substitution to prenatal sex selection; (2) the OCP strictness measure (penalty/birth quota) omits non-pecuniary sanctions (e.g., employment consequences), likely underestimating policy effects; (3) potential reporting biases in self-reported BCBH data, including underreporting of abandoned girls, retrospective inaccuracies, and differing propensities to report by gender or family circumstances; (4) concurrent macro changes (marketization, urbanization, rising education/incomes) may confound estimates despite controls and fixed effects; and (5) health status of children could influence abandonment and reporting, though robustness checks excluding disease/disability cases showed stable results.