Introduction

Child abandonment is among the least-investigated consequences of China’s fertility control policy (Johnson,1996), partially because direct abandonment records are not publicly available to scholars. Previous studies have used small-scale survey data (Johnson,1996; Kay et al., 1998; Zhang, 2006) or indirect estimation (Chen et al., 2015) to explore this topic. Additionally, due to the limited data availability, little is known about how culture, policies, and economic forces affect abandonment behaviors. The rapid development of mobile internet in recent decades has provided new channels for collecting self-reported abandonment information. For example, the Baby Come Back Home website (www.baobeihuijia.com, in short BCBH), the largest self-reported “people finding” NGO internet platform in China, contains information on over one hundred thousand cases. In each case, the poster describes in detail when, where, and how the abandonment occurred. A growing number of studies are using this data source to explore related topics, such as informal adoption (Wang et al., 2018; Ma et al., 2020). Here, data from the BCBH website are used to create a temporal-spatial dataset to investigate how socioeconomic factors and public policies affect the number of child abandonments and whether the effects are gender-differentiated.

The BCBH data show that China experienced a wave of child abandonment from the late 1970s to the early 2000s, which peaked in 1990 and has since continued to fall. Girls were more likely to be victims of abandonment during this wave. In a study of eight national welfare homes for children, Shang et al. (2005) found that a large proportion of abandoned children are female or disabled. This finding is consistent with other studies indicating that more girls than boys were “missing” during this period (Ebenstein 2010; Kay et al., 1998; Lid et al. 2004). The high number of “missing” girls is due to the joint influence of China’s patriarchal culture and fertility-control policies (Chen et al., 2015; Almond et al., 2018).

Economists consider the one-child policy (OCP) as an implicit consumption tax (Huang et al., 2023), meaning that Chinese parents must pay for each additional birth once they have exceeded the policy-assigned birth quota. In theory, how an increase in tax rate shifts demand depends on the consumer’s preferences. For example, in a patriarchal culture, parents have stronger preferences for sons. An increase in the marginal tax rate on an over-quota birth would induce two effects. First, parents have fewer children or become more likely to abandon children as their children become more expensive. Second, as budget constraints become stricter, the practice of selective abortion would be more prevalent, especially among parents with strong son preferences who found their fetuses to be female. Therefore, whether stricter OCP implementation makes child abandonments more gender-selective is unclear. This is because stricter fertility restrictions also induce more abortions of female fetuses, which distorts the sex ratio at birth and can replace the practice of abandoning girls.

We then examine the moderating role of the traditional son-preference culture in the relationship between policy strictness and the gender composition of abandoned children. The patriarchal lineage, which prefers sons, has a profound and lasting influence in East Asia. On the other hand, the socioeconomic environment has experienced substantial changes since the late 1970s. The reform and opening-up policies have brought steady growth in education, urbanization, and family incomes in China (Lu et al., 2019). Meanwhile, the tradition of gender-selective abandonment still exists. Furthermore, as ultrasound technology for prenatal diagnosis diffuses, the practice of gender-selective abortion increases. Figure 1 displays the theoretical framework of our study. In short, we focus on the impact of the OCP on abandonment behaviors in Chinese society and further examine how this impact depends on the child’s gender and the presence of clan culture.

Fig. 1: Research Framework.
figure 1

This is the research framework for the full paper, with question marks marking the mechanisms of influence on which we focus.

Full size image

Since the central government authorized provincial governments to formulate specific OCP regulations according to their local conditions, the indicator of policy strictness can be constructed at the province level. We then aggregate the BCBH data and examine our hypotheses. The Chinese government has gradually relaxed the OCP since 2011 and gave supplemented birth quotas to families satisfying certain conditions. For instance, if both spouses are the only child of their families, they are permitted to have a second birth without punishment. The coverage of such an exemption was extended in 2013 to families in which one of the spouses is the only child and then in 2015 to all families. Therefore, we chose 2012 as a breakpoint to compare the impacts before and after these policy changes. In general, our findings contribute to our understanding of how social contexts shape the impact of fertility policies on abandonment behaviors.

Here, a web-crawling approach was used to obtain a corpus of large-scale individual data on parent‒child separation in China and construct a self-reported dataset of abandonment through text mining. This approach can overcome the data limitations in previous studies and provide a direct and detailed description of abandonment cases in China. Although these self-reported data inevitably include underreporting bias and retrospective mistakes, most reports were motivated by a genuine desire to find parents/children and thus provided detailed information about the abandonment event and people involved. Meanwhile, due to China’s mobile phone ownership rate (116.3 mobiles per 100 people in 2021Footnote 1) and the widespread use of the internet in China (1.067 billion internet users in 2022, a penetration rate of 75.6%Footnote 2), the data have representativeness to some extent and can be further used to address research questions contextualized in the sociocultural and demographic milieu of China. This study extends the existing literature by focusing on how patriarchal culture interacts with fertility restriction policies to shape abandonment behaviors. Another contribution is identifying gender-differentiated patterns.

The remainder of the report is organized as follows: the data and the model design are described in the methods section, the results section shows the regression results and discussion, and the conclusion section discusses the conclusion, contribution, and limitations.

Methods

Data source

The data in this article were constructed from an extensive database by appending two subdatabases from the BCBH website. One includes parents looking for abandoned or lost children, while the other focuses on children who seek their parents. The quality of this active family search data is relatively high because most registrants provided true information and as many details as possible to improve the probability of finding their child/parent. A total of 107,679 cases of missing persons from 1980 to 2020 were collected through web crawling (as of March 8, 2022) and text mining. Second, the province information in the addresses was extracted and calibrated for accuracy. Four cases that did not contain gender information were removed from the sample. The cases in which children were over 12 years old when they left their parents were also removed because they were more likely to be runaways than abandoned children. The description section of the database is divided into two columns and was designed by the website to allow reporters to leave as much information as possible. Using text fuzzy identification methodology, we screened self-reported situation descriptions in the database by using text recognition to exclude nonvoluntary cases such as children who were kidnapped (guai), stolen (tou), robbed (qiangjie, means robbed by force), abducted (youpian), trafficked (mai), and coerced (qiangpo, means by force) (Ma et al., 2020). We also conducted an additional robustness check.

We extended the coverage of Ebenstein’s (2010) measure of the provincial-level OCP penalty rate to 1980–2020. The OCP penalty rates (illustrated by per capita annual income by province) were collected from the official documents covering provincial-level administrative measures of family planning policies. The ratio of the minimum requirement of social maintenance fees for overbirthing was calculated in the per capita disposable income of residents in each province. Provinces have diverse birth control policies for different groups, and the number of provincial birth quotas is obtained by weighting the population base of provinces. In addition, fertility decline is a comprehensive result of OCP, which may lead to an intermediate effect on prenatal gender discrimination (Jayachandran, 2017). We used the province-level birth rate from the China Population and Employment Statistics Yearbook as a control variable.

We used two cultural measurements. One is the number of genealogies to population ratio in each province (Murphy et al., 2011), and genealogy information is available from the Shanghai Library (https://jiapu.library.sh.cn/#/), which has the largest collection of genealogical books in China. The other is the proportion of the Han ethnic population to the provincial population in 2000 since son preference is traditional in this population. We used the latter measure for a robustness check.

We also include socioeconomic variables, such as average schooling, urbanization rate, and income per capita, to control for province-level heterogeneity in economic development and rural-urban inequality (Fan et al., 2011; Knight 2014). These measures of development factors were obtained from the China Statistical Yearbooks. The illiteracy rate in the National Census is calculated with the population of 6-year-olds and above, which may not be an appropriate indicator. Therefore, we lagged it by 3 years to represent a more mature group.

Data description and hypotheses

As shown in Fig. 2, there have been waves of child abandonment since 1949, and the proportion of abandoned girls accounted for more than half in most years. The first surge occurred before the introduction of family planning policies during China’s Great Famine from 1958 to 1961. The abandonment incidence increased after the introduction of the family planning policy in the late 1970s, peaking in 1990 and subsequently falling, with the gender ratio leveling off. A slight rise can be seen in 2015 and 2016 when China modified its OCP and shifted into the universal two-child policy era.

Fig. 2: Evolutionary characteristics of abandoned children over time in China, 1949–2020.
figure 2

Cases registered on the Baby Come Back Home website up to 2022-03-08.

Full size image

Family planning policies and son-preference cultures vary across provinces. Likewise, child abandonment has a divergent regional profile by province (Fig. 3). In provinces with strict OCP implementation and a strong clan culture, such as Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Hubei, Hunan, Guangdong, and Sichuan, the abandonment of girls occurred more in the 1990s. In contrast, there is a pattern of abandoning more boys emerging in Guizhou. Considering the provincial-level differences in population size, the density of abandoned children was calculated with the 2000 census data. We also present the trend of the SRB in Fig. 4, which illustrates the relationship between the SRB and the abandonment sex ratio in high clan culture provinces such as Jiangxi, Fujian, and Guangdong.

Fig. 3: Provincial distribution of abandoned boys and girls in China, 1980–2020.
figure 3

Cases registered on the Baby Come Back Home website up to 2022-03-08.

Full size image
Fig. 4: Sex ratio at birth (SRB) in various provinces of China, 1980–2020.
figure 4

Cases registered on the Baby Come Back Home website up to 2022-03-08.

Full size image

Furthermore, we divided the provinces into corresponding groups according to the strictness of OCP implementation and the strength of clan culture. According to Gu et al. (2007), China’s family planning policies are divided into 4 groups: one child, 1.5 children, two children and three children, accounting for 35.4%, 53.6%, 9.7% and 1.3% of the sample, respectively. In the manuscript, we further combined the two-child and three-child groups into an OCP_Loose policy zone, defining one child and 1.5 children as OCP_Strict and OCP_Medium, respectively. Figure 5 exemplifies that the abandonment of boys and girls varies across regions with assorted levels of OCP strictness, although they all undergo a process of rising and then falling. Clan culture zones are divided by their median value to ensure comparable numbers for the 2 subsamples. As shown in Fig. 6, the low clan culture group does not demonstrate a gender difference. However, in the high clan culture group, the gender difference was significant before 2000. Therefore, this study also contributes to the literature on China’s child abandonment rate by considering the complexity of China’s divergent cultural traditions and policy implementation.

Fig. 5: Abandoned boys and girls in different policy zones.
figure 5

According to Gu et al. (2007), China’s family planning policies are divided into 4 groups: one child, 1.5 children, two children and three children, accounting for 35.4%, 53.6%, 9.7% and 1.3% of the sample, respectively. In the manuscript, we further combined the Two children and Three children groups into an OCP_Loose policy zone. One child and 1.5 children were defined as OCP_Strict and OCP_Medium, respectively. The OCP_Strict zone includes Bei**g, Tian**, Shanghai, Jiangsu, Chongqing, and Sichuan; the OCP_Medium zone includes Hunan, Guangdong, Shanxi, Liaoning, Jilin, Heilongjiang, Tibet, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, and Hubei, and the OCP_Loose zone includes Hebei, Inner Mongolia, Henan, Guangxi, Guizhou, Shaanxi, Gansu, Hainan, Yunnan, Qinghai, Ningxia, and **njiang. Cases registered on the Baby Come Back Home Website up to 2022-03-08; 2. According to Gu et al. (2007), China’s family planning policies are divided into 4 groups: one child, 1.5 children, two children and three children, accounting for 35.4%, 53.6%, 9.7% and 1.3% of the sample, respectively. In the manuscript, we further combined the Two children and Three children groups into an OCP_Loose policy zone. One child and 1.5 children were defined as OCP_Strict and OCP_Medium, respectively. The OCP_Strict zone includes Bei**g, Tian**, Shanghai, Jiangsu, Chongqing, and Sichuan; the OCP_Medium zone includes Hunan, Guangdong, Shanxi, Liaoning, Jilin, Heilongjiang, Tibet, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, and Hubei, and the OCP_Loose zone includes Hebei, Inner Mongolia, Henan, Guangxi, Guizhou, Shaanxi, Gansu, Hainan, Yunnan, Qinghai, Ningxia, and **njiang.

Full size image
Fig. 6: Abandoned boys and girls in different cultural zones.
figure 6

Clan culture zones are divided by their median value to ensure comparable numbers for the 2 subsamples; Clan culture_Low zone includes Tianjian, Shaanxi, Inner Mongolia, Hainan, Gansu, Guizhou, Heilongjiang, Chongqing, Qinghai, Ningxia, Shanxi, Guangxi, Hebei, Sichuan, Jilin and Yunnan; Clan culture_High zone includes Hubei, Jiangxi, Fujian, Bei**g, Liaoning, Zhejiang, Shanghai, Guangdong, Anhui, Henan, Shandong, Jiangsu, and Hunan. 1. Cases registered on the Baby Come Back Home website up to 2022-03-08; 2. Clan culture zones are divided by their median value to ensure comparable numbers for the 2 subsamples; Clan culture_Low zone includes Tianjian, Shaanxi, Inner Mongolia, Hainan, Gansu, Guizhou, Heilongjiang, Chongqing, Qinghai, Ningxia, Shanxi, Guangxi, Hebei, Sichuan, Jilin and Yunnan; Clan culture_High zone includes Hubei, Jiangxi, Fujian, Bei**g, Liaoning, Zhejiang, Shanghai, Guangdong, Anhui, Henan, Shandong, Jiangsu, and Hunan.

Full size image

Hypothesis 1 The strictness of OCP is positively associated with the number of children abandonments.

Hypothesis 2 In the provinces with high son-preference culture, more boys are abandoned.

In a society with a son-preference culture, only boys are considered family heirs, while girls are treated as the future brides of other families. Hence, the logic of Hypothesis 2 was that in the provinces where the clan culture was strong, sons would receive preferences over daughters. However, the birth quota, typically one child for most Chinese families, often cannot satisfy the need for a male heir. Therefore, for an extra penalty-free try for sons, some families might hide their birth records and abandon earlier-born daughters (Ebenstein, 2010). Chen et al. (2013) showed that as ultrasound diagnosis machines spread nationwide, prenatal sex selection became much more prevalent in China. As shown in Fig. 4, we found a highly skewed sex ratio at birth using census data as the evidence.

The OCP comprises two parts. First, the local governments assign a birth quota to each family according to provincial-level regulations and the conditions of the family. Second, if the number of children exceeds the quota, the spouses must pay a monetary penalty for each illegal birth. Therefore, the policy indicator is a composite indicator (see Eq. 1) of the provincial-level allowable birth quotas (Gu et al., 2007) and penalty rates (Ebenstein, 2010). Thus, higher penalty rates and lower birth quotas led to stricter birth control. As a result, we divided the provinces into three groups according to the toughness of OCP implementation. The OCP_Strict zone contains provinces with the strictest birth quota, restricting each reproductive couple from having more than one child. The OCP_Medium zone includes the provinces that allow the reproductive couple to have a second birth quota if the first child is a girl. The OCP_Loose zone contains the provinces where the birth quota was two or above.

$$pol = penalty\,rate/birth\,quota$$
(1)

We collected the OCP policy factor Pol (see Eq. 1, when Pol is larger, the implementation of OCP is stricter), the son-preference culture factor Cul (the number of genealogy books divided by the size of provincial population) and Han (the share of Han Chinese population) as IV, and other provincial-level control variables (disposable income per capita, illiteracy ratio, urbanization ratio, and birth rate). Table 1 shows the descriptive statistics. Each variable is collected by province and year. As shown in Table 1, the abandonment of girls has been more prevalent than that of boys. In parallel, the other variables show little difference between these two groups by observing the mean, standard deviation, minimum, and maximum values.

Table 1 Descriptive statistics.
Full size table

Model design

In previous studies, the OCP and son-preference culture have been cited as reasons for the increase in the number of abandoned children, girls in particular (Poston et al.,1997; Shang et al., 2005; Murphy et al., 2011; **ong (2023)). However, these factors do not explain the abandonment of boys versus the overall decrease in abandonment. Moreover, the interaction between policy and culture has yet to be determined. We follow Chung and Das Gupta (2007) and consider the impact of developmental factors (education, urbanization, and economic growth) on fertility as control variables (Chung and Das Gupta, 2007; Cohen et al., 2008; Wang and Chi, 2017), in addition to the effects of state policies (which have strengthened patriarchy since the 1950s in South Korea) and son-preference clan culture (Zhang and Ma, 2017).

To examine the causes and to clarify the role of policy and culture, we further constructed a multiple regression model (Eq. 2) at the provincial scale. The following equation is estimated using the OLS model with robust standard errors.

$$Y_{i,t} = \alpha _0 + \alpha _1Pol_{i,t} + \alpha _2Pol_{i,t} \ast Cul_i + X_{i,t} + \eta t + \zeta _{\it{i}} + \varepsilon _{i,t}$$
(2)

The dependent variable is the density (total abandoned cases weighted by each provincial population of 2000) in province i and year t. Pol represents the family planning policy variable. Poli,t * Culi represents the interaction between policy and cultural variables. X are control variables, including birthrate, illiteracy rate, family income, the square of family income, and urbanization rate. To avoid omitting variables, the model controls for time-fixed effects ηt and province-fixed effects ζi. εi,t is a vigorous standard error. We pooled all observations from 1980 to 2020. To further explore policy and cultural factors, we adopted three classifications for grouped regressions according to the stringency or leniency of OCP implementation. Furthermore, we performed group regression for the girl and boy groups.

Results

Table 2 shows the complete sample and diverse gender regression results, and Table 3 shows the regression results after 2012, when the OCP was relaxed. Overall, OCP had a more severe impact on abandonment, which tests Hypothesis 1. Moreover, the policy influence combined with the culture was even graver in these three groups. When we see Models 1, 3, and 5, which only show policy impact, the OCP positively influenced the total and boy groups and was a valuable supplement to previous studies that focused on the impact of OCP on girls (Johansson, Nygren (1991); Chen et al., 2015). However, the coefficient of “Pol” in Model 5 of the girl group was positive, and the scale is much smaller than the same coefficient, which is the coefficient of “Pol” in Model 3 of the boy group. However, the coefficient of Model 5 did not show significance, in contrast to common sense and previous son-preference literature that states that girls experience more sex discrimination under the OCP and a son-preference culture (Ebenstein 2010; Chen et al. 2015; Almond et al., 2018). One possible explanation is that girl abandonment is the typical method of postnatal sex discrimination. In contrast, due to the spread of prenatal ultrasound diagnosis, sex selection behavior shifts to prenatal sex abortion (Chen et al., 2013).

Table 2 Regression results of the total sample & separate gender groups.
Full size table
Table 3 Regression results before and after 2012.
Full size table

We added the son-preference culture factor into Models 2, 4, and 6 to moderate the effect of the policy. In Model 2, although the coefficient of Pol * Cul is −1.706, the absolute value is higher than that of the coefficient of Pol, and the mean of Cul is 0.371, indicating that the average impact of Pol (1.355*3.04 = 4.12) is larger than the moderating effect of the Pol * Cul interaction (1.706*0.371*1.355 = 0.86). However, we can presume a substantial moderating effect of culture since the standard deviation is 0.63, which is more significant than the mean of 0.371.

Moreover, in Model 4, the interaction culture factor does not influence the boys= group, suggesting that culture shows no interaction with the policy. Boys’ welfare has been solely damaged by the policy. Meanwhile, in Model 6, the interaction of policy and culture is negative, showing that the impact of the OCP penalty on girl abandonments becomes weaker in province with strong clan culture. However, these results do not suggest that the OCP or the son-preference culture reduces sex discrimination against girls. From Ebenstein (2010), we find that the sex ratio at birth becomes more skewed as the OCP becomes stricter, especially in the 2nd or later birth order. Moreover, Chen et al. (2013) show that with the spread of diagnostic ultrasound in China, which significantly reduces the cost of prenatal sex selection, the sex ratio at birth becomes more abnormal each year. Furthermore, the census given by national statistics shows the consistently highly skewed provincial sex ratio at birth, confirming our opinion that the interaction of OCP and clan culture has made couples choose to abort girls rather than abandon them.

Table 2 shows the change in OCP: Models 7, 9, and 11 present the strict OCP period, and Models 8, 10, and 12 present the relaxed OCP period. The Chinese government has changed its OCP since 2011, when one family could have a second baby if both parents were the only children in their family. In 2015, the policy was extended to an unconditional 2-child birth quota. Finally, the policy was diminished to an unconditional 3-birth quota for one family in 2021. Hence, we choose 2012 as the breakpoint for testing. The results hold stable before 2012, while the policy and culture show no impact after 2012. Model 12 in Table 3 illustrates the same result as Model 6 in Table 2. The negative “Pol*Cul” interaction means that girl abandonment has reduced as “Pol*Cul” increased. However, it could be that the reproductive couple chooses prenatal sex selection instead of postnatal girl abandonment.

The results of the strict OCP period are consistent in Table 2. The relaxed period results show that the OCP and its interaction with culture no longer impacted child abandonment.

In summary, OCP implementation is associated with child abandonment, in particular of boys. One possible explanation is that the demand for male heirs is so strong that the easing of OCP implementation in some provinces makes supply possible, and illegal adoption continues between zones; i.e., Central China provides the on-demand supply of boys from East China (Huang and Weng, 2019). Another possible explanation is that the OCP has increased acceptance of girls, since in the one-and-half-child zone, many reproductive couples have been granted a second childbirth quota if the first is a girl. Thus, the OCP could be recognized as an eclectic approach and a revision of the traditional son-preference culture.

Nevertheless, there may be an underestimation of abandoned girls. Despite the child-reporting dataset, the mortality rate is higher for girls in the 0–9 age group (Anderson and Ray, 2010). In addition, most abandoned girls are adopted as child brides in Central China (Chen et al., 2015), and abandoned girls receive less investment in their education (Chen et al., 2015); thus, an adopted girl may not have the desire to find her home because she has been trained to accept this fate. In addition, some underreporting of abandoned girls is possible because, by tradition, females are not treated as heirs, reducing the impulse to locate them.

Robustness tests

We performed 3 groups of regressions to test the model’s robustness. The results are shown in the Appendix. First, considering whether the number of genealogies is representative of fertility culture, we changed the proxy indicator from the number of genealogy books to the proportion of Han Chinese in each province. The larger the Han population is, the more likely it will be that Han culture will be dominant. The total sample and subgender results hold almost the same as the base model. However, the consequences of diverse OCP implementation zones show some alterations from the base model results, potentially due to the Han density and the genealogy density. The differences do not change the main findings.

Second, previous studies have demonstrated that most abandoned boys are disabled (Kay et al., 1998). Our primary regression did not consider the health status of the children. To further examine whether health conditions (illness and disability) affect the results, we manually identified 2486 cases of disease and disability. After removing them from the sample, we reran the regression. The results were unchanged.

Third, we considered whether boys’ parents would be more likely to regret abandonment. Who feels more shame? Patriarchal culture and female infanticide were once prevalent. Therefore, we used a dataset combination strategy to prevent the respective bias. To verify the resulting possible systematic differences, we ran the two original datasets separately (dataset 1: children seeking their parents; dataset 2: parents seeking their children). All the influences remained at a similar significance level in the regressions, and the core findings remained unchanged. Nevertheless, the two results are not identical, and the channels of influence for certain factors differ. As stated in the BCBH data description, each dataset has its bias, and all have an underreporting tendency for abandoned girls. We found results comparable to those obtained with the baseline model.

Conclusion

Child abandonment has existed in human societies for hundreds of years. In East Asian countries, girls are more likely to be abandoned than boys due to the strong patriarchal culture. Previous studies focus on “missing girls” and seldom consider abandoned boys. Above all, little is known about how fertility control policies have shifted gender-selective desertions since the 1970s. Using a new self-reported dataset, we can better explain the evolution of abandonment in China. More notably, OCP and culture significantly impact the evolutionary characteristics of abandoned children.

The OCP was responsible for boys’ abandonment but with a gentler impact. Girls suffer more from the OCP and the son-preference culture. The number of abandoned girls is not the result of cultural protection. In contrast, there might be a sad story: many female fetuses have been aborted to avoid OCP penalties since the nationwide spread of the ultrasound technique in the 1980s. By grou** 29 provinces (excluding Hong Kong, Macau, Taiwan, **njiang, and Tibet) into assorted policy zones, we examined the channels of influence of these three factors in particular regions, controlling for development factors (family income, illiteracy rate, and urbanization rate), birthrate, year, and spatial effects. The results are consistent with the baseline regression results.

Based on the empirical results, we argue that the phenomenon of child abandonment should be channeled at its source. The birth rate in China has slumped to its lowest level since records started in 1949, and the OCP has been further liberalized in 2021, the third childbirth quota. Moreover, the population started to experience negative growth in 2022. This finding implies that the policy factor influencing abandonment behavior no longer exists.

Our study could be expanded in several ways. First, we consider gender-selective abortion as a potential explanation for the gender-differentiated impact. Nevertheless, because of the lack of abortion data, we cannot directly examine this explanation. Second, we measure the toughness of OCP implementation with birth quotas and monetary fines. However, they are only one component of the OCP penalties. For instance, parents working in the public sector may lose their employment for violating the OCP. Therefore, our study might underestimate the effects of the OCP. Last, in the same period, China also experienced substantial transitions in other domains, such as marketization and urbanization. These alterations also change abandonment behaviors.