Descriptive statistical analysis
Table 1 indicates that the overall health status of migrant women of childbearing years is relatively favorable. More than 97% of these women perceive themselves to be in a healthy or basically healthy state, with a relatively small proportion of women reporting as unhealthy or cannot take care of themselves. This suggests that the health state of women of childbearing age plays a significant role in the decision-making process for migration. As the number of children increases, the proportion of unhealthy women of childbearing age gradually rises, while the proportion of healthy or basically healthy women of childbearing age gradually declines. Moreover, the self-rated health of women of childbearing age is found to be negatively associated with the number of children born preliminary.
Table 1
Number of children born and health status (unit: %)
Number of children born | Not able to take care of oneself | Unhealthy, but able to take care of oneself | Basically Healthy | Healthy |
---|
0 | 0.05 | 0.65 | 7.03 | 92.27 |
1 | 0.03 | 1.02 | 9.48 | 89.48 |
2 | 0.02 | 1.57 | 10.62 | 87.79 |
3 and above | 0.07 | 2.99 | 13.47 | 83.47 |
Data source: 2018 CMDS data |
Table 2 presents the means or frequencies of the main variables, with descriptive statistical analyses grouped based on whether childbearing occurred during this migration. The results indicate that the health status of migrant women of childbearing age is relatively good, with a mean value of 3.87. Furthermore, the average number of births is 1.43, significantly lower than the replacement level of 2.1. Regarding personal characteristics, the average number of years of education is 10.20 years; The proportion of women aged 25–34 is 47.61%, accounting for the largest share, followed by 35–44 years old, 45–49 years old, and 15–24 years old, respectively; 91% of women are Han Chinese, 83% are in agricultural household registration, and 96% are in marital status.
Among the health characteristics, it was found that 94% of women have at least one kind of medical insurance. Additionally, 53% of women have received reproductive health education within their communities or work units. Furthermore, 31% of women have established resident health records, and only 14% of women have contracted with a local doctor.
In terms of migrant characteristics, the largest share of the migration distance is inter-provincial migration, which is 48.60%, followed by inter-city within provinces and inter- county within cities. Concerning reasons for migration, 77.68% of women of childbearing age migrated for work and business, accounting for the largest proportion, followed by migration for marriage and other reasons. As for the areas of migration, 45.12% of women of childbearing age migrated to the eastern region of China, whereas 29.19% have migrated to the western region and 19.69% to the central region, and the least 5.99% to the northeast region.
The sample was classified based on whether or not childbearing occurred during the migration. The results indicate that women of childbearing age who gave birth during the migration have higher mean values for health status, the number of children born, years of education, and household income compared to those who did not give birth during this migration; moreover, the percentage of women who were married, received health education, established health records, and contracted with local doctors are also higher; in contrast, the percentage of Han Chinese and agricultural households are lower.
Compared to the women who did not conduct fertility behavior during migration, the proportion of women who conducted fertility behavior during migration is larger in the younger age group (15–44 years old) and lower in the senior age group (45–49 years old); the proportion of inter-city within the province is higher, while the inter-provincial migration and inter-county within the cities are lower among the migration distance; the percentage of individuals migrating for marriage and other reasons is more significant, while the proportion of that migrating for work and business is relatively more minor; the proportion of those who migrate to the eastern and western regions is relatively high, and those who migrate to the central and northeastern regions are relatively low.
Table 2
Descriptive statistical analysis
Variables | Definition | Total | Birth migration | No birth migration |
---|
Dependent variable | | | | |
Self-rated health | 1 = Not able to take care of oneself; 2 = Unhealthy, but able to take care of oneself; 3 = Basically healthy; 4 = Healthy | 3.87 | 3.89 | 3.86 |
Independent variable | | | | |
Fertility behavior | Number of children born | 1.43 | 1.56 | 1.32 |
Moderating Variables | | | | |
Education level | Years of education | 10.20 | 10.95 | 9.58 |
Household income | Logarithmic processing | 8.82 | 8.88 | 8.78 |
Personal characteristics | | | | |
Age | 15–24 years | 7.43% | 9.70% | 5.54% |
| 25–34 years | 47.61% | 59.43% | 37.82% |
| 35–44 years | 32.31% | 26.63% | 37.02% |
| 45–49 years | 12.65% | 4.24% | 19.61% |
Ethnicity | 1 = Han Chinese; 0 = Minority | 0.91 | 0.90 | 0.92 |
Hukou | 1 = Agricultural; 0 = Non-agricultural | 0.83 | 0.81 | 0.85 |
Marital Status | 1 = In marriage; 0 = Not in marriage | 0.96 | 0.98 | 0.95 |
Table 2
Descriptive statistical analysis (continued)
Health characteristics | | | | |
---|
Medical insurance | 1 = With insurance; 0 = Without insurance | 0.94 | 0.93 | 0.94 |
Health education | 1 = Received; 0 = Not received | 0.53 | 0.56 | 0.50 |
Health record | 1 = Established; 0 = Not established | 0.31 | 0.33 | 0.29 |
Local doctor | 1 = Contracted; 0 = Not contracted | 0.14 | 0.15 | 0.12 |
Migrant characteristics | | | | |
Migration distance | Inter-provincial | 48.60% | 47.95% | 49.14% |
| Inter-city | 33.54% | 34.55% | 32.71% |
| Inter-county | 17.85% | 17.51% | 18.14% |
Reasons for migration | Work and business | 77.68% | 70.25% | 83.83% |
| Marriage | 20.24% | 27.55% | 14.19% |
| Other reasons | 2.08% | 2.21% | 1.98% |
Migration areas | Eastern region | 45.12% | 46.8% | 43.73% |
| Central region | 19.69% | 18.81% | 20.42% |
| Western region | 29.19% | 29.30% | 29.10% |
| Northeast region | 5.99% | 5.09% | 6.75% |
[Table 2 about here]
Baseline regression analysis
Table 3 shows the results of the baseline regression analyses. The first model incorporated the moderating variable of years of education, the second model incorporated the moderating variable of household income, and the third model included both moderating variables. The results reveal that the increased number of children born significantly decreases the health status of migrant women of childbearing age. Models 1 to 3 are all at the 1% significance level, thereby hypothesis 1 is confirmed.
Models 1 to 3 indicate that the interaction term between the socioeconomic status variables (years of education and household income) and the number of children born positively affects the self-rated health of migrant women of childbearing age; this suggests that the socioeconomic status of migrant women positively moderates their health status and that as their socioeconomic status increases, the negative impact of having more children on their health is significantly reduced. Thus, hypothesis 2 is verified.
Regarding control variables, among the personal characteristics, it is found that the health status of women in the other age groups is poorer compared to women in the 15–24 age group; Han Chinese, agricultural households, and married women show better health statuses. Among the health characteristics variables, it is observed that possessing medical insurance, receiving reproductive health education, establishing health records, and contracting with a local doctor can significantly improve the health status of migrant women of childbearing age. Among the variables of migration characteristics, compared to inter-provincial women, the health status of inter-city and inter-county women is poorer; women who migrate for marriage and other reasons are less healthy than those who migrate for work and business. Compared to women who migrate to eastern China, women who migrate to the western and northeastern regions have poorer health status, and women who migrate to the central region have better health status.
[Table 3 about here]
Table 3
Baseline regression results
Variables | Model 1 N = 51286 | Model 2 N = 51286 | Model 3 N = 51286 |
---|
| Estimate | Std.error | Estimate | Std.error | Estimate | Std.error |
---|
Fertility behavior | -0.034*** | 0.009 | -0.133*** | 0.042 | -0.120*** | 0.041 |
Years of education | 0.004*** | 0.001 | | | 0.003** | 0.001 |
Fertility×education | 0.003*** | 0.001 | | | 0.003*** | 0.001 |
Household income | | | 0.024*** | 0.007 | 0.024*** | 0.007 |
Fertility×income | | | 0.014*** | 0.005 | 0.010** | 0.005 |
Age group | | | | | | |
15–24 years | | | | | | |
25–34 years | -0.029*** | 0.005 | -0.025*** | 0.005 | -0.029*** | 0.005 |
35–44 years | -0.091*** | 0.006 | -0.095*** | 0.006 | -0.090*** | 0.006 |
45–49 years | -0.194*** | 0.008 | -0.208*** | 0.008 | -0.191*** | 0.008 |
Ethnicity | 0.032*** | 0.007 | 0.037*** | 0.007 | 0.029*** | 0.007 |
Hukou | 0.018*** | 0.005 | 0.005 | 0.004 | 0.021*** | 0.005 |
Marital status | 0.030*** | 0.010 | 0.017* | 0.010 | 0.016 | 0.010 |
Medical insurance | 0.013* | 0.007 | 0.013* | 0.007 | 0.010 | 0.007 |
Health education | 0.027*** | 0.003 | 0.030*** | 0.003 | 0.027*** | 0.003 |
Health record | 0.019*** | 0.004 | 0.021*** | 0.004 | 0.019*** | 0.004 |
Local doctor | 0.019*** | 0.005 | 0.020*** | 0.005 | 0.020*** | 0.005 |
Migrant distance | | | | | | |
inter-provincial | | | | | | |
inter-city | -0.008* | 0.004 | -0.004 | 0.004 | -0.007 | 0.004 |
Inter-county | -0.015*** | 0.005 | -0.007 | 0.005 | -0.010* | 0.005 |
Migration reasons | | | | | | |
Work and business | | | | | | |
Marriage | -0.047*** | 0.005 | -0.042*** | 0.005 | -0.043*** | 0.005 |
Other reasons | -0.074*** | 0.015 | -0.069*** | 0.015 | -0.071*** | 0.015 |
Migration areas | | | | | | |
Eastern region | | | | | | |
Central region | 0.006 | 0.011 | 0.019* | 0.011 | 0.020* | 0.011 |
Western region | -0.052*** | 0.016 | -0.045*** | 0.016 | -0.034** | 0.016 |
Northeast region | -0.106*** | 0.018 | -0.092*** | 0.018 | -0.084*** | 0.018 |
Provincial control | Yes | | Yes | | Yes | |
Table 3
Baseline regression results (continued)
Constant term | 3.839*** | 0.021 | 3.677*** | 0.068 | 3.639*** | 0.06 |
---|
R2_a | 0.055 | | 0.056 | | 0.059 | |
Note: *P < 0.10; **P < 0.05; ***P < 0.01 |
Robustness test
Replacement model
As the dependent variable is an ordinal variable, the regression model is replaced with an ordered probit (Oprobit) model for robustness testing. Models 4 and 5 contain the moderating variables of years of schooling and household income, respectively. Model 6 includes the two moderating variables. The findings in Table 4 demonstrate that the key explanatory variable remains significantly negative, similar to the results in Table 3; this implies that as the number of children born increases, the health status of migrant women of childbearing age significantly declines.
In addition, the interaction term between the years of education and fertility behavior had a significant positive impact on self-rated health in both Model 4 and Model 6, indicating that the impact of fertility behavior on health status is positively moderated by years of education. Similarly, the interaction term between household income and fertility behavior in Model 5 has a significantly positive moderating effect on health status. However, the household income moderating effect was not significant in Model 6. In general, the robustness of the conclusions can be obtained through the method of replacement model.
Table 4
Robustness tests (Oprobit model)
Variables | Model 4 N = 51286 | Model 5 N = 51286 | Model 6 N = 51286 |
---|
| Estimate | Std.error | Estimate | Std.error | Estimate | Std.error |
---|
Fertility behavior | -0.052* | 0.027 | -0.255** | 0.125 | -0.241* | 0.123 |
Years of education | 0.026*** | 0.005 | | | 0.021*** | 0.005 |
Fertility×education | 0.006** | 0.003 | | | 0.005** | 0.003 |
Household income | | | 0.108*** | 0.024 | 0.094*** | 0.024 |
Fertility×income | | | 0.025* | 0.014 | 0.022 | 0.014 |
Control variables | Yes | | Yes | | Yes | |
Provincial control | Yes | | Yes | | Yes | |
Note: *P < 0.10; **P < 0.05; ***P < 0.01 |
Propensity score matching
As fertility behavior is a decision made by women of childbearing age, it is a non-random event, and various other factors can influence fertility behavior. Therefore, the estimated coefficients may be biased due to the self-selection problem. To further ensure the robustness of the results, this study adopts the propensity score matching method (PSM) to deal with the endogeneity issue, which can effectively address the problems of omitted variables and sample self-selection bias. The fundamental concept of PSM is to construct a counterfactual framework and an approximate "randomized experiment" to eliminate significant bias resulting from the observable characteristics of the treatment and control groups and calculate the average treatment effect.
Given that the independent variable in this paper is continuous, we have categorized the samples whose number of children born exceeds the mean value of 1.43 as the treatment group, and those below 1.43 as the control group. We further examined the robustness of the results through four matching methods: K-nearest neighbor matching, radius matching, kernel matching, and intra-caliper K-nearest neighbor matching.
Table 5 presents the average treatment effects of fertility behaviors on the health of migrant women of childbearing age under different matching methods. We used the Bootstrap self-sampling methods to adjust the possible bias because of single matching. The results in Table 5 indicate that the average treatment effects obtained from various matching methods are significant at the 1% level. Migrant women with an actual number of children born above the mean value have a significantly worse self-rated health status of 0.014–0.017 compared to those born below the mean value. The ATT values are consistent across different matching methods, further supporting the robustness of the results.
Table 5
Robustness tests (Propensity score matching)
Matching method | Average treatment effect | Bootstrap std.error | T-value |
---|
K-Nearest neighbor matching | -0.017 | 0.005 | -3.51*** |
Radius matching | -0.014 | 0.004 | -3.39*** |
Kernel matching | -0.014 | 0.004 | -3.43*** |
Intra-caliper k-nearest neighbor matching | -0.017 | 0.004 | -4.16*** |
Note: K nearest neighbor matching set K = 4; radius matching set caliper value = 0.01; kernel function and bandwidth of kernel matching use their default values; intra-caliper K-nearest neighbor matching set K = 4, caliper value = 0.01; Bootstrap sampling number is 100 |
Heterogeneity analysis
The analysis above reveals that an increase in the number of children born significantly reduces the self-rated health of migrant women of childbearing age. However, there are notable distinctions among the childbearing behaviors of different groups of women of childbearing age, leading to varying impacts on women's health. Tables 6 and 7 explore the heterogeneity in the impacts of fertility behavior on the health of migrant women of childbearing age with varying household registration and migration distance, respectively.
As a result of China's dualistic economic institution, there exists a notable difference in the social resources and welfare accessible to women of childbearing age during the process of childbirth. Table 6 shows that an increase in the number of children significantly reduces the self-rated health of rural migrant women. However, this effect is insignificant for migrant women with non-agricultural household registration.
Table 6
Fertility behavior and health of migrant women of childbearing age in different household registration
Variables | Rural N = 42533 | Urban N = 8753 |
---|
| Estimate | Std.error | Estimate | Std.error |
---|
Fertility behavior | -0.127*** | 0.046 | -0.093 | 0.095 |
Moderating variables | Yes | | Yes | |
Control variables | Yes | | Yes | |
Provincial control | Yes | | Yes | |
Constant term | 3.676*** | 0.078 | 3.576*** | 0.128 |
R2_a | 0.060 | | 0.056 | |
Note: *P < 0.10, **P < 0.05, ***P < 0.01 |
This paper further conducts group regression based on the migrant distance; the results show that further of the migration distance, the negative impact of fertility behavior on women's health is more serious. Specifically, an increase in the number of children born can significantly reduce the health status of women who migrate inter-provincially and inter-city. In contrast, there is no significant adverse effect of fertility behavior on the health of inter-county migrant women. This situation could be attributed to the notion that as a woman migrates farther away from her place of origin, she may receive less support from her family and encounter more incredible difficulty in obtaining reproductive-related benefits.
Table 7
Fertility behavior and health of women of childbearing age with different migrant distance
Variables | Inter-provincial N = 24926 | Inter-city N = 17203 | Inter-County N = 9157 |
---|
| Estimate | Std.error | Estimate | Std.error | Estimate | Std.error |
---|
Fertility behavior | -0.129** | 0.053 | -0.161** | 0.082 | 0.032 | 0.096 |
Moderating variables | Yes | | Yes | | Yes | |
Control variables | Yes | | Yes | | Yes | |
Provincial control | Yes | | Yes | | Yes | |
Constant term | 3.737*** | 0.083 | 3.634*** | 0.132 | 3.306*** | 0.176 |
R2_a | 0.046 | | 0.070 | | 0.076 | |
Note: *P < 0.10, **P < 0.05, ***P < 0.01 |
Further Analysis
The concept of fertility behavior includes more than one indicator of the number of children born. Next, we investigated the impacts of different fertility behaviors on the health of migrant women of childbearing age in four aspects: whether to give birth during this migration, the number of boys born, the number of girls born, and the gender structure of fertility (number of boys divided by the number of girls).
The findings presented in Table 8 demonstrate that giving birth during migration significantly weakens the self-rated health of women of childbearing age, indicating that migrant women would face additional challenges if they chose to give birth during migration because it is difficult for them to access the same welfare benefits that local women can more easily obtain. Ultimately, this fertility behavior hurts the health of migrant women of childbearing age. Hypothesis 3 is verified.
Furthermore, models 8 to 10 reveal that a rise in the number of female children considerably diminishes the self-rated health of mothers. Conversely, the increase in the number of male children and the proportion of boys in the gender structure does not negatively impact the health of migrant women of childbearing age; meanwhile, the corresponding coefficient values are positive. The potential reason for this phenomenon may be the persistence of the "son preference" among the migrants. With the birth of more boys in migrant families, mothers can receive better healthcare and nutrition support from the family. Furthermore, the birth of male children can provide mothers with greater psychological satisfaction, ultimately resulting in no significant negative impact on their self-rated health.
Table 8
Effects of different fertility behaviors on the health of migrant women of childbearing age
Variables | Model 7 N = 51286 | Model 8 N = 47471 | Model 9 N = 47471 | Model 10 N = 27363 |
---|
| Estimate | Std.error | Estimate | Std.error | Estimate | Std.error | Estimate | Std.error |
---|
Birth migration | -0.009** | 0.003 | | | | | | | |
Number of boys | | | 0.002 | 0.003 | | | | | |
Number of girls | | | | | -0.005** | 0.003 | | | |
Fertility structure | | | | | | | 0.005 | 0.005 | |
Control variables | Yes | | Yes | | Yes | | Yes | | |
Provincial control | Yes | | Yes | | Yes | | Yes | | |
Constant term | 3.457*** | 0.036 | 3.426*** | 0.038 | 3.430*** | 0.038 | 3.444*** | 0.050 | |
R2_a | 0.058 | | 0.060 | | 0.060 | | 0.060 | | |
Note: *P < 0.10, **P < 0.05, ***P < 0.01 |
Given the continued decline in China's fertility rate in recent years, coupled with the substantial size of the migrant population of childbearing age, increasing the fertility rate of the migrants has become a crucial factor in improving the overall fertility level. The subsequent content centers on the impact of the health status of migrant women of childbearing age on their future fertility intentions. The binary dependent variable is whether or not migrant women plan to give birth in the following two years. The key independent variable is the self-rated health status of migrant women of childbearing age. Table 9 employs Logit and Probit models to perform the empirical analysis. The results demonstrate that enhancement in health status significantly raises the probability of migrant women’s intention to have children within the next two years.
As increasing the number of children born significantly negatively impact the health status of women of childbearing age, the negative health effect is one of the reasons that women forgo giving birth to a higher parity. This situation is manifestly incongruous with China's intention to elevate the national fertility levels.
The findings presented in Table 9 can enlighten Chinese policymakers. Specifically, the government ought to fully consider women regarding the negative health effect of fertility-related actions when it attempts to increase the overall fertility level. If migrant women of childbearing age could receive adequate support and assistance from their families and society during the phases of childbirth and child-rearing, and their health status is optimally protected, then it would significantly increase their desire to have more children in the future. This result presents novel insights for prospective policymakers who aim to raise fertility levels.
Table 9
Self-rated health and future childbearing intentions of mobile women of childbearing age
Variables | Logit N = 50002 | Probit N = 50002 |
---|
| Estimate | Std.error | Estimate | Std.error |
---|
Self-rated health | 0.106* | 0.059 | 0.058* | 0.031 |
Control variables | Yes | | Yes | |
Provincial control | Yes | | Yes | |
Constant term | -4.644*** | 0.690 | -2.806*** | 0.385 |
Note: *P < 0.10, **P < 0.05, ***P < 0.01 |