Introduction

With the comprehensive and in-depth development of industrialization, marketization, urbanization, and internationalization of China's economy and society, people's concept of fertility has undergone a major change. Women of childbearing age have continued to decline in their willingness to have children. The average number of children planned to have dropped from 1.76 in 2017 to 1.64 in 2021. The total fertility rate of the whole society has dropped sharply from 5.6 in 1970 to 1.06 in 2022, which is not only lower than the world average fertility level of 2.5 but also below the population replacement level of 2.1 and the international alert level of 1.5 [1, 18,19]. As an important public service policy in urban and rural areas, how does urban public health service affect the fertility intentions of urban and rural residents, especially rural migrant women? What mechanism will affect the fertility intentions of rural migrant women? At present, these issues have received little attention and need further discussion.

Therefore, the main purpose of this paper is to discuss the internal mechanisms and main effects of urban public health services on rural migrant women's fertility intentions.The following parts of this paper are arranged as follows: The second part is literature review and research hypothesis, which is mainly to sort out relevant literature results and put forward theoretical assumptions; The third part is the research design, mainly including data sources, variable selection and description, and empirical regression model setting. The fourth part is the analysis of empirical results. Using the data from the 2018 China Mobile Population Monitoring Survey, the mechanism and effect of urban public health services on rural women's fertility intentions are determined; The fifth part is the conclusion and discussion.

Literature review and research hypothesis

Literature review

The scale of rural women's mobility in China is huge, and because of the barrier between the urban–rural dual registered residence system and the unbalanced development, various economic and noneconomic costs in urban and rural areas are growing rapidly, and the pressure on urban and rural survival and development is increasing. Rural migrant women can neither enter the city nor return to the countryside for a long time. They have repeatedly demonstrated "amphibious" mobility between urban and rural areas [20], and their fertility concept has always fluctuated between rural socialization and urban socialization. Their fertility intention has always been dynamically evolving under the influence of the urban–rural dual economy, society, life, values, etc [21,22,23,24]. Scholars generally believe that the change in rural migrant women's fertility intentions is affected by both the common determinants of population fertility and the personality factors of rural migrant women.

Scholars in mainstream demography, sociology, economics, and other fields believe that there are many common factors in population fertility decision-making. Economic and social development and the growth of the means of living affect the macro-cyclical changes of fertility intentions [6, 25]. The social status of the floating population, the concept of marriage and childbearing culture, the level of income and wealth, the cost of childbearing, the value of children, education and occupation, the level of physical health, age, and the sex of children The preference for quantity and quality directly determines the fertility choice [26,27,28,29,30,31,32,33,34]. The government's family planning and public service policies have a significant regulatory effect on fertility intentions [35,36,37]. A large number of theoretical and empirical studies in economics, sociology, and demography also found that women's childbearing and work conflict led to women's employment informalization, employment interruption, employer discrimination, workplace unfairness, depreciation of human capital, etc., resulting in complex "fertility punishment" [38], and in different stages of social development, women's "fertility punishment" effects at different stages of life were different, leading to the increase of women's fertility costs and affecting their fertility intentions. It is estimated that the "fertility punishment" effect of British women is between 4.4% and 20% [38], and the "fertility punishment" of American women is about 5% to 20% [39], while the "fertility punishment" of Chinese women leads to a decrease of women's hourly wage by 7% to 18% [23], which has a significant social adaptation effect on migrant fertility [44, 51]. Fertility intention was also significantly negatively affected [10]. However, as rural migrant women's urban social adaptation has increased, their pressure for survival and development has decreased, their sense of happiness has increased, and the fertility compensation effect has occurred, potentially leading to an increase in fertility intentions and adaptability [12].

As a common factor influencing women's fertility, public health services have a significant impact on this. On the one hand, some scholars believe that public medical and health services will change people's conception of fertility and reduce their willingness to procreate. Research shows that the underdeveloped public health services in Europe in the nineteenth century led to high fertility in Europe [52]. Jiang et al. found that the public medical subsidies in China's planned economy period directly reduced people's desire to have children [53]; On the other hand, public medical and health services will improve women's fertility intentions because they will improve women's fertility safety [18, 19]; Third, urban public health services can enhance rural migrant women's urban social adaptability, enhance their willingness to stay in the city for a long time, reduce their mobility costs, and reduce their fertility burden and uncertainty. Their fertility intentions could be modified [62]. Based on the above analysis, this paper proposes the following research hypotheses:

  1. (1)

    Urban public health services directly affect the fertility intentions of rural migrant women.

  2. (2)

    Urban public health services affect the fertility intentions of rural migrant women through the health mechanism.

  3. (3)

    Urban public health services affect the fertility intentions of rural migrant women through the willingness to stay mechanism.

The logic of urban public health services, affecting the fertility intentions of rural migrant women is shown in Fig. 1.

Fig. 1
figure 1

Logic of urban public health services, affecting the fertility intentions of rural migrant women

Full size image

Research design

Data sources

The empirical data presented in this article was taken from the China 2018 National Migrants Dynamic Survey (CMDS), which was organized and implemented by the National Health Commission using a hierarchical, multi-stage, and scale proportional Probability Proportionate to Size Sampling (PPS) method. This survey covered 31 provinces (autonomous regions and municipalities directly under the Central Government) and the **njiang Production and Construction Corps and analyzed the public health services and the fertility intentions, health, residence willingness, employment, and other information of the migrating population. All of these were consistent with authoritative, scientific, large-scale, and targeted characteristics, and the research theme of the study. Since this focused on the fertility intentions of rural migrant women, only the women between the ages of 15 and 49, who were in their occasional childbearing age and resided in the urban area for six months or more, including those having an agricultural hukou, were included in this study. After screening, a total of 34,092 samples were included in this study.

Selection of variable and description

Interpreted variables

This study was interpreted as "fertility intention". The variables were taken from the CMDS 2018 questionnaire. For the question, "Do you have any plans to have children this year and next?" The answers "yes", "no", and "did not think well" respectively indicated that the respondents were willing to have children this or next year, did not have the willingness to have children in this or next year, and did not know their willingness to have children this or next year. These fertility intentions were measured as a binary variable, if the respondent selected "Yes", the variable was assigned to 1, while the value of "No" or "did not think well" is 0. Based on this, a total of 3623 samples were assigned 1, accounting for 10.63% of the valid samples included in this study.

Explanatory variables

Public health education and implementing the health file management in migrating populations were considered the key construction contents of urban public health services issued by The National Health and Family Planning Commission in the 2013 "Pilot Work Plan for Equalization of Basic Public Services for Health and Family Planning of Floating Populations". In this study, "public health education" and "health record management" were set as the core explanatory variables [19].

Public health education

For the question in CMDS 2018 questionnaire, "Have you received health education in your current community/workplace in the past year?", the public health education items, received by each respondent included "occupational disease prevention", "infectious disease prevention, and treatment", "reproductive health and maternal and child health", "chronic disease prevention and treatment", "mental health", "public emergency self-help", and "other aspects", scored from 0 to 7 respectively [35]. According to CMDS 2017 data, "other aspects", included health education on smoking control. On average, each migrant female received at least two public health education items, and 18.56% of rural migrant women did not receive any public health education.

Health record management

For the question in CMDS 2018 questionnaire, "Has the resident health file been established for you", the answers received included "Yes, has been established", "Not built, not heard", "Not built, but heard of", and "Do not know". If the respondent selected "Yes, has been established", the variable was assigned 1, while the other options were assigned as 0. The study variable received a total of 10,220 answers assigned 1, accounting for 29.98% of the total respondents.

Covariates

In this study, a series of covariates, reflecting the characteristics of rural migrant women, spouses, families, mobility, and inflow areas were empirically controlled. Among these, the individual characteristics were reflected in controlling the age, ethnicity, years of education, marital status, political status, employment status, participation in urban medical insurance, and the number of children of the rural migrant women. The family characteristics were reflected in controlling the nature of spousal hukou, family ethnic characteristics, family income level, and family housing burden of the rural migrant women. The mobility characteristics were reflected in variables, such as controlling the movement range of rural migrant women and the time duration they spent in the newly migrated place. The newly migrated places by the rural migrant women were characterized as the provincial areas.

Mechanism of variables

The theoretical part demonstrated the effects of urban public health services on the fertility intention of rural migrant women to have children via health mechanisms, labor intensity mechanisms, and residence willingness mechanisms. For the health variable in the questionnaire, "In the last year, have you been sick (injured) or unwell?", the answers included "Yes, the most recent occurred within two weeks", "Yes, the most recent occurred two weeks ago", and "No". For the first two answers, the assigned variable was 0, while for the last answer, it was 1. The residency willingness variable was measured from the question "If you intend to stay in the local area, how long do you expect to stay there?". If the respondent was willing to stay for 5 years or more, the variable was assigned as 1, while it was 0 for any other answer. The specific definitions and descriptive statistics of each major variable are listed in Table 1.

Table 1 Definitions and descriptive statistics of main variables
Full size table

Empirical regression model setting

Benchmark regression model

In this study, a logit model was used to investigate the impact of urban public health services on the fertility intentions of rural migrant women. Due to the binary nature of explanatory variables, the model expression was calculated as given in Eq. (1).

$${FD}_{i}={\beta }_{0}+ {\beta }_{1}{PHS}_{i}+ {\beta }_{2}Control+ {\varepsilon }_{i}$$
(1)

where \({FD}_{i}\) indicated the willingness of rural migrant women to have children, \({PHS}_{i}\) indicated the public health services, and \({Control}_{i}\) indicated the control factors, such as personal endowment characteristics, family characteristics, and mobility characteristics of rural migrant women.\({\beta }_{0}\), \({\beta }_{1}\), \({\beta }_{2}\) were the estimated coefficients in this study, among which, \({\beta }_{1}\) reflected the effects of public health services on the fertility intention of rural migrant women, and \({\varepsilon }_{1}\) indicated the random disturbance items.

Sample selectivity bias

The rural migrant women did not have random access to the public health services, relying on their self-selection of public health service needs to some extent. Therefore, the heterogeneity within this group would result in a large selective bias in the benchmark model estimates. In this regard, Rosenbaum & Rubin recommended the use of propensity score matching (PSM) to correct the sample selectivity bias in order to achieve an effect similar to that of the randomized trials [63]. However, Diamond & Sekhon argued that the PSM method did not assess the equilibrium of covariates well and that the estimated PSM was not reliable [64]. In order to solve this issue, Hainmueller & Xu proposed a solution as the entropy equilibrium matching method [65], which had many advantages over PSM. First, it could ensure that the processing and control groups were balanced in the sample characteristics and retained the useful information of all samples. Second, in the second phase of estimation, the model setting was more flexible. Third, the average difference test outcomes of covariates matched by this method were more reliable.

The basic idea of the entropy equilibrium matching method was constructing a binary dummy variable of the rural migrant women's access to the urban public health services, \({PHS}_{i}=\left\{\mathrm{0,1}\right\}\) which \({PHS}_{i}=0\) represent control groups and processing groups, respectively. By introducing the \(j\) covariate and construction matrix, the sample average effect expression was obtained, which is given in Eq. (2).

$$ATT=E\left[{Y}_{1}|PHS=1\right]-E\left[{Y}_{0}|PHS=1\right]$$
(2)

where \(E\left[{Y}_{0}|PHS=1\right]\) represented a counterfactual result. The entropy equilibrium directly estimated the weights using the potential large equilibrium constraint set. The counterfactual results were obtained using Eq. (3).

$$E\left[{Y}_{0}|PHS=1\right]=\frac{\sum \left\{i|D=0\right\} {Y}_{i}{w}_{i}}{\sum \left\{i|D=0\right\}{w}_{i}}$$
(3)

where \({w}_{i}\) was the entropy equilibrium weight selected using the control group. This study selected these weights using Eq. (4), which minimized the entropy distance metric.

$$\underset{{w}_{i}}{\mathrm{min}}H\left(w\right)=\sum_{\left\{i|D=0\right\}}{w}_{i}\mathrm{log}\left({w}_{i}\right./\left.{q}_{i}\right)$$
(4)

Equation (4) obeyed both Eqs. (5, 6, 7) for the averaging and normative constraints.

$$\begin{array}{cc}\sum\limits_{\left\{i\vert D=0\right\}}w_ic_{ri}\left(X_i\right)=m_r,r\epsilon1,&R\end{array}$$
(5)
$$\sum_{\left\{i|D=0\right\}}{w}_{i}=1$$
(6)
$${w}_{i}\ge 0,\left\{i|D=0\right\}$$
(7)

where \({q}_{i}\) was the benchmark weight, \({c}_{ri}\left({X}_{i}\right)={m}_{r}\)\({c}_{ri}\left({X}_{i}\right)={m}_{r}\) 1 was applied as a set of R equilibrium constraints on the covariate moment of the weighted control group. This step first weights the covariates in the scheme using Eq. (5) to impose the constraints on each covariate. The constraints generally included first-order moments (mean), second-order moments (variance), and third-order moments (skewness). This ensured that the covariate distribution moments of the reweighted control and processing groups were consistent.

Using these equilibria, normative and non-negative constraints, the entropy equilibrium method minimized the entropy distance between \(W\) and benchmark weight vector in Eq. (4) by searching for and selecting a set of unit weights\(W={\left\{{w}_{i}, ,{w}_{n}\right\}}^{T}\). Finally, this weight was used to perform a weighted least squares regression analysis on the effects of urban public health services on the fertility intentions of rural migrant women [42].

Analysis of empirical results

Analysis of baseline regression results

This study first analyzed the effects of urban public health services on the fertility intentions of rural migrant women using logit regression. The regression results of the health records and effects of health education on the fertility intention of rural migrant women are listed in Table 2(a) and (c). Given that the logit regression was a nonlinear model, the obtained parameters were not the true marginal effects of the indicators. The marginal effects of health records and health education, affecting the fertility intention of rural migrant women are also listed in Table 2(b) and (d). It can be seen that the participation of rural migrant women in health records increased their fertility intentions significantly and also enhanced their probability of planning to have children in the next two years by 1.6%. Each additional health education, received by the rural migrant women increased their fertility intentions by increasing their probability of planning to have children in the next two years by 0.1%. These results suggested that the urban public health services were effective in increasing the fertility intentions of rural migrant women, thereby confirming Hypothesis 1.

Table 2 Impact of the urban public health services on the fertility intentions of rural migrant women
Full size table

Table 2 also shows that the fertility intentions of rural migrant women were not only affected by the urban public health services, but were also closely related to their personal, family, and mobility characteristics. For example, the characteristics, such as age, marital status, employment status, number of children, spouse’s household nature, and mobility range, greatly decreased the fertility intentions of rural migrant women, while other characteristics, such as political status, family ethnicity, family income level, and length of residence in the newly migrated place significantly increased the fertility intentions of rural migrant women. In particular, an increase in the age of rural migrant women per unit reduced their chances of having children in the next two years by 0.6%. As compared to the remarried rural migrant women, the first-married rural migrant women showed a 5.7% lower probability of having children. Employment reduced the probability of having children by about 1% in the next two years for rural migrant women. An increase in the number of children per unit decreased the probability of having children in the next two years by about 17.2% for the rural migrant women. Those with an agricultural spouse were also less likely to have children, showing a 0.9% decrease in the probability of having children in the next two years. As compared to the rural migrant women, who moved across provinces, the probability of having children in the next two years decreased by 0.8% among those, who moved within provinces. The rural migrant women with party membership showed a 1.7% increase in the probability of having children in the next two years. Either spouse in a household who was an ethnic minority can increase the probability of rural migrant women having children by 0.9% in the next two years. The fertility intentions of rural migrant women increases with increased household incomes, an increase in the unit household income level variable increased the probability of having children in the next two years by 2.9%. The increase in the time duration of migration increased the probability of having children in the next two years by 0.1%.

Analysis of entropy equilibrium matching results

Receiving public health services by the rural migrant women was based on their "self-selection" to a certain extent and affected by the characteristics of individual covariates. A comparison of the fertility intentions of rural migrant women, seeking public health services, should be performed to analyze the systematic differences between the two groups. In order to alleviate the problem of between-group bias in the assessment of effectiveness, this study used the entropic equilibrium matching method to set constraints on the first-order moments (mean), second-order moments (variance), and third-order moments (skewness) for the covariates in the regression analysis. This ensured that the control and treatment group samples had equal-weighted means on each covariate, which helped in alleviating the problem of between-group bias in the assessment of effectiveness. As compared to the PSM, this method was more conducive to achieving a balanced distribution of variables among the groups and was less affected by the measurement error while achieving the accurate matching of samples. The binary nature of the core explanatory variables required the division of control and treatment groups. As a core explanatory variable in this study, “health records” was a binary variable, but public health education was not a binary variable. Therefore, this study, using different methodologies, made two attempts to adapt health education to a binary variable. First, the number of health education items received by the rural migrant women were used as the criteria for dividing the rural migrant women into treatment and control groups; those, having at least one health education, were assigned 1 and included in the treatment group, while those, lacking health education, were assigned 0 and included into a control group. This logic generated binary variables with 1, 2, 3, 4, 5, 6 health education items. Second, the rural migrant women were divided into treatment and control groups based on the contents of health education they received. For example, if the rural migrant women had received reproductive, maternal, and child health education, they were considered as the treatment group and assigned a value of 1, while those, lacking any of these education contents were considered as the control group and assigned a value of 0. The binary variables for each type of health education were generated using this logic. It was worth noting that, by adjusting the first-order matrix, the binary variables could be effectively matched [42], and no further second and third-order moment adjustments were required for such variables. Table 3 reports the Average value, variance and matching test results of main covariates before and after entropy equalization.

Table 3 Matching test for main covariates
Full size table

The data after entropy equilibrium matching were re-regressed. Table 4 enlists the results of entropy equilibrium matching estimates obtained from the effects of public health services on the fertility intentions, as well as the comparison of results before and after treatment. The magnitude and significance of the coefficients changed due to the use of entropic equilibrium matching. In particular, kee** the health records had a significant positive effect on the fertility intentions of rural migrant women and increased the coefficient of fertility intentions by 0.135 as compared to those, who did not keep health records. The health education variable had a significant and favorable impact on the fertility intentions of rural migrant women only if they had received at least four or more health education content. The effects of public health services on fertility intentions were related to the contents of health education received by the rural migrant women. Only the reproductive health, maternal, child health, and self-education for public emergencies affected the fertility intentions of rural migrant women, while the other aspects of health education had a significant negative effect on the fertility intentions of rural migrant women.

Table 4 Comparison of the urban public health services on the fertility intention after entropic equilibrium matching
Full size table

Mechanism test

For investigating the in-depth mechanism of the effect of urban public health services on the fertility intentions of rural migrant women and verifying the existence of the theoretical transmission channel between the health status and fertility intentions, the test models were established, which are given in Eqs. (8) and (9).

$${mediator}_{i}= {a}_{0}+ {a}_{1 }{PHS}_{i}+ {a}_{2}Control+ {\varepsilon }_{1}$$
(8)
$${FD}_{i}={\gamma }_{0}+{\gamma }_{1}{mediator}_{i}+{\gamma }_{2}{PHS}_{i}+{\gamma }_{3}Control+{\epsilon }_{2}$$
(9)

where \({mediator}_{i}\) was the mediating variable. This study chose health status and willingness to describe the results based on the theoretical analysis. The specific test steps were as follows. First, a model was used to see if there was a positive promoting effect on the correlation between the urban public health services and mediating variable (8). Based on its coefficient being significantly positive, the mediating variable was introduced into the basic empirical model (1) to obtain model (9) and determine if it met the criteria for a mediating mechanism based on its coefficient and significance.

The results of the mediating mechanism test are listed in Tables 5 and 6, where Table 5 enlists the results of the health status mediating mechanism test and Table 6 shows the results of the willingness to stay mediating mechanism test. In Table 5, the results of urban public health services (health record management) show that the health level of rural migrant women improved by 20.2% due to health records. The coefficient of the effect of urban public health services (health record management) on the fertility intentions of rural migrant women after the introduction of health status was 24.7%, which was lower than the baseline regression coefficient of 25.2% without the introduction of health status, thereby passing the mediating effect test. The results of urban public health services (health education) showed that the health level of rural migrant women improved by 5.7% through health education. The coefficient of the urban public health services (health education) effects on the fertility intentions of rural migrant women was 2.1% after the introduction of health status. This was lower than the baseline regression coefficient of 2.2% without the introduction of health status, thereby passing the mediating effect test. These findings showed that the improvement of health status was an important influencing mechanism of the urban public health services to increase the fertility intentions of rural migrant women.

Table 5 Mechanism test and health status of the rural migrant women
Full size table
Table 6 Mechanism test and willingness to stay in the rural migrant women
Full size table

Table 6 shows the result of the intermediary mechanism test for the residency intention of rural migrant women. The health records could increase their residency intention by 33.6% after introducing the residency intention variable. The coefficient of the effects of health records on the fertility intentions of rural migrant women was 24.6%, which was lower than the baseline regression result of 25.2%, thereby passing the mediating effect test. Health education could increase the stay intention of rural migrant women by 1.8%, and the coefficient of the effect of health education on rural migrant women’s intention to have children was 2.2% after the introduction of the intention to stay variable. This was not significant in comparison to the decrease in baseline regression coefficient due to the retention of decimal places in the reported results, but it still passed the mediating effect test. As evidenced by the findings above, an increase in the willingness to stay was an important influencing mechanism of the urban public health services to improve the fertility intentions of rural migrant women.

Further analysis

These studies confirmed the positive effects of urban public health services on the intentions of rural migrant women and verified their transmission pathways and mechanisms. The direct mechanisms of action and transmission pathways might be heterogeneously different between the various groups of rural migrant women due to the differences in the demand for and access to urban public health services. Therefore, this study deeply explained the heterogeneity in terms of pregnancy experience, income level, and the duration of inflow into residence.

These results showed that the women, who had previous pregnancy experience, were more likely to rely on their own experiences rather than urban public health services as compared to those, who had no previous pregnancy experience. The results of the heterogeneity analysis of pregnancy experience are listed in Table 7. The interaction term between health records and pregnancy experience (HR_PE) was -0.445, and that between health education and pregnancy experience (HE_PE) was -0.086, both of which were significant at the 1% level, indicating that women's pregnancy experience weakened the promotion effect of urban public health services on their fertility intention. The results of the heterogeneity analysis regarding the availability of pregnancy experience showed that urban public health services were more effective in increasing the fertility intentions of women with no pregnancy experience.

Table 7 The heterogeneity analysis about pregnancy experience
Full size table

Considering the impact of urban public health services on rural migrant women's fertility intentions from a cost perspective, then rural migrant women's high or low income would be an important moderating factor influencing urban public health services to promote their fertility intentions. This paper argues that with higher income, rural migrant women can maintain their health status through other channels and also face less life stress, thus the effect of urban public health services on the promotion of rural migrant women's fertility intentions will continue to decrease. The results of the heterogeneity analysis of income levels are shown in Table 8. The interaction term between health records and income level (HR_IL) was -0.82, which was significant at the 1% level, while the interaction term between health education and income level (HE_IL) was -0.047, but did not pass the significance test. The results of the heterogeneity analysis regarding income level showed that as the income level of rural migrant women increased, the promotion effect of health records on their fertility intentions gradually decreased, and the promotion effect of health education on their remaining intentions was not significantly affected. This is due to the fact that health records have an impact on health primarily at the level of health care cost savings, while health education has an impact on health primarily at the level of health awareness.

Table 8 The heterogeneity analysis about income level
Full size table

The length of time spent in the place of residence is an important influencing factor on rural migrant women's identity, environmental integration and fertility intentions. This paper argues that as the time of inflow to the place of residence increases, on the one hand, the long period of residence is conducive to the establishment of health records, and on the other hand, rural migrant women gradually have a stronger sense of belonging to the city and are more willing to accept health education. Thus the longer rural migrant women have been flowing into their place of residence, the stronger the effect of urban public health services in promoting their fertility intentions. The results of the heterogeneity analysis regarding the inflow time are shown in Table 9. The coefficient of the interaction term between health records and length of inflow to residence (HR_LIR) was 0.215 (significant at the 5% level), and the coefficient of the interaction term between health education and length of inflow to residence (HE_LIR) was 0.062 (significant at the 1% level). The results indicate that the promotion effect of urban public health services on rural migrant women's fertility intentions gradually increases as their time of inflow to their place of residence increases.

Table 9 The heterogeneity analysis about length of inflow to residence
Full size table

Conclusions and policy recommendations

This paper uses the 2018 CMDS data to study the impact of urban public health services on the fertility intentionsingness of rural migrant women and draws the following conclusions: (1) Urban public health services promote rural migrant women's fertility intentions through direct mechanisms, and the conclusion is still valid after using entropy equilibrium matching; (2) Through the mediation mechanisms of health status and willingness to stay, urban public health services promote the fertility intentions of rural migrant women; (3) Urban public health services have a better effect on improving the fertility desire of rural migrant women who have no pregnancy experience, a low income level, and a short residence time in the inflow area. The preceding study conducted an in-depth analysis of the mechanism and heterogeneity of urban public health services on rural migrant women's fertility intentions, providing reference empirical evidence for the government to formulate and improve the urban public health service system, improve the health status of rural migrant women and their level of citizenship and fertility intentions, and promote the homogenization and development of China's public health services.

The fertility behavior of rural migrant women is determined not only by the common factors of population fertility, such as age, gender, health, marriage and childbearing cultural concepts, income and wealth, fertility cost, child value, education, and occupation [25,26,27,28,29,30,31,32,33,34], but also by individual factors such as mobility interruption and social adaptation [42,43,44, 49,50,51]. Most scholars regard public health services as common determinants of women's fertility and point out that public health services may affect women's fertility intention asymmetrically by changing fertility concepts, reducing fertility costs, promoting health levels, reducing fertility risks, and improving satisfaction [52,53,66], but at the same time promote the fertility intentions of rural migrant women through the direct mechanism and the intermediary mechanism between health status and residence desire. Because of the long-term "amphibious" movement of rural migrant women in urban and rural areas, the rapid growth of various economic and non-economic costs in urban and rural areas, and the inability to effectively express their fertility intentions due to the huge pressure of urban and rural survival and development [21,22,23,24], while urban public health services such as health education and health archives, through various forms of information dissemination and targeted behavioral intervention, promote the efficiency and fairness of the supply of medical and health services in the whole society, On the one hand, it can reduce the cost burden of rural migrant women using medical and health services, alleviate the pressure on rural migrant women's own survival and development due to childbirth, and promote the expression of rural migrant women's fertility intentions; On the other hand, urban public health services can help rural migrant women avoid fertility risks by improving their health literacy and their health status, which can greatly reduce the economic and non-economic costs that rural migrant women bring to themselves because of fertility, thus hel** to release their fertility intentions; In addition, urban public health services are usually provided locally in the inflow area, which helps to enhance rural migrant women's sense of identity and belonging to the local area and makes them more willing to integrate into the inflow area and make long-term living arrangements, which can greatly reduce the flow cost of rural migrant women, reduce their reproductive burden and uncertainty, and improve their fertility intentionsingness. The study also shows that urban public health services have a better effect on the fertility intentions of rural migrant women who have no pregnancy experience, a low income level, and a short residence time in the inflow area. This is because rural migrant women without pregnancy experience can further release their convergent fertility under the constraints of lack of fertility experience, high fertility risk uncertainty, and large fertility punishment by using urban public health services; In terms of income level, rural migrant women with higher incomes can not only maintain their health status through other channels but also face less pressure for survival and development. Therefore, the promotion effect of urban public health services on rural migrant women's fertility intentions continue to decline, and the effect will be more obvious for rural migrant women with lower income levels; In terms of residence time in the inflow area, the longer rural migrant women live in urban areas, the greater the impact of urban economic, social, and cultural life and values, the lower their reliance on their children, the narrower the demand space for childbearing, and the naturally smaller impact of urban public health services on their fertility intentions.

Based on the research findings in this study, the following policy recommendations were recommended. First, the supply of public health services should be strengthened. Even though urban public health services could effectively increase the fertility intentions of rural migrant women, the government should increase the financial investment in public health services, open up social financing channels, improve the level of public health services, broaden the basic coverage of public health services, and further eliminate the worries of rural migrant women about giving birth. Second, the quality of public health services should be improved. The quality and depth of health education services and flexible customization of the health education service curricula based on the acceptance ability and needs of rural migrant women should be improved while increasing the efficiency of health education services. Third, the construction of health file informatization should be accelerated. The inadequate construction of health file informatization in some remote areas and the existence of health file information barriers were not conducive to the timely access and application of the health file information, which greatly reduced the convenience of applying health files to the migrating population. Local governments should expedite the development of a health file management model, providing full coverage to both the urban and rural residents, high informatization, high accuracy, and weak barriers, as well as develo** the homogenization and standardization of health file management services.