Health literacy is one of the important factors affecting people’s health and well-being, encompassing the ability to obtain, understand, and utilize health information and services to make behaviors and decisions that benefit their own health. As shown in Figure 4, sufficient and equal access to BPHS can enhance health literacy of rural migrant workers, leading to better health outcomes, reduced medical expenses and precautionary savings (29). Firstly, by providing health education and pamphlets, BPHS encourage timely medical treatment and ultimately reducing potential health risks. As a kind of human capital, good health condition ensures work efficiency and improves wage income, thus increasing household consumption (30, 31). Secondly, free physical examinations and medical insurance can alleviate the economic burden of medical expenses and increase the willingness to seek medical attention, resulting in more efficient utilization of medical resources, controlled health risks, and reduced crowding-out effect of medical spending on household consumption (15, 32). Finally, good health literacy can mitigate health anxiety and uncertainty, stabilize future income expectations, and reduce precautionary household savings and consumption volatility (14, 27, 33). Therefore, we propose the following research hypothesis:

Hypothesis 2: The enhancement of BPHS can promote rural migrant workers’ household consumption by improving health literacy.

Based on the push-pull theory, high quality public services act as positive factors that attract population inflow (34). As shown in Figure 4, BPHS are effective forces to attract talents. When rural migrant workers feel cared in inflow cities, the psychological and social distance will be shortened, and the sense of belonging and identity will be enhanced, thus increasing their willingness to become citizens. As for the definition of citizenization willingness, scholars commonly use long-term residence intention and hukou intention to measure it. Besides, some scholars further introduce psychological identity (28). Therefore, to some extent, the citizenization willingness in this paper is significantly related to citizenization behavior. The increase of citizenization willingness undoubtedly boosts household consumption. Firstly, as consumption environment is one of the key factors for consumption decision, cities with diverse consumption scenarios for goods enhance consumption expenditure and quality compared to rural areas (22, 35). Secondly, in the process of citizenization, the traditional consumption concept, consumption consciousness, and cultural psychology of rural migrant workers are imperceptibly impacted by urban lifestyle, they may unconsciously learn and imitate the consumption habits and modes of urban residents. Finally, household registration restrictions can be effectively solved after citizenization (36, 37). In other words, rural migrant workers can receive equal protection in employment, pension, medical care, education, housing and other aspects. The improvement of consumption power and the change of consumption willingness inevitably increase the household consumption demand and their marginal propensity to consume, especially in aspects such as culture, entertainment, transportation and communication (38). Therefore, this study proposes the following research hypothesis:

Hypothesis 3: The enhancement of BPHS can promote rural migrant workers’ household consumption by increasing citizenization willingness.

Household consumption of rural migrant workers includes all kinds of expenditure in production and life, such as food expenditure, medical expenditure, and housing expenditure. In this study, we took the logarithmic of households’ total annual consumption expenditure (Consumption) as the main dependent variable. The converted amount of food and housing covered by employment units were also included. The consumption excluding the above conversion costs (Consumption_exclude) was used for the robustness test. We also collected three items as alternative measures for household consumption of rural migrant workers, includingConsumptionrate, Averagepersonconsumption, and Averagelaborconsumption. Considering that high housing prices may affect household consumption decisions, we further tested the relationship between BPHS and Consumptionwithouthousing.

We conducted an indicator system to measure BPHS. Table 2 showed the specific indicators and their definition. The variable BPHSindex was calculated as follows. Firstly, we defined xij to be the actual benefit level of rural migrant worker i in the index j. In other words, the index measured whether rural migrant workers enjoyed the same level of BPHS as urban residents. Therefore, we used 0–1 dummy variables for representation. Secondly, we adopted the entropy method to determine the weight of each indicator (39). wj represented the weight of index j. Finally, we calculated BPHS based on the formula BPHSindexi=∑j=17wjxij.

A-F double critical value method is always regarded as a multidimensional poverty analysis tool (40). This study reevaluated BPHS according to this method for robustness test. Referring to existing studies (4), and considering the median value and mean value of BPHSindex (Table 3), we set the critical value of access to BPHS as 0.30, that is, when BPHSindex was larger or equal to 0.30, rural migrant workers were judged to be in the multidimensional access to BPHS, and the variable BPHS_equalequaled one. In other words, rural migrant workers basically enjoyed the same rights as city dwellers. Besides, we changed the critical value of access to BPHS to 0.40 for further test in order to avoid the difference of results caused by the subjective selection of critical values.

Based on the existing literature and field experience, we further included three levels of control variables (5, 13, 41). First, in terms of respondent characteristics, basic information and social attitude were included, such as Gender, Age, Education, marital status (Notmarried, Married, Divorcedorwidowed), Partymember, and working status (Employed, Employer, Self−employed). Second, at the household level, economic and demographic characteristics were taken into consideration. Specifically, including Familysize, number of households living with the respondent in the country of departure (Localsize), population burden ratio (Oldshare,Childshare), and the logarithmic of annual household income (Income). Third, at the macro level, we mainly considered city fixed effects.

To further overcome the potential endogeneity problems in OLS estimations, we employed the propensity score matching (PSM) method. PSM is mainly applicable when the core explanatory variable is a binary variable. Based on this, we used BPHS_equal for further test.⁶ In this study, 1:2 nearest neighbor matching (NNM) and 0.05 caliper matching (CM) were adopted.

Figures 5, 6 show the kernel density of propensity scores before and after matching of NNM and CM, respectively. The horizontal axis represents the propensity score and the vertical axis represents the density value. We can find that the kernel density curves of the treatment group and control group before matching are significantly different, while the kernel density equation curves of two groups are much closer after matching. These changes indicate that the observable feature differences between the two groups are significantly reduced after matching, and the matching effects are good.

Table 5 reports the estimated results of PSM. The empirical results show that after correcting the self-selection bias of BPHS by using PSM method, BPHS has a significant positive effect on household consumption of rural migrant workers. To be specific, the estimated average treatment effect is 1.01 and 1.51% using the 1:2 nearest neighbor matching (NNM) and 0.05 caliper matching (CM) respectively. As a conclusion, the above results are consistent with the theoretical analysis, verifying the hypothesis 1.

The instrument variable (IV) approach was also employed to address the potential endogeneity problems caused by omitted unobservable variables and reverse causality. As discussed in Supplementary material S2, we believed that County−levelindexwas an appropriate instrumental variable. The IV estimation results are shown in Table 6, where we include the same control variables and city fixed effects as in Table 4 column D. The t value of IV is significant at the 1% level and the first-stage F-statistics in the IV regression is 145.88, suggesting that this IV is a strong instrument, since it exceeds the conventional “rule of thumb” of 10 for an F-statistic.

The results in Table 6 indicate that BPHS has a significantly positive impact on household consumption of rural migrant workers at the 1% level, which is consistent with the results of OLS estimations. While, it should be pointed out that the value of the IV coefficient is quite larger than that in OLS regressions, which is quite usual (45). The household consumption of rural migrant workers increases by 4.50% for every 1 unit increases in BPHS. Hence, we conclude that our main aforementioned conclusions are still confirmed with mitigating potential endogeneity of BPHS.

Health literacy refers to the ability of rural migrant workers to access, understand, and process health information and services to make decisions that benefit their own health. In other words, health behavior is a decision-making behavior based on the improvement of information acquisition ability and health cognition, which reflects health literacy to a certain extent. Based on the existing literature and data availability in the CMDS, respondents were asked to answer “when you were last sick (injured) or unwell, where did you first go for medical attention?,” the option “nowhere (no treatment)” was assigned to the value of 0 and others equaled to 1 (46). In order to make a more intuitive comparison of the differences in health literacy between rural migrant workers and urban residents, we further selected urban assimilation of health habits to discuss (47). To be specific, the questionnaire item “there is a big difference between my health habits and those of local citizens” was used to measure this indicator, options including fully agree, basically agree, disagree, and completely disagree, with a value of 1, 2, 3, and 4 in order. A higher score indicates higher health literacy.

Tables 9, 10 report the results of the mediation effect test of health literacy, which is measured by health behavior and health habit, respectively. We have already found the welfare effects of BPHS in column D of Table 4, with the magnitudes of 3.46%. As presented in column A of Tables 9, 10, BPHS affects health literacy positively and significantly at the 1% level. The coefficient on BPHSindex, Healthbehavior, and Healthhabitare all significantly positive at the 1% level in columns B, which is consistent with economic theoretical inferences and implies that the enhancement of BPHS does boost consumption by improving rural migrant workers’ health literacy. Besides, compared with the direct effect, the positive change after introducing health literacy is smaller. In conclusion, the hypothesis 2 is verified.

Scholars commonly use psychological identity, long-term residence intention and hukou intention to measure citizenization willingness. In order to avoid errors and limitations in the measurement of a single index, this study selected indicators to measure the citizenization willingness of rural migrant workers from three aspects. (1) There were eight questions related to psychological identity in the CMDS.⁸ We divided agreement into four levels, ranging from completely disagree to completely agree on a scale of 1–4. The fifth through seventh sections were graded in reverse. Using equal weights, we then obtained the psychological identity score for each respondent, which ranged from 1 to 4; (2) In the CMDS 2017, statistics were made on the expected residence years of respondents in the local area. If rural migrant workers had the intention to live for more than 5 years, the value of Long−termresidenceintention was 1; otherwise, the value was 0; (3) According to the questions, “If you meet the requirements for local hukou, are you willing to move your hukou into the local government?,” if the answer was “yes,” the key variable Hukouintention equaled 1; otherwise, the value was 0. In terms of emotional level,Psychologicalidentity, Long−termresidenceintention, and Hukouintention show a progressive relationship.

Tables 11–13 summarize the outcomes of the mediation effect test of citizenization willingness, which is measured by psychological identity, long-term residence intention, and hukou intention, respectively. We have already verified that the enhancement of BPHS positively and significantly affects the household consumption of rural migrant workers. As presented in columns A of Tables 11–13, BPHS is positively and significantly associated with citizenization willingness which is measured by psychological identity, long-term residence intention, and hukou intention at the 1% level, with magnitudes of 0.257, 0.173, and 0.106, respectively. The coefficients on Psychologicalidentity, Long−termresidenceintention, and Hukouintention are all significantly positive at the 1% level in columns B. The results indicate that increasing citizenization willingness is one of the main channels through which the enhancement of BPHS boosts household consumption. Besides, after introducing citizenization willingness, the marginal effects of BPHSindex to Consumption are smaller than that in column D of Table 4. With reference to the mediation effect test procedures put forward by Baron and Kenny (43), we can conclude that the enhancement of BPHS can affect household consumption of rural migrant workers through the partial mediation effect of citizenization willingness, which verifies hypothesis 3.

It would be interesting and clear to present a table with the hypotheses where it is established which are accepted. Table 14 summarizes the hypothesis test results and indicates that all the hypotheses we proposed above have been verified.

The theory of New Economic Geography emphasizes the important role of regional geographical location in the migration (48). Labor mobility is driven by rational economic decision-making, as individuals weigh the costs and benefits to maximize their own utility (49). Typically, workers in economically underdeveloped regions are more likely to migrate to areas with more job opportunities, higher incomes, and better public services, particularly across provincial or city boundaries (50). However, in some populous provinces with high numbers of rural migrant workers, those who only hold NRCMS in their hometowns face significant barriers to accessing medical services in other locations due to insurance reimbursement restrictions. In contrast, urban migrants are often able to more easily adapt and integrate into new cities, requiring less assistance from the host community. Fortunately, with the continuous advancement in BPHS, rural migrant workers are increasingly able to access the same medical insurance policies as urban residents in their destination cities, and the practice of settling medical insurance in different places has become more widespread across the country.

This study divided rural migrant workers into trans-provincial, intercity within the province and inter-country within the city according to their flow patterns, and explored the heterogeneity of the flow patterns of rural migrant workers’ household consumption caused by BPHS. Table 15 represents that the promotion effect of BPHS on household consumption of rural migrant workers mainly exists in trans-provincial and intercity within the province rural migrant workers, and the impact on trans-provincial rural migrant workers is higher than the overall average effect. However, the enhancement of BPHS does not promote household consumption among inter-country within the city migration. Taken together, these results indicate that the enhancement of BPHS has a higher marginal effect on improving the utility and welfare of the trans-provincial migration.

Intergenerational difference is an important perspective in the study of rural migrant workers. Previous studies have divided rural migrant workers into two groups, the “old generation” and the “new generation,” based on whether they were born before or after 1980 (51). The “old generation” typically has lower levels of knowledge, skills, education, and weaker health conditions compared to the “new generation” (52). With sufficient provision of BPHS in inflow areas, rural migrant workers in the “old generation” can not only reduce their out-of-pocket health expenses but also alleviate their human capital deficiencies to a large extent (15), which ultimately leads to an increase in income and a decrease in precautionary savings, thereby improving their consumption level (14, 26). In contrast, the “new generation” faces more pressure to support their families and pay for mortgages. Although the enhancement of BPHS can improve their health literacy and increase their willingness to become citizens, their low income and high-pressure result in a lack of consumption power.

According to the conventional definition, we divided sample into old generation and new generation. The results of heterogeneous effects by generations are presented on Table 16. The coefficients on BPHSindex are significantly positive in both two groups, indicating that the enhancement of BPHS stimulates household consumption. In addition, the estimated average treatment effect is 4.81 and 1.31% in old generation and new generation respectively, reflecting the potentially positive role of BPHS for vulnerable groups with old generation.

There exists a positive correlation between income and consumption, which means that rural migrant workers’ households with higher income tend to have fewer budget constraints on consumption. These households are able to allocate more funds to areas such as culture, entertainment, and personal development (13). However, the primary goal of BPHS is to ensure the fundamental right of health, which is not directly linked to these consumption categories. Additionally, having more wealth provides greater certainty for the future, which can impact the welfare effect of BPHS (26). Therefore, while BPHS can have an impact on income and consumption, the marginal changes in income and consumption are not always apparent. As a result, it can be concluded that the enhancement of BPHS has a greater effect on improving the household consumption of lower-income levels compared to higher-income households.

Considering the large differences in the per capita disposable income of urban residents in different provinces, we compared the per capita disposable income of rural migrant workers with the income of urban residents in the destinations, and those higher than the income in the inflow cities were included into the higher-income group, the others were included into the lower-income group. Table 17 displays the results of heterogeneous effects by income. BPHS is positively and significantly associated with household consumption at the 1% level, with magnitudes of 3.67, and 2.53% in higher-income group and lower-income group, respectively. These findings suggest that the enhancement of BPHS plays a more significant role in improving the welfare of rural households with lower income, which reflects the inclusiveness of BPHS.