Care-receivers’ expected intergenerational support in aging rural China: A perspective of sustainable livelihoods and the sandwich effect

ABSTRACT

Expected intergenerational support reflects how potential care-receivers estimate the future supply of and demand for intergenerational support. The population of rural China is aging quickly, and future care-receivers in rural China are today’s young and middle-aged, who are now experiencing a decline in fertility as well as urbanization and migration. These changes influence the distribution of resources across generations within families and may cause the young and middle-aged to revise their expectations of intergenerational support. We focus on a “sandwich effect,” in which the young and middle-aged care for children and elderly parents simultaneously. Using survey data, we find that with respect to the expected supply of intergenerational support there are three categories; we call them “traditional concern,” “urbanized concern,” and “aging risk concern.” Expected demand for intergenerational support also has categorical features: A minority expects high demand with a preference for financial and instrumental support, while the majority expects low demand but focused on emotional support. Families’ sustainable livelihoods have significant impacts on individuals’ expectations of intergenerational support. The “sandwich effect” plays moderating roles on the relationships between families’ sustainable livelihoods and individual expectations.

KEYWORDS:

• Care-receivers
• Aging society
• Expected intergenerational support
• Sustainable livelihoods
• Sandwich effect

1. Introduction

The fraction of China’s population over 60 years old was 18.7% in the 2020 census and is expected to increase steadily in the coming decades. Much more quickly than other developed countries, China is becoming a predominantly elderly society (Djundeva et al., 2019; Hvistendahl, 2013). Due to the out-migration of young rural labor and the return-migration of middle-aged rural workers, rural China will age faster than the nation as a whole (Li & Lin, 2016). Compared to the Western world, rural China is a typical example of aging before wealth: Neither the pension system nor medical insurance satisfies people’s needs, and there is a pervasive lack of institutional care due to the rural history of disadvantaged living conditions (Johansson & Cheng, 2016; Zhang et al., 2012). This is why the aged in rural China depend on their adult children’s support rather than public welfare (Giles et al., 2010). In addition, both adult children and elderly parents in rural China take intergenerational support for granted due to the persistence of multiple-generation residency and the norm of filial piety (Cong & Silverstein, 2008a; Li & Lin, 2016).

With rapid urbanization and modernization, today’s young and middle-aged—the care-receivers in the future—may be confronted with aging challenges unprecedented in China’s history. Compared to older generations, who have experienced the traditional lifestyle of rural China, the expectation of intergenerational support among the younger generation is likely to change (Zeng et al., 2008). The current young and middle-aged live in a transitional society where fertility is declining, the non-agricultural economy is growing, and traditional norms are weakening (Johansson & Cheng, 2016). In quantifying how public policy and family support will combine to form a comprehensive aging support system in future China, the expectation of intergenerational support among future care-receivers is an index that reflects the future supply of and demand for family resources in aging support (Chen & Han, 2016; Janus & Koslowski, 2020).

Rural individuals’ expectations of family support vary according to the availability of family resources (Canda, 2013). In rural China, adult children take care of parents using family resources that have accumulated across generations, and this leads to parents’ dependence on adult children (Liang et al., 2014). At the family level, sustainable livelihoods, including the family’s natural, physical, social, financial, and human resources, reflect how families accumulate and distribute resources across generations (Li et al., 2012). Sustainable livelihoods determine the connection between families’ resources and their intergenerational relationships, and may contribute to an individual’s expectation of intergenerational support. Under China’s rapid economic and social development, the young and middle-aged are more likely to develop greater independence from their parents and children (Qin et al., 2020), and with improvement in families’ sustainable livelihoods, they may change the amount and kind of intergenerational support that they expect to have as care-receivers in the future.

The young and middle-aged with children in China are a “sandwich” generation, in that family resources must eventually be shared with both their parents and their children (Giles et al., 2010). Since the distribution of family resources among three generations affects intergenerational relationships, this sandwich effect may play a moderating role on the relationship between families’ sustainable livelihoods and an individual’s expectation of intergenerational support (Warmenhoven et al., 2018), and in this study we analyze this moderating role among the young and middle-aged population. We use latent class analysis and a multilevel regression mixture model with data from a survey of rural families and individuals. First, a classification of individuals’ expected intergenerational support is made. Then this classification is studied in terms of the family’s sustainable livelihoods. The moderating role of the sandwich effect on the relationship between families’ sustainable livelihoods and individual expectation is then evaluated.

2. Literature

2.1. The current intergenerational support in rural China

Theories of intergenerational relationship, including intergenerational exchange theory, intergenerational altruism theory and intergenerational solidarity theory, are often used to describe how intergenerational support works (Bengtson & Roberts, 1991; Chou, 2011; Cong & Silverstein, 2008b, 2011, 2014; Fingerman et al., 2011; Fu et al., 2020). Financial support, instrumental support, and emotional support are the key components of intergenerational support (Cong & Silverstein, 2008a). Most elders in rural China currently have no reliable pension, and financial support from adult children supplies their basic survival needs when, after age 60, they become unable to obtain stable labor income (Warmenhoven et al., 2018). Due to the tradition of multiple-generation residency, instrumental support, which involves adult children taking care of elderly parents in person, is considered indispensable (Pei & Cong, 2020). The third is emotional support, by which elderly parents maintain stable connection with their adult children (Liu, 2017). Intergenerational support not only originates from filial piety and multiple-generation residency, but also involves the distribution of family resources and cooperation across generations in providing this support (Feng et al., 2012). In current rural China, the legal system ensures that supplying intergenerational support is the obligation of adult children (Chou, 2011). Even though rural families in China are living with rapid urbanization and massive migration, both culture and law ensure that intergenerational support remains stable (Phillips & Feng, 2015). However, the young and middle-aged in today’s rural China are experiencing significant demographic and social transitions, including rapid urbanization, decrease in fertility, and an increase in individualism, and, in light of these changes, whether expectations of intergenerational support among future care-receivers will be consistent with existing theories is not clear.

2.2. Expected intergenerational support for aging-society care-receivers

An individual’s expectation of family resources has two parts: One is the expected supply, which reflects whether or not there are resources that can be acquired from the family (Zhan, 2004); the other is the expected demand, which is the individual’s need for resources from his or her family (Wu & Li, 2014). Therefore, expected intergenerational support not only reflects whether the individual believes they will receive such support based on current family resources, but also whether they will have to depend on intergenerational support in the future (Cong & Silverstein, 2014). Under current urbanization, modernization, and migration in rural China, however, the supply of such support may be uncertain since family arrangement and rural lifestyles are changing. First, aged parents are less likely to obtain all their supports from their adult children since more adult children are living independently and are investing in their children’s education instead of support for their aged parents (Wu et al., 2016). Second, due to the separate living arrangements, some aged parents who are left behind in villages have to look for new solutions to their needs, such as relying on themselves and their communities rather than their children for care (He & Ye, 2014). Third, with the decrease of filial piety, elderly parents feel lonely and have less confidence that they will receive children’s emotional support since their children usually migrate into cities for jobs (Liu, 2017). The new communication platforms, such as video calls, connect left-behind elders virtually with migrant children and can maintain stable emotional connections (Hu et al., 2022). However, compared to urban elders who are used to an independent lifestyle, rural elders with migrant children still prefer to see their children in person since the rural culture emphasizes that children should talk regularly with elders in person instead of via video calls (Hu et al., 2022). In addition, for the rural elderly the inconvenience of using smart phones decreases the positive feelings of video calls and makes them more dependent on children’s in-person emotional support (Nie et al., 2021). Today, elderly parents no longer regard financial support from their children as the only way to obtain financial resources; instead, the pension system in some parts of rural China has improved, which helps elderly parents decrease their dependence on children’s finances (Su et al., 2017). In some rural communities in developed parts of China, children are able to hire professional help for their elderly parents, who no longer regard their children as their only acceptable caregivers (Zhu & Walker, 2018). However, no matter how the residential pattern or intergenerational relations change, emotional support from children will still be demanded by elderly parents (Croll, 2008).

As family structure and intergenerational relations change, the young and middle-aged, as well as their children, will develop new patterns of filial piety that could lead to different expectations of intergenerational support (Croll, 2008). Even though the majority of Chinese still observe filial piety, the demands for intergenerational support have become different (Sun, 2013). Take the urban case for example: Elders in cities today have retirement pay and public services, both of which effectively reduce their need for resources from children (Chen & Han, 2016). As the mechanism of resource exchanges related to filial piety changes, urban elders are likely to become more independent of their children and expect less intergenerational support (Chou, 2011). In developed rural regions where public services are similar to those in cities, the young and middle-aged, as well as their children, are more urbanized and have less psychological and material dependence on each other (Xu & Chow, 2011). Generally speaking, however, the welfare system in urban China is better developed than in rural China, where elderly parents are barely able to survive on resources from outside of the family (Chen & Han, 2016). It is unclear whether the young and middle-aged in rural China have the same expectation of intergenerational support as urban residents.

The contents of intergenerational support are also changing due to smaller family size and the resulting new family lifestyles among the young and middle-aged. Their expectation to acquire and to need intergenerational support may still exist, but may fall into various categories. With the multiplicity of rural family structures and intergenerational relations, the rural young and middle-aged vary in their perspectives as future care-receivers (Warmenhoven et al., 2018). In eastern rural China, for instance, young residents with more financial resources prefer intergenerational emotional connections rather than financial connections, while the latter are more prevalent among those in western rural China who are less well off. (Chen & Han, 2016). Further, expectations of intergenerational support among future care-receivers may vary due to improvement of living conditions. For instance, in regions where rural families’ living conditions are improving, the younger elderly residents expect emotional rather than financial support due to their better economic and health status, whereas the older elderly expect all forms of support due to their disadvantaged living conditions (Huang, 2012; Zhan, 2004). The literature referred to above indicates that intergenerational support may still be expected by the young and middle-aged as they contemplate becoming future care-receivers. However, the status and contents of the expected supply and demand may vary among individuals. The first purpose of our study is to explore categories of expectations for intergenerational support that depend on variation in family resources and family relations (Nylund et al., 2007).

2.3. Sustainable livelihoods: How individuals acquire family resources and expect intergenerational support

Sustainable livelihoods, including natural, physical, social, financial, and human capitals, are important in quantifying the impacts of family resources on individual expectations (Liu & Zong, 2018). A family’s sustainable livelihood represents the accumulation of family resources across generations and may define patterns of intergenerational exchange among the young and middle-aged and their children (Lin & Pei, 2016). The impacts of sustainable livelihoods on individuals’ expected intergenerational supports are not consistent across rural China. An increase in natural and physical capital will continue to supply resources for intergenerational exchange when the young and middle-aged grow old (Liu & Zong, 2018). But in developed regions, enhancement of natural and physical capital also allows rural elders to become more independent of their children and to expect less intergenerational support (Liu, 2017). With respect to social capital, rural residents in urbanized regions are more likely to get non-agricultural jobs, in which they develop larger social networks outside their villages (Lin & Pei, 2016). These social networks reflect the enhanced social capital that attracts rural residents to migrate to cities and decreases their loyalty to elders in intergenerational exchange (Mao et al., 2020; Wu & Li, 2014). But how the young and middle-aged assess the likelihood that they will receive intergenerational support and their own needs for intergenerational support when they will become aged is unknown. With respect to financial capital, future intergenerational exchange may differ from the past since the young and middle-aged may have more financial resources and decrease their dependence on children when they become aged (Qin et al., 2020). However, the traditional attitude of “yang er fang lao” (parents raise offspring in order to ensure support when they are old) still exists across rural regions (Zavoretti, 2006), and it is unclear whether the new generations will change their expectations for intergenerational exchange. Considering human capital, the current young and middle-aged have fewer children and may expect less support from these children in the future (Guo et al., 2012). Meanwhile, the quality of human resources among rural families, for example, in education level, has improved giving rural parents more confidence that they will have their children’s financial support in the future (Pei & Cong, 2020). This inconsistency indicates that the detailed impact of changing human capital is still unclear. Considering the current rural circumstance where families’ sustainable livelihoods are changing rapidly, the second purpose of our study is to explore how the young and middle-aged, who will be the majority of the future aged population in China, assess their likelihood of receiving intergenerational support and their own needs for intergenerational support based on the status of families’ sustainable livelihoods.

2.4. The sandwich effect: A moderating role on the relationship between family resources and individual expectations

In assessing future needs for intergenerational support, it is necessary to consider the particular role of those cohorts who are simultaneously parents of children and adult children of elderly parents (Guo, 2014). This sandwich effect, namely the young and middle-aged sharing resources with elderly parents and children simultaneously, is inevitable since the young and middle-aged will always be care-givers as the “middle generation” in rural China (Warmenhoven et al., 2018). Considering the current low fertility and that Chinese parents have child-oriented values and pay more attention to children than elderly parents, this sandwich effect may no longer be a stable structure for intergenerational support but may act as a moderator on the relationship between families’ sustainable livelihoods and individual expected intergenerational support (Lin & Pei, 2016; Zhang et al., 2019).

If the middle generation perceives a decrease of well-being due to the burden of supplying intergenerational support, they may reduce their sharing of resources with parents (Djundeva et al., 2019). In eastern China where residents benefit from the developed economy and have greater ambitions for their children’s future, the middle generation with more resources reduce support for their parents while investing more in raising children (Zhu, 2016). Today’s young and middle-aged adhere less to “yang er fang lao” due to their independence and the decline in family size (Falkingham et al., 2020). The sandwich effect may then contribute to reducing expectations that the young and middle-aged will acquire support from their children in the future. When the middle generation considers the possibility that their children will support them, the nature of any moderating effect on the relationship between sustainable livelihoods and individual expected supply of intergenerational support is unclear.

Some elderly parents live better if they simultaneously obtain aging support from both public services and adult children, who may believe that they will also need similar support in the future if they want to age more comfortably (Fingerman et al., 2011). Even in families where adult children share limited resources with elderly parents, under the sandwich effect filial piety still obtains and may lead the younger generation to believe that they will need support from their children in the future when they are old (Cong & Silverstein, 2011). The sandwich effect may cause the young and middle-aged with more family resources to believe that they will need aging support from their children in the future. When the young and middle-aged consider their future need for support, it is not clear what the sandwich effect on the relationship between sustainable livelihoods and individuals’ expected demands for intergenerational support will be. The third purpose of our study is to address the potential moderating role of the sandwich effect on the relationship between sustainable livelihoods and individual expected supply of and demand for intergenerational support.

3. Methods

3.1. Data collection

The data used in this study come from a self-report survey (“2018 Hubei Survey of Family Development and Governance in Urbanization,” supported by Family Development Department of the National Health Commission of China). At the end of 2017, the fraction of the population over 60 years old in Hubei was 18.8%, which was higher than the average for the nation (17.3%). Counties were selected as sampling sites since the rural population is distributed across counties instead of cities (Liu & Xu, 2016). To make sure that the final sample represented the general aging situation in Hubei, Huangmei County and Jiayu County were randomly selected as the sampling sites; these represent counties with a population over one million and below one million, respectively.

In the first step, the town is the primary sampling unit. It has been suggested that the sampling ratio of the primary sampling unit in the site with larger population should be higher than that in the site with smaller population (Scott & Wild, 2001). Based on the population distribution, we applied 2:1 as the sampling principle; accordingly, 12 of 16 towns (75%) in Huangmei and three of eight towns (37.5%) in Jiayu were selected. Huangmei is a typical agricultural region of mainly rural households, and twelve towns were randomly selected there (Awuah et al., 2017). In Jiayu, where urbanized communities and agricultural villages coexist, three towns were selected, including one that relies mainly on agriculture while the other two rely mainly on non-agricultural industry. The purpose of selecting more urbanized towns in Jiayu is to focus on the urbanized respondents, since the focus of the survey was family issues in urbanized rural China. In the second step, due to the homogeneity among agricultural villages in rural China, one community was randomly selected from each town in Huangmei as the secondary sampling unit (Wu et al., 2018). In Jiayu, we applied 1:4 as the second-step sampling principle: two communities were randomly selected in the agricultural town due to the more scattered distribution of households there; three communities were randomly selected in each of the two non-agricultural towns to boost the number of urbanized individuals, since those with non-agricultural jobs were difficult to find. In the third step, 15 households were randomly selected from each selected community in both counties. At least two individuals from each household were invited to be the ultimate sampling units to answer the questionnaire. Ethical approval of the survey was provided by the Family Development Department of the Health Commission of Hubei Province, Shaanxi Normal University and Xi’an Jiaotong University. All individual and household responses are anonymous.

The aim of this survey is to collect nested data, including the family and the individual, in order to study the cross-level effects from family resources to individual perceptions and behaviors (Awuah et al., 2017). At least two individuals from each household were invited to answer the questionnaires. All respondents from the same family were asked who was responsible for family decisions. This identified individual answered both a questionnaire concerning family resources and one that addressed individual perceptions. Other respondents from the same family only answered the individual questionnaire. Some families had only one person who agreed to participate, and this individual was invited to answer both the family and individual questionnaires. The final dataset includes 1,032 individuals from 702 households, but only part of this total sample was utilized. First, since natural capital is the key contributor to rural livelihoods, only families who still rely on agriculture were retained. Second, individuals from age 18 to age 59 were included as representing the typical young and middle-aged population. Third, samples without children were deleted to ensure that respondents considered intergenerational support based on having intergenerational interactions. The final data used in this study included 296 households and 436 individuals.

3.2. Measures

3.2.1. Expected intergenerational support

Expected intergenerational support is measured in the individual questionnaire. There is no available scale to measure individual expectation of intergenerational support. In a previous study on rural residents of China, the expectations of family resources involved both the expected supply of and expected demand for resources (Luo & Zhan, 2012). We take that study as a reference, and use the expected supply of intergenerational support (ESIS) and the expected demand for intergenerational support (EDIS) to measure individual expectations of intergenerational support. Detailed measurements of ESIS and EDIS are presented in Table 1.

Table 1. Properties of the sample (N = 436 individuals).

Variables		%/Mean	SD(Min/Max)
Expected supply of intergenerational support (ESIS)
Worry about financial support	No	42.7%	–
	Yes	57.3%	–
Worry about instrumental support	No	56.9%	–
	Yes	43.1%	–
Worry about emotional support	No	58.9%	–
	Yes	41.1%	–
Expected demand for intergenerational support (EDIS)
Dependent on financial support	No	73.8%	–
	Yes	26.2%	–
Dependent on instrumental support	No	74.3%	–
	Yes	25.7%	–
Dependent on emotional support	No	58.3%	–
	Yes	41.7%	–
Age		38.252	8.448(18/59)
Education duration (years)		8.968	2.850 (0/16)
Gender: Female		54.8%	–
Male		45.2%	–
Marriage status: no marriage experience		3.9%	–
with marriage experience		96.1%	–

Data: 2018 Hubei survey on family development and governance in urbanization.

For ESIS, the extent of uncertainty in terms of worry about the future can measure individual expectation of resource supplies (Flinkfeldt, 2020). However, worry about intergenerational relationships is usually defined in terms of psychological status rather than expectation (Xu et al., 2017). Since the future supply of intergenerational support is uncertain, and the expected supply of resources reflects individuals’ concerns about whether they will have sufficient resources, we decided to use absence (“no”) or presence (“yes”) of worry about each of the three items in Table 1 as the measure of ESIS (Fingerman et al., 2011; He & Ye, 2014). Since the expected demand for resources reflects how individuals evaluate their future dependence on resources, questions about future dependence are included (Chou, 2011). For EDIS in today’s China, elderly parents are “mainly” rather than “fully” dependent on adult children (Su et al., 2017). Hence three questions related to individual needs in Table 1 were used to measure EDIS. All six of these individual variables are binary. There are no missing data in ESIS. In EDIS, the percentage of missing data for the first question, the second question, and the third question are 0.38%, 0.48%, and 0.38%, respectively. Missing data were recorded as “0,” namely “no,” effectively assuming that all those who did not answer questions of EDIS would have had negative answers.

3.2.2. Individual information

Details about individual information used as control variables are presented in Table 1. Both age and years of education are measured as continuous variables in years. Ages range from 18 to 59, averaging 38. Since the age of those who raise children and support elders simultaneously is about 40, these samples represent the “sandwich effect” population (Cong & Silverstein, 2014). Education duration ranges from 0 to 16 and the average is nine years (graduated from middle school), which is consistent with what is known about rural China (Luo & Zhan, 2012). Gender is measured as binary variable (0=male; 1=female). Marital status in the survey is recorded by seven options, including unmarried, cohabitating, engaged, married, remarried, divorced, or widowed. The definition of marriage today includes not only registered marriage but also stable relationship with a potential partner (Esteve et al., 2020). Considering the existing marriage norm and the strong connection between marriage and intergenerational relationship in rural China, this study does not use the new definition of marriage to define marital status. First, the new definition of marriage is not generally accepted in rural China due to the strong Confucian culture, in which marriage only refers to the registered relationship (Obendiek, 2016). Thus, rural individuals of the sandwich generation who experience intergenerational relationships are usually officially married (Qin et al., 2020). In addition, members of the sandwich generation who are divorced or widowed are usually middle-aged, and their attitudes to intergenerational relationship are influenced more by marriage norms, such as expecting married couples to support elders, than by the experience of being single (Xu et al., 2017). Second, those who are not registered as married but live together as partners in rural China are a small minority, and this relationship is unlikely to be accepted by the whole family (Obendiek, 2016). Within the norm of filial piety, officially registered couples or couples who had a public marriage ceremony are accepted family members in intergenerational connections, and experience strong pressure to cooperate in supplying aging support as the sandwich generation (Falkingham et al., 2020). To control the impact of marriage norms in our discussion of expected intergenerational support, we divide marital status into two groups based on whether the individual is officially married or not. Therefore, the first three options are combined as the group with no marriage experience (code = 0, four percent) and others are combined as the group with marriage experience (code = 1, 96%). Since all respondents in this study have children, the distribution of this definition is consistent with the situation in rural China, where individuals are unlikely to have children if they have no marriage experience (Sun, 2013). There are no missing data for age, education, or gender. 0.58% of cases did not answer the question on marriage status. Since all respondents had non-zero answers to the later question “How many children do you and your spouse have?”, missing data here were recorded as 1 due to the low rate of birth outside marriage in rural China.

3.2.3. Sustainable livelihoods and the sandwich effect

Sustainable livelihoods and the sandwich effect are assessed in the family questionnaire. Five indexes define sustainable livelihood: natural capital (X₁), physical capital (X₂), social capital (X₃), financial capital (X₄), and human capital (X₅). In ecologically fragile regions, where families are sensitive to change in the environment, natural capital is more likely to be related to land, soil, and water supply (Pandey et al., 2017). In rural central China, the value of natural resources among families is more likely to refer to agricultural land and related profits, which supply food for survival and money for living expenses (Liu & Xu, 2016). Thus natural capital here was measured by (1) annual agricultural income (X₁₁) and (2) acreage of farmland/water (X₁₂). Both items reflect the family’s natural resources related to aging support. The family’s rents received from farmland (X₁₃) were included as another contributor to natural capital that can help to support the elderly. Physical capital—the family’s immovable property, machinery, and animals—was measured by (1) the area of the house (X₂₁), (2) rents from the house (X₂₂), and (3) family-owned facilities such as a tractor, harvester machine, motorcycle, car, and cattle (X₂₃). For social capital, compared to urban families who rely on social networks in their communities and workplaces, rural families in China mainly acquire social support from their communities and relatives (Li et al., 2012). In this study, three measures of social capital are included: (1) the number of the family’s reliable friends (X₃₁); (2) the number of the family’s reliable neighbors in the community (X₃₂); (3) the number of the family’s reliable relatives (X₃₃). These three measures of social capital reflect social resources from the society, the community and the family, respectively. Financial capital is measured by (1) the percentage of the family’s annual income that is banked (X₄₁); (2) the total amount of the family’s income that is banked (X₄₂), (3) the family’s money available for investment (X₄₃), and (4) the family’s financial arrears (debts) (X₄₄). Both X₄₁ and X₄₂ reflect the family’s financial assets that could be shared across generations in intergenerational support (Ding et al., 2018). X₄₃ reflects the family’s additional financial capacity, which could be used to improve the well-being of all family members (Li et al., 2012). Since current rural families usually have debt, which may significantly decrease the sharing of family resources across generations, X₄₄ is introduced as an additional financial index, and its value is negative in the calculation (Liu & Xu, 2016). Human capital refers to laborers in the family who are able to contribute to the family’s support. Here we use (1) the number of agricultural laborers within the family (X₅₁), (2) the number of laborers aged from 16 to 65 who are legal laborers (X₅₂), and (3) the number of migrant workers within the family (X₅₃), to measure human resources that are related to intergenerational support. For X₅₁ and X₅₃, some elders who are older than the legal labor age may be considered as part of human capital due to their participation in agriculture and migration (Wu et al., 2018). However, the number of migrating elders aged 65+ has significantly decreased in recent decades (Liu et al., 2020). The number of these elders involved in agricultural production is also decreasing due to the prevalence of agricultural automation (Huang et al., 2016). X₅₁ and X₅₃ then represent mainstream workers in the family’s labor force.

The values for each capital are aggregated to form a total score (Ding et al., 2018). The first step is standardization, where the original value of each item is transformed to take a value from 0 to 1 according to X’_ij =(X_ij- X_ijmin)/(X_ijmax - X_ijmin) (Pandey et al., 2017). In Table 2, the X_i are the final values of the five capitals, which are calculated based on the simple average of the X’_ij values (e.g., X₁=X₁₁ *0.33+ X₁₂ *0.33+ X₁₃ *0.33) (Liu & Xu, 2016). In our regression analysis, scores for the five capitals are combined, and the sum of the scores is our comprehensive index of sustainable livelihood. The sandwich effect is also measured at the family level, and the question in the family questionnaire is “Are there parents who are over 60 years old in your family?” Since all target samples have children, this question addresses the sandwich effect, where the young and the middle-aged raise children and support elderly parents simultaneously (Phillips & Feng, 2015). Table 2 reports the final scores for the five capitals of sustainable livelihoods, the total score for sustainable livelihood and the distribution of the sandwich effect. Detailed calculations of the scores of the five capitals are shown in Appendix A. The percentages of missing data from X₁₁ to X₅₃ are 4.38%, 4.10%, 88.71%, 0%, 0%, 0%, 1.83%, 0%, 97.27%, 1.83%, 1.46%, 3.47%, 2.92%, 6.20%, 0.92% and 2.19%, respectively. Since variables related to families’ sustainable livelihoods are numerical, the missing data for X_ij were recorded as “0.” There are two especially high rates of missing data, namely X₁₃ (88.71%) and X₂₃ (97.27%), referring to the rentals from lands and houses. Since rural residents in China have few chances to obtain such income, it is reasonable to record the missing data for X₁₃ and X₂₃ as “0” (Wu et al., 2016).

Table 2. Sustainable livelihood and the sandwich effect (N = 296 households).

Sustainable livelihood	Mean/%	SD	Min/Max
Natural capital (X1)	0.034	0.047	0/0.343
Physical capital (X2)	0.197	0.086	0.007/0.495
Social capital (X3)	0.086	0.079	0/0.612
Financial capital (X4)	0.042	0.076	−0.195/0.410
Human capital (X5)	0.277	0.141	0/0.825
Sandwich effect
Taking care of elders: Yes	73%	–	–
No	27%	–	–

Data: 2018 Hubei survey on family development and governance in urbanization.

X’_ij= (X_ij-X_ijmin)/(X_ijmax -X_ijmin).

3.3. LCA and MRMM methodology

Latent class analysis (LCA) is used here to identify categories of ESIS and EDIS. LCA classifies samples into categories and estimates the probability of each category; the conditional probability of each index within each category shows properties of the categories (Muthén & Muthén, 2017). The first step is to assume there is one category and estimate the statistical parameters. Then two categories are assumed and the statistical parameters are estimated again. This process continues until an optimal classification is found (Clark et al., 2015), according to the values of AIC (Akaike information criterion), BIC (Bayesian information criterion) and ABIC (sample-size-adjusted Bayesian information criterion) (Lee et al., 2014). If the values of AIC, BIC, and ABIC in the model with two categories are significantly lower than those in the model with one category, then the two-category model is regarded as better. This process continues until the values of AIC, BIC, and ABIC in the last model are not significantly lower than those in the previous model; then the previous model is regarded as the optimal classification (Nylund et al., 2007). An additional statistical measure is entropy, which ranges from 0 to 1. The closer entropy is to 1, the more reliable is the partition into latent classes (Henry & Muthén, 2010).

We use multilevel regression mixture modeling (MRMM) to assess potential causal connections between the identified category and covariates (Henry & Muthén, 2010). The identified categories of ESIS and EDIS at the individual level are included as the dependent variables; sustainable livelihoods and the sandwich effect at the family level are included as covariates; individual information is included as control variables. Scores for the five capitals of sustainable livelihoods in MRMM are combined as the total score for sustainable livelihood to give a comprehensive index representing how family resources from livelihoods influence individual expectation. To confirm the moderating effect of the sandwich effect, we apply the step regression strategy in MRMM. First, control variables, sustainable livelihoods and sandwich effect are introduced into MRMM to find the net effect. Then all variables mentioned above, as well as an interaction term that is the product of the sustainable livelihood and the sandwich effect, are introduced to assess the moderating effect of the sandwich effect. We use M-plus for computations of LCA and MRMM.

4. Results

The process of identifying the expected supply of intergenerational support (ESIS) is shown in Table 3. The statistical parameters from one to three classes are significantly decreasing, while the statistical parameters from three to four classes are slightly increasing, which violates the basic rule for accepting the optimal classification model (Nylund et al., 2007). Entropy in the model of four classes is extremely low (0.506). These results indicate that the model with three classes is optimal for ESIS. In identifying categories of expected demand for intergenerational support (EDIS) in Table 3, the statistical measures from one class to two classes decrease significantly, but the measures from two classes to three classes increase slightly. Entropy in the model of three classes is extremely low (0.586). As a result, the model with two classes is regarded as optimal for EDIS.

Table 3. LCA results.

Model	One-class model	Two-class model	Three-class model	Four-class model
	Number of ESIS categories
AIC	1787.532	1765.501	1748.033	1751.868
BIC	1799.765	1794.044	1792.887	1813.033
ABIC	1790.244	1771.830	1757.979	1765.431
Entropy	--	0.842	0.942	0.506
	Number of EDIS categories
AIC	1596.358	1418.761	1426.747	--
BIC	1608.591	1447.304	1471.601	--
ABIC	1599.070	1425.090	1436.693	--
Entropy	--	0.917	0.586	--

Data: 2018 Hubei survey on family development and governance in urbanization.

4.1. Categories of expected intergenerational support

Figure 1 shows the categories of ESIS that were identified, including the categorical proportion of each class (CLP in Figure 1) and the conditional probabilities of detailed indexes (indexes related to financial support, instrumental support, and emotional support) within each identified class (COP in Figure 1) (Nylund et al., 2007). In the group pictured on the left of Figure 1, the highest conditional probability is emotional support, which indicates that the young and middle-aged are more likely to worry that their children will be unable to supply them with emotional support in the future. This reflects 40 years of rural transition in which families have become urbanized and parents are more likely to worry about a lack of emotional support than about financial or instrumental support from adult children (Giles et al., 2010). As a result, we name the group on the left of Figure 1 “urbanized concern.” The proportion of this category among total individual samples is 24.5% (Muthén & Muthén, 2017).

Figure 1. Identified categories of ESIS (N = 436 individuals).

Data: 2018 Hubei Survey on family development and governance in urbanization.

In the middle group of Figure 1, the young and middle-aged worry that children will be unable to supply all the support they will need. Compared to the majority of concerns about the supply of intergenerational support in rural China, this group reflects the typical high-risk expectation related to aging (Zhu, 2016). Thus we call this group “aging risk concern,” and the proportion in this category is 16.5% (the lowest of the total individual samples). This is consistent with a previous finding that only a small fraction of the rural population perceives high risks associated with aging (Zhang et al., 2019). In the group on the right of Figure 1, the conditional probability of financial support is the highest and the conditional probability of emotional support is nearly zero, which are typical features of traditional rural intergenerational relations: the greatest shortage of intergenerational support is finance, which is also the most frequent shortage of resources to support the elderly among rural families (Song et al., 2012). Hence we call this group “traditional concern,” and its proportion is 59.0% (the largest among the total individual samples).

Figure 2 shows the identified categories of EDIS. In the group on the left of Figure 2, conditional probabilities of financial support, instrumental support, and emotional support are below 0.3. In the group on the right of Figure 2, conditional probabilities of financial support, instrumental support, and emotional support are above 0.6. Thus the groups on the left and right of Figure 2 are called “low expected demand” and “high expected demand,” respectively. The proportion of low expectation is 76%, which indicates reduced prevalence of psychological dependence on children among the young and middle-aged (Cong & Silverstein, 2014). Within the category of low expectation, the highest conditional probability of emotional support indicates that the young and middle-aged prefer to receive emotional support from their children, who will be irreplaceable in rural China (Guo, 2014). The demand for instrumental support in that category has the lowest conditional probability, which reflects that the young and middle-aged are less likely to rely directly on their children in the future. The priority in the category of high expectation is instrumental support. Most individual samples belong to the category of low expectation, which indicates that when young and middle-aged become old, they expect to decrease their dependence on children.

Figure 2. Identified categories of EDIS (N = 436 individuals).

Data: 2018 Hubei survey on family development and governance in urbanization.

4.2. Correlations between sustainable livelihoods and expected intergenerational support

The correlations between sustainable livelihoods and ESIS are shown in Table 4, where we see that natural, physical, and human capital at the family level are significantly associated with ESIS at the individual level. With more natural capital, the young and middle-aged are more likely to have aging-risk concerns. With more physical capital, the young and middle-aged are more likely to have urbanized concern and worry that they may not receive emotional support from their children in the future. With more human capital, the young and middle-aged are more likely to have traditional concerns.

Table 4. Correlations of ESIS with sustainable livelihoods (N = 436 individuals).

	Class 1 (Urbanized concern)	Class 2 (Aging risk concern)	Class 3(Traditional concern)
Sustainable livelihood	Mean (S.E.)	Mean (S.E.)	Mean (S.E.)	Chi^2 (p-value)
Natural capital	0.029 (0.003)	0.045 (0.006)	0.033 (0.003)	6.057 (0.048*)
Physical capital	0.220 (0.008)	0.174 (0.008)	0.194 (0.006)	19.597 (0.000***)
Social capital	0.095 (0.008)	0.087 (0.008)	0.078 (0.006)	3.347 (0.188)
Financial capital	0.048 (0.006)	0.032 (0.007)	0.041 (0.006)	3.083 (0.214)
Human capital	0.285 (0.012)	0.227 (0.012)	0.298 (0.012)	16.468 (0.000***)

* p < .05, ** p < .01, *** p < .001.

Data: 2018 Hubei Survey on family development and governance in urbanization.

Table 5 presents the correlations between sustainable livelihoods and EDIS. Only physical and financial capital are significantly different among the categories of EDIS. With more physical capital in the family, the young and middle-aged are more likely to have higher expectations, especially for emotional support. With more financial capital in the family, the young and middle-aged are also more likely to have higher expectations.

Table 5. Correlations of EDIS with sustainable livelihoods (N = 436 individuals).

	Class 1 (High expectation)	Class 2 (Low expectation)
Sustainable livelihood	Mean(S.E.)	Mean(S.E.)	T-test(P-value)
Natural capital	0.032 (0.004)	0.035 (0.003)	0.220 (0.639)
Physical capital	0.216 (0.009)	0.194 (0.005)	4.745 (0.029*)
Social capital	0.092 (0.009)	0.085 (0.004)	0.382 (0.536)
Financial capital	0.055 (0.007)	0.038 (0.004)	4.025 (0.045*)
Human capital	0.277 (0.012)	0.276 (0.008)	0.000 (0.986)

* p < .05, ** p < .01, *** p < .001.

Data: 2018 Hubei survey on family development and governance in urbanization.

4.3. Effects of sustainable livelihoods and the sandwich effect

We use a multilevel regression mixture model (MRMM), in which categories of individual’s ESIS and EDIS are dependent variables. In our multilevel models, level 1 is for the individual and level 2 is for the family. Sustainable livelihoods are independent variables, and individual information provides control variables. The sandwich effect, as well as the Interaction term of sustainable livelihood and sandwich effect, are introduced into the model to detect any moderating properties. The results of MRMM for ESIS are presented in Table 6. Compared to standard errors (SEs) that are usually reported in multilevel regression models, cluster robust standard errors (CR-SEs) have been proposed to account for heteroscedasticity (Jackson, 2020). However, the basic assumption that must hold in reporting CR-SE is that there is no higher level of the hierarchy to ensure that observations between clusters are not related (McNeish et al., 2017). In the present study, the sampling strategy is a typical three-level investigation. All families in this study come from 20 communities. Since individual attitudes concerning aging support are influenced by the local conditions, such as the community, individuals may still be relevant between families since they belong to the same community (Canda, 2013). It is not appropriate to report CR-SEs since the individual-family dataset here come from the community-family-individual dataset. Thus, we report SEs instead of CR-SEs. Models 1–1 and 1–2 compare urbanized concern and traditional concern. The results for model 1–2 indicate that the main effect of sustainable livelihood is significant and the sandwich effect has a significant moderating effect on the relationship between sustainable livelihood and individual ESIS. Those who have more sustainable livelihoods and experience the sandwich effect will have urbanized concern, which entails more worries that in the future their children will be unable to supply emotional support. Models 2–1 and 2–2 compare aging risk concern with traditional concern. The net effect of sustainable livelihood in model 2–1 indicates that those with more family resources are more likely to have traditional concerns. The significant main effect of sustainable livelihood in model 2–2 (OR = 0.566) indicates that the young and middle-aged from families with more resources are more likely to have traditional concern, namely worries about being unable to receive financial support from children in the future, when they are not experiencing the sandwich effect. However, the moderating effect indicates that those from rural families with more livelihoods will have aging risk concern if they are experiencing the sandwich effect (OR = 1.404).

Table 6. Main effects and moderating effect on ESIS.

	Model 1-1			Model 1-2			Model 2-1			Model 2-2
	Class 1(urbanized concern)			Class 1(urbanized concern)			Class 2(aging risk concern)			Class 2(aging risk concern)
Reference: Class 3 (Traditional concern)	O.R	S.E.	P	O.R	S.E.	P	O.R	S.E.	P	O.R	S.E.	P
Between level
Sustainable livelihood	1.473	0.273	0.083	1.149	0.003	0.000***	0.673	0.163	0.046*	0.566	0.001	0.000***
Sandwich effect	2.304	0.517	0.012*	2.277	0.010	0.000***	1.103	0.180	0.565	1.093	0.002	0.000***
Sustainable livelihood * Sandwich effect	–	–	–	1.617	0.004	0.000***	–	–	–	1.404	0.001	0.000***
Within level
Age	1.005	0.025	0.828	1.006	0.000	0.000***	1.012	0.008	0.132	1.012	0.000	0.000***
Education (years)	1.284	0.102	0.006**	1.285	0.002	0.000***	0.914	0.031	0.005**	0.915	0.000	0.000***
Gender: male (ref: female)	1.138	0.282	0.604	1.163	0.009	0.000***	0.858	0.046	0.002**	0.862	0.001	0.000***
Marriage status: with marriage experience (ref: no marriage experience)	2.130	2.090	0.589	2.201	0.008	0.000***	1.272	1.575	0.863	1.310	0.003	0.000***
AIC	1728.448			1729.920			1728.448			1729.920
BIC	1822.233			1831.861			1822.233			1831.861
ABIC	1749.243			1752.524			1749.243			1752.524
df	23			25			23			25
Entropy	0.762			0.760			0.762			0.760
N (individuals)	436

* p < .05, ** p < .01, *** p < .001.

Data: 2018 Hubei survey on family development and governance in urbanization.

The modeling strategy of MRMM for EDIS is the same as that for ESIS and the results are presented in Table 7. In model 3–1, the net effect of sustainable livelihood is significant: the young and middle-aged from families with more livelihoods are more likely to demand intergenerational support. In model 3–2, the sandwich effect has a significant moderating impact on the relationship between sustainable livelihood and individual EDIS. The main effect of sustainable livelihood in model 3–2 shows that those from families with more sustainable livelihoods are more likely to have high expectations. This likelihood is increased when those from families with more sustainable livelihoods experience the sandwich effect.

Table 7. Main effects and moderating effect on EDIS.

	Model 3-1			Model 3-2
	Class 1 (low expectation)
Reference: Class 2 (high expectation)	O.R	S.E.	P	O.R	S.E.	P
Between level
Sustainable livelihood	0.442	0.012	0.000***	0.553	0.001	0.000***
Sandwich effect	0.322	0.144	0.000***	0.375	0.000	0.000***
Sustainable livelihood * Sandwich effect	–	–	–	0.766	0.002	0.000***
Within level
Age	0.984	0.046	0.723	0.984	0.000	0.000***
Education (years)	0.504	0.126	0.000***	0.553	0.001	0.000***
Gender: male (ref: female)	0.308	0.069	0.000***	0.352	0.001	0.000***
Marriage status: with marriage experience (ref: no marriage experience)	0.562	0.813	0.590	0.650	0.004	0.000***
AIC	1745.540			1747.400
BIC	1798.549			1804.487
ABIC	1757.294			1760.058
df	13			14
Entropy	0.504			0.494
N (individuals)	436

* p < .05, ** p < .01, *** p < .001.

Data: 2018 Hubei survey on family development and governance in urbanization.

5. Discussion

The above results have several theoretical implications. First, intergenerational exchange theory explains how elderly parents build up reciprocity with adult children to maintain the stable connections that establish a basis for family support (Cong & Silverstein, 2011). However, the young and middle-aged in the recent decades do not rely fully on family relationships, unlike the older generations who were born before the 1980s and rely strongly on family members (Cong & Silverstein, 2014; Fu et al., 2020; Lin & Pei, 2016). In the present study, the categories of ESIS and EDIS indicate that the tradition of intergenerational exchange is no longer stable; the general system of intergenerational support may still exist in the future, but the content and the pattern of support are expected to be different due to changes in family resources and resource-sharing mechanisms. Second, intergenerational altruism theory suggests that adult children should share family resources with their elderly parents, since the latter are to some extent needy (Fingerman et al., 2011). Our results indicate that this family norm is weakening. In terms of filial piety, the young and middle-aged believe that even though they will in the future be the needy elderly, their children may not be fully committed to supporting them because of changes in family structure and in attitudes towards family obligations (Qin et al., 2020). Third, intergenerational solidarity theory suggests that resource-sharing enables a strong network of intergenerational support in which family members cooperate with each other to support elders (Bengtson & Roberts, 1991). Our results confirm that family resources from sustainable livelihoods are still likely to be indispensable for the new generation to have such support in the future. However, the availability of family resources may lead new generations to have levels of intergenerational support that differ from those in the past, in terms of both supply and demand.

Our LCA analysis produced three categories of the expected supply of intergenerational support (ESIS), which we named “traditional concern,” “urbanized concern,” and “aging risk concern.” Traditional concern is the most frequent and reflects that most of young and middle-aged in rural China are worried that their children may be unable to provide financial support for them in the future. It indicates that finance is still the most reliable support from children in rural China, where the social security system is underdeveloped (Zhang et al., 2019). Urbanized concern indicates that in the future some children may be unable to supply emotional support to their elderly parents. This reflects the current urbanization and migration, in which adult children are more likely to live separately and independently (He & Ye, 2014). Aging risk concern comes from the young and middle-aged having little confidence that they will receive any support from their children. Even though the fraction in this category is the lowest, it reveals that some respondents have no confidence in their children in the coming aging society, where filial piety may decrease and intergenerational independence may be more prevalent (Janus & Koslowski, 2020).

The expected demand for intergenerational support (EDIS) is categorized as high expectation or low expectation. There is a significantly low prevalence of needing instrumental support directly from children in the group with low expectation. The probable explanation is that there are significant improvements in commercial and community aging services, which may be considered as partly substituting for children’s instrumental support since rural China will become more developed in the future (Guo et al., 2017; Zhang et al., 2019). High expectation reflects a preference for financial and instrumental support from children in the future. However, the lower proportion with high expectation suggests that today’s young and middle-aged expect be less dependent on children in the future. The existence of high expectation represents the likely outcome of aging in rural China: Young and middle-aged who are psychologically dependent on children are a minority in their generation. They still believe that support will come directly from adult children, but they will be confronted with challenges: Their children are less likely to supply care directly due to family members becoming more independent and being more likely to live separately in the future (Zhu & Walker, 2018).

The correlation analysis indicates that those who worry about a potential lack of intergenerational support in the future come from families with more natural capital. Since families with more natural capital are likely to have multiple generations in residence, the young and middle-aged from these families are more likely to have stronger traditional norms, be more dependent on intergenerational relations, and to worry about the lack of intergenerational support in the future when they become elderly (Guo, 2014). Those who worry about not having emotional support from their children come from families with more physical capital. The likely reason is that the young and middle-aged whose families benefit from migration and have more physical capital are reducing their emotional connections to their parents and may worry about emotional connection with their children in the future (Giles et al., 2010; Liu & Zong, 2018). Respondents who worry about receiving financial support from their children come from rural families with more human capital. Families with more adult children in rural China usually have conflicts about aging support, especially financial and instrumental support, due to greater intergenerational independence and separate living arrangements (Warmenhoven et al., 2018). Thus the young and middle-aged whose families have more human capital, are likely to be concerned about a lack of financial and instrumental support when they become old.

With more physical capital in the family, the young and middle-aged are more likely to have high expectation; that is, a strong need for children’s’ support, especially emotional support. The likely reason is that the young and middle-aged from rural families with more physical capital are usually migrant workers who supply less support to elderly parents and expect more emotional support from their children in the future (Liu & Zong, 2018). With more financial capital in the family, the young and middle-aged are also more likely to have high expectations. There are two possible explanations for this. Compared to those whose families are in worse financial conditions, the young and middle-aged from families with better financial conditions are more likely to experience independence and loneliness and may expect to have more emotional connections with children in the future (Qin et al., 2020). Also, increased financial resources among rural families is usually related to rural-urban migration, as a result of which the young and middle-aged reduce their emotional support of elderly parents (Zhan, 2004). When the young and middle-aged have to deal with challenges by themselves, they may recall the advantages of family support and expect to need this support when they are elders in the future (Luo & Zhan, 2012).

The significant net effects of sustainable livelihoods in MRMM models suggest that the young and middle-aged expect that the future intergenerational support will depend on the amount of family resources. In the comparison between aging risk concern and traditional concern, the young and middle-aged with more livelihoods are more likely to worry about the future lack of financial support; they fit the category of traditional concern. This may reflect the current intergenerational relationship, in which the middle generation have more financial resources but share more of these resources with their children than with their elder parents (Zhang et al., 2019). The middle generation may worry that they will be confronted with this situation when they themselves become elder parents in the future. In the comparison between high and low expected demand for intergenerational support, those from families with more livelihoods are more likely to have high demand. This is consistent with rural reality: Under a welfare shortage in villages, improving the family’s livelihood will help to contribute resources for support from adult children to elder parents (Xu & Chow, 2011). Another explanation may be related to the mechanism of rural-urban migration: if a rural family’s economic condition is improved, their children will not have to migrate and will stay close to elderly parents, who will expect intergenerational support more frequently due to the convenient location of their children (Liu, 2017), while those from families with poorer livelihoods are more likely to have low demand. Considering that conflicts about supporting elders usually exist among members of poorer rural families, the probable explanation is that the disadvantaged families’ livelihoods make the young and middle-aged have little expectation from their children for aging support, since they have already experienced these conflicts (Warmenhoven et al., 2018).

The sandwich effect plays a moderating role on the relationship between sustainable livelihoods and ESIS. In the comparison between urbanized concern and traditional concern, if the young and middle-aged share resources with their children and elderly parents simultaneously, an increase in sustainable livelihoods is associated with more concern about the shortage of emotional support when they become old. The reason may be that some young and middle-aged who live with the sandwich effect invest less in their parents emotionally when they work harder and have limited time, and may have no confidence that they will have frequent emotional connection with their own children when they are old (Zhan, 2004). In the comparison between aging risk concern and traditional concern, if the young and middle-aged share resources with their children and their elderly parents simultaneously, an increase in sustainable livelihoods is associated with greater concern about all patterns of support, namely aging risk concern. However, those who are not subject to the sandwich effect will only be concerned about financial support, namely traditional concern, if there is an increase in sustainable livelihoods. A possible reason for this is that members of the current sandwich generation experience the difficulties of sharing resources up and down, and are less likely to believe that their children will do the same in the future (Canda, 2013). Furthermore, Chinese parents have child-oriented values and, due to the decline in fertility, pay more attention to their children than to their elderly parents (Zhang et al., 2019). The results above indicate that in the current rural China, where fertility is declining, urbanization is increasing, and migration is frequent, the young and middle-aged from better-off families are uncertain that they will receive support from their children in the future if they share resources up and down as the middle generation.

The sandwich effect plays moderating roles on the relationship between sustainable livelihoods and EDIS. When the young and middle-aged experience an increase in sustainable livelihoods, both those with and without the sandwich effect are likely to need support from adult children in the future. The moderating result indicates that this likelihood is increased among the former group. A possible reason is that due to the welfare shortage in rural areas, the young and middle-aged perceive a future need for support from children, especially instrumental and emotional support (Pei & Cong, 2020). If they have more sustainable livelihoods and simultaneously support their children and elderly parents as caregivers, they may have less time to take care of parents directly and hence they will emphasize the necessity of direct care from their children in the future (Zhang et al., 2019). The sandwich effect here indicates that improvement in family resources doesn’t necessarily lead to a decrease in psychological dependence on children. Instead, those with more resources have high expectation of being supported directly by their children. If they share resources with children and parents simultaneously, the extent of this high expectation will increase.

6. Conclusion

Our study shows that even though family resources in rural China are improving, there will be conflicting challenges in the future when the current young and middle-aged will become a large aged population but will need their children’s support. However, they expect that their children will probably be unable to fully support them then. The young and middle-aged in rural China will not only become a huge elderly population, but will also have uncertain intergenerational support due to changes in family structure and family relations. Since rural families may benefit both from China’s continuous development and the tradition of intergenerational support, it is necessary to combine intergenerational support and welfare for the elderly as a comprehensive system to respond to the future challenges facing China as its population ages.

7. Limitations

There are several limitations in this study to explain expected intergenerational support among future care-receivers in the coming aging China. The sampling strategy, including the selection of sampling sites and respondents, is limited and may not explain the national situation since social and economic development across the country is rapid and diverse. The survey does not include the quality of intergenerational relationships, which reflects the connections between the current young and middle-aged and their children and may significantly influence the former in their consideration of intergenerational support as future care-receivers. More targeted nested data should be collected as a multilevel dataset and cluster robust standard error should be reported, which is a more convincing way to describe heteroscedasticity in multilevel studies. Further studies including representative samples, covariates that are accurately related to intergenerational relationship, and methodology that improves the robustness of multilevel models are desirable.

Ethics statement

To make sure this study is implemented according to the rule of Ethical Approval, the survey and investigation in this study has been approved by the Family Development Department of the Health Commission of Hubei Province, Shaanxi Normal University and Xi’an Jiaotong University. All the individual and household samples in this study are anonymous.

Additional information

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article (1) National Natural Science Foundation of China [grant number 72074177] (2) Humanities and Social Sciences Research Project of the Ministry of Education (22YJC840034) (3) Shaanxi Social Science Fund (2022F008)

 

Appendix 1. Measurements of sustainable livelihood (N=296 households).

Sustainable livelihood (family level)	Mean	SD	Min/Max	Weighting strategy
Natural capital (X1)	0.034	0.047	0/0.343	X1=X’11*0.33+X’12*0.33+X’13*0.33
Family’s income (yuan) from agriculture (X11)	13787.16	18065.48	0/200000
Family’s acreage (mu) of land/forest/ pond (X12)	14.680	59.255	0/900
Family’s rental (yuan) from land/forest/ pond (X13)	170.055	750.132	0/8000
Physical capital (X2)	0.197	0.086	0.007/0.495	X2=X’21*0.33+X’22*0.33+X’23*0.33
Family’s acreage (m^2) of house (X21)	146.000	70.860	20/500
Total score of possess (0=no,1=yes) of Harvester machine, tractor, car/motorbike and cattle (X22)	1.323	0.734	0/4
Family’s rental (yuan) from house (X23)	576.992	8039.843	0/150000
Social capital (X3)	0.086	0.079	0/0.612	X3=X’31*0.33+X’32*0.33+X’33*0.33
Family’s number of reliable friends (X31)	7.447	9.426	0/95
Family’s number of reliable neighbors in the community (X32)	8.348	12.766	0/108
Family’s number of reliable relatives (X33)	18.110	17.021	0/150
Financial capital (X4)	0.042	0.076	-0.195/0.410	X4=X’41*0.25+X’42*0.25+X’43*0.25+ X’44*(-1)*0.25
The percentage of the family’s annual income that is banked (X41)	14.539	20.394	0/99
The total amount of the family’s income that is banked (yuan) (X42)	27333.97	169843.6	0/3110000
Family’s available money (yuan) for investment (X43)	36562.58	58056.53	0/500000
Family’s debt money (yuan) (X44)	36908.58	95124.28	0/800000
Human capital (X5)	0.277	0.141	0/0.825	X5=X’51*0.33+X’52*0.33+X’53*0.33
Number of agricultural laborers within the family (X51)	1.984	0.908	1/7
Number of laborers from age 16 to age 65 within the family (X52)	2.986	1.157	1/6
Number of migrant workers within the family (X53)	1.352	1.076	0/5
The total score of sustainable livelihoods	0.459	0.339	0/1.368	X1 + X2+ X3+ X4+ X5

Data: 2018 Hubei survey on family development and governance in urbanization.

Note: X’_ij= (X_ij-X_ijmin) / (X_ijmax -X_ijmin)