Ageing and familial support: a three-generation portrait from urban China

Abstract

Research on ageing in China has been preoccupied with the unsolved question of whether traditional filial piety is eroded or sustained by societal modernisation. This article engages with the ongoing debate on modernisation and family change, but seeks to go beyond the prevailing dichotomous conclusion. Rather than focusing on one intergenerational relationship between ageing parents and their adult children – a common formula in the existing literature – this article draws upon 120 life-history interviews, involving both genders and three generations in three cities, and examines how old-age support practices have shifted across three generations, as well as between sons and daughters across time. The findings indicate that while there has been a decline in everyday financial and instrumental support by adult children for their parents across all three generations, crisis-induced intergenerational solidarity has remained intact. As the market economy has matured, differences in ageing experience have widened between working-class and affluent families. The article also reveals that care for bilateral parents has characterised the behaviour of the three urban generations. The complex shifts and continuities are the outcome of a combination of state policies, evolving filial norms, gender and demographic forces, as well as reflecting the broader structural consequences of China's shift to a market economy. By systematically comparing old-age practices by generation and gender in both Mao and post-Mao eras, the article makes a major empirical contribution to the study of ageing in urban China. From a theoretical perspective, it contributes to the global debate on modernisation and ageing by emphasising the uneven processes in which social change interacts with family life within a single country, when viewed through generational and gender prisms. In so doing, it highlights the ways in which old-age support trajectories are firmly grounded in local history and cultural, economic and demographic forces.

Introduction

Modernisation theory argues that industrialisation, rural–urban migration and the growth of institutions such as the welfare state have led to declining familial support for older people (Goode, 1963; Cowgill, 1974). However, research in developing countries suggests social change through modernisation has not necessarily resulted in declining family ties (Bawdekar and Ladusingh, 2012). Despite this, the modernisation and ‘abandonment of older people’ framework continues to dominate gerontology scholarship in the developing world (Aboderin, 2004). Research on ageing in China is no exception and this theoretical preoccupation is reflected in the unsolved question of whether traditional filial piety is eroded or sustained by societal modernisation. The question has added urgency for China as it enters a period of ‘super ageing’ (Joseph and Phillips, 1999), with the most recent census finding that the number of people aged 60 and above has reached 260 million or 18.7 per cent of the population (National Statistical Bureau, 2021).

This article engages with existing debate on modernisation and family change, but seeks to go beyond the current dichotomous conclusion. Drawing upon 120 life-history interviews (involving both genders and three generations in three cities) from a larger research project on Chinese family life, it investigates the shifting old-age familial support in recent decades. Rather than focusing on one intergenerational relationship between ageing parents and their adult children – a common formula in existing research – this article answers the call of Aboderin (2004) and examines the family practices of old-age support in three successive generations. This comparison enables an understanding of past practices rather than simply assuming a ‘traditional’ baseline prescribed by Confucian discourse of filial piety. Further, since Chinese patrilineal and patrilocal customs mean that filial piety is a gendered concept, the analysis also reveals how sons' and daughters' support practices have evolved.

Modernisation framework and the scholarship on China

According to modernisation and ageing theory, older people will experience a decline in support from their families as society modernises. This is due to changes in normative values: industrialisation and urbanisation cause the traditional extended family to be replaced by the nuclear family and so orientation towards the conjugal family strengthens while willingness to provide old-age support weakens (Aboderin, 2004). The reasoning, supposedly having its origins in Euro-American social reality, has been challenged for its failure to reflect historical or contemporary western experience (O'Rand, 1990) and for its assumption that societal development follows a uniform path. Nevertheless, the modernisation framework has dominated the study of ageing in China (Ikels, 2006). In discussing existing empirical scholarship on China, this article explains how the modernisation formula fails to capture continuities and shifts in old-age support.

The Confucian discourse of filial piety – defining how older people should be respected and taken care of by younger generations – has been widely considered the normative guidance for old-age familial support in China (Chow, 1995). It demands ‘unconditional’ and all forms of support to parents (Lin, 1992). Essential aspects are ‘support, obedience and continuing the family line’ (Ikels, 2006: 389). Filial piety is also highly gendered, emphasising the importance of father–son relationships since it is the son who continues the family line in a patriarchal society (Hsu, 1971). Daughters should serve their natal families before marriage, but subsequently join their husband's descent line, providing his parents with care and bearing grandchildren (Wolf, 1972).

Given the ageing population in China and absence of comprehensive social welfare provision, the commitment of the young to fulfil normative obligations and care for their parents in old age has been the focus of existing literature. Survey-based studies suggest that while submission to parental authority is no longer an essential component of filial piety (Cheung and Chan, 2006), there is still widespread and continuing endorsement of filial support (Whyte, 1997; Hu and Scott, 2016). But while adjustments are being made in intergenerational living arrangements and financial transfer, existing empirical findings are mixed and patchy.

Focusing on the Maoist era and the 1980s, Davis-Friedmann (1991) found multigenerational living the norm and that state policies had stabilised family life, as well as guaranteeing a subsistence level for older people. However, Davis-Friedmann also suggested there was a pronounced urban–rural divide, with the urban older people having access to far better benefits and a state pension. Qualitative fieldwork in rural villages in the 1990s found early household division resulting from the marriage and departure of adult children to be a key change weakening filial piety (Yan, 2003; Miller, 2004; Zhang, 2004). These studies indicated a decline in parental authority and painted a dismal picture of old-age life, inferring a ‘breakdown of the traditional mechanism of elder support’ (Yan, 2003: 171). In particular, Yan's study led to his argument for the rising ‘individualisation’ of Chinese families, represented by the ‘disembodiment of the individual’ from ‘the patriarchal order’ (Yan, 2011: 208), so ‘the younger generation in rural China is no longer bound by the social contracts of Confucian culture’ (Qiu et al., 2022: 1428).

In post-Mao urban China, nuclear family living is common (Ma et al., 2011). The fact that different generations live in separate households does not necessarily mean that they are functionally separate: cooking, care and transfers may take place within networked families that are geographically separate (Unger, 1993 on Tianjin; Whyte, 2003 on Baoding; Shen, 2013 on Shanghai). Nevertheless, comparing qualitative data from Guangzhou in 1987 and 1998, Ikels (2004a: 350) found that housing reform and increased job mobility could jeopardise the principle of kin proximity and lead to ‘delay in the activation of support by some children until parental need levels are perceived as high’. The one-child policy (introduced in 1979) also poses challenges: Chinese cities are ageing rapidly and yet strict family planning has produced a 2–4–8 family structure in which the urban couple (generally adults of the one-child generation) must support four parents and eight grandparents (Ikels, 2006).

A number of recent studies of intergenerational support in China have been underpinned by intergenerational exchange theory as an alternative framework to the modernisation thesis (see Lee and Xiao, 1998; Cong and Silverstein, 2011; Luo and Zhan, 2012; Guo et al., 2013). This theory stresses that intergenerational transfers are reciprocal and gravitate towards a balance that maximises a family's overall wellbeing (Cong and Silverstein, 2011). One prominent variation of exchange theory is the conceptual framework of intergenerational family solidarity, a multi-dimensional construct incorporating associational, affectual, consensual, functional and normative solidarity, as well as family structure (Bengston and Roberts in Falkingham et al., 2020). Studies guided by these alternative frameworks have generally found a continuous flow of intergenerational exchange and intergenerational functional solidarity in Chinese families, providing contrasting evidence to the scholarship concerned with the erosion of filial piety. Nevertheless, the emphasis of these studies is upon micro-level intergenerational interactions and therefore reference and links to wider social-structural changes across generations in China is limited.

There are several gaps in the existing literature. First, modernisation theory has epistemological limitations due to a lack of empirical evidence on family life in pre-industrial societies, its theorisation being grounded in a structural functionalist interpretation of an a priori conception of social life in pre-industrial societies (Aboderin, 2004). Similarly, existing scholarship on China predominantly draws upon data comprising one set of intergenerational relations. As there are no comparable data from previous generations, the implicit yardstick existing research employs is the repertoire of behaviour prescribed in Confucian discourse of filial piety. There is an epistemological risk in comparing contemporary practices with the traditional discourse of filial piety which adds up to a form of ‘generational rhetoric’ (Luscher, 2011: 192) and this partly explains why existing findings on China are patchy and often contradictory. A generation-sequence design (Aboderin, 2004) is better positioned to compare the life experiences and perspectives of successive generations vis-à-vis familial support.

Second, there is a general assumption underpinning modernisation theory that old-age support is mainly driven by obligatory norms. This assumption leads ultimately to an explanation of the decline in old-age support through a change in values (Burgess, 1960). An alternative perspective draws attention to the material constraints on people and their ability to provide support for older people (De Lehr, 1992). Whilst not engaging with modernisation theory, Finch (1989) highlights demography and argues family support is governed to a considerable extent by prevailing demographic conditions as well as the economic situation. In China, ageing literature tends to consider the cultural value of filial piety to be the sole foundation for old-age support. Although Ikels (2006) warned that filial value may not be sufficient to counterbalance the economic forces unleashed by the post-Mao reforms and demographers tried to model the impact of number of children on support for the older people (Zimmer and Kwong, 2003), how filial norms in combination with changing material and demographic circumstances shape the shifts and continuities in support practices has seldom been examined in the Chinese context.

This paper uses generation and gender as conceptual tools to account for the variation in old-age support practices in urban China. Generation is an important social structural factor because it embodies a particular set of formative and socio-historical experiences that define life chances and material circumstances (Alwin and McCammon, 2003). Generation matters particularly in the context of China since it witnessed various state-engineered political, economic and demographic mass movements in the second half of the 20th century, each of which impacted disproportionately on a particular generation, with sequential consequences for individual livelihood and family practices. The generational approach broadens existing theoretical perspectives by capturing the social background of a particular time which can include normative value, demographics as well as economic circumstances.

Gender is a key structural factor of family life: maintenance of family welfare and intergenerational relations is premised upon an implicit gender contract (Ginn and Arber, 2000). In China, gendered filial piety differentiated sons' and daughters' responsibilities to their parents, and whilst women's liberation has been a core component of the Chinese Communist Party's rhetoric since 1949, care work remains a woman's job. An analysis of generation and gender should therefore capture how values and the economic and demographic circumstances in which men and women are situated at a particular time shape their old-age support practice as well as to explain shifts and continuities across three successive generations.

This study is the first comprehensive qualitative research to examine empirically the changing process of old-age support across three successive generations of men and women in multiple sites in China. The two specific research questions this article addresses are:

1. (1) What are the changes and continuities of old-age support practices across three generations of men and women in urban China?

2. (2) How do wider economic, cultural and demographic circumstances shape old-age support practices in urban China?

The article therefore seeks to contribute to the global debate on modernisation and ageing. Rather than oscillating between whether or not modernisation has led to the decline in family support, it calls for an examination of the uneven processes in which social change interacts with family life within a single country structured by generation and gender. It thereby highlights the ways in which old-age support trajectories are firmly grounded in local history and culture, economy and demography.

Methodology

A life-history approach helps to establish a connection between the past and present and weaves individual lives and wider social processes together. In family research, the life-history approach is particularly useful where an individual's biography overlaps with other family members' trajectories, offering an insight into how family members interact with each other over a lifecourse. This article draws upon life-history interviews from a larger project which adopted a multi-sited and comparative research design to examine how family life (parent–child relations, conjugal relations, sexual relations and intergenerational relations) shifted across three generations of men and women in multiple sites in China.

Between October 2016 and October 2018, the author led a team of local research assistants in order to conduct life-history interviews in multiple locations in China. Purposive non-random sampling techniques were utilised to identify and recruit interviewees. This sampling approach allowed the project team to target particular groups of the population. Four key criteria were utilised in this sampling process: location (urban; rural; semi-rural/semi-urban, i.e. suburban counties and villages), socio-economic background, age and gender. Due to limited space, this article focuses on family data collected from the 120 interviews undertaken only in urban sites (see Table 1).

Table 1. Characteristics of interviewees from urban three-generational families and two-generational families

Note: 1. When referring to interviewees born between the 1930s and the 1950s, the occupation indicates the main occupation before their retirement.

The urban sites included in this project are Tianjin, Xi'an and Guangzhou. In each city, the project sought to recruit interviewees from different occupational backgrounds:¹ namely managerial/professional occupations (civil servants, managers, doctors, teachers, etc.) and working-class occupations (factory workers, service workers, taxi drivers, etc.). It was originally hoped that it would be possible to interview men and women from three successive generations in each family unit, with the oldest generation targeting those born in the 1930s. However, this was not always possible: securing the agreement of one family member to be interviewed did not guarantee that other family members would also be willing to be interviewed. As a result, interviewee samples in each location contained a mixture of three-generational and two-generational families (see Table 1). While the team was not always able to secure an equal number of male and female interviewees from each family unit, there was an approximate gender balance in each generation across all families. With the aid of social media advertisements, informal networks and snowballing, 40 urban families were eventually recruited for inclusion in the project. One-to-one private interviews were conducted with various family members in each family unit. Particular care was taken to assure each interviewee that their responses would be treated in total confidence and not be revealed to other family members. The project strictly complied with the Research Ethics Framework of the funding body and the governance procedures of the Research Ethics Committee at SOAS University of London, where the author is based, including ongoing informed consent (participants could withdraw at any stage of the research), confidentiality and anonymity.

Each interviewee was asked to recall his or her childhood first and then encouraged to take the lead. If not covered during the natural course of the ensuing conversation, specific questions regarding relationships with family members were asked. All interviews were recorded and fully transcribed. The older generation (G1) was born between the early 1930s and early 1950s, with the majority in the 1940s. The middle generation (G2) was born in the 1960s and the early 1970s. The younger generation (G3) was born between the mid-1980s and 2000s with the majority in the 1990s. The parents of G1 (majority born in the 1910s) are coded G0 here. A Grounded Theory approach (Glaser and Strauss, 1967) was used to identify themes and concepts which emerged from the individual participants' life narratives. Similar concepts were then combined into key themes for a higher level of understanding followed by a further in-depth analysis of intergenerational relations for each household. Finally, key themes from individuals and families were analysed according to the project's key analytical foci: generation, gender and location.

Intergenerational relations, narrated by interviewees, covered a broad range of experiences.² The findings here focus predominantly on one aspect of international support,³ that is, old-age support practices (including financial, emotional, instrumental support and personal care). Two inter-linked extended families, covering a range of scenarios (e.g. residence pattern, siblings' formation, educational/occupational difference), have been selected to describe the generational and gendered changes (Figure 1). The discussion of these selected cases is supplemented by other family data to capture common features across three generations. All names and any identifying characteristics have been altered to protect the anonymity of the interviewees.

Figure 1. Family structure of Ms Chen and Ms Li (those in bold were interviewed).

The representativeness of an in-depth qualitative study of Chinese families is inevitably limited by its non-probability-based sampling strategy. However, the multi-sited and generational sequence design generated comparable and multigenerational family data in different sites to identify common themes and practices across generations among urban families, filling a significant gap in the existing scholarship of Chinese family studies. Further, as indicated in the Discussion below, this study found various generation-specific support practices confirmed by quantitative studies in other regions, strengthening the overall transferability of the findings to the understanding of urban family life in the context of rapid social transformations in China.

Findings: old-age support

G1 vis-à-vis G0

G0 had five or six surviving children on average. The father of most G1s worked in the socialist work unit before retiring with a state pension. The mothers of most G1s were housewives with no pension. Regardless of whether both parents had a pension or not, the practice of G1s giving a proportion of their wage to their parents every month, throughout their life, was widely reported. Before marriage, when a G1 started paid employment, they handed all, or most, of their income to their parents. A male G1 explained:

At that time, everyone behaved like this. No matter how much you earned, you had to hand it over to your parents (G0). I haven't heard of anyone in my generation who didn't give their monthly wage to their parents.

After marriage, G1's everyday financial, emotional and instrumental support was closely tied to G0's living arrangements. In the 1950s housing was nationalised, with privately owned rental properties transferred to local government, and state enterprises built public housing for their employees (Wang, 1995). This change meant that for G1, accommodation provided at a very low rent by the husband's workplace became the most common source of marital housing (see also Liu, 2007). However, housing pressures in urban China forced some G1s (ranging from 20 to 30% of interviewees, depending on location) to move in with the husband's family upon marriage – sometimes by building a mezzanine floor within an existing room.

Parents who lived alone received a financial transfer from their children every month. Ms Chen (G1) recalled:

after marriage, we started by giving 10 yuan per month, which we increased to 20, then 30 and eventually 50 yuan in the 1980s and 1990s. My mother was very economical and saved this money. We also bought her clothes as well as the things she needed for day-to-day use. After she died, she left us 8,000 yuan, which she had saved during all these years.

Proximity was the dominant factor in determining which children would take on the main responsibility of visiting and providing instrumental support. Ms Chen's youngest sister recalled:

I took on the most responsibility because I lived closest to my parents. I was only 5 minutes’ walk away. I went to visit them every day, helping with washing clothes, cooking and carrying coal for the stove. If I missed a day, my mother would scold me. If I didn't help with the chores, she would also criticise me.

Her mother and older siblings also played a critical role in persuading her to marry her husband because his family lived close by.

In the case of G0 who lived with a son's family, the co-residing G1 married couple were expected to provide everyday emotional and instrumental support. Some G1 recalled a ritualistic practice for showing respect, imposed by parents, which required the younger generation to address the older generation before and on their return from work. Most daughters-in-law (G1) expressed dissatisfaction with their mothers-in-law (G0), although they did – to varying degrees – assist with housework. A frequent source of dispute was the way in which the younger generation of women (G1) looked to their husbands to assist with housework whilst the older generation (G0) took the view it was solely the daughter-in-law's duty.

G0s' non-resident married children were required to send a proportion of their monthly earnings to their parents. Ms Li's parents lived with her younger brother's family, she recalled:

every month we had to give some money to my mother – 10 yuan per month [when her salary was 38 yuan]. She called this the ‘breastmilk fee’. All of us were required to do this. My father was still working at that time.

Non-resident children were expected to visit their parents during important festivals, and many G1s visited at least once a month. Those who were particularly attached to their parents visited more frequently; Ms Li recalled: ‘I couldn't put my parents out of my mind. I was working close by [15 minutes’ walk] … So I formed a habit of visiting them every day’.

When G0s needed care because of illness or old age, various arrangements were available. Regardless of whether the couple lived alone or with married children, the healthy spouse was always the first choice. Because of women's longer average life expectancy and tendency to be the younger party, wives were more likely to look after their husband than the reverse. Ms Chen's father regularly drank and died of bronchitis in 1987. She recalled:

the doctors suspected he had a tumour, but decided against operating on him because he was too old. He was in bed at home for several months before he died. It was mainly my mother who looked after him. We also visited, sometimes changing his bedsheets.

If the spouse was not in good health, children undertook a care rotation system. While still living with the youngest son's family in the early 1990s, Ms Li's father developed dementia and became paralysed. Ms Li's mother was not in good health, but fully mobile. The mother therefore cared for her husband during the day whilst the children's families provided night-time care – one family per night, helping with changing, washing bedsheets/clothes and toilet trips – for nearly two years until he died. Although he was retired, Ms Li's father's work unit covered his medical expenses as was the norm of this time.

If widowed G0s had no mobility issues they continued to live on their own or opted to spend part of their time with one particular child's family (usually the one they were emotionally closest to). If the widowed parent became in need of constant care, rotation among children's families (both sons and daughters) was the most common arrangement across all sites. Equal shared responsibility was a priority, although children's emotional attachment to their parents affected the quality of care. For example, after her husband died, Ms Chen's mother was in reasonable health until the late 1990s when she developed heart problems and became bedbound until her death three years later. During this period, her four children rotated her care for 24 hours each at a time. Since the mother had not worked for a work unit, her children needed to share her medical costs. When it was Ms Chen's turn, she provided care during the day, while her two children (G2) took turns to cover the night shift. The same pattern was followed by the families of her two sisters. Ms Chen's youngest sister elaborated:

I constantly changed my mother's clothes and sheet as there were no [adult] nappies available at that time. We put a layer of thick cloth underneath the bedsheet. Normally every day there was a whole bowl of washing to do.

However, when it was her brother's turn to help, his family sent meals but did not assist with washing/changing sheets, and no one stayed overnight. According to Ms Chen, her brother's family felt her mother had failed to prioritise bringing up her son's children over those of her daughters, as was the social norm at that time.

From the early 2000s, two key developments changed old-age care arrangements. First, housing allocation by the work unit officially ended in 1998 and a commercial market for urban housing rapidly developed. There was a general consensus that if parents decided to pass on their accommodation to a particular child, that adult child's family would be mainly responsible for parents' old-age care. Ms Chen's mother-in-law liked her youngest son best and after selling her city accommodation, she bought a suburban two-bedroom flat for his family, with whom she lived until she died. The other adult children only visited Ms Chen's mother-in-law during important festivals. If all children (including daughters) looked after their parents on a rotation basis, the parents' property was expected to be divided equally among the sons after their deaths.

Second, paid carers became part of families' rotation plans. This was widely reported by Guangzhou families due to the city's more advanced development and commercialisation of care (Li and Hu, 2009). Ms Zhao's father cared for his wife, who had dementia, for a number of years until her death in 2000. During his widowerhood, he lived with his youngest son's family and used his pension to cover the cost of hiring a carer to undertake daily chores. In his final months before death, a paid carer assisted him full time and his children's families took turns to visit to ensure he saw one member of the family every day. Since Ms Zhao's three siblings worked, it was necessary to hire a paid carer – and the fact that many workers had been laid off at that time meant there was increased availability.

G0s therefore received regular financial transfer from adult children – a practice which continued after their marriage. The provision of emotional support and instrumental support was closely linked to G1s' parents' living arrangement, with the co-residing sibling or the sibling who lived closest to the parents taking on the main responsibility of providing instrumental support. Adult children who felt particularly close to their parents also visited more frequently. In the case of those requiring constant care, a healthy spouse was always the first choice of carer, and if this was not possible care responsibilities were equally shared among all adult children on rotation. The commercialisation of G0s' housing as well as the availability of paid care services were new factors shaping the old-age care arrangements from 2000 onwards.

G2 vis-à-vis G1

G1s born in the 1930s had four children on average, while G1s born in the 1940s were influenced by the mid-1960s local initiative of ‘later marriage, birth spacing and fewer children’ so had two children on average. The practice of handing monthly pay over to parents before marriage still existed in the 1980s and early 1990s, but had become much less common by the late 1990s. Ms Li's older daughter handed over her entire monthly wage to her parents when she started to work in 1988, but when her sister graduated from university in the late 1990s she did not hand over any money. Variations in family practice between older and younger siblings of the same family were closely related to broader structural changes taking place as China transitioned from a planned economy to a market economy, especially the formal ending of the state job allocation system in the mid-1990s (Solinger, 2006).

After marriage, adult G2s were no longer required to give parents a share of their monthly wage. Instead, it became the norm to give money to their parents during important festivals. The amount of money given to parents was dependent on the economic capacity of the adult children. As economic reform and market competition accelerated, income gaps among urban residents widened, including between siblings. At the Spring Festival, Ms Li's elder daughter (a laid-off worker who became an office clerk) gave each of her parents 200 yuan while her sister (a well-paid civil servant) gave them 1,000 yuan each (and a further 200 yuan to Ms Li's husband whenever she visited as she had a particularly strong relationship with him). Ms Li had a close relationship with her elder daughter and felt sorry for her:

since my older daughter's economic circumstances were not very good, when she bought something for me, I would normally give her the money. Whenever possible, I would also secretly give her some money.

G2s experienced a fundamental shift in the way urban families obtained marital housing. Those who started work and married in the 1980s generally benefited from housing allocated by the husband's work unit. However, for those married in the 1990s, public housing allocation became increasingly precarious because at this time state enterprises were forced to restructure and lay off staff (Solinger, 2006). At least initially, the younger cohort within G2 generally moved in with the husband's family upon marriage. Then, in the late 1990s, urban housing become a tradeable commodity, banks could offer mortgages, and many G2s moved to a flat for their own nuclear family. A key reason for moving out was quarrels between different generations of women over everyday chores. At the time of the fieldwork, most G1s lived alone, but were close to at least one adult child's family (where ‘close’ ranged from occupying different floors of the same building or belonging to the same/neighbouring residential community to a car journey of less than 30 minutes).

With the rapid expansion in mobile technology, phone/video calls have become a major source of emotional support to parents who live alone. Whilst children who had a good relationship with their parents were more likely to call them, frequency of calls alone does not capture the complex dynamics of intergenerational relationships. Ms Chen's daughter lived close (10 minutes by car) to Ms Chen and so according to the proximity principle that had operated in G1's generation, Ms Chen's daughter was expected to visit more often than her brother who lived further away. However, Ms Chen's daughter did not feel close to her mother who had always favoured her brother and whose apartment was expected to pass to her brother. As a result, the daughter made a phone call to her mother every day for half an hour, but only visited her once every three or four months. When Ms Chen became ill with a fever, the daughter continued to phone her, but did not take her to see a doctor as adult children who felt emotionally close to their parents would have done.

Unlike G1s who provided instrumental support when visiting their parents, the provision of instrumental support by G2s depended on the age and/or health of G1s. For parents aged over 80, most G2s would undertake tasks such as cooking and washing dishes on their visits. For parents in their seventies and in good health, their children did not always help with the chores. Ms Li and her husband quarrelled over whether their daughters should help with the washing up when they visited. Ms Li felt that her children had worked hard during the week days and so should relax when they came to visit their parents, but her husband felt that they ought to follow customary practice and assist with the chores.

With regard to personal care, less well-off families continued the model adopted by G1: that is, spousal care as the first choice followed by children's care on a rotation basis. Ms Chen's husband suffered a stroke in 1995, but was mobile until 2002 when he fell and suffered brain damage. From then on, he was mentally disabled and bedbound for two years until his death in 2004. Sixty per cent of Mr Chen's medical costs were reimbursed by the urban resident's medical insurance scheme, but the remaining 40 per cent had to be paid out of their savings. Everyday care fell upon Ms Chen. She recalled:

I often cried in the toilet. Sometimes I even wanted to die as it was both physically and mentally tough to look after him. I moved him five times a day in order to prevent bedsores. But as he was mentally disabled, he hit and scolded me while I did this.

Her son lived close by at that time and visited every day for 10–20 minutes to see if there was anything he could do to help (e.g. during these visits he cut his father's hair and shaved his beard). Ms Chen's daughter, who lived far away at the time, visited once a week and assisted with washing dirty clothes. Ms Chen's son had been close to his mother since childhood and spoke emotionally of this experience:

I was really grateful to my mother. She shouldered too much. At that time, my mother said to me, ‘I can't help you with the school pickups of your daughter. But leave your father with me, I'll care for him’. She lightened the burden of the younger generation. I am really touched by what she did.

This intense care experience came almost straight after Ms Chen had undertaken a three-year rotation of care for her own mother.

When Ms Chen became widowed in 2004, both her son and daughter agreed to give her 100 yuan per month. After one year Ms Chen declined this payment because she felt ‘their conditions were not very good, both had been laid off from work, and were doing odd jobs’. When asked about her plan for old-age care, she commented ‘care from children by rotation. But I really hope that day will never come’. She felt quite lonely living on her own; her daughter who lived only 10 minutes away by car called her every day but only visited every three or four months. Her son did not live close, but because he was emotionally close, he visited her every two or three weeks. When she entered her seventies, she became nervous of living on her own, worrying about what would happen if she died without anyone knowing. In recent years, Ms Chen had occasionally felt heart pain in the evenings and kept close to her medication in case of emergency. She mentioned this to her children, and her son suggested taking turns with his sister to stay with her during the night but Ms Chen turned this down, feeling that such an arrangement was premature and wanting to delay a rotation plan until she absolutely needed it.

In contrast to Ms Chen's family in which both generations held working-class occupations, families with better resources adopted new kinds of arrangement. For example, Ms Li's second daughter experienced upward mobility and had become a highly paid senior civil servant. Because she had more money, this daughter was able to design more innovative old-age arrangements for her parents and her in-laws:

I know what most older parents need is our companionship. But I cannot satisfy this. My workplace is already like a battlefield and requires all my energy. I tried my best to earn money to buy accommodation for them so that they could live near my older sister's flat.

In effect, she attempted to provide financial support to replace the instrumental and emotional support demanded of her. In doing so, she also transferred care responsibility to her sibling, whom she felt had a less-important career than herself.

A common struggle that G1s had gone through while looking after their own parents was the care arrangement during G0s' hospitalisation: family members needed to cover various tasks such as accompanying the patient for medical examinations and toilet trips, preparing and dispatching meals three times a day and providing company during the night. In order to relieve family members of the care burden, in major Chinese cities a new service has been introduced which obviates the need for family members to stay with a patient during hospitalisation. However, coverage varies considerably between locations. As Tianjin was the first city nationwide to pioneer this new form of care arrangement it therefore has a much wider coverage, in terms of both illness and the number of hospitals (more than 20) included in the scheme. Ms Qian (G1) had first-hand experience of the new service:

every day when it was meal time, hospital staff always delivered the food to me on time. If I had any issues, I just pressed a button to call them. It was very convenient.

Her younger son accompanied her to the hospital and paid the hospitalisation fees on the spot (as her older son was working in Guangdong), but Ms Qian insisted on returning the money as she felt her son's life was not easy either. During the four days of Ms Qian's hospitalisation, the younger son (G2) and the older son's daughter (G3) still visited her every day, but no longer had to undertake other care tasks as the previous generation had done.

Both before and after marriage, G2s' parents no longer received regular financial support from their adult children; instead, adult children gave money to their parents during important festivals and the amount was dependent on their economic capacity. Most G1s lived alone but physically close to at least one adult child's family. The provision of instrumental support was more closely linked to the care needs of G1 than living proximity and phone/video calls had become a major channel for G2 to provide emotional support. With respect to the provision of personal care, the gap between working-class families and affluent families had widened: while the former followed a care rotation among siblings, the latter had more innovative old-age plans or used paid care.

G3 vis-à-vis G2

As a result of the one-child policy, more than 90 per cent of the urban G3 participants were an only child. Among G3s, the practice of handing over their monthly income to parents before marriage was extremely rare. Only a small fraction of those who were born in the 1980s and none of those born in the 1990s followed this custom. But whilst no regular financial transfers were made from G3s to G2s, other forms of financial support were provided by G3s. For example, the purchase of additional private insurance for parents was common among those with stable and well-paid jobs. In Xi'an, a 26-year-old nurse purchased private medical insurance in case her parents became seriously ill.

Neolocal residence has become the trend among G3s upon marriage. The social norm is that when a couple marry, the husband's family is responsible for providing the conjugal residence upon marriage (Liu, 2016a). Some G2 parents purchased the marital flat for their son well in advance, e.g. Mr Yan in Xi'an bought an additional flat in the block in which he lived when his son was still in middle school (and the price was relatively low). But G2s who were not economically well off or had not benefited from the property boom struggled to pay the full price of a flat although they tried hard to meet the acceptable social norm by paying the deposit (at least 30% of the total price). Where the wife is an only child, her family sometimes also contributes to the cost of the marital flat, although there is no obligation to do so. Ms Chen's daughter paid almost half of the 600,000 yuan needed to buy a one-bedroom flat that was a 10-minute bicycle ride from her own accommodation, for her son upon his marriage in 2016. As her son worked as an administrator with a low wage of 2,000 yuan per month, she continued to contribute 1,000 yuan per month towards his mortgage payment after marriage, leaving him to pay the remaining 700 yuan. Ms Chen's daughter continued working to make enough money to help provide financial support for her son's family.

As parents are responsible for purchasing the marital flat, the neolocal residence tends to be close to at least one set of parents. Despite living separately, G3s continued to go to one set of parents for daily meals as they were considered to be busy with work and lacked experience in cooking. Regardless of the children's occupational background, G2s commonly covered the meal costs of the stem family when eating together. In return, many married G3s continued the G2 practice of giving money or gifts to parents during festivals and/or organised paid holidays for them. Mr Tian (G2) explained:

my son and his wife are both working in white-collar jobs and paid relatively well. But I don't want any of their money, not a single penny. Every month, we cover their daily expenditure, rice, noodles, oil and salt. We eat together [though living separately with a 15-minute car drive between them]. We even pay for all their fruit. We don't use any of their money. But during the festivals, he would give us something. For example, in the Spring Festival, they gave us 2,000 yuan, and at the Mid-Autumn Festival they gave us another 2,000 yuan. They also organised a trip for all four of us plus their child to Hainan Island [a holiday resort]. They paid for the flight tickets and accommodation. We stayed there for a week.

At the point in their life trajectory when the interviews were conducted, G2s were support providers to G3s both financially and instrumentally (undertaking housework and looking after grandchildren). When asked about future old-age plans for their parents, the 1990s cohort showed no hesitation in their willingness to undertake their filial responsibility. Speaking of future old-age support for his own and his wife's parents, Ms Chen's grandson commented:

it is definitely us. My parents only have me as the only child. Her parents only have her as the only child … When they become immobile, I will definitely look after them. What else could I do? If necessary, we will move to live with them or they move to live with us.

Because their parents were older, the 1980s cohort had to think more carefully in planning for their old-age support. Economic resources played a critical role in shaping their calculations. One man (a salesperson born in 1984 and living 20 minutes by car from his parents) wanted to live much closer to his parents, but was constrained by the high price of housing. One woman (a manager, born in 1987), having witnessed carers' abuse of older people at retirement homes she had visited, was strongly opposed to the idea of putting her parents in a home with the exceptions of a few top-quality institutions. Thanks to her higher income, she managed to buy a flat for her parents in the residential compound in which she also lived. She was also exploring new residential retirement communities that boasted five-star hotel facilities in their provision of meals, daily support, medical care and leisure activities. These communities for the older people have begun to appear in all three cities in the last five years, however, they are all targeted at high-income families.

When asked about their own old-age plans, parents of the 1980s cohort acknowledged the difficulties facing their children with the following common concern:

whenever I think about this issue, I worry. Nowadays each family has only one child. If you expect them to look after you, they still have to make a living and provide for their own family. (Ms Yin, born in 1957, Tianjin)

Some mothers also pointed out that because of the absence of siblings, the one-child generation lacked the hands-on experience of caring for others, unlike G1s and G2s who had helped with domestic chores and care rotation plans in their youth:

my son is not the type of person who can wait on others. I still wait on him. I've spoilt him since his childhood so he didn't do anything except for studying. I don't expect anything from my daughter-in-law either: she doesn't know how to cook and hasn't touched her kitchen for years. (Ms Zhong, born in 1962, Xi'an)

Most parents of the 1990s cohorts were also anxious to avoid being a burden on their children, although some affluent families were more creative in planning for their old age. Mr and Mrs Guo (G2) – a school headmaster and owner of a kindergarten respectively, had an only daughter busy preparing to study abroad. When I asked Mr Guo what his plans for old age would be if his daughter were to settle abroad, he spoke of the possibility of living collectively with other relatives:

some of us who get on well could find a flat or house together – for example, my brother, my sister and my wife's sister. Four families living together. Ideally, the children of the four families would take turns to visit us: one day per adult children's family.

G2s were support providers to G3s both financially and instrumentally. In return, G3s gave money or gifts to parents during festivals and also pledged to undertake their filial responsibility in future. However, the time pressures placed upon G3 by the market economy conflict with the care responsibilities for the ageing parents. G2s' old age would diversify further dependent upon the economic status of the family and involve a mixture of familial and institutional care.

Discussion

Generation

The preceding comparative analysis points to a decline in financial and instrumental support from adult children to their parents, at the everyday level, across three generations in urban China. G0s aged in an era when family resources were limited but similar across different families. Differences in the quality of old age for G1 and G2 widened between affluent and working-class families. While G3 attempted to care for their G2 parents in old age, time constraints and their limited capability to carry out personal care tasks suggest a future of combined familial and institutional care. Are these shifts in old-age support practices a result of eroding filial piety, as predicted by the modernisation framework? The findings paint a complicated picture. When asked about family tradition, almost 90 per cent of interviewees across the three generations quoted filial piety as a major element. While such a declaration may be the by-product of an interview setting, filial piety as an ideology is deeply ingrained in everyday Chinese psyche – albeit one that translated into different support practices across the three generations.

The generational variation in old-age support practices is closely associated with the dramatic social changes that have taken place in urban China. Since 1949, multiple politically driven mass economic and demographic campaigns have taken place, including nationalisation of housing in the 1950s, the mid-1960s ‘later marriage, birth spacing and fewer births’ initiative, the one-child policy from 1979, and economic restructuring of state enterprises and privatisation of housing in the 1990s. Each of these state-led modernisation campaigns affected one particular generation disproportionately. Under the planned economy, most G1s were guaranteed a lifetime of employment and many were allocated marital housing by the husband's workplace. With the transition to a market economy, many G2s were laid off within a few years of starting work with a state employer and, in 1995, the ‘iron rice bowl’ (de facto permanent employment) was abolished. From 2000 onwards, when the real-estate market started to expand, many G2s were still able to purchase marital housing by using their own savings to cover the deposit. Not only did all graduating G3s have to find their own jobs (rather than them being allocated by the state), but also, thanks to skyrocketing housing prices, they had to rely on parental financial support, without which it was virtually impossible for them to purchase marital housing.

The old-age support practices of each generation were shaped by the prevailing social norms and economic and demographic circumstances of the time. Ms Jing (G1) in Tianjin summarised how the shift in family practices reflects broader structural adjustments in Chinese society:

in the past, although we were working, whenever my [extended] family needed me I just took a day off. If my family required my help for a longer period, I could easily take more leave. At that time, we could take up to six months’ leave without affecting anything. Some of my colleagues took six months' leave and then came back to work. After working for a while, they started taking leave again. Would this be possible now? Of course not. Nowadays if the younger generation take leave, their wage would be affected. If you keep taking leave, you would actually be fired. So we should avoid adding to the burden of younger generations. Nowadays, the pressure in the society is huge. It is not easy to earn money and raising children is so expensive. This is so different from our time. Although our wages in the past were low, it was the same for everyone. If one family could eat fish, the families in the whole lane could eat fish. But today, if another family's child has extra tuition to learn something new, we need to provide the same tuition for our grandchild in order not to be left behind. The result is that the focus of the younger generation has to be on making money.

Ms Jing points to at least four layers of contrast between the past and the present. First, the planned economy guaranteed job security for life; as a result, G1s were able to respond more easily to the needs of their extended family. G1s old-age support for G0s largely confirmed the findings of a 1994 survey in Baoding by Whyte (2003: 309) that ‘filial obligations are alive and well in urban China’. By contrast, in a market economy, where precarity is a key feature of employment, it is increasingly difficult for G2s and G3s to devote substantial amounts of time to their extended family without suffering financial consequences. This is likely to have contributed to the reduction in instrumental support provided on a daily basis by the younger generations.

Second, albeit at a much lower living standard, the planned economy generated a high degree of egalitarianism among urban citizens and families. Except for government officials, G0s' old-age experiences varied little between workers' families and those with professional careers. By contrast, although it led to improvements in material welfare, market force also accelerated the widening of wealth and income gaps among siblings as well as families. Increased resources available to G2s and G3s enabled them to explore and establish more innovative care arrangements for their parents. While filial piety prepares adult children with the will to care, greater financial resources have equipped them with a stronger ‘capacity to care’ (Baldassar and Merla, 2013). The findings of G2s' old-age support for G1 confirm Ikels' (2006) prediction that the economic forces unleashed by the post-Mao reforms would lead to inequalities among older people. However, Polanyi's (1944) claim that market forces foster individualistic tendencies that have an undermining impact on family obligations is not supported by the data here. While G2 and G3's capacity to care was affected by growing economic constraints, participants from younger generations demonstrated strong adherence to the value of filial piety, something also confirmed by Zhan's (2004) survey findings of strong will among school and college students to fulfil filial responsibilities.

Third, the value of children changed across the three generations. G1s and G2s had siblings – often many of them – and therefore were not indulged as children. Under the one-child policy, G2s were allowed only a single child, who therefore acquired a priceless status in the extended family (Liu, 2022). Draconian implementation of the one-child policy, combined with the core element of filial piety that emphasises the importance of having children, dramatically increased the value of children in Chinese society. This has tilted the power balance between generations and, paradoxically, sabotaged the other core element of filial piety – submission to parental authority. The outcome is that China has shifted from a society focused on the older people to one focused on the child. In turn, each family has made every effort to provide its only child with the best possible chance to succeed in the competitive market economy – something also documented in other urban regions (see Fong, 2004; Liu, 2007).

Fourth, closely linked to the increasing value of children, resource allocation has tilted towards the younger generation. Ms Jing felt that the older generation (G1) should avoid burdening their children (G2) so that they (G2) could focus on making money in order to support their own children (G3 and G3's children) – a sentiment widely held by G1s. Modernisation theory interprets the shift in resource allocation as a result of normative change that prioritises the conjugal family. Intergenerational justice literature based on western social norms refers to this process as the ‘transitive order’ of justice between generations: ‘that is, we “pay” the generosity of the preceding generation by giving in turn to our successors … Whatever claims older people may have are limited by this overriding transitive obligation across the chain of generations’ (Moody, 1993: 229). These explanations are, however, insufficient to account for the shift in the Chinese context.

In traditional filial piety, an inability to continue the family line was considered the most unfilial outcome of all (Ikels, 2004b). Enforced implementation of the one-child policy made it impossible for urban families to guarantee a male heir so the only child of urban families, whether a boy or girl, was the sole means of continuing the family blood line. This was not simply the responsibility of the conjugal family, but was a collective goal for the multigenerational extended family and it was widely felt among interviewees that grandparents and parents ought to make sacrifices to ensure the wellbeing and success of the only child. The combination of traditional filial ideology, which emphasises having children, and the strict implementation of the one-child policy in urban China, accentuated this transitive obligation and produced a hierarchy of priority among generations. Although priority responsibility has shifted towards the younger generation, the older generation has not been forgotten (see the Findings section). The consequence is that G2s have become the ‘sandwich generation’ (Falkingham et al., 2020), bearing a care responsibility in both directions for the entire extended family.

The three-generational comparative analysis not only points to changes in attitudes and practices, but also highlights continuities. Crisis-induced intergenerational solidarity, reflected in arrangements for hospitalisation and terminal care, remained intact in the first two sets of intergenerational relations (G1s towards G0s, and G2s towards G1s). Despite parental concern over their availability to lend assistance, it is likely that G3s will continue to honour obligations of crisis-induced solidarity, given this generation's greater dependency upon their parents and their financial indebtedness towards them. Further, with skyrocketing house prices in urban China, parents' own housing and how its disposition after death will serve as an important bargaining chip in future intergenerational negotiations. With increasing care services available in urban settings, men and women of the one-child generation are most likely to be the organisers of their parents' care, with the day-to-day work carried out by paid carers.

Emotional support remained relatively consistent across the three generations: the child who felt emotionally closest to parents was likely to have the closest contact with them. Younger generations increasingly use technology (virtual calls and social media) to provide emotional support at a distance. Within each generation, the degree of emotional attachment to parents is reflected in the quality of care. As shown in the analysis of G1 and G2 generations, there is a qualitative difference between the intimate care behaviour of children who felt close to their parents and that of children who merely felt an obligation to care for them (see also Liu, 2017 on old age support practices in rural China).

Gender

Gender is key to understanding intergenerational support as studies based on varied cultural settings show that women and men are assigned different tasks within the family (Finch, 1989). The Chinese situation is distinctive because of the patrilineal family norms inscribed in the discourse of filial piety. Traditionally, male heirs and their wives were expected to be the primary providers of support for parents in old age, while married daughters were obligated to provide support to their in-laws rather than to their own parents (Lee and Xiao, 1998). This gendered discourse is so powerful that it often disguises daughters' actual involvement in their own parents' care. Among the G1s whose grandparents settled in the cities when they were young, women of their parents' generation had been closely involved in their own parents' care. For example, Ms Qian (G1) in Tianjin recalled:

when my maternal grandmother was ill, my mother looked after her in our home [her mother had three elder brothers, all of whom were fully employed while her mother had no job]. When my grandmother realised that she was close to death, in the middle of the night she insisted upon going to her son's house: ‘I won't die in your house. I have sons, I am going to die in my son's house.’

This practice of relocation to the son's house before death was reported among urban G1s' and rural G2s' parents and confirmed by findings from other regions of China (on rural Henan, see Cao, 2019). Dying at a daughter's house deviates from the gendered filial norm and would potentially affect the reputation and public standing of the extended family. The relocation tactic therefore highlights the performative element of filial piety which interviewees referred to as ‘surface work’. However, this tactic has made daughters' involvement in parents' old-age care obscured by the surface work performed by the sons.

Among urban G1s, although married sons were far more likely than married daughters to live with their parents, all the women interviewees had been actively involved in both the care of their own parents as well as that of their in-laws. All the married children – both brothers and sisters – made the same payments⁴ to their parents regularly (see the quotes in the Findings section regarding Ms Li's and Ms Chen's mothers), a key reason being that women of this generation had a stable income from their state-allocated job. This confirmed the findings of the 1994 Baoding survey regarding the provision of monetary assistance to parents by married daughters and married sons: ‘there is not a statistically significant difference between the assistance provided by daughters versus sons’ (Whyte and Xu, 2003: 179).

There are differences between sons and daughters in how they provided emotional and instrumental support. First, the gendered notion of emotionality perceives women as being more emotionally competent and expressive (Illouz, 1997); daughters were therefore more likely to be the key players in delivering emotional support. But this does not mean that sons were emotionally distant from their parents. Because the son preference was prevalent when G1s grew up, some male G1s spoke of their intense emotional attachment to their parents, in particular to their mother since she was the one who spent most time with the children. Second, due to the biological determinist understanding of gender and the gendered association of technology, women (daughters-in-law and daughters) took charge of cooking and washing clothes, while men (sons and sons-in-law) were expected to help with heavy lifting and carrying, repairing furniture and appliances, etc. Third, despite participating in paid work, women workers were still socially considered as having a stronger family orientation than their male counterparts (Liu, 2007). Indeed, since care work has remained primarily a woman's responsibility, it has more often been women rather than their husband or brothers who have taken leave from work in order to see to the needs of bilateral extended families. Despite daughters' disproportionate contribution to parents' old-age care and their legal rights of inheritance, G1 women have in practice largely been excluded from the patrilineal inheritance custom. In the extremely rare cases where a married daughter stayed with her parents (usually because her husband was from another city and/or unable to secure housing from his work unit) and inherited her parental home, a breakdown in relations between her and her brothers often followed.

In contrast with the seeming altruism towards their own parents, women generally enacted surface work in their interactions with in-laws. Ms Li summarised her relationship with her mother-in-law:

when all is said and done, our relationship never went sour, although it was lukewarm. Honestly speaking, how could a woman treat her mother-in-law the same as her own mother? As long as there is no quarrelling and fighting, one is a good daughter-in-law.

Most G1 women were dissatisfied with their mother-in-law, but the majority chose to keep on superficially good terms whilst complaining about them to their husband in private. Despite dissatisfaction with their in-laws, most daughters-in-law complied with the filial obligation to assist them in old age. By contrast, in accordance with patrilocal tradition, there is no normative expectation for a son-in-law to offer support (although those who had a good relationship with their in-laws did assist in care rotation plans). Some G1 husbands frowned upon their wives' intense involvement with their natal families, but mostly acquiesced even if they took no part themselves. For example, when Ms Li invited her widowed mother to stay in the marital home, her husband did not speak to his mother-in-law once during her stay.

Within the urban G2 generation two shifts have taken place. First, daughters who have assumed an equal (or more) responsibility in caring for their own parents now expect their involvement to be recognised in the division of parental assets after death. The rapid increase in the value of Chinese real estate has dramatically increased the stakes. Further, the state's mass mobilisation of women into paid employment has conferred upon urban Chinese women a distinctive working identity, which, through their employment and earning capacity, has significantly increased their power within their conjugal family as well as their own natal family. This shift from the principle of equal division among brothers to the principle of equal division among children – male and female – has also been noted in Shanghai (Shen, 2020). Where they are excluded from the future division of parental property, daughters may simply undertake surface work in lending support to their own parents, as shown in the case of Ms Chen's daughter.

Second, there has been a shift in care responsibility for the husbands' parents from daughters-in-law to their husbands and his siblings. Unlike the G1, in which a daughter-in-law was still expected to undertake a certain amount of surface work to maintain a superficially harmonious relationship and provide instrumental support, these expectations have lessened in G2 and subsequent generations. The supposition that an absence of a blood relationship generates less-than-intimate relations between daughter-in-law and mother-in-law is now a widely shared consensus. This belief is so powerful that it has significantly shaped women's relations with their in-laws. The fact that encounters with their ageing in-laws were restricted to important festivals or hospital visits reflects the lukewarm nature of these relationships. In families in which relationships had broken down entirely, avoidance of any contact between both parties was often the most appropriate solution. According to Ms Chen's son, for example, his wife harboured such a deep grudge against his mother from the time when they were all living together that she wished her dead. Ms Chen's son therefore visited his mother without informing his wife. For members of this generation, each spouse took responsibility for their own parents' old-age care; the involvement of the other spouse was largely dependent upon the relationship between him/her and the relevant parents.

A new pattern has emerged among the one-child generation characterised by the apparent prioritisation of everyday ties with the wife's parents over those with the husband's parents. This trend is closely linked to the widespread belief that women get on better with their own mother than with their in-laws, as a result of which a husband tends to acquiesce in his wife's encouragement of more frequent interactions with her own parents. One example of this is the new normative trend whereby G3s' children are more frequently brought up by the maternal, rather than paternal, grandparents. This practice has been incorporated into the Tianjin local idiom as a new tradition – ‘once you have a child, it'll be the maternal grandmother who will bring it up’. All G3 men expressed willingness to look after their own parents in addition to those of their wives.

Conclusion

Against the backdrop of broader socio-cultural, demographic and economic changes in the 20th century, old-age support practices have changed considerably in urban China. Rather than being eroded, filial piety has been reinvented under a new guise. Paradoxically, strict implementation of the one-child policy, combined with the traditional filial discourse which places a high premium on having children, has dramatically increased the value of children in Chinese society and facilitated a shift from a society centred on older people to a child-centred one. In spite of the tilted power dynamics between generations, the ideology of filial support remains strong, not least because being filial is ‘to be properly Chinese’ (Miller, 2004: 52), as well as being officially endorsed in public discourse (Ikels, 2006). Nevertheless, translating the filial motivation to support into solid and intimate care practices requires material resources as well as prior emotional commitment.

As a result of increasing economic and time pressures imposed by the rigours of a market economy, everyday financial and instrumental support towards parents has declined among younger generations. Driven by the rapid development of China's market economy, differences in the quality of the ageing experience between working-class and affluent families has widened. However, crisis-induced intergenerational solidarity has remained intact among urban families. Rather than a decline of intergenerational ties, as predicted by the modernisation thesis, intergenerational interdependence has strengthened for the one-child generation because of their greater dependency on their parents. As the value of their housing has increased, urban parents have continued to exert considerable power in their negotiations with the one-child generation. As the availability of urban care services expands, the principal role of men and women of the one-child generation is increasingly likely to be that of organising their parents' care with the day-to-day work being undertaken by paid care workers.

With respect to gender differences in old-age support practices, the findings of this article reveal that despite the gendered emphasis of filial piety that places married women on the filial map of their in-laws, a married daughter's care for her own parents has, in practice, been in evidence for at least three generations. Although earlier generations still practised patrilineal inheritance, daughters' significant contribution to their parents' old-age care has gradually been recognised, at least within younger generations, through inclusion in their own parents' inheritance decisions. As a result of the draconian implementation of the one-child policy, as well as greater participation in paid work by urban women and their consequent increased power and influence in both conjugal and natal families, within three generations urban Chinese families have transitioned from a patrilineal institution to one that offers a bilateral system of support and inheritance. These complex shifts and continuities in old-age support practices of urban men and women across three generations are the outcome of a combination of evolving filial norms, gender issues and demographic forces, as well as of the broader structural consequences of China's shift from a planned to a market economy.