Who cares and who pays? Financial sources and care preferences among older adults in China Authors: Shibei Ni, Xiaorui Cheng Published: 2026 ABSTRACT AbstractBackground and ObjectivesChina’s rapid population aging challenges traditional family-based care arrangements. This study examines heterogeneity in home-based care preferences by assessing how financial support sources, number of children, and children’s gender composition are associated with care preferences among older adults in Southwestern China.Research Design and MethodsData were drawn from the 2023 Chengdu-Chongqing Economic Circle Social Dynamics Survey. The analytic sample included 1,812 adults aged 50 years and older with children. Multinomial logistic regression models estimated associations between financial resources, family structure, and 4 care preference categories: self/spousal home care, child-provided home care, mixed family care at home, and other arrangements.ResultsFinancial reliance patterns were strongly associated with care preferences. Dependence on adult children for financial support was associated with a higher likelihood of preferring child-provided care, with this association stronger among women than men. Children’s gender composition was not significantly associated with care preferences. Larger family size (four or more children) was associated with mixed family care preferences.Discussion and ImplicationsThese findings suggest that financial resources may play a more prominent role than children’s gender in shaping care preferences in this population. Women’s greater reliance on children is consistent with gendered economic disadvantage in later life. Policies that strengthen pension adequacy and support diverse home-based care arrangements may help address evolving care needs. This study revisits traditional gender-based care perspectives by showing that financial sources are more strongly associated with care preferences than caregiver gender, marking a shift toward pragmatic filial piety in China, using new data from a less investigated area with a large aging population. These findings translate into critical policy actions: bolstering financial independence through improved pensions, particularly for women, and reinforcing support for home-based care. By prioritizing the sustainability of aging-in-place over institutional models, stakeholders can better address the needs of aging populations amidst shrinking family sizes, thereby improving societal and individual well-being. Background and objectives The world is undergoing an unprecedented demographic shift characterized by rapid population aging, largely driven by sustained declines in fertility and significant increases in life expectancy. According to the World Health Organization, the proportion of the world’s population aged 60 years and older is expected to nearly double from 12% in 2015 to 22% by 2050. Furthermore, the number of people aged 60 years and older is projected to grow from one billion in 2020 to 1.4 billion by 2030. While population aging began in high-income nations, such as Japan (where nearly 30% of the population is over 60 years old), the fastest rates of change are now occurring in low- and middle-income countries, which are expected to cover two-thirds of the global older population by 2050. This global trend presents both the societal triumph of longer, healthier lives and a significant challenge for social security, labor markets, and healthcare systems worldwide. As a populous country, China is facing an overwhelming aging trend. By the end of the 20th century, older adults in China accounted for 10% of the population aged 60 years and above, and 7% aged 65 years and above, indicating that the country had entered an aging society. Between 2000 and 2020, the size of China’s older adult population and old-age dependency ratio gradually increased, while the degree of aging also steadily intensified. Currently, population aging is still accelerating. By the end of 2023, China’s population aged 60 years and above reached 296 million, accounting for 21.1% of the total national population; among them, the population aged 65 years and above reached 216 million, accounting for 15.4% of the total national population (Xu et al., 2024). Under these circumstances, old-age care has become an increasingly urgent concern as people live longer. Guaranteeing stable support for older adults and maintaining the functioning of the family remain significant challenges, particularly given that home-based care is China’s traditional and predominant care model. While definitions of older adults vary across contexts, frequently operationalized as aged 60 years and older in global health frameworks (e.g., the World Health Organization) and 65 years and older in many high-income settings (Orimo et al., 2006). However, in the Chinese context, age 50 is widely used as a policy-relevant threshold due to the statutory retirement age for women in many sectors. Accordingly, this study adopts a 50+ definition in the empirical analysis while retaining the conventional 60+ definition when referring to prior literature. Conceptualizing old-age care modalities Late-life care modalities fundamentally address the core questions of where and how older adults receive support. Literature delineates these modalities along two primary axes. The first categorizes care based on the source of support, distinguishing between family-, society-, and self-supported modes (Li et al., 2024). The second adopts a locational framework drawn from aging policy, classifying care into home-based, community-based, and institutional modes (Huang et al., 2023; Sun et al., 2022). Within this locational typology, home-based care refers to older adults aging in place, either by co-residing with adult children or living independently (Liu et al., 2016), whereas community-based care denotes formal or semi-formal support services delivered in community settings without institutional residence. Amidst ongoing economic development and social transformation, traditional family structures have changed, leading to greater diversification of care arrangements (Du et al., 2016; Mu & Zhu, 2019). Despite this shift, the family remains a crucial domain for care provision, and home-based care persists as the predominant and preferred modality in China. Research has consistently indicated that older adults, influenced by traditional beliefs and cultural reservations about institutionalization, demonstrate a stronger preference for home-based care than community or institutional options (Liu et al., 2016; Lu & Zhang, 2018; Shao & Li, 2023). This preference for “aging in place” is also widely observed in Western countries (van Bilsen et al., 2008) and is supported by policy initiatives such as community-based aging programs in Singapore (Thang & Hong, 2015). At the same time, demographic pressures, including rising life expectancy and declining fertility, have prompted the development of more diverse socialized care modes, including formal and non-family-based arrangements (Gu et al., 2017). However, countries have pursued these developments through distinct institutional pathways rather than a single convergent trajectory. In many European contexts, including the United Kingdom, long-term care systems have shifted emphasis from institutional provision toward community- and home-based services (Scheibl et al., 2019). In the United States, broader policy frameworks have increasingly emphasized home- and community-based services, alongside the development of market-oriented residential models such as continuing care retirement communities that primarily serve financially advantaged older adults. Meanwhile, Japan has implemented a universal long-term care insurance system integrating formal services with family care (Watanabe et al., 2018). These cross-national variations indicate that the expansion of non-family or formal care modes reflects diverse welfare regimes and policy designs rather than a uniform global transition. Although the evolution of diverse care modes is inevitable, the pivotal role of the family in China persists (Chen, 2020), which has led to significant differences within the family-based care practices. Therefore, a more granular analysis of home-based modalities is essential to accurately capture the characteristics and factors influencing older adult care in China. Financial sources for old-age care Financial resources are a critical foundation for older adults’ quality of life and a key determinant in the choice of care modality. In developed countries, pension systems form the core of retirement security frameworks (Ju et al., 2017). In contrast, in developing countries, the protective function of pensions remains underdeveloped, leaving family support an essential component of economic sustenance. In China, old-age economic support has traditionally rested on three pillars: financial transfers from children or relatives, personal labor income, and pensions (Yi, 2015). The relative importance of these pillars has transformed over time. Before the 2009 New Rural Pension Insurance, older adults in rural areas relied primarily on family and labor income, whereas older adults in urban areas depended more on pensions. Between 2010 and 2015, reliance on pensions increased modestly, whereas family support gradually declined, although family transfers remained dominant in rural areas (Ding & Qu, 2019). Following the major social security reforms in 2014 and 2015, data from the 2020 national census revealed that pensions had, for the first time, become the primary income source for China’s older adult population (Du, 2023). Nevertheless, benefit levels, particularly under the urban–rural residents’ scheme, are often insufficient. Consequently, economic support from family and personal labor income remains a crucial supplement, especially in rural areas (Zhou et al., 2025). Significant disparities in economic support persist in China. Research indicates a substantial gender gap in pensions, with men receiving, on average, 1.9 times the pension benefits of women, thereby increasing older women’s dependence on private family transfers (Jiang & Zheng, 2013; Zhan, 2020). Family structure also plays an important role. Intrahousehold income redistribution, such as co-residence with a spouse, helps narrow income disparities among older adults. While living with adult children can provide financial safeguards, the presence of co-residing grandchildren may dilute the resources available to older adults (Jia et al., 2021). In this study, these financial sources are treated not merely as a form of support but as key predictors that structure the feasibility of different care arrangements. Factors influencing the choice of old-age care modes A review of the literature reveals different focal points based on context. Research in developed countries typically emphasizes the influence of social policies, living arrangements, long-term care insurance, and access to public services (Hajek et al., 2017; Sugisawa et al., 2020). In contrast, studies in the Chinese context categorize influencing factors into individual characteristics, family factors, and broader environmental conditions. At the individual level, key determinants include gender, age, education, health, and economic status (Yamanaka et al., 2022; Zhao et al., 2021). Younger, more highly educated, healthier, and economically advantaged older adults demonstrate a stronger propensity for autonomous and socially engaging care arrangements (Datta Gupta & Larsen, 2010; Liu et al., 2016). Economic resources directly influence care decisions (Gu, 2014): abundant resources enable preferences for self-supported or institutional care (Liang, 2020), whereas receiving financial transfers from adult children reinforces the choice of home-based care (Shang & Zhao, 2022). Gender also significantly affects preferences (Ji, 2022), with some studies suggesting that men show a greater preference for home-based care (Jia, 2020). Family-level factors such as marital status, family structure, and the number and co-residence of children exert a profound influence on care preferences. Regarding the number of children, older adults from one-child families, having long been accustomed to nuclear family living, often prefer independent arrangements and may deliberately maintain spatial distance from their adult child to avoid “imposing trouble” (Wang, 2016). Conversely, those with multiple children tend to favor home-based care, facilitated by the potential of rotating caregiving responsibilities among siblings (Tao & Liu, 2019). Generally, having a larger number of children is associated with greater caregiving resources (Wang et al., 2025; Wu & Wu, 2022), whereas having fewer children is linked to a stronger willingness to opt for institutional care (Wang et al., 2020). This pattern is particularly pronounced in East Asia, where cultural norms of filial piety mediate the relationship between family size and care preferences (Choi et al., 2018). In contrast, studies from Europe suggest that the number of children is not a critical determinant; factors such as education and spousal presence are more strongly associated with informal care provision (Albertini & Mantovani, 2022; Hajek et al., 2017). Nonetheless, some Chinese studies complicate this narrative by indicating that family structure (e.g., living with a spouse) may influence preferences more than the number of children (Wu, 2012), underscoring the multifaceted nature of these determinants. Furthermore, literature has highlighted that the gender structure of the offspring is also critical in the choice of older adult care modes. Older adults with sons exhibit a lower probability of choosing independent support (Long & Yang, 2024), while those without sons demonstrate a greater propensity to utilize formal care services (Huang et al., 2023). Within the Chinese context, traditional patrilineal norms are evolving toward “bilaterality,” where daughters play an increasingly significant caregiving role; however, patrilineal traditions persist, especially in rural areas (Mao & ZHu, 2017). Research in China indicates that gender differences in intergenerational support are multidimensional. Daughters tend to provide more day-to-day and instrumental caregiving to aging parents (Hu, 2017; Zhang et al., 2023), whereas their contributions to financial support have also increased in recent cohorts. Consequently, the gender pattern of monetary transfers has become more complex rather than strictly son-dominant (Xie & Zhu, 2009; Zhang et al., 2023). This dynamic differs notably from other cultural contexts. In India, married sons are typically the primary caregivers (Kadoya & Khan, 2019). A qualitative study in Italy has highlighted further divergence: Chinese respondents favored sons, Filipino participants favored daughters, and native Italians and Maghrebi respondents distributed their expectations almost equally. The same study found that Chinese families were more accepting of co-residence, whereas Italian families preferred to employ paid caregivers (Albertini & Mantovani, 2022). At the societal and environmental levels, the influencing factors for care mode include service availability, pension policies, and cultural norms. Greater access to community-oriented services promotes a preference for community-based care (Wang et al., 2022), whereas a wider range of available options is associated with a stronger preference for self-supported care (Wang et al., 2022). Moreover, government pensions and insurance policies (Engelhardt et al., 2005; Zhang, 2019), as well as traditional cultural values, such as filial piety (Xu et al., 2023), continue to shape care choices. Finally, structural factors, such as household registration (hukou) and geography, are relevant: individuals with rural registration demonstrate a stronger preference for family-based care, whereas their urban counterparts rely more on pension (Tang et al., 2020; Wang, 2016). Despite regional divergence, home-based care remains the dominant choice nationwide. Although the literature on old-age care is extensive, some limitations have been observed. First, the research has concentrated on the eastern, central, and northwestern regions of China, whereas the southwestern regions remain underinvestigated. Second, despite significant differences in care modes, cultural norms, and resource availability between urban and rural China, comparative analyses of the urban-rural divide remain scarce (Tang et al., 2020). Third, the internal heterogeneity within home-based care—the predominant mode—is often overlooked. Nuanced distinctions such as aging in place through self-support, relying primarily on children, or utilizing a combination of both within the home have received insufficient scholarly attention. Fourth, while determinants such as demographics and economics have been well studied, the role of offspring gender structure has been less explored, with most studies limited to simple binary indicators rather than nuanced categorical comparisons. Accordingly, this study contributes to the literature on care modalities by disaggregating home-based care into distinct family-based arrangements and examining how financial dependence, number of children, and children’s gender composition are associated with these modalities. Using survey data from southwestern China, this analysis provides regionally specific evidence on later-life care preferences, with particular attention to differentiation within home-based care alongside more open or non-home-based orientations. Given persistent gender inequalities in pension entitlements, lifetime earnings, and caregiving roles in China, men and women may differ in their resources and expectations for later-life care. We therefore examine gender heterogeneity in care preferences. Research design and methods Data and sample This study used the Chengdu-Chongqing Economic Circle Social Dynamics Survey, initiated and conducted by the Research Institute of Social Development at Southwestern University of Finance and Economics. Data were collected from November to December 2023 and covered 142 counties (including county-level cities) within the Chengdu-Chongqing Economic Circle. The total sample size was finalized as 3,600 households, a figure established after comprehensive consideration of precision, cost, and survey feasibility, drawing upon several major domestic social surveys. A stratified four-stage probability proportional to size sampling design was employed. In the first stage, counties, including county-level cities, served as the Primary Sampling Units (PSU). To enhance representativeness, these county-level units were stratified into five sampling frames according to regional socioeconomic characteristics: the Chengdu metropolitan area, the southern Sichuan urban agglomeration, the northern Sichuan urban agglomeration, the Chongqing central urban area, and other districts and counties within Chongqing. Twenty PSUs were selected from these strata, with allocation proportional to the population size of each frame. The second stage involved selecting streets and townships as the Secondary Sampling Units (SSUs). The composition of the SSUs within each selected PSU was determined based on the urbanization level of the respective PSU. In the third stage, three Tertiary Sampling Units (TSUs)—villages or residential communities—were randomly selected from each chosen SSU using a simple random sampling method. The fourth and final stages entailed the selection of respondents from each TSU. Thirty residents from different households were recruited for the survey. A team of 150 professionally trained interviewers of postgraduate students from the university was recruited to collect the data. Systematic door-to-door visits were conducted to identify and recruit eligible participants. The target population comprised residents aged 18 years or older who had been living locally for at least six months. The interviewees provided verbal consent and were able to quit at any point. All data were kept anonymous. Data were collected via face-to-face interviews. A total of 3,551 completed surveys were received. This analysis focused on older adults and their old-age care plans. The analytic sample includes individuals aged 50 and above, consistent with China’s institutional retirement age for women, which marks an early transition into later-life economic and caregiving arrangements. To better understand old-age care preferences and filial piety practices in contemporary China, this study involved only respondents with children. Of the 3,551 respondents, 2,099 were aged 50 years or older, and 1,962 had at least one child, constituting the target population for this study. After excluding cases with missing information on analytical variables, the final sample included 1,812 respondents (92.4% of eligible cases). Variables Old-age care preferences The dependent variable in this study was individuals’ old-age care preferences, measured using responses to the survey question: “Which type of old-age care do you intend to choose for yourself?” The original design for this question was multiple choice, in which respondents selected all that applied, offering 10 options. These first included five modes of “home care”: self-care or spousal care, care provided by paid caregivers, care provided by children, care assisted by artificial intelligence, and home-based care supplemented by community services. Other options included institutional care, living in a retirement community, and reciprocal older adult care. Because respondents could select multiple care options, a rule-based coding procedure was applied to construct mutually exclusive categories. First, we identified respondents expressing exclusive family-based home care preferences: those selecting only self/spousal care at home were coded as “self/spousal home care,” those selecting only child-provided care at home were coded as “child-provided home care,” and those selecting both self/spousal and child care (and no other options) were coded as “mixed family care at home.” All remaining response patterns—including any selections involving non-family caregivers, institutional care, community services, or multiple combinations beyond these family-based options—were coded as “other.” We acknowledge the heterogeneity within this “other” category, which comprises both mixed care arrangements (i.e., family care combined with formal services) and exclusive non-family care. However, grouping these diverse patterns into a single category was a necessary analytical choice to serve as a contrast group. This procedure ensured that each respondent was assigned to one and only one mutually exclusive category while preserving meaningful distinctions between exclusive family care preferences and more diversified care orientations. The analysis therefore focuses on heterogeneity within family-based care preferences while distinguishing them from more diversified or non-exclusive care orientations. Explanatory variables This study involved three main explanatory variables: the financial source of old-age support, the number of children, and the gender composition of the children. Financial support Financial support for old-age care was measured using the survey question: “What is the financial source for your old-age care?” The original item included eight single-choice options distinguishing reliance on one’s own pension/savings, reliance on adult children, combinations of self and filial support, continued employment income, and other subsidies. Given limited observations in some categories, responses were recoded into five mutually exclusive groups: (1) self-reliance based on pension or savings, (2) reliance on adult children, (3) combined self and child support, (4) continued employment, and (5) other sources. Continued employment was retained as a distinct category because labor income in later life reflects ongoing labor force participation and social engagement, which differ conceptually and empirically from retirement- or pension-based self-reliance. Number and gender of children Children’s status was another dimension of interest as the number and gender of children were included as variables in the analysis. Children’s gender data were derived from questions regarding the number of sons and daughters. This variable distinguished two mutually exclusive types, including “at least one son” and “daughters only.” By distinguishing families with any sons from those with daughters only, the classification reduces ambiguity present in multi-category gender compositions (e.g., both sons and daughters) and allows a clearer test of whether the presence of sons is associated with differences in preferred care arrangements. The number of children was categorized to reflect distinct fertility regimes in China (one-child, two-child, and larger families), which correspond to differences in potential caregiver availability. Covariates The analysis further incorporated a set of covariates for individual demographic and household characteristics. Individual demographic attributes included gender, age, marital status, self-reported health status, educational attainment, perceived economic status, and hukou type (agricultural or nonagricultural household registration). To be noted, China’s household registration (hukou) system creates a profound urban-rural institutional divide that shapes access to social welfare, healthcare coverage, and pension benefits. Urban residents typically receive more generous pension entitlements, higher-quality healthcare services, and greater availability of formal care resources, whereas rural residents rely more heavily on family-based support in later life. These structural disparities substantially influence older adults’ financial resources and care preferences in China’s aging context. In addition, this study defines the older population as individuals aged 50 years and above. This threshold reflects China’s institutional retirement system, in which the statutory retirement age for women in many sectors is 50 years. As retirement marks a critical transition in income sources, social roles, and family dependence, age 50 represents a meaningful boundary for examining old-age care preferences in the Chinese context. These variables were integrated into the analytical models to enhance the robustness of the estimation and provide a more comprehensive understanding of the factors shaping older adult care preferences. Data analysis The associations between financial support in old age, children’s status, and preferences for old-age care modes were assessed using a multinomial logistic regression, adjusting for demographic characteristics. Covariates were assessed for multicollinearity, and no variables were excluded from the models due to multicollinearity (all VIFs <1.5). To account for potential unobserved heterogeneity across local contexts, standard errors were clustered at the county level in all multinomial logistic models. The substantive conclusions remained unchanged when standard errors were clustered at the county level, suggesting that county-level contextual variation does not drive the observed associations. Results from two subsamples were compared to explore differential results for older men and women. All covariates were retained in the analysis, rather than using a backward elimination procedure. Statistical analyses were conducted using Stata version 16, and multinomial logistic regression results are reported as relative risk ratios (RRR) with 95% confidence intervals (CIs). A stepwise modeling strategy was adopted to distinguish the association of financial resources from that of family structure. Model set 1 included financial sources, the primary explanatory variables of interest, together with sociodemographic covariates. Model set 2 additionally incorporated children’s characteristics (number and gender composition) as adjustment covariates to capture variation in family structure and potential filial support availability. Comparing the two model sets allowed us to assess whether the associations of financial sources persisted after accounting for family structure. Finally, the independence of irrelevant alternatives (IIA) assumption was tested using Hausman tests based on multinomial logit models without cluster adjustment. The assumption was not violated. Results Respondent statistics The analytic sample included 1,812 adults aged 50 and above with at least one child. Overall, self/spousal home care (35.7%) and other care modes (36.7%) were the most common preferences, followed by child-provided (14.8%) and mixed family home care (12.8%). Nearly half relied primarily on their own pension or savings (47.0%), while smaller proportions depended on adult children (8.4%), combined sources (15.3%), or continued employment (5.6%). The sample was predominantly aged 50–59 (47.2%) and married (87.3%), with low educational attainment (76.4% junior high or below). Most respondents had at least one son (70.9%) and one or two children (89.2%), and 55.6% held agricultural hukou. As all variables involved are categorical variables, differences between men and women were tested using χ2 tests. Several characteristics differed significantly by sex (Table 1). Women were more likely than men to report other care preferences (40.8% vs 32.9%) and less likely to rely on self/spousal home care (34.1% vs 37.2%). Women also showed greater reliance on self-based financial sources (49.9% vs 44.3%) and were more often widowed (12.4% vs 5.7%). In family structure, women were less likely to have at least one son (68.0% vs 73.7%) and more likely to have only daughters (32.0% vs 26.4%). Women were also more concentrated in younger-old ages (50–59: 50.4% vs 44.2%) and in nonagricultural hukou (55.1% vs 34.5%). No significant sex differences were observed in education or self-rated health. Table 1Descriptive statistics of the sample by sex.CharacteristicsOverallMenWomen p N % n % n % Old-age care preferences .003 Self/spousal home care64735.7135037.1929734.10 Child-provided home care26814.7914315.2012514.35 Mixed family care at home23212.8013814.679410.79 Other care modes66536.7031032.9435540.76 Financial sources for old age .099 Self-reliance (pension or savings)85247.0241744.3143549.94 Reliance on adult children1538.44838.82708.04 Combined self and child support64015.3235537.7328532.72 Continued employment1025.63565.95465.28 Others653.59303.19354.02 Age group .000 50–5985547.1941644.2143950.40 60–6957931.9529130.9228833.07 70 and over37820.8623424.8714416.53 Level of education .534 Primary school and below65235.9832634.6432637.43 Junior high school73240.4038741.1334539.61 Senior high school and equivalent30016.5615616.5814416.53 Undergraduate (equivalent) and above1287.06727.65566.43 Marital status .000 Married1,58187.2585490.7572783.47 Widowed1628.94545.7410812.40 Others693.81333.51364.13 Gender of children .008 At least one son1,28570.9269373.6559267.97 Daughters only52729.0824826.3527932.03 Number of children .000 One child92751.1643245.9149556.83 Two children68938.0238741.1330234.67 Three children1417.78869.14556.31 Four children and over553.04363.83192.18 Hukou type .000 Agricultural1,00755.5761665.4639144.89 Nonagricultural80544.4332534.5448055.11 Self-reported health status .237 Good20011.049810.4110211.71 Average56030.9127929.6528132.26 Poor1,05258.0656459.9448856.03 Perceived economic status .028 Upper-middle1206.62757.97455.17 Middle77942.9938941.3439044.78 Lower-middle61834.1133435.4928432.61 Low29516.2814315.2015217.45 Descriptive statistics of the sample by sex. Table 2 indicates significant variation in characteristics across care preferences. Financial sources differed markedly (p < .001): self/spousal home care was most common among the self-reliant (57.5%), child-provided home care among those relying on children (23.1%), and mixed family care among those with combined support (48.3%). Age and gender distributions also varied, with younger respondents and women more represented in other care modes, and older adults and men in family-based care. Lower education and agricultural hukou were concentrated in child-provided or mixed care, whereas higher education and nonagricultural hukou were more common in self/spousal or other modes. Family size differed significantly; other factors showed limited variation. Table 2Sample characteristics by care preferences (%).CharacteristicsSelf/spousal home care (n = 647)Child-provided home care (n = 268)Mixed family care at home (n = 232)Other care modes (n = 665) p Financial sources for old age .000 Self-reliance (pension or savings)57.5030.9730.6049.02 Reliance on adult children5.7223.1310.344.51 Combined self and child support29.6840.3048.2834.29 Continued employment4.173.739.056.62 Others4.941.871.725.56 Age group .000 50–5942.6644.7840.0955.04 60–6935.7032.4629.7428.87 70 and over21.6422.7630.1716.09 Gender .003 Male54.1053.3659.4846.62 Female45.9046.6440.5253.38 Level of education .000 Primary school and below36.3253.3643.5326.02 Junior high school38.1834.3343.9743.76 Senior high school and equivalent17.9310.078.6220.60 Undergraduate (equivalent) and above7.572.243.889.62 Marital status .031 Married87.6486.1989.2286.62 Widowed9.2711.199.057.67 Others3.092.611.725.71 Gender of children .321 At least one son71.7273.5172.8468.42 Daughters only28.2826.4927.1631.58 Number of children .000 One child49.9248.8834.9158.95 Two children39.5737.6946.9833.53 Three children7.739.339.916.47 Four children and over2.784.108.191.05 Hukou type.000 Agricultural56.2671.2765.0945.26 Nonagricultural43.7428.7334.9154.74 Self-reported health status .067 Good9.7413.8112.5010.68 Average33.6924.6326.2932.33 Poor56.5761.5761.2156.99 Perceived economic status .116 Upper-middle6.655.605.177.52 Middle43.2847.3938.7942.41 Lower-middle34.1632.8432.3335.19 Low15.9214.1823.7114.89 Sample characteristics by care preferences (%). Main regression model Tables 3 and 4 present the multinomial logistic regression results, reporting the RRRs for factors associated with old-age care preferences. The models predicted the likelihood of older adults choosing child-provided home care, mixed family care at home, or other care modes, relative to the reference category of spousal or self-home care. With other variables controlled for, the source of financial support for old age was a highly significant predictor of care preferences (Table 3). Compared with self-reliance, older adults who relied on adult children had substantially higher risks of desiring child-provided home care (RRR = 6.30, 95% CI: 3.66–10.86) and mixed family care at home (RRR = 3.25, 95% CI: 1.67–6.32). Reliance on both self and adult children was associated with significantly higher likelihoods of all three care modes. Continued employment was not significantly related to child-provided home care but predicted mixed family care (RRR = 3.95, 95% CI: 1.24–12.57). Table 3Factors associated with child-provided home care, mixed family care at home and other care modes among older respondents.VariablesChild-provided home careMixed family care at homeOther care modes Financial sources for old age Self-reliance (pension or savings) (ref) Reliance on adult children6.30*** [3.66, 10.86]3.25** [1.67, 6.32]1.40 [0.63, 3.09] Combined self and child support2.24*** [1.48, 3.39]2.99*** [1.99, 4.48]1.63*** [1.33, 2.01] Continued employment1.33 [0.59, 3.04]3.95** [1.24, 12.57]2.32 [0.77, 6.98] Others1.05 [0.34, 3.20]1.15 [0.48, 2.73]2.56*** [1.53, 4.27] Age group 50–59 (ref) 60–690.78 [0.60, 1.03]0.90 [0.64, 1.26]0.64*** [0.50, 0.81] 70 and over0.69 [0.48, 1.00]1.25 [0.90, 1.74]0.65** [0.50, 0.84] Gender Male (ref) Female1.04 [0.82, 1.31]0.844 [0.69, 1.03]1.27* [1.04, 1.55] Level of education Primary school and below (ref) Junior high school0.75 [0.48, 1.16]1.16 [0.92, 1.45]1.51* [1.10, 2.07] Senior high school and equivalent0.52 [0.27, 1.02]0.55* [0.35, 0.88]1.42* [1.01, 2.00] Undergraduate (equivalent) and above0.27** [0.11, 0.68]0.57 [0.32, 1.01]1.43 [0.92, 2.22] Marital status Married (ref) Widowed0.94 [0.53, 1.41]0.77 [0.50, 1.19]0.90 [0.62, 1.29] Others0.98 [0.34, 2.83]0.59 [0.20, 1.74]1.67 [0.85, 3.28] Hukou type Agricultural (ref) Nonagricultural0.95 [0.68, 1.31]1.17 [0.77, 1.79]1.60*** [1.23, 2.07] Self-reported health status Good (ref) Average0.62 [0.35, 1.10]0.73 [0.51, 1.08]0.77 [0.54, 1.09] Poor0.94 [0.53, 1.69]1.12 [0.74, 1.69]0.83 [0.55, 1.25] Perceived economic status Upper and upper-middle (ref) Middle1.21 [0.633, 2.301]1.22 [0.55, 2.69]0.86 [0.58, 1.29] Lower-middle1.04 [0.536, 2.008]1.27 [0.53, 3.07]0.86 [0.55, 1.36] Low0.97 [0.471, 2.015]2.21 [0.97, 5.04]0.77 [0.45, 1.33]* p < .05.** p < .01.*** p < .001. Factors associated with child-provided home care, mixed family care at home and other care modes among older respondents. p < .05. p < .01. p < .001. Table 4Factors associated with child-provided home care, mixed family care at home and other care modes among older respondents (main model).VariablesChild-provided home careMixed family care at homeOther care modes Financial sources for old age Self-reliance (pension or savings) (ref) Reliance on adult children6.44*** [3.76, 11.04]3.18** [1.61, 6.29]1.42 [0.64, 3.13] Combined self and child support2.25*** [1.49, 3.40]3.02*** [2.02, 4.53]1.63*** [1.32, 2.02] Continued employment1.42 [0.65, 3.11]3.92* [1.27, 12.13]2.37 [0.78, 7.25] Others1.08 [0.35, 3.30]1.16 [0.49, 2.79]2.58*** [1.53, 4.34] Gender of children At least one son (ref) Daughters only1.00 [0.70, 1.44]1.21 [0.86, 1.69]1.04 [0.85, 1.28] Number of children One child (ref) Two children0.71* [0.51, 0.97]1.46* [1.03, 2.07]0.84 [0.65, 1.07] Three children0.75 [0.48, 1.18]1.27 [0.77, 2.11]0.89 [0.60, 1.34] Four children and over1.02 [0.38, 2.75]3.54** [1.46, 8.60]0.47 [0.19, 1.16] Age group 50–59 (ref) 60–690.79 [0.61, 1.04]0.91 [0.65, 1.27]0.64*** [0.50, 0.81] 70 and over0.73 [0.47, 1.11]1.06 [0.73, 1.55]0.70* [0.53, 0.93] Gender Male (ref) Female1.03 [0.83, 1.28]0.85 [0.68, 1.04]1.26* [1.03, 1.54] Level of education Primary school and below (ref) Junior high school0.73 [0.47, 1.14]1.19 [0.95, 1.51]1.48* [1.08, 2.02] Senior high school and equivalent0.49* [0.26, 0.96]0.59* [0.38, 0.92]1.36 [0.98, 1.89] Undergraduate (equivalent) and above0.26** [0.10, 0.65]0.60 [0.33, 1.08]1.37 [0.88, 2.13] Marital status Married (ref) Widowed0.92 [0.62, 1.37]0.73 [0.46, 1.18]0.91 [0.63, 1.30] Others0.95 [0.33, 2.72]0.61 [0.21, 1.78]1.64 [0.84, 3.20] Hukou type Agricultural (ref) Nonagricultural0.89 [0.64, 1.24]1.28 [0.87, 1.89]1.53** [1.15, 2.03] Self-reported health status Good (ref) Average0.61 [0.34, 1.09]0.72 [0.48, 1.07]0.76 [0.53, 1.09] Poor0.95 [0.52, 1.73]1.11 [0.73, 1.70]0.83 [0.55, 1.25] Perceived economic status Upper and upper-middle (ref) Middle1.20 [0.66, 2.18]1.33 [0.60, 2.98]0.84 [0.57, 1.24] Lower-middle1.04 [0.54, 2.00]1.43 [0.60, 3.42]0.84 [0.54, 1.30] Low0.97 [0.51, 1.85]2.46* [1.06, 5.72]0.75 [0.44, 1.27] Note. Values report relative risk ratios, with confidence intervals in parentheses.* p < .05.** p < .01.*** p < .001. Factors associated with child-provided home care, mixed family care at home and other care modes among older respondents (main model). Note. Values report relative risk ratios, with confidence intervals in parentheses. p < .05. p < .01. p < .001. Table 4 adds children’s characteristics to assess whether family structure mediates or confounds the associations observed in Table 3. The role of financial resources remains largely stable, indicating that financial independence is associated with care preferences beyond differences in family composition. Specifically, the gender of children (relative to at least one son) showed no statistically significant association with any of the care preferences. Conversely, the number of children was a significant predictor of mixed family care preferences. Compared to older adults with one child, those with two children (RRR = 1.46, 95% CI: 1.03–2.07) and especially four children or more (RRR = 3.54, 95% CI: 1.46–8.60) were significantly more likely to prefer mixed family care arrangements. Having more children was not, however, significantly associated with desiring pure child-provided home care. Several control variables were also significant: a higher level of education was strongly associated with lower likelihood of preferring child-provided home care (e.g., undergraduate and above: RRR = 0.26, 95% CI: 0.10–0.65); hukou type was a factor for other care modes, with nonagricultural respondents having higher likelihood (RRR = 1.53, 95% CI: 1.15–2.03) than agricultural respondents; and low perceived economic status was a significant predictor of mixed family home care (RRR = 2.46, 95% CI: 1.06–5.72) relative to upper-status respondents. Although age differed significantly across care categories in descriptive analyses (Table 2), this variation was primarily driven by the “other” care group; age was not significantly associated with differences among home-based arrangements in multinomial models. Self-rated health showed marginal descriptive variation (p = .067) but was not significant in the main models. Stratified analysis The association of financial support with care preferences revealed a clear gender disparity in Table 5. Wald tests based on gender-predictor interaction terms revealed significant gender heterogeneity in the association between financial sources of old-age support and preferred care arrangements (χ2 = 64.94, p < .001). In contrast, the association between children’s gender composition and care preferences did not differ significantly by gender (χ2 = 0.92, p = .821). These results indicate that financial dependence structures operate differently for men and women, whereas the influence of children’s gender appears comparable across genders. Table 5Stratified analysis by sex.VariablesFemaleMaleChild-provided home careMixed family care at homeOther care modesChild-provided home careMixed family care at homeOther care modes Financial sources for old age Self-reliance (pension or savings) (ref) Reliance on adult children17.25***6.61***3.27*3.29***1.980.80[6.10, 48.76][2.97, 14.71][1.30, 8.24][2.15, 5.03][0.79, 4.98][0.32, 1.97] Combined self and child support2.92***3.72***2.17***1.83*2.45**1.23[1.72, 4.98][2.20, 6.29][1.66, 2.84][1.08, 3.08][1.40, 4.29][0.87, 1.75] Continued employment2.873.493.680.763.64*1.53[0.82, 10.04][0.74, 16.40][0.91, 14.85][0.28, 2.05][1.34, 9.86][0.55, 4.28] Others1.102.064.14**1.040.871.76[0.20, 6.12][0.64, 6.59][1.80, 9.52][0.30, 3.57][0.21, 3.63][0.83, 3.74] Gender of children At least one son (ref) Daughters only0.881.190.961.171.251.14[0.58, 1.34][0.81, 1.77][0.76, 1.20][0.72, 1.90][0.74, 2.12][0.84, 1.53] Number of children One child (ref) Two children0.56**1.83**0.900.861.380.81[0.37, 0.86][1.19, 2.81][0.62, 1.31][0.56, 1.31][0.88, 2.15][0.61, 1.08] Three children0.421.680.801.121.110.98[0.17, 1.02][0.76, 3.67][0.43, 1.46][0.65, 1.91][0.61, 2.01][0.59, 1.62] Four children and over0.167.98***0.771.942.620.34[0.02, 1.01][2.76, 23.04][0.17, 3.53][0.75, 5.07][0.81, 8.52][0.10, 1.19] Age group 50–59 (ref) 60–690.851.330.63**0.700.630.65*[0.57, 1.27][0.86, 2.06][0.46, 0.85][0.44, 1.10][0.39, 1.02][0.44, 0.96] 70 and over0.961.110.780.610.920.66[0.42, 2.18][0.51, 2.42][0.45, 1.36][0.37, 1.02][0.57, 1.50][0.44, 1.02] Level of education Primary school and below (ref) Junior high school0.571.411.430.811.041.50***[0.29, 1.12][0.87, 2.27][0.84, 2.44][0.47, 1.41][0.72, 1.50][1.19, 1.91] Senior high school and equivalent0.551.201.600.430.30**1.15[0.22, 1.36][0.51, 2.83][0.91, 2.80][0.17, 1.12][0.14, 0.65][0.81, 1.62] Undergraduate (equivalent) and above0.21*0.231.370.280.701.41[0.05, 0.97][0.03, 1.81][0.77, 2.45][0.08, 1.01][0.33, 1.45][0.85, 2.36] Marital status Married (ref) Widowed1.110.700.820.690.700.98[0.61, 2.01][0.36, 1.36][0.46, 1.46][0.37, 1.31][0.36, 1.36][0.47, 2.03] Others1.440.591.050.570.602.85*[0.33, 6.18][0.15, 2.36][0.40, 2.74][0.13, 2.43][0.14, 2.67][1.14, 7.13] Hukou type Agricultural (ref) Nonagricultural1.051.862.11***0.770.931.17[0.63, 1.75][0.94, 3.66][1.41, 3.16][0.54, 1.11][0.56, 1.54][0.81, 1.70] Self-reported health status Good (ref) Average0.581.020.990.54*0.540.58[0.22, 1.53][0.54, 1.94][0.52, 1.87][0.30, 0.98][0.25, 1.19][0.28, 1.21] Poor1.001.431.030.820.890.65[0.38, 2.67][0.84, 2.44][0.53, 2.01][0.45, 1.50][0.44, 1.78][0.31, 1.37] Perceived economic status Upper and upper-middle (ref) Middle1.432.350.951.091.080.71[0.46, 4.47][0.54, 10.34][0.55, 1.63][0.52, 2.28][0.38, 3.07][0.43, 1.18] Lower-middle1.042.170.731.191.370.96[0.29, 3.72][0.49, 9.54][0.33, 1.60][0.49, 2.91][0.50, 3.77][0.60, 1.52] Low0.945.32*0.841.131.760.65[0.31, 2.84][1.29, 22.03][0.43, 1.67][0.53, 2.38][0.54, 5.71][0.33, 1.29] Note. Values report relative risk ratios, with confidence intervals in parentheses.* p < .05.** p < .01.*** p < .001. Stratified analysis by sex. Note. Values report relative risk ratios, with confidence intervals in parentheses. p < .05. p < .01. p < .001. For women, relying on adult children (vs. self-reliance) was associated with a substantially higher likelihood of preferring child-provided home care (RRR = 17.25, 95% CI: 6.10–48.76), mixed family care at home (RRR = 6.61, 95% CI: 2.97–14.71), and other care modes (RRR = 3.27, 95% CI: 1.30–8.24). Among women, reliance on adult children for financial support shows a pronounced association with preferring child-provided home care. The wide confidence interval suggests some imprecision in the exact magnitude. Rather than indicating a definitive effect size, the finding may reflect the close alignment between financial dependence and anticipated caregiving source. For men, the association was significant but notably weaker: reliance on children increased the likelihood of child-provided home care (RRR = 3.29, 95% CI: 2.15–5.03) but showed no statistically significant association with mixed family care or other care modes. Continued employment showed no statistically significant association with women’s care preference, whereas among men it was a significant predictor of preferring mixed family care (RRR = 3.64, 95% CI: 1.34–9.86). Children’s gender (daughters only vs. at least one son) showed no statistically significant association with any care preference for either older women or men, thus providing limited evidence for a systematic “daughter-care advantage” in this sample. The number of children exhibited a non-linear relationship: compared to one-child families, women with two children were less likely to prefer pure child-provided care (RRR = 0.56, 95% CI: 0.37–0.86) but more likely to prefer mixed family care (RRR = 1.83, 95% CI: 1.19–2.81). In particular, having four or more children strongly predicted women’s preference for mixed family care (RRR = 7.98, 95% CI: 2.76–23.04), while the corresponding association for men was not statistically significant. Several socioeconomic covariates also differentiated the gender-specific models. Undergraduate-level education reduced women’s likelihood of preference over child-provided home care (RRR = 0.21, 95% CI: 0.05–0.97), whereas senior high school education reduced men’s likelihood of preferring mixed family care at home (RRR = 0.30, 95% CI: 0.14–0.65). Nonagricultural hukou increased women’s likelihood of other care preferences (RRR = 2.11, 95% CI: 1.41–3.16) but showed no similar result for men. We conducted several sensitivity analyses to assess the robustness of our findings. First, we re-estimated the main model using a binary logistic regression. The outcome was collapsed into family-based (self/spousal, child, or mixed) versus non-family care, where non-family care corresponds to the “other” category, including community, institutional arrangements, and so forth. We also tried another categorization to turn the outcome variable into child-provided home care versus the others. Second, we tested the IIA assumption, which was not violated (p > .10). The assumption was not violated. All analyses produced results that were substantively consistent with the main findings, underscoring the robustness of our results. Discussion and implications This study examined the factors influencing old-age care preferences among older adults in China, with a particular focus on the interplay between financial support systems and intra-family dynamics, such as the number and gender of children. Our analysis differentiates among self/spousal home care, child-provided home care, mixed family care at home, and other care modes, thereby addressing several gaps identified in the existing literature, particularly concerning the internal heterogeneity of home-based care and the nuanced role of offspring structure. The findings revealed that financial interdependence was the primary factor associated with care type choice, and this association varied significantly by gender. Second, the gender of the children was found to be statistically irrelevant, whereas the number of children (specifically four or more) was a strong predictor of mixed family care. The primacy of financial interdependence in shaping care arrangements The main finding of this study was the strong link between financial support and older adults’ care preferences. Those relying on adult children for financial support were more likely to prefer child-provided care and mixed family care than self-reliant individuals. This aligns with previous research that identified economic factors as critical determinants of care decisions (Gu, 2014; Shang & Zhao, 2022). However, our findings quantify the sheer magnitude of this dependency, demonstrating that financial and care arrangements are not separate but deeply intertwined. This financial-care nexus is gendered. Among women, reliance on children (vs. self-reliance) was associated with a huge increase in the likelihood of preferring child-provided home care. For men, this association was significant but weaker. This disparity aligns with existing research on the gender gap in pension benefits (Zhan, 2020) and supports the argument that older women’s comparatively limited economic resources (Jiang & Zheng, 2013) heighten their dependence on intergenerational transfers. For older women, preferring child-provided care reflects both a norm and constrained financial autonomy. Conversely, older men—who are more likely to receive higher pensions or have a spouse available for caregiving—appear less constrained by financial reliance on their children in shaping their care preferences. This analysis refines the economic perspective. Prior literature identifies three main pillars of economic support in old age: pensions, transfers from adult children, and personal labor income (Yi, 2015). Our findings related to “continued employment” correspond directly to this third pillar. Continued employment among older adults is not significantly linked to child-provided home care; rather, it is a strong predictor of mixed family care at home and other care modes. This suggests that reliance on children and continued labor income are two distinct economic pathways. While reliance on children anchors care within the family, continued employment offers a more flexible approach, allowing older adults to combine care sources and reduce reliance on any one option. Deconstructing familial obligations: revisiting the role of children A central, and perhaps counterintuitive, finding of this study was the non-significant association between children’s gender and care preferences. Contrary to the “daughterly-care hypothesis” and the deeply embedded cultural norm of “raising sons for old age” (Long & Yang, 2024; Mao & ZHu, 2017), we found no statistical difference between having at least one son and having daughters only. This diverges from prior literature that positions child gender as a key factor in older adults’ care preferences. This result may reflect evolving family norms and is consistent with the concept of “bilaterality”. Previous studies document both traditional son preference and the growing involvement of daughters in parental care (Zhou et al., 2025). Our analysis suggests that the practical availability of children, rather than their gender, is more salient in shaping care preferences. This pattern indicates a possible weakening of gender-differentiated expectations in contemporary China, compared with earlier East Asian evidence of strong gendered patterns (e.g., Albertini & Mantovani, 2022; Kadoya & Khan, 2019). While the gender of children was not a significant factor, the number of children was important. Having more children, particularly four or more, was associated with a higher likelihood of preferring mixed family care rather than purely child-provided care. This aligns with research showing that larger sibships expand caregiving resources (e.g., Wang et al., 2025; Wu & Wu, 2022). By contrast, single-child families may face concentrated responsibility, often resulting in exclusive child-provided arrangements or a preference for independence to avoid burdening the only child (Wang, 2016). Families with multiple children can distribute responsibilities more flexibly, supporting shared caregiving (Tao & Liu, 2019). The prevalence of mixed-care preferences in these families underscores heterogeneity within home-based care, often treated as a single category in prior research. Regarding socioeconomic status, higher educational attainment was associated with lower odds of preferring child-provided home care, consistent with research linking education to greater autonomy and openness to nonfamilial care options (Liang, 2020; Yamanaka et al., 2022; Zhao et al., 2021). Similarly, nonagricultural hukou was associated with a higher likelihood of preferring other care modes, reflecting urban–rural differences in care contexts (Tang et al., 2020). Self-rated health was not significantly associated with care preferences, differing from studies identifying health as a primary determinant (Datta Gupta & Larsen, 2010; Liu et al., 2016). This may reflect the subjective nature of self-rated health and the possibility that care preferences are influenced more by long-term financial and family structures than by current health status. Health may become more influential following acute shocks not captured in cross-sectional data, highlighting an area for future research. Limitations and directions for future research This study had several limitations. First, the cross-sectional data failed to capture actual care behaviors, and we can only identify associations, not causation. Therefore, longitudinal data are needed to track how these preferences evolve into reality, especially following health or financial shocks. Second, our sample was economically concentrated in the middle-to-lower-middle strata and reported a high prevalence of poor health, which may limit the generalizability of our findings to wealthier or healthier cohorts of older adults. Meanwhile, information on current employment status contained substantial missing values, preventing its inclusion in the main models. Another limitation concerns the heterogeneity of the “other” care preference category. Although we clarified its composition, this category still combines several distinct arrangements, including institutional care, community-based care, and mixed care forms. As a result, it should not be interpreted as a substantively unified preference type. Rather, it functions as a residual category capturing preferences that are not exclusively family-based. This grouping may mask meaningful variation across non-family or mixed care orientations. Finally, although our quantitative models reveal significant factors, they cannot explain the mechanisms behind the relationships. One interesting finding was the relationship between older adults’ continued employment and care arrangements. The reasons for this could be older adults’ expectations of more financial autonomy or financial dependence on children, which are not available. Therefore, qualitative research is needed to explore family functions and negotiations behind such phenomena. Conclusion and implications This study provided a nuanced and empirically grounded analysis of old-age care preferences in China. Our findings demonstrate that while the family remains the central pillar of older adult support, the mechanisms underlying that support are evolving. Care preferences are closely linked to financial interdependence, a pattern that may place older women at greater risk of dependence. The key contribution of this study is that it revisits traditional gender-based perspectives on care. We find that the who of caregiving (son vs daughter) may matter far less than the how (financial reliance) and the how many (number of children). This suggests a shift toward a more pragmatic, resource-based, and gender-neutral model of filial piety and reflects the reality that older adults’ perceptions are changing with modernization. For policymakers, these findings highlight the urgent need to bolster the financial independence of older adults, particularly women, through adequate pension systems. Furthermore, policies must recognize the heterogeneity of home-based care, as it remains the preferred care choice of older adults in China. Rather than shifting emphasis toward other care models, stakeholders should prioritize strengthening and supporting home-based care arrangements. Owing to smaller family sizes, ensuring the sustainability and effectiveness of home-based care may become increasingly critical.