Gender in old care: comparing sons’ and daughters’ role in parental functional and psychological well-being Authors: Xuyang Du, Hualin Wei, Xianbo Zhang, Conghui Bai Published: 2025 ABSTRACT China’s rapidly aging population and evolving socio-economic landscape, particularly shifting gender roles, are reshaping traditional caregiving dynamics. However, limited research has examined the associations between caregiver gender and the functional and psychological health of older parents, particularly in the context of gender matching between caregivers and care recipients. This study addresses this gap by utilizing four waves of CLHLS data, employing a Logit model for estimation, and conducting robustness checks using the PSM method. The empirical results suggest that parents with sons as caregivers exhibit better functional health but poorer psychological well-being. Additionally, when the caregiver and care recipient share the same gender, the care recipient exhibits significantly better psychological health. Further analysis shows that the positive association between sons’ caregiving and functional health is statistically significant only in low-income families. while the psychological benefits of daughters’ caregiving are only evident in high-income families. Moreover, caregiving patterns show greater variation in urban areas compared to rural areas. This study contributes theoretically by developing an analytical framework that integrates gender role theory, social exchange theory, and the homophily hypothesis. The framework is applied to examine associations between caregiver gender and the health of older adults. From a practical perspective, the study identifies heterogeneous caregiving patterns across income groups and urban–rural contexts. These findings provide empirical insights for strengthening family caregiving and for designing more targeted aging policies. Introduction Global population aging poses profound challenges for health and social care systems worldwide. By 2050, the proportion of individuals aged 65 + will surge to 16% worldwide, with developing countries facing compounded pressures due to rapid demographic shifts and constrained fiscal capacities [1, 2]. In OECD nations, 60% of healthcare budgets prioritize reactive medical treatments over preventive care—a model ill-suited to aging populations requiring long-term support [3]. This systemic mismatch exacerbates elder care deficits, particularly in contexts where family-based care remains indispensable for offsetting institutional shortfalls [4–7]. China represents a particularly important case for examining these demographic and social challenges. China’s eldercare system exemplifies the collision between persistent cultural frameworks and rapid socioeconomic modernization [8–13]. China has the world’s largest aging population, projected to reach 400 million by 2035. This demographic shift poses significant challenges for eldercare, particularly due to persistent rural–urban disparities and entrenched gendered caregiving norms [9]. In rural regions, patrilineal traditions institutionalize sons as primary economic providers, yet outmigration of male labor forces has created 42 million “empty nest” older adults who predominantly rely on daughters for daily care [14, 15]. This reliance starkly contrasts with the traditional metaphor of “married daughters as spilled water,” which symbolizes the perceived attenuation of daughters’ familial obligations post-marriage [11, 16]. These systemic tensions—rural dependence on culturally marginalized care from daughters and urban reliance on fragmented market mechanisms—reveal fundamental institutional misalignments. These contradictions perpetuate intergenerational inequalities in resource allocation and limit older adults’ capacity to maintain functional health [16, 17]. They reflect a mismatch between family-based caregiving expectations and the structural changes brought by economic and demographic transition. This mismatch contributes to poorer health outcomes among older adults and indicates the importance of incorporating gender considerations into policy design. The psychological implications of caregiving are increasingly recognized as critical for understanding health outcomes among older adults. Evidence indicates that daughters’ involvement in care is associated with more favorable psychological outcomes, such as higher life satisfaction and lower depressive symptoms [17]. However, these benefits have received limited recognition in research, where the primary focus has been on functional and physical health indicators [16]. Much of this emphasis has been linked to the role of sons in providing household economic support, leaving the psychological contributions of daughters relatively undervalued [16–19]. At the same time, shifting gender roles are reshaping caregiving dynamics. Women’s rising economic agency, supported by universal education and reduced gender barriers in the labor market, has enabled daughters to match sons in household economic support [20, 21], even surpassing them in psychological care provision [19]. Therefore, examining gendered caregiving patterns is essential for explaining variations in health outcomes among older adults. Recent studies have also shown that older adults’ expectations regarding future caregiving are associated with health and well-being outcomes in East Asian contexts [22, 23]. These findings indicate that research on caregiving should consider not only actual arrangements but also expectations about the respective roles of sons and daughters. This perspective provides a useful starting point for the present study, which is similarly concerned with how gendered caregiving patterns shape older adults’ health. However, most existing studies have concentrated on functional health indicators, such as activities of daily living (ADL) and instrumental activities of daily living (IADL), with limited attention to psychological well-being [14, 19, 24–28]. There is also a lack of research on how caregiver gender differences are associated with the psychological and functional health of fathers and mothers, respectively. Finally, although many studies address rural-urban differences, they often overlook the significant role of family income in shaping caregiving outcomes [29, 30]. As a result, the existing literature lacks consistent conclusions regarding the associations between caregiver gender and the physical and psychological health of care recipients of different genders. It also shows limited clarity on how these relationships vary across rural–urban settings and income groups. Based on the gaps identified in existing research, this study uses longitudinal data from CHARLS and advanced statistical methods to estimate the following research objectives: (1) To investigate the associations between caregiver gender and older parents’ health outcomes; (2) To assess whether the gender match between caregivers and older adults (same-gender vs. different-gender relationships) is linked to variations in health outcomes. (3) To explore how these associations differ across rural–urban contexts and income groups. Literature review and theoretical framework Caregiver gender and health Caregiver gender and functional health (ADL/IADL) A large body of research shows that children’s caregiving improves parents’ functional health, typically measured by ADL and IADL [18, 30–33]. In China, Zhao & Shangguan (2024) observe that the absence of migrant children corresponds to worse health outcomes among rural and low-income older parents, with no similar relationship identified for urban and higher-income families [27]. Building on this, Chu (2021) further reports that greater geographical distance between children and parents is associated with reduced care provision, with daughters more often linked to improvements in physical health and sons to variations in psychological health [18]. However, cross-national evidence complicates this narrative: Kikuzawa (2016) finds no significant caregiver-gender differences in Japan [34], and Schulz et al. (2024) reach similar conclusions for Germany [19]. Such inconsistencies point to the role of broader cultural and institutional contexts. In East Asian societies shaped by Confucian traditions, norms of filial piety and the expectation of “raising sons for old-age security” position family members—particularly children—at the core of elder support [17]. In this case, caregiving roles are strongly gendered: sons are expected to provide financial assistance, while daughters are more involved in daily and emotional care. In contrast, welfare-state contexts such as Germany rely heavily on institutional and community-based services, which reduce the relative importance of family support [6]. Japan represents a distinctive case. Family caregiving traditions coexist with an expanding public support system, which may reduce the extent to which caregiver gender is associated with variations in functional health [34]. These cultural and institutional contrasts help explain why the associations of caregiver gender on ADL and IADL varies across countries and cultures. Caregiver gender and psychological health Empirical evidence indicates that children’s caregiving contributes to parents’ psychological well-being by reducing depression, loneliness, and social isolation [16, 17, 19, 30]. For example, Zhao & Shangguan (2024) show that the absence of migrant children significantly harms the psychological health of rural and low-income parents in China [27]. Chu (2021) reports that caregiving by daughters is more strongly associated with improvements in psychological outcomes, whereas the corresponding associations for sons are weaker [18]. Lin et al. (2024) report that long-distance children in the United States are linked to higher rates of depression and isolation [35], whereas Shulyaev et al. (2024) show that the presence of children reduces depression among hospitalized older adults in South Korea [33]. Gap Despite substantial evidence that children’s caregiving improves both functional and psychological health, the literature reveals two important gaps. First, findings on functional outcomes (ADL/IADL) are inconsistent across cultural contexts, reflecting variation in caregiving norms and practices. Second, psychological health has received far less attention, with most studies focusing narrowly on functional indicators. Together, these gaps underscore the importance of examining caregiver-gender differences in relation to both functional and psychological health outcomes of older parents. Hypothesis H1 Caregiver gender is associated with both the functional (ADL/IADL) and psychological health of older parents. Caregiver–Parent gender interaction and health Caregiver–Parent gender interaction and functional health (ADL/IADL) Empirical studies suggest that the alignment of caregiver and parent gender is associated with functional outcomes. Schulz et al. (2024) report that daughters are more effective than sons in improving parents’ IADL [19]. Lin et al. (2024) find that fathers particularly value financial assistance from sons, which contributes to maintaining functional health [35]. These findings highlight how gender congruence between caregivers and parents may improve functional outcomes through better alignment of caregiving supply and parental expectations. Caregiver–Parent gender interaction and psychological health Empirical studies also show that caregiver–parent gender interaction has important psychological implications. Zhang & Harper (2022) find that daughters improve mothers’ psychological well-being more effectively than sons [30]. Duan & Chen (2023) report that daughters’ emotional support alleviates parental depression, while sons’ contributions are weaker [31]. Mafla-España et al. (2024) find that daughter–mother dyads reduce loneliness and emotional distress [36], while Cook et al. (2018) emphasize that same-gender emotional exchanges enhance caregiving quality [37]. Gap Although existing studies suggest that caregiving outcomes may differ when caregiver and parent genders align, limited research has estimate whether same-gender dyads produce better results across both functional (ADL/IADL) and psychological health outcomes. Hypothesis H2 Same-gender caregiver–parent dyads yield better health outcomes—both functional (ADL/IADL) and psychological—than cross-gender dyads. Heterogeneity in older adults’ health outcomes by Urban–Rural and income differences Heterogeneity in functional health outcomes (ADL/IADL) Older adults’ functional health, measured through ADL and IADL, is significantly shaped by socioeconomic conditions such as income, culture, family composition, and residential setting [16, 19, 24–26, 30]. These factors contribute to substantial heterogeneity across income groups and urban–rural contexts, resulting in variations in the observed associations between caregiving and health outcomes [17, 24, 27, 29, 30]. From an urban–rural perspective, caregiving strategies differ due to disparities in economic development and cultural expectations [14]. In rural areas, traditional norms assign sons the role of primary economic providers, and their financial support has a stronger association with parents’ functional health [27]. By contrast, urban families emphasize access to health services and psychological support, both of which contribute to improved IADL [38]. From an income perspective, higher-income households tend to prioritize financial assistance and formal care resources, which are more effective in sustaining parents’ ADL and IADL [16]. Heterogeneity in psychological health outcomes Psychological health outcomes, such as depressive symptoms and life satisfaction, also exhibit clear heterogeneity across income and residential settings. In low-income families, daughters’ direct emotional support and hands-on care are critical for alleviating parental psychological distress [30]. Non-financial forms of support, such as companionship and daily assistance, play a particularly important role in these contexts [29, 30]. n multi-child rural households, sons typically assume economic caregiving responsibilities, while daughters’ emotional support substantially enhances parents’ psychological well-being [38]. In urban settings, caregiving tends to be more diverse, with daughters increasingly central in providing psychological support. Longitudinal studies demonstrate that daughters’ involvement significantly reduces depressive symptoms and improves psychological well-being, particularly among parents with chronic illnesses [39, 40]. These findings suggest that the associations between caregiving and health outcomes differ markedly across residential settings and income groups. In rural areas, limited access to formal health services and stronger adherence to patrilineal norms mean that sons’ financial support remains particularly important for sustaining parents’ functional health, while daughters’ contributions are often confined to daily and emotional care. In urban settings, however, better-developed medical infrastructure and more flexible family arrangements enable daughters to play a broader role, with their support improving both ADL/IADL and psychological outcomes. Income level also shapes caregiving strategies: higher-income households can rely on financial transfers and formal services to maintain functional capacity, whereas lower-income households, with fewer external resources, depend more heavily on daughters’ hands-on and emotional support to safeguard psychological well-being. Gap Existing research has examined heterogeneity in caregiving outcomes by urban–rural setting or income level separately, but has seldom considered how caregiver-gender patterns in health outcomes vary across their combined categories. Hypothesis H3 Caregiver-gender patterns in both functional (ADL/IADL) and psychological health outcomes differ across four socioeconomic contexts: urban high-income, urban low-income, rural high-income, and rural low-income families. Theoretical discussion This study draws on three theoretical perspectives to explain why caregiver gender and gender matching are expected to be associated with parents’ functional and psychological health. Social Exchange Theory conceptualizes caregiving as a reciprocal process in which different types of support—financial, emotional, or daily assistance—are exchanged between generations. Fathers often prioritize financial transfers, while mothers tend to place greater value on emotional and daily support [13]. This framework helps clarify why sons and daughters are associated with different patterns in parental health outcomes. It provides theoretical grounding for H1, which posits that caregiver gender is associated with both functional and psychological health. It also supports H2, which emphasizes that gender-matched dyads may be more effective because they align with parents’ gendered expectations. Gender Role Theory emphasizes how prevailing social norms shape caregiving behaviors. In many societies, men are expected to serve primarily as financial providers. Women, by contrast, are more closely associated with emotional support and nurturing roles [11]. These normative expectations shape the type and quality of support delivered by sons and daughters, leading to systematic differences in caregiving outcomes. This theory therefore reinforces H1 by explaining gender-based variation in caregiving outcomes, and supports H2 by suggesting that congruence between caregiver and parent gender facilitates a closer match between care needs and care provision. The Homophily Hypothesis posits that social interactions are more effective when participants share similar characteristics, such as gender. Same-gender relationships foster trust, emotional closeness, and smoother communication, which can enhance both functional and psychological support [41]. This perspective provides a direct rationale for H2, which predicts that same-gender caregiver–parent dyads yield better health outcomes than cross-gender dyads. Data and methodology Data and sample This study utilizes data from four waves of the China Health and Retirement Longitudinal Study (CHARLS) conducted in 2013, 2015, 2018, and 2020. The dataset provides information on family structure, employment, income, and health for individuals aged 45 and above across 28 provinces in China [42]. In 2013, there were 18,612 observations. Follow-up surveys were conducted in 2015, 2018, and 2020, with new participants added each year to compensate for attrition due to respondent death or loss. As shown in Fig. 1, the number of observations was 21,097 in 2015, 19,816 in 2018, and 19,395 in 2020, resulting in a total of 78,920 observations across the four waves. Missing data on all variables were addressed using listwise deletion, meaning that cases with missing values on any of these variables were excluded. This choice was made because missingness in CHARLS is largely due to item non-response during interviews, which occurs randomly and is not systematically related to observed characteristics [43]. Under such conditions, imputing values with linear prediction or other model-based methods would not be appropriate. Although listwise deletion reduces the effective sample size and may bias estimates if data are not missing completely at random, we mitigated this concern in two ways. First, we explicitly reported the number of cases excluded due to missing values in the Fig. 1 to ensure transparency. Second, we conducted robustness checks to confirm that our main results are not sensitive to the missing-data strategy [44]. Accordingly, while this approach inevitably reduces statistical power, it provides a conservative and transparent way to estimate the associations between caregiving and health outcomes. Since this study aims to estimate gender differences in caregiving patterns and their associations with parents’ ADL, IADL, and psychological health, the analytic sample was restricted to parents exclusively cared for either by sons or by daughters. As shown in Fig. 1, the pooled sample from four CHARLS waves consisted of 78,920 person-wave observations. Among these, 73,734 cases reported no child-provided care and 1,253 cases reported joint caregiving by both sons and daughters; both groups were excluded because they do not allow a clear identification of gender-specific caregiving patterns or their associations with health outcomes. We also excluded cases involving missing values on variables. After these exclusions, the final sample comprised 2,814 parents aged 60–91, including 2,029 solely cared for by sons and 785 solely cared for by daughters. Fig. 1Cleaning Data Flowchart Cleaning Data Flowchart As shown in Table 1, the dependent variables in this study are ADL, IADL, and CES-D. According to the official CHARLS report, ADL reflects an individual’s functional status, particularly their ability to perform activities of daily living and overall health condition. IADL represents reflects an individual’s functional status, particularly cognitive function and decision-making ability. CES-D assesses an individual’s psychological status, specifically the degree of depression. Higher scores on these variables indicate worse outcomes [42]. The key independent variables are the Gender of Caregiver (GC) and the Gender of Caregiver and Parents (GCP). The GC is coded as 0 to indicate that the caregiver is a daughter, and 1 to indicate that the caregiver is a son. The variable GCP is coded as 0 to represent cases where the caregiver’s gender differs from the parent’s gender, and 1 to represent cases where the caregiver’s gender same as the parent’s gender. For the control variables, based on existing studies [16, 17, 30, 45], the caregiver’s education level, marital status, household income, household size, and urban-rural status are included as control variables. Given the complexity and inconsistencies in calculating household income in CHARLS, as well as evidence from prior studies, this paper uses per capita household consumption as a proxy for household income [46]. Additionally, because age is closely associated with variation in ADL, IADL, and CES-D, the age of the parents is included as a control variable in this study [47, 48]. Table 1Explanation of variablesVariableSymbolDescription Dependent variables: Activities of daily livingADLThe ADL score ranges from 0 to 6, with higher scores indicating worse functional status, particularly their ability to perform activities of daily living and overall health condition. Instrumental activities of daily livingIADLThe IADL score ranges from 0 to 5, with higher scores indicating worse functional status, particularly cognitive function and decision-making ability. Center for Epidemiological Studies-DepressionCES-DThe CES-D score ranges from 0 to 29, with higher scores indicating worse psychological status, specifically the degree of depression.Key independent variables: Gender of caregiverGCA score of 0 indicates caregiving by a daughter, while a score of 1 indicates caregiving by a son. Gende of caregiver and parentsGCPA score of 0 indicates that the caregiver’s gender is different from the care recipient’s gender, while a score of 1 indicates that the caregiver’s gender same as the care recipient’s gender.Control variables: Education levelEduThe proportion of caregivers who have completed junior secondary education. MarriageMarThe proportion of caregivers who are married. Household consumption per capitaConAnnual per capita household consumption, measured in RMB. AgeAgeThe age of the parents. Family sizeFSThe number of families RuralRuralUrban is coded as 0, and rural is coded as 1. Table 1Explanation of variablesVariableSymbolDescription Dependent variables: Activities of daily livingADLThe ADL score ranges from 0 to 6, with higher scores indicating worse functional status, particularly their ability to perform activities of daily living and overall health condition. Instrumental activities of daily livingIADLThe IADL score ranges from 0 to 5, with higher scores indicating worse functional status, particularly cognitive function and decision-making ability. Center for Epidemiological Studies-DepressionCES-DThe CES-D score ranges from 0 to 29, with higher scores indicating worse psychological status, specifically the degree of depression.Key independent variables: Gender of caregiverGCA score of 0 indicates caregiving by a daughter, while a score of 1 indicates caregiving by a son. Gende of caregiver and parentsGCPA score of 0 indicates that the caregiver’s gender is different from the care recipient’s gender, while a score of 1 indicates that the caregiver’s gender same as the care recipient’s gender.Control variables: Education levelEduThe proportion of caregivers who have completed junior secondary education. MarriageMarThe proportion of caregivers who are married. Household consumption per capitaConAnnual per capita household consumption, measured in RMB. AgeAgeThe age of the parents. Family sizeFSThe number of families RuralRuralUrban is coded as 0, and rural is coded as 1. Explanation of variables Methodology To examine the association between caregiver gender and both the functional and psychological health of older parents, the following benchmark regression is specified (Research Objective 1):\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \:\begin{array}{c}\begin{array}{c}{\text{\:ADL\:}}_{i,t}={\beta\:}_{1}{\text{\:GC\:}}_{i,t}+{\beta\:}_{2}{\text{\:Edu\:}}_{i,t}+{\beta\:}_{3}{\text{\:Mar\:}}_{i,t}+{\beta\:}_{4}{\text{\:Con\:}}_{i,t}\\\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:+{\beta\:}_{5}{\text{\:Age\:}}_{i,t}+{\beta\:}_{6}{\text{\:FS\:}}_{i,t}+{\beta\:}_{7}{\text{\:Rural\:}}_{i,t}+{u}_{i}+{u}_{t}+{u}_{i,t}\\\:\end{array}\#\left(3.1\right)\end{array} \end{aligned}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \:\begin{array}{c}\begin{array}{c}{\text{\:IADL\:}}_{i,t}={\beta\:}_{1}{\text{\:GC\:}}_{i,t}+{\beta\:}_{2}{\text{\:Edu\:}}_{i,t}+{\beta\:}_{3}{\text{\:Mar\:}}_{i,t}+{\beta\:}_{4}{\text{\:Con\:}}_{i,t}\\\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:+{\beta\:}_{5}{\text{\:Age\:}}_{i,t}+{\beta\:}_{6}{\text{\:FS\:}}_{i,t}+{\beta\:}_{7}{\text{\:Rural\:}}_{i,t}+{u}_{i}+{u}_{t}+{u}_{i,t}\\\:\end{array}\#\left(3.2\right)\end{array} \end{aligned}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \:\begin{array}{c}\begin{array}{c}{\text{\:CESD\:}}_{i,t}={\beta\:}_{1}{\text{\:GC\:}}_{i,t}+{\beta\:}_{2}{\text{\:Edu\:}}_{i,t}+{\beta\:}_{3}{\text{\:Mar\:}}_{i,t}+{\beta\:}_{4}{\text{\:Con\:}}_{i,t}\\\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:+{\beta\:}_{5}{\text{\:Age\:}}_{i,t}+{\beta\:}_{6}{\text{\:FS\:}}_{i,t}+{\beta\:}_{7}{\text{\:Rural\:}}_{i,t}+{u}_{i}+{u}_{t}+{u}_{i,t}\\\:\end{array}\#\left(3.3\right)\end{array} \end{aligned}$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:ADL\:}}_{i,t}$$\end{document}represents the functional status, particularly their ability to perform activities of daily living and overall health condition of individual i in wave t;\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:IADL\:}}_{i,t}$$\end{document} represents the functional status, particularly cognitive function and decision-making ability of individual i in wave t༛\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:CESD\:}}_{i,t}$$\end{document} represents the psychological status, specifically the degree of depression of individual i in wave t. All of these are ordinal categorical variables, with higher scores indicating worse outcomes in each respective condition [42]. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:GC\:}}_{i,t}$$\end{document}, where 0 indicates that individual i is cared for by a daughter at wave t, and 1 indicates that individual i is cared for by a son at wave t. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:Edu\:}}_{i,t}$$\end{document} represents the caregiver’s education level, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:Mar\:}}_{i,t}$$\end{document} represents the caregiver’s marital status, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:Con\:}}_{i,t}$$\end{document} is the annual per capita household consumption, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:Age\:}}_{i,t}$$\end{document} is the age of the care recipient, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:FS\:}}_{i,t}$$\end{document} is the number of household members, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:Rural\:}}_{i,t}$$\end{document} represents the urban-rural status [16, 17, 30, 45]. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{u}_{i}$$\end{document} means individual effect, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{u}_{t}$$\end{document} means time effect, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\epsilon\:}_{i,t}$$\end{document} means error term. Given that both the independent and dependent variables consist of numerous categorical variables, a Logit model is used for estimation [49]. Additionally, for robustness checks, the propensity score matching (PSM) kernel matching method is applied, followed by re-estimating the same equation using the Logit model [50]. Caregiver type (care provided by sons vs. daughters) was treated as the binary treatment variable, and health outcomes (ADL, IADL, and life satisfaction) were specified as dependent variables. The propensity score was estimated with a logit model including parental age, marital status, educational attainment, household per capita consumption, household size, and urban–rural residence as covariates. These covariates were selected because they are widely recognized as influencing both caregiving arrangements and health outcomes [16, 17, 30, 45]. Kernel matching with common support was applied to construct comparable treatment and control groups, and the average treatment effect (ATE) was calculated [50]. Balance diagnostics were conducted, which confirmed that after matching, all covariates showed standardized mean differences below 10% and no statistically significant differences across groups, indicating that balance was achieved. To verify robustness, nearest-neighbor and radius matching were also tested and yielded substantively consistent results. To examine how the gender match between the caregiver and the care recipients (same-gender vs. different-gender relationships) is associated with both the functional and psychological health of older parents, the following benchmark regression is constructed (Research Objective 2):\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \:\begin{array}{c}\begin{array}{c}{\text{\:ADL\:}}_{i,t}={\beta\:}_{1}{\text{\:GCP\:}}_{i,t}+{\beta\:}_{2}{\text{\:Edu\:}}_{i,t}+{\beta\:}_{3}{\text{\:Mar\:}}_{i,t}+{\beta\:}_{4}{\text{\:Con\:}}_{i,t}\\\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:+{\beta\:}_{5}{\text{\:Age\:}}_{i,t}+{\beta\:}_{6}{\text{\:FS\:}}_{i,t}+{\beta\:}_{7}{\text{\:Rural\:}}_{i,t}+{u}_{i}+{u}_{t}+{\epsilon}_{i,t}\\\:\end{array}\#\left(3.4\right)\end{array} \end{aligned}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \:\begin{array}{c}\begin{array}{c}{\text{\:IADL\:}}_{i,t}={\beta\:}_{1}{\text{\:GCP\:}}_{i,t}+{\beta\:}_{2}{\text{\:Edu\:}}_{i,t}+{\beta\:}_{3}{\text{\:Mar\:}}_{i,t}+{\beta\:}_{4}{\text{\:Con\:}}_{i,t}\\\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:+{\beta\:}_{5}{\text{\:Age\:}}_{i,t}+{\beta\:}_{6}{\text{\:FS\:}}_{i,t}+{\beta\:}_{7}{\text{\:Rural\:}}_{i,t}+{u}_{i}+{u}_{t}+{\epsilon}_{i,t}\\\:\end{array}\#\left(3.5\right)\end{array} \end{aligned}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \:\begin{array}{c}\begin{array}{c}{\text{\:ADL\:}}_{i,t}={\beta\:}_{1}{\text{\:GCP\:}}_{i,t}+{\beta\:}_{2}{\text{\:Edu\:}}_{i,t}+{\beta\:}_{3}{\text{\:Mar\:}}_{i,t}+{\beta\:}_{4}{\text{\:Con\:}}_{i,t}\\\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:+{\beta\:}_{5}{\text{\:Age\:}}_{i,t}+{\beta\:}_{6}{\text{\:FS\:}}_{i,t}+{\beta\:}_{7}{\text{\:Rural\:}}_{i,t}+{u}_{i}+{u}_{t}+{\epsilon}_{i,t}\\\:\end{array}\#\left(3.6\right)\end{array} \end{aligned}$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\text{\:GCP\:}}_{i,t}$$\end{document}s a binary variable, with 0 indicating that at wave t, the caregiver’s gender differs from the gender of care recipient i, and 1 indicating that the caregiver’s gender matches the gender of care recipient i. The other variables are the same as those previously described. Similarly, given that both the independent and dependent variables consist of numerous categorical variables, a Logit model is used for estimation [49]. Additionally, for robustness checks, a PSM kernel matching method is applied, followed by re-estimating the equation using the Logit model [50]. Result Table 2 shows that among the 2,814 observations, 2,029 care recipients were attended to by sons, while 785 were cared for by daughters. The average age of care recipients was approximately 67 years, with about 70% living in rural areas, underscoring the necessity of urban-rural heterogeneity analysis. ADL, IADL, and CES-D scores were higher for care recipients cared for by daughters, indicating poorer corresponding health outcomes. Over half of the caregivers had not completed junior high school. This reflects a generally low level of educational attainment among both sons and daughters. Marital status showed minimal variation, with approximately 84% of caregivers being married. Household size was also similar, averaging around three members, consistent with China’s family planning policies [51]. Additionally, households where daughters acted as caregivers reported significantly higher consumption levels than those with sons, reflecting women’s growing financial independence and earning capacity [20, 21]. Table 2Descriptive statisticsFull sampleBy care giver’s genderSonDaughterN = 2814N = 2,029N = 785MMM(SD)(SD)(SD)ADL1.3987(1.674)1.344012(1.658443)1.540127(1.707944)IADL1.9033(1.431)1.865944(1.434214)2(1.420513)CES-D12.2271(7.110)12.02267(7.024505)12.75541(7.304514)GC0.7210(0.449)1(0)0(0)GCP0.4058(0.491)0.2789552(0.448596)0.733758(0.4422741)Edu0.3890(0.477)0.380828(0.4757987)0.4100637(0.4814633)Mar0.8406(0.353)0.8403154(0.3536684)0.8414437(0.3497337)Con1.38e + 04(14415.739)13265.91(14102.38)15249.96(15111.35)Age67.8394(10.196)68.38886(10.04757)66.41911(10.44306)FS3.1063(1.613)3.173484(1.643668)2.932484(1.519189)Rural0.6997(0.458)0.7289305(0.4446215)0.6242038(0.4846366) Table 2Descriptive statisticsFull sampleBy care giver’s genderSonDaughterN = 2814N = 2,029N = 785MMM(SD)(SD)(SD)ADL1.3987(1.674)1.344012(1.658443)1.540127(1.707944)IADL1.9033(1.431)1.865944(1.434214)2(1.420513)CES-D12.2271(7.110)12.02267(7.024505)12.75541(7.304514)GC0.7210(0.449)1(0)0(0)GCP0.4058(0.491)0.2789552(0.448596)0.733758(0.4422741)Edu0.3890(0.477)0.380828(0.4757987)0.4100637(0.4814633)Mar0.8406(0.353)0.8403154(0.3536684)0.8414437(0.3497337)Con1.38e + 04(14415.739)13265.91(14102.38)15249.96(15111.35)Age67.8394(10.196)68.38886(10.04757)66.41911(10.44306)FS3.1063(1.613)3.173484(1.643668)2.932484(1.519189)Rural0.6997(0.458)0.7289305(0.4446215)0.6242038(0.4846366) Descriptive statistics 1.3987 (1.674) 1.344012 (1.658443) 1.540127 (1.707944) 1.9033 (1.431) 1.865944 (1.434214) 2 (1.420513) 12.2271 (7.110) 12.02267 (7.024505) 12.75541 (7.304514) 0.7210 (0.449) 1 (0) 0 (0) 0.4058 (0.491) 0.2789552 (0.448596) 0.733758 (0.4422741) 0.3890 (0.477) 0.380828 (0.4757987) 0.4100637 (0.4814633) 0.8406 (0.353) 0.8403154 (0.3536684) 0.8414437 (0.3497337) 1.38e + 04 (14415.739) 13265.91 (14102.38) 15249.96 (15111.35) 67.8394 (10.196) 68.38886 (10.04757) 66.41911 (10.44306) 3.1063 (1.613) 3.173484 (1.643668) 2.932484 (1.519189) 0.6997 (0.458) 0.7289305 (0.4446215) 0.6242038 (0.4846366) Table 3 reports the logit estimates and PSM robustness checks for Eq. 3.1, 3.2, and 3.3, examining the associations between caregiver gender and the functional and psychological health of older parents (Research Objective 1). Both the first column (logit) and the second column (robustness check) indicate that parents cared for by sons have a 0.289 lower probability of experiencing worse ADL compared to those cared for by daughters. Similarly, the third column (logit) and the fourth column (robustness check) reveal that parents cared for by sons have a 0.311 lower probability of experiencing worse IADL than those cared for by daughters. These findings suggest that functional health is better for parents under the care of sons. However, the fifth column (logit) and the sixth column (robustness check) show that parents cared for by sons have a 0.474 higher probability of experiencing worse CES-D scores compared to those cared for by daughters. This indicates that psychological health is better for parents when cared for by daughters. Additionally, an increase in household size significantly reduces the probability of worsening psychological health for care recipients, with a coefficient of 0.113. Table 3The association between caregiver gender and functional and psychological health of older parentsLogit modelPSM robustLogit modelPSM robustLogit modelPSM robustADLADLIADLIADLCES-DCES-DGC−0.289***−0.293***−0.311**−0.304**0.474**0.489**(0.087)(0.088)(0.122)(0.122)(0.222)(0.222)Edu0.0300.034−0.121−0.109−0.152−0.124(0.082)(0.083)(0.110)(0.110)(0.213)(0.214)Mar−0.400***−0.412***−0.012−0.006−0.313−0.303(0.113)(0.113)(0.148)(0.148)(0.348)(0.349)Con0.0000.0000.000**0.000**−0.000**−0.000**(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)Age0.012***0.013***−0.009*−0.009*−0.034***−0.035***(0.004)(0.004)(0.005)(0.005)(0.011)(0.011)FS−0.0000.0030.056*0.053−0.113**−0.142**(0.024)(0.025)(0.033)(0.035)(0.055)(0.060)Rural−0.053−0.042−0.151−0.1490.373*0.332(0.086)(0.086)(0.118)(0.118)(0.218)(0.219)_cons−0.029−0.1132.341***2.326***6.068***6.249***(0.314)(0.323)(0.403)(0.415)(0.886)(0.925) N 281427782814277828142778* p < 0.1, ** p < 0.05, *** p < 0.01 Table 3The association between caregiver gender and functional and psychological health of older parentsLogit modelPSM robustLogit modelPSM robustLogit modelPSM robustADLADLIADLIADLCES-DCES-DGC−0.289***−0.293***−0.311**−0.304**0.474**0.489**(0.087)(0.088)(0.122)(0.122)(0.222)(0.222)Edu0.0300.034−0.121−0.109−0.152−0.124(0.082)(0.083)(0.110)(0.110)(0.213)(0.214)Mar−0.400***−0.412***−0.012−0.006−0.313−0.303(0.113)(0.113)(0.148)(0.148)(0.348)(0.349)Con0.0000.0000.000**0.000**−0.000**−0.000**(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)Age0.012***0.013***−0.009*−0.009*−0.034***−0.035***(0.004)(0.004)(0.005)(0.005)(0.011)(0.011)FS−0.0000.0030.056*0.053−0.113**−0.142**(0.024)(0.025)(0.033)(0.035)(0.055)(0.060)Rural−0.053−0.042−0.151−0.1490.373*0.332(0.086)(0.086)(0.118)(0.118)(0.218)(0.219)_cons−0.029−0.1132.341***2.326***6.068***6.249***(0.314)(0.323)(0.403)(0.415)(0.886)(0.925) N 281427782814277828142778* p < 0.1, ** p < 0.05, *** p < 0.01 The association between caregiver gender and functional and psychological health of older parents * p < 0.1, ** p < 0.05, *** p < 0.01 Table 4 reports the logit estimates and PSM robustness checks for Eq. 3.4, 3.5, and 3.6. These models evaluate whether gender matching between caregivers and care recipients is associated with differences in functional and psychological health (Research Objective 2). The results in columns 1 through 4 indicate that the gender match (GCP) shows no significant association with ADL or IADL. This suggests that whether the caregiver and care recipient are of the same gender is not associated with differences in the functional health of care recipients. However, the results in column 5 (logit) and column 6 (robustness check) reveal that when the caregiver and care recipient are of the same gender, the probability of a decline in the care recipient’s psychological health is significantly reduced by 0.397. Additionally, consistent with the findings in Table 3, an increase in household size significantly lowers the probability of a decline in the care recipient’s psychological health, with a coefficient of 0.105. Table 4Associations between caregiver–care recipient gender match and functional and psychological healthLogit modelPSM robustLogit modelPSM robustLogit modelPSM robustADLADLIADLIADLCES-DCES-DGCP0.0960.0930.0920.092−0.397**−0.423**(0.078)(0.078)(0.103)(0.104)(0.212)(0.213)Edu0.0280.013−0.124−0.116−0.154−0.125(0.082)(0.082)(0.110)(0.110)(0.213)(0.214)Mar−0.390***−0.403***−0.0010.002−0.315−0.297(0.113)(0.113)(0.147)(0.147)(0.348)(0.348)Con0.0000.0000.000**0.000**−0.000**−0.000**(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)Age0.011***0.011***−0.011**−0.011**−0.032***−0.031***(0.004)(0.004)(0.005)(0.005)(0.011)(0.011)FS−0.006−0.0060.0500.052−0.105*−0.108**(0.024)(0.024)(0.032)(0.032)(0.055)(0.055)Rural−0.080−0.080−0.180−0.1840.426**0.402*(0.085)(0.086)(0.117)(0.118)(0.216)(0.218)_cons−0.153−0.1342.200***2.198***6.360***6.327***(0.313)(0.314)(0.406)(0.406)(0.884)(0.882) N 281427912814279128142791* p < 0.1, ** p < 0.05, *** p < 0.01 Table 4Associations between caregiver–care recipient gender match and functional and psychological healthLogit modelPSM robustLogit modelPSM robustLogit modelPSM robustADLADLIADLIADLCES-DCES-DGCP0.0960.0930.0920.092−0.397**−0.423**(0.078)(0.078)(0.103)(0.104)(0.212)(0.213)Edu0.0280.013−0.124−0.116−0.154−0.125(0.082)(0.082)(0.110)(0.110)(0.213)(0.214)Mar−0.390***−0.403***−0.0010.002−0.315−0.297(0.113)(0.113)(0.147)(0.147)(0.348)(0.348)Con0.0000.0000.000**0.000**−0.000**−0.000**(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)Age0.011***0.011***−0.011**−0.011**−0.032***−0.031***(0.004)(0.004)(0.005)(0.005)(0.011)(0.011)FS−0.006−0.0060.0500.052−0.105*−0.108**(0.024)(0.024)(0.032)(0.032)(0.055)(0.055)Rural−0.080−0.080−0.180−0.1840.426**0.402*(0.085)(0.086)(0.117)(0.118)(0.216)(0.218)_cons−0.153−0.1342.200***2.198***6.360***6.327***(0.313)(0.314)(0.406)(0.406)(0.884)(0.882) N 281427912814279128142791* p < 0.1, ** p < 0.05, *** p < 0.01 Associations between caregiver–care recipient gender match and functional and psychological health * p < 0.1, ** p < 0.05, *** p < 0.01 Table 5 reports the income heterogeneity results on the associations between caregiver gender differences and parents’ functional and psychological health. The first column shows the estimates for high-income families, while the second column focuses on low-income families. Panel A, B, and C summarize how caregiver gender differences are related to parents’ ADL, IADL, and CES-D, respectively. Income groups were classified using the median split method [52]. The results reveal that associations between caregiver gender and parents’ health are statistically significant only in low-income families, with no significant associations observed in high-income families. Specifically, for low-income families, caregiving by sons significantly reduces the likelihood of a decline in parents’ functional health, with coefficients of 0.4927 for IADL and 0.3494 for ADL. Moreover, caregiving by sons also significantly lowers the probability of a decline in parents’ psychological health (CESD), with a coefficient of 0.181. In summary, caregiving by sons in low-income families significantly reduces the likelihood of a decline in both functional and psychological health among parents. Table 5Income heterogeneity resultsVariableHigh IncomeLow IncomePanel A:ADL−0.0745−0.4927***(−0.7089)(−4.4191)Observations14061408Panel B:IADL−0.1119−0.3494***(−1.1076)(−3.0968)Observations14061408Panel 3:CESD−0.1338−0.1810*(−1.2859)(−1.6645)Observations14061408t statistics in parentheses* p < 0.1, ** p < 0.05, *** p < 0.01 Table 5Income heterogeneity resultsVariableHigh IncomeLow IncomePanel A:ADL−0.0745−0.4927***(−0.7089)(−4.4191)Observations14061408Panel B:IADL−0.1119−0.3494***(−1.1076)(−3.0968)Observations14061408Panel 3:CESD−0.1338−0.1810*(−1.2859)(−1.6645)Observations14061408t statistics in parentheses* p < 0.1, ** p < 0.05, *** p < 0.01 Income heterogeneity results t statistics in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01 Table 6 reports the urban–rural heterogeneity in the associations between caregiver gender differences and parents’ functional and psychological health. The first column reports estimate for urban residents, while the second column focuses on rural residents. In Panel A, the caregiving by sons is associated with a significant decline in parental ADL scores in both urban (−0.2110, p < 0.05) and rural areas (−0.3312, p < 0.05). The larger negative coefficient in rural areas suggests that caregiving by sons leads to a higher likelihood of improved ADL outcomes for rural parents compared to their urban counterparts. In Panel B, the probability of improved IADL outcomes shows a significant association with caregiving by sons in urban areas (−0.2122, p < 0.05), while no statistically significant relationship is observed in rural areas (−0.1898, p > 0.1). This indicates that sons’ caregiving is more effective in enhancing instrumental functional health in urban contexts. In Panel C, caregiving by sons shows no significant correlation with improvements in CES-D scores in urban areas (−0.0806, p > 0.1). However, in rural areas, caregiving by sons is associated with a significant reduction in the probability of poorer psychological health (−0.2610, p < 0.05), highlighting its importance in mitigating mental health challenges in these regions. Table 6Urban-rural heterogeneity resultsVariableUrbanRuralPanel A:ADL−0.2110**−0.3312**(−2.2208)(−2.5313)Observations1969845Panel B:IADL−0.2122**−0.1898(−2.3112)(−1.4625)Observations1969845Panel C:CESD−0.0806−0.2610**(−0.8888)(−1.9859)Observations1969845t statistics in parentheses* p < 0.1, ** p < 0.05, *** p < 0.01 Table 6Urban-rural heterogeneity resultsVariableUrbanRuralPanel A:ADL−0.2110**−0.3312**(−2.2208)(−2.5313)Observations1969845Panel B:IADL−0.2122**−0.1898(−2.3112)(−1.4625)Observations1969845Panel C:CESD−0.0806−0.2610**(−0.8888)(−1.9859)Observations1969845t statistics in parentheses* p < 0.1, ** p < 0.05, *** p < 0.01 Urban-rural heterogeneity results t statistics in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01 Table 7 reports the urban–rural and income-based heterogeneity in the associations between caregiver gender differences and parental functional and psychological health. For ADL (Panel A), caregiving by sons is associated with a significant reduction in the likelihood of functional limitations among low-income groups in both rural (−0.4322, p < 0.01) and urban areas (−0.5993, p < 0.01). This indicates that sons’ caregiving provides greater health benefits for lower-income parents, likely because these families have fewer alternatives to rely on. However, it is not statistically significant for high-income groups in either rural (−0.0537, p > 0.1) or urban (−0.1190, p > 0.1) areas, which indicate that higher-income parents rely on alternative resources or support systems that reduce the observable association with sons’ caregiving. For IADL (Panel B), sons’ caregiving significantly improves functional outcomes for rural low-income (−0.3163, p < 0.05) and urban low-income parents (−0.4332, p < 0.05). However, it is not statistically significant for high-income groups in either rural (−0.1480, p > 0.1) or urban (0.0124, p > 0.1) areas. This again highlights the crucial role of sons’ caregiving in lower-income families where formal care resources may be limited. For CES-D (Panel C), an interesting pattern emerges. In rural high-income families, caregiving by sons is associated with a significantly increased likelihood of psychological distress (0.7859, p < 0.05). Similarly, in urban high-income families, sons’ caregiving is marginally associated with increased psychological distress (0.7836, p < 0.1). In contrast, caregiving by sons does not have a statistically significant associated with CES-D scores for low-income families in either rural (−0.3773, p > 0.1) or urban (0.1316, p > 0.1) areas. This suggests that in high-income families, the caregiving provided by sons may lead to role conflicts or unmet expectations, resulting in psychological strain. Table 7Urban-rural and income heterogeneityVariableRural Low IncomeRural High IncomeUrban Low IncomeUrban High IncomePanel A:ADL−0.4322***−0.0537−0.5993***−0.1190(−3.0196)(−0.4150)(−3.2160)(−0.6468)Observations985984423422Panel B:IADL−0.3163**−0.1480−0.4332**0.0124(−2.2701)(−1.1832)(−2.1646)(0.0701)Observations985984423422Panel C:CESD−0.37730.7859**0.13160.7836*(−0.6704)(2.0751)(0.2235)(1.9108)Observations985984423422t statistics in parentheses* p < 0.1, ** p < 0.05, *** p < 0.01 Table 7Urban-rural and income heterogeneityVariableRural Low IncomeRural High IncomeUrban Low IncomeUrban High IncomePanel A:ADL−0.4322***−0.0537−0.5993***−0.1190(−3.0196)(−0.4150)(−3.2160)(−0.6468)Observations985984423422Panel B:IADL−0.3163**−0.1480−0.4332**0.0124(−2.2701)(−1.1832)(−2.1646)(0.0701)Observations985984423422Panel C:CESD−0.37730.7859**0.13160.7836*(−0.6704)(2.0751)(0.2235)(1.9108)Observations985984423422t statistics in parentheses* p < 0.1, ** p < 0.05, *** p < 0.01 Urban-rural and income heterogeneity t statistics in parentheses * p < 0.1, ** p < 0.05, *** p < 0.01 Discussion For Research Objective 1, the findings show that parental functional health is better when care is provided by sons, whereas psychological health is more favorable when care is provided by daughters. This result aligns with existing studies [16–18, 27, 30], and is consistent with gender role theory which suggests that societal expectations for men primarily center on economic support, while women are more inclined to provide emotional care [11]. Therefore, sons may contribute more directly to functional health outcomes because improvements in ADL and IADL often require access to material resources, including medical care, assistive devices, and household modifications [16–19]. This reflects the persistence of traditional expectations. Sons take greater responsibility for material support, particularly in families that still value their role in providing economic security. As gender role theory emphasizes, men are traditionally expected to act as financial providers, and this role positions them to mobilize economic resources that address parents’ functional limitations [11]. In contrast, daughters’ involvement shows a stronger association with psychological well-being. This is because depressive symptoms and life satisfaction are more strongly shaped by emotional support, companionship, and the quality of interpersonal interaction. Empirical evidence shows that social support and affective exchanges are protective against depression and loneliness in older adults [16–19], which aligns with the caregiving roles more frequently undertaken by daughters. These results also show that caregiving in China remains shaped by cultural norms that separate material and emotional responsibilities. Daughters’ frequent contact and affective involvement strengthen family bonds and provide psychological comfort, which are essential for older adults’ mental health. Overall, these findings support gender role theory while revealing its persistence in modern family structures. Despite social and economic change, gendered caregiving expectations continue to influence both the form and outcomes of intergenerational care. For Research Objective 2, gender matching between caregivers and care recipients is not significantly associated with functional health. However, it shows a positive association with psychological health. These findings are consistent with the homophily hypothesis, which highlights that similarity in sociodemographic traits strengthens social bonds and facilitates communication. Same-gender caregiver–parent dyads may thus enhance psychological health by fostering empathy, mutual understanding, and emotionally resonant interactions [13, 41]. In the context of family caregiving, emotional support often depends on shared communication patterns and mutual recognition of needs [13, 41]. When caregivers and parents share the same gender, they are more likely to understand each other’s emotional cues and stress responses, which reduces psychological strain for both sides. This emotional congruence also reflects broader socialization patterns in which men and women develop distinct ways of expressing care and seeking comfort. For older parents, especially in traditional societies, receiving care from a same-gender child may feel more natural and less intrusive, reinforcing dignity and autonomy. In contrast, improvements in functional health are more closely tied to material resources and professional caregiving skills. These outcomes are less dependent on interpersonal similarity between caregivers and care recipients. Functional assistance, such as helping with mobility, hygiene, or medical visits, relies on physical ability and economic support rather than emotional connection. Therefore, while same-gender pairing enhances psychological well-being through relational comfort, it has limited relevance to functional recovery, which depends on external resources and institutional support. This distinction explains why same-gender matching strengthens psychological but not functional outcomes [53]. It also underscore that emotional and functional dimensions of older care are governed by different perspectives—one rooted in social relationships, the other in economic and practical capacities [53]. For Research Objective 3, the heterogeneity analysis indicates significant associations between caregiver gender differences and both functional and psychological health, particularly in low-income households. In these families, limited resources heighten parental dependence on children, and sons, as traditional economic providers, are better positioned to mobilize financial and material support, thereby improving functional health outcomes [19]. This pattern reflects the persistence of resource-based caregiving in constrained environments, where financial assistance directly translates into better access to nutrition, medical care, and assistive devices [19]. In such contexts, caregiving operates less as an emotional exchange and more as an economic function within the household, reinforcing the traditional role of sons as providers of material stability. By contrast, in high-income households, where basic material needs are already met, the marginal contribution of sons’ economic support is reduced. Instead, daughters’ emotional and relational support becomes more salient, contributing to improvements in psychological health [30]. In these families, the meaning of care shifts from financial provision to emotional presence. Parents in wealthier settings tend to value companionship, communication, and emotional understanding more than direct material assistance [30]. Daughters’ caregiving, often characterized by frequent contact and empathetic interaction, fulfills these psychological expectations more effectively [13, 41]. This suggests that emotional capital becomes a more critical form of intergenerational resource when financial security is already ensured. This pattern suggests that income level is the central determinant of whether caregiver gender differences translate into measurable health benefits. In essence, economic inequality shapes the very function of caregiving. In poorer households, care is a substitute for missing markets and public services. While in richer households, it is a medium of emotional exchange. This duality mirrors the broader social transition in aging societies—from survival-oriented to well-being-oriented older care. Recognizing this shift is important for policy design, as it implies that interventions to improve older adults’ health must address both the economic and emotional dimensions of family support. The magnitude of these associations varies across urban and rural settings. Urban families have access to more developed healthcare infrastructure and services, which may reinforce the observed association between sons’ economic contributions and parents’ functional outcomes [27]. In rural areas, however, despite stronger dependence on children due to limited institutional alternatives, the effectiveness of caregiving is constrained by resource scarcity and weaker service availability [14, 16, 27]. These contextual contrasts help explain why functional health improvements are more pronounced in urban settings, whereas psychological health benefits associated with daughters’ care emerge more clearly in resource-rich households across both urban and rural areas. Further analysis of the interaction between income and urban–rural context shows patterns consistent with the separate heterogeneity analyses. Together, these findings highlight that sons’ caregiving primarily benefits older adults’ functional health in resource-constrained households, while in resource-rich contexts it may generate psychological tensions that undermine well-being. Limitation and future research Although the analyses were carefully conducted and robustness checks were performed to ensure the reliability of the findings, some limitations remain and should be acknowledged. First, because this study focused specifically on gender differences in caregiving, the sample was restricted to parents solely cared for by sons or solely by daughters. Parents receiving joint caregiving from multiple children were excluded, which may limit the generalizability of the findings. Future research could incorporate mixed-care arrangements to capture a broader spectrum of caregiving patterns. Second, although this study extends the literature by examining both functional and psychological health, the psychological dimension was measured using a general depression index. More nuanced indicators—such as anxiety, cognitive status, or subjective well-being—could be employed in future analyses to provide a more comprehensive assessment. Third, the methodology employed in this study was chosen to fit the characteristics of the pooled cross-sectional data. However, if long-term panel data become available in the future, approaches such as two-way fixed effects or dynamic models could be applied to strengthen causal inference by accounting for both time and individual effects. After outlining these limitations, the next section turns to a conclude the findings and their implications for policy. Conclusion China’s rapidly aging population and evolving socio-economic landscape, particularly shifting gender roles, are reshaping the traditional belief that raising sons ensures old-age support. However, existing research provides limited evidence on the associations between caregiver gender and parents’ functional and psychological health. with most studies overlooking differences based on caregiver-care recipient gender match. Based on this background, this study aims to examine (1) how caregiver gender is associated with the functional and psychological health of care recipients, and (2) how gender match between caregivers and care recipients relates to these health outcomes. Using four waves of CLHLS data, a Logit model is employed for estimation, with robustness checks conducted through the PSM method. This study makes two key contributions. Theoretically, this study integrates gender role theory, social exchange theory, and the homophily hypothesis to build a framework for analyzing how caregiver gender relates to functional and psychological health. Empirically, it highlights heterogeneous patterns across income groups and urban–rural contexts, providing new directions for comparative research. The empirical results can be directly mapped to the three research objectives. For RO1, parents cared for by sons exhibit significantly better functional health (ADL and IADL), whereas parents cared for by daughters experience better psychological health. For RO2, caregiver–parent gender matching shows no significant association with functional outcomes. However, it is associated with improved psychological well-being, suggesting that same-gender dyads facilitate more effective emotional support. For RO3, the associations of caregiver gender with health outcomes are heterogeneous: better functional health is observed for parents in low-income households when cared for by sons, whereas less favorable psychological outcomes are observed mainly in high-income households. These income-based patterns are further shaped by the urban–rural context, with stronger effects observed in urban families than in rural ones. Based on the findings of this study, the following policy recommendations are proposed. First, since caregiving by sons contributes more to parents’ functional health, the government can reinforce this effect through targeted financial and tax support. Reducing household tax burdens and providing care subsidies for families with functionally impaired older adults would allow sons to offer practical assistance more effectively. At the same time, as caregiving by daughters is more closely related to psychological well-being, policies should expand flexible work arrangements and caregiving leave for women to help them maintain emotional engagement while balancing work and family responsibilities. Second, because same-gender caregiver–parent relationships are associated with better psychological outcomes, the government should encourage community programs that offer free counseling and gender-sensitive communication workshops for opposite-gender caregivers to improve empathy and understanding. In institutional care settings, assigning caregivers and older adults of the same gender could also strengthen emotional connection and improve psychological outcomes. Third, additional financial support should be directed to low-income and rural households to ensure access to essential medical services and improve functional health. This could include expanding insurance coverage and increasing benefit levels to reduce barriers to healthcare access. In contrast, in urban and high-income families, where emotional needs are more pronounced, policies should focus on developing community centers, mental health counseling, and social participation programs to enhance psychological well-being.