Measuring supply–demand coupling coordination in China’s older adult care services: spatiotemporal evolution, spatial spillover effects, and driving factors based on the spatial Durbin model Authors: Lingxiao Guo, Yiwen Wei, Xinru Zhang, Hua Wei, Qunshan Tao, Kunyu Chen Published: 2026 ABSTRACT AbstractBackgroundThe coordinated development of older adult care service resource supply and the needs of older adults is not only pivotal for advancing the construction of China’s older adult care service system and improving the social security framework, but is also essential for the effective implementation of the national strategy for actively addressing population aging. However, a structural imbalance currently exists between older adult care service supply and the needs of older adults, with pronounced developmental disparities across the eastern, central, and western regions, underscoring an urgent need for in-depth analysis of the evolutionary patterns of supply–demand coordination and its spatial transmission mechanisms from a systems coupling perspective. This study theoretically transcends the conventional paradigm of unidirectional supply-side or demand-side analysis by constructing an analytical framework for supply–demand coordination in older adult care services from a systems coupling perspective, thereby extending the application boundaries of coupling coordination theory to the social services domain; empirically, it employs multiple spatial econometric methods to systematically reveal the spatiotemporal evolutionary patterns, spatial agglomeration characteristics, and regionally heterogeneous driving mechanisms of the supply–demand coupling coordination degree, providing a quantitative basis for the scientific formulation of differentiated older adult care policies.MethodsBased on panel data from 31 provinces and municipalities in China spanning 2014 to 2023, the entropy weight method was applied to measure supply and demand levels, the coupling coordination degree model was employed to evaluate the level of coordinated development, Moran’s Index was used to characterize spatial agglomeration features, and a spatiotemporal dual fixed-effects Spatial Durbin Model was constructed to analyze the direct, indirect, and total effects of the influencing factors.ResultsThe findings indicate that the supply–demand coupling coordination degree of older adult care service resources in China has followed a steadily upward yet slowly progressing trend, remaining at an overall low level of coordination. Significant regional differentiation is evident, manifesting as a spatial pattern of “high in the east, low in the west,” with this divergence continuously intensifying. Significant spatial autocorrelation has given rise to relatively stable high–high and low–low clustering zones, reflecting that the structural imbalance between supply and demand constitutes the fundamental constraint on coordinated development. Multi-dimensional driving factors — including population aging, economic development, and fiscal investment — exhibit marked regional heterogeneity in their effects on the coupling coordination degree, with considerable differences in development dynamics across regions.ConclusionsOlder adult care service resources display a structural imbalance in which supply quality lags behind the upgrading of demand. The eastern region maintains a relative lead in coupling coordination, the central region is in a phase of steady catch-up, and the western region continues to face fundamental bottlenecks. Each region should, in accordance with its own stage of development and regional characteristics, pursue differentiated supply–demand coordination pathways by optimizing fiscal investment structures, establishing regional vertical cooperation mechanisms, and implementing targeted reforms in resource allocation, so as to effectively advance the dynamic equilibrium between older adult care service supply and the needs of older adults.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12913-026-14913-4.Keywords: Older adult care services, Supply–demand coupling, Spatial analysis, Influencing factors, Regional heterogeneity Introduction Population aging represents one of the most profound social challenges confronting the world in the twenty-first century. According to the United Nations World Population Prospects 2024, the global population aged 65 and above is projected to reach 2.2 billion by the late 2070s, at which point the number of older persons will, for the first time, surpass that of individuals under the age of 18; by 2070, the global aging rate is expected to exceed 20%, and the majority of countries will enter what has been termed a “super-aged society.” This trend has already prompted profound social policy responses in numerous developed countries [1]. Japan was the first to implement a long-term care insurance system in 2000, incorporating both home-based and institutional care into its mandatory social insurance framework and delivering tiered care services to older adults through standardized needs assessments [2]. Germany had already established long-term care insurance as the fifth pillar of its social security system as early as 1995, constructing a differentiated subsidy mechanism centered on a five-level care classification [3]. The Nordic countries have long maintained universally accessible home-based and institutional care systems financed through public resources to ensure service accessibility for older adults [4]. Nevertheless, even these countries — which were among the first to build comprehensive institutional frameworks — continue to confront persistent structural challenges: care worker shortages are widespread, with Japan’s Ministry of Health, Labour and Welfare projecting a shortfall of more than 300,000 care workers before 2025, while imbalances in the urban–rural distribution of services have become increasingly prominent at the local government level across Nordic countries. This demonstrates that, even in countries with relatively well-developed institutional frameworks, the supply–demand coordination of older adult care service resources remains an ongoing challenge. How to achieve dynamic coordination among supply quality, resource accessibility, and regional equity against a backdrop of continuously growing demand constitutes a fundamental question shared by older adult care service policies worldwide, and provides an important international reference point for examining China’s situation from a comparative perspective. Against this backdrop, China’s aging trajectory warrants particular attention, given its profound implications for the social security system, the provision of public services, and national economic development [5]. According to data from the National Bureau of Statistics, by the end of 2023, China’s population aged 60 and above had reached 297 million, accounting for 21.1% of the total national population, while the population aged 65 and above stood at 186 million, representing 13.2%. Concurrently, the number of older adults with disabilities or partial functional impairment has exceeded 40 million, with this figure continuing to grow at an accelerating pace. The challenges facing China’s older adult care service system are distinctive in both their scale and structural complexity. From a rural–urban structural perspective, data from the Seventh National Population Census reveal that the aging rate among the rural population aged 60 and above reached 23.81%, significantly higher than the 15.82% recorded in urban areas. This urban–rural reversal is particularly pronounced, rooted fundamentally in large-scale population mobility — in 2020, China’s floating population reached 376 million, with large numbers of working-age adults continuing to concentrate in urban areas, thereby intensifying aging pressures in rural regions and giving rise to a structural contradiction in which weak rural older adult care service supply coexists with robust aging-related demand [6]. From a household structural perspective, census data show that China’s average household size shrank from 3.10 persons in 2010 to 2.62 persons in 2020. The trends toward smaller households and growing numbers of older adults living alone have accelerated the erosion of traditional family-based care functions, making the tension between expanding demand for socialized care services and insufficient effective supply increasingly pronounced [7]. Against this challenging backdrop, the structural contradictions within China’s older adult care service system have become increasingly apparent: on the supply side, multiple obstacles persist, including imbalanced resource allocation, a severe shortage of professional care workers (statistics indicate that there are only approximately 500,000 practitioners in the older adult care workforce nationwide, against an actual demand exceeding 6 million — an enormous shortfall), and insufficient depth of integration between medical and care services [8]. On the demand side, the needs of older adults exhibit multi-tiered and diversified characteristics, owing to the large scale of the older adult population, marked heterogeneity in health status, and the rapidly growing number of individuals with disabilities or cognitive impairment. This structural misalignment between supply and demand has severely constrained improvements in resource allocation efficiency [9]. Particularly noteworthy is the significant disparity in older adult care service development across the eastern, central, and western regions, which poses serious challenges to regionally coordinated development [10]. In order to advance the scientific development of older adult care undertakings, the central government has successively issued a series of important policy documents. The National Medium- and Long-Term Plan for Actively Addressing Population Aging (2021–2050), issued by the central government in 2021, provides long-term strategic guidance for responding to population aging. The 14th Five-Year Plan for National Aging-Related Services Development and the Older Adult Care Service System, released by the State Council in 2022, explicitly sets out the development goal of “building an older adult care service system in which home-based, community-based, and institutional care are mutually coordinated, and in which medical care, health preservation, and nursing care are integrated.” The Opinions on Advancing the Construction of the Basic Older Adult Care Service System, issued by the General Office of the State Council in 2023, further reinforced the equitable provision of basic older adult care. These policy documents collectively establish a comprehensive institutional framework at the strategic, planning, and implementation levels, converging on a single core objective: to achieve a dynamic equilibrium between older adult care service supply and the needs of older adults, and to build a care and welfare network that covers both urban and rural areas and accommodates multiple service tiers [11]. Unlike the policy approaches adopted by developed countries such as Japan and Germany — which emphasize safeguarding individual care entitlements through mandatory long-term care insurance and regulating service supply through market mechanisms — China’s older adult care policy places greater emphasis on government-led resource coordination and regionally equitable service provision, with narrowing the older adult care service gap between the eastern, central, and western regions incorporated as a core policy objective [12]. This institutional orientation not only reflects China’s locally grounded approach to addressing population aging but has also profoundly shaped the regional differentiation of the supply–demand landscape for older adult care service resources, constituting an important policy context for this study’s focus on supply–demand coordination at the provincial level. Domestic and international academic research has accumulated a rich body of findings on the development of older adult care service systems; however, notable differences persist in research perspectives and methodological approaches, giving rise to several research gaps that have yet to be adequately addressed. In the domain of supply-side optimization research, both domestic and international scholars have developed relatively systematic understanding across the dimensions of institutional development, workforce allocation, and the integration of medical and care services. Ouyang et al. demonstrated that the expansion of older adult care facility scale and the integration of service models can significantly enhance service accessibility and care quality through synergistic effects [13]. Drawing on social-ecological systems theory to examine the operational mechanisms of community-based home care services that integrate medical and care provision, Gao found that a grassroots healthcare integration model characterized by “bidirectional embedding” can effectively consolidate primary healthcare resources with community-based older adult care systems, offering a practical solution for addressing the fragmentation of medical-care integration resources [14]. At the level of international comparison, Tamiya et al.’s research on Japan’s long-term care insurance demonstrated that systematically designed institutional frameworks can effectively respond to shifts in demand within aging societies [15]. Studies on Germany’s tiered care system indicated that differentiated supply strategies can substantially improve resource allocation efficiency [16]. Kyomugisha et al. highlighted the pivotal role of community service networks in enhancing the accessibility of older adult care services [17]. Setia et al.’s examination of Singapore’s older adult care reforms showed that innovative supply models can effectively reshape demand structures and promote service utilization [18]. Nevertheless, the aforementioned international studies share two common limitations. First, their analytical frameworks are uniformly focused on single-country institutional design and supply-side efficiency evaluation, implicitly treating supply–demand relationships as unidirectional and linear in nature, thereby overlooking the bidirectional dynamic characteristics of supply and demand as an interacting coupled system. Second, all of these studies were conducted at the national aggregate level, lacking quantitative identification of spatial heterogeneity at the provincial or sub-national scale, and are therefore ill-suited to serve as direct methodological references for China’s context — a country characterized by vast territorial extent and highly uneven regional development. In the domain of spatial effects and supply–demand coupling research, Li et al. employed spatial econometric models to demonstrate that well-planned older adult care facility distribution not only improves the life satisfaction of older adults but also generates significant regional spillover effects [19], marking the first introduction of a spatial association perspective into older adult care service research. Subsequently, domestic scholars increasingly attended to the spatial distribution patterns of older adult care service resources: Chen et al., applying standard deviational ellipse analysis and hotspot analysis methods, investigated the evolutionary patterns of older adult care facilities in the Yangtze River Delta region and found that regional integration policies facilitated a spatial shift in the distribution of care facilities from high concentration toward broader dispersal, with particularly pronounced enabling effects in the peripheral areas of metropolitan regions [20]. Building on these foundations, domestic scholars further introduced coupling coordination theory into the analysis of supply–demand systems. Bian et al., adopting an institutionally stratified analytical framework and applying a coupling coordination evaluation model, conducted a systematic assessment of the coordinated development of resource allocation and service utilization across older adult care institutions and community-based care service organizations in 31 provinces in China, finding that the coupling coordination degree of both types of institutions exhibited a spatial pattern of “high in the east, low in the west,” with the relative development orientation shifting overall from resource-allocation-lagging to service-utilization-lagging [21]. Feng et al. measured and analyzed the temporal evolutionary characteristics of the coupling coordination between the smart health care industry and high-quality economic development, finding that the coupling coordination degree exhibited a steadily upward trajectory at the national level, while remaining at a relatively low level at the provincial scale, with inter-regional disparities constituting the primary source of overall variation; the principal obstacles affecting coupling coordination included the foundational support capacity of the smart health care industry and coordinated and shared economic development [22]. Gu et al. incorporated the supply–demand dynamics of healthcare resources and changes in the older adult population into a “four-force coupling” holistic governance analytical framework, using Hefei as a practical case study to reveal a spatial pattern in which overall coordination levels trended upward but regional disparities remained pronounced [23]. Liu et al. examined the coupling coordination between digital technology and older adult care services, finding that internet use can significantly enhance the service consumption levels of older adults through pathways including improved health outcomes, enhanced social interaction, and increased reliance on smart care services [24]. While these studies have laid an important foundation for establishing the analytical paradigm of supply–demand coupling coordination, three systematic limitations remain, collectively constituting the core research gaps that this study seeks to address. First, the bidirectional coupling mechanism between supply and demand requires further elaboration. The majority of existing studies emphasize unidirectional influence mechanisms, overlooking the bidirectional dynamic coupling relationship between supply and demand and its spatiotemporal differentiation characteristics. In practice, the expansion and upgrading of demand also exerts a significant driving effect on the improvement of the supply system: the deepening of population aging directly stimulates market demand for older adult care services, providing impetus for supply-side development [25]; the rising proportion of older adults with disabilities generates urgent demand for specialized services, which in turn promotes optimization of the supply system [26]. Second, the comprehensiveness of indicator frameworks requires expansion. Supply-side indicators in existing research are largely confined to conventional hardware dimensions such as the number of care institutions and bed capacity, while demand-side indicators seldom encompass key dimensions reflecting the structure of contemporary older adult care demand — such as digital participation and actual service utilization — making it difficult to fully capture the true state of the current supply–demand landscape, and particularly failing to reflect the profound structural changes in supply and demand driven by the rapid development of smart care services [27]. Third, the spatial spillover effects and regionally heterogeneous driving mechanisms require systematic identification. Existing studies predominantly employ standard panel data models or static spatial description methods, lacking systematic identification of the spatial dependence and spillover effects of inter-provincial supply–demand coordination. Even where regional differences are considered, analyses tend to remain at the level of broad descriptions of the eastern, central, and western regions, with insufficient quantitative characterization of the differential pathways through which each driving factor operates across regions, thereby falling short of providing the precise empirical support needed for differentiated policy formulation [28]. In response to the three research gaps identified above, this study offers targeted methodological responses in its design: (1) to address the insufficient treatment of bidirectional supply–demand coupling mechanisms, this study treats supply and demand as an interacting coupled system and constructs a bidirectional dynamic analytical framework, transcending the unidirectional transmission paradigm of existing research; (2) to address the insufficient comprehensiveness of indicator frameworks, this study constructs a comprehensive supply–demand indicator system incorporating emerging dimensions such as medical-care integration, smart older adult care services, and digital participation, thereby enabling a more thorough assessment of supply–demand development levels for older adult care service resources at the provincial level; (3) to address the insufficient characterization of spatial spillover effects and regional heterogeneity, this study employs a spatiotemporal dual fixed-effects Spatial Durbin Model to systematically identify the spatiotemporal evolutionary patterns, spatial agglomeration characteristics, and direct, indirect, and spatial spillover effects of the driving factors of the supply–demand coupling coordination degree of older adult care service resources across 31 provinces and municipalities in China from 2014 to 2023, with particular emphasis on revealing the heterogeneous driving mechanisms of the eastern, central, and western regions. The principal contributions of this study are twofold: (1) Theoretical contribution: transcending the unidirectional analytical paradigm of supply-side or demand-side research, this study establishes a theoretical model of supply–demand coordinated development from a systems coupling perspective, systematically elucidating the bidirectional interaction mechanisms and dynamic regularities of the two subsystems, and incorporating emerging dimensions such as smart older adult care services and digital participation into a comprehensive evaluation framework, thereby extending the application boundaries of coupling coordination theory to the social services domain. (2) Empirical contribution: through multidimensional and multi-method quantitative analysis, this study comprehensively reveals the spatiotemporal evolutionary patterns, spatial agglomeration characteristics, and regional heterogeneity of supply–demand coupling coordination in older adult care service resources, addressing the methodological limitations of existing research in the identification of spatial spillover effects and the quantitative characterization of regionally heterogeneous driving mechanisms, and providing a scientific basis for the formulation of regionally differentiated and precisely targeted older adult care policies. Materials and methods Research methods Entropy weight method The entropy weight method is an objective weighting approach grounded in information theory, which determines the weight of each indicator by measuring the degree of dispersion in its data, thereby avoiding the biases associated with subjective weighting [29]. In this study, the entropy weight method was applied to assign weights to the supply and demand indicators of older adult care service resources, ensuring that indicators with greater information content received correspondingly higher weights and accurately capturing the relative importance of each indicator within the evaluation framework. The specific calculation procedure is as follows: 1) Normalize the raw data using the range standardization method. For positive indicators, the following formula is applied: For negative indicators, the following formula is applied: 2) Calculate the relative weight proportion of each indicator 3) Calculate the information entropy of each indicator 4) Calculate the weight coefficient of each indicator 5) Calculate the composite score In the above calculations, denotes the raw data value for the indicator of the province or municipality, and is the corresponding standardized value. represents the proportion of the standardized value of that indicator for a given province or municipality relative to the sum of standardized values for that indicator across all provinces and municipalities, such that values sum to 1 across all provinces and municipalities. is the information entropy of the indicator, reflecting the uniformity of its data distribution; a lower information entropy value indicates greater variability in the indicator’s data and thus a richer effective information content. is the final weight coefficient, with all weights summing to 1. denotes the total number of evaluation units, and denotes the total number of indicators. The composite score , obtained through weighted summation, represents the overall supply or demand score of older adult care service resources for each province or municipality, with values ranging from ; higher values indicate a higher level of development along the corresponding dimension [30]. Coupling coordination degree model The coupling degree reflects the degree of association between two systems, while the coordination degree measures the harmony of their joint development. In this study, the coupling coordination degree model was employed to quantitatively assess the degree of coupling and the level of coordinated development between the two subsystems of older adult care service resource supply and demand, thereby comprehensively revealing the interaction mechanisms and co-evolutionary characteristics of the supply–demand system. The calculation procedure is as follows: 1) Calculate the coupling degree of the two subsystems 2) Calculate the composite development index of the two subsystems 3) Calculate the coupling coordination degree In the above calculations, denotes the composite supply score of older adult care service resources for the province or municipality, and denotes the corresponding composite demand score, both of which were derived through the entropy weight method. is the coupling degree, with values ranging from ; higher values indicate a greater degree of coupling between the two systems. α and β are weight coefficients satisfying ; in this study, was adopted, assigning equal weight to both the supply and demand subsystems [31, 32]. This specification is based on the following considerations: in terms of research orientation, this study aims to evaluate the level of coordinated development between the supply and demand systems rather than to privilege either side, and the equal-weight specification is consistent with this analytical objective; in terms of measurement, the entropy weight method had already been applied to objectively assign weights to indicators across all supply and demand dimensions, and the equal-weight treatment at the composite development index level avoids the redundant introduction of subjective weights, thereby ensuring satisfactory objectivity and cross-provincial comparability of the model. is the composite development index, reflecting the overall development level of the supply–demand system. is the coupling coordination degree, with values ranging from ; it serves as the core indicator for measuring the coordinated development of older adult care service supply and demand, with higher values indicating a higher level of coordination. In accordance with the widely adopted convention in existing research of dividing values into 10 levels at intervals of 0.1 [30, 33], the classification of coupling coordination degree levels adopted in this study is presented in Table 1. Table 1Classification of coupling coordination degree levelsNo.Coupling Coordination Degree (D)Coupling Coordination Level1[0.00−0.10)Extreme dissonance2[0.10–0.20)Severe dissonance3[0.20–0.30)Moderate dissonance4[0.30–0.40)Mild dissonance5[0.40–0.50)Near dissonance6[0.50–0.60)Marginally coordinated7[0.60–0.70)Primarily coordinated8[0.70–0.80)Moderately coordinated9[0.80–0.90)Well coordinated10[0.90−1.00]Highly coordinated Classification of coupling coordination degree levels Spatial autocorrelation analysis Spatial autocorrelation analysis is used to examine whether the distribution of a given attribute across geographic spatial units exhibits spatial dependence and agglomeration characteristics. In this study, the Global Moran’s Index and the Local Moran’s Index were employed to measure, respectively, the overall spatial association characteristics and the local spatial agglomeration patterns of the supply–demand coupling coordination degree of older adult care services, with a view to revealing the underlying regularities of its geographic distribution [34]. The selection of the spatial weight matrix should be aligned with the inherent logic of the research context. In this study, the geographic distance matrix was adopted as the benchmark spatial weight matrix, with the economic-geographic nested matrix introduced as a supplementary verification tool (the construction of the latter is detailed in Section “Spatial durbin model”). The geographic distance matrix measures inter-provincial spatial proximity using the inverse of the spherical distance between provincial capital cities, enabling a continuous characterization of the attenuation effects of geographic friction on policy transmission, information diffusion, and resource flows. Compared with binary adjacency matrices, the inverse-distance matrix provides a more refined depiction of the distance-decay pattern governing inter-provincial spatial interaction intensity, and is therefore more suitable for research conducted at the provincial scale [35]. In the context of this study, the diffusion of older adult care service policies, the mobility of professional care workers, and cross-regional reference to and learning from service models are all contingent upon geographic proximity; the intensity of inter-provincial information transmission and factor flows diminishes as spatial distance increases. The geographic distance matrix is capable of continuously capturing this attenuation pattern, making it well-suited to the spatial transmission mechanisms of supply–demand coordination in provincial-level older adult care services. The matrix element is defined as , where denotes the spherical distance (in kilometers) between the provincial capital cities of province and province ; diagonal elements are set to , and the matrix is row-standardized so that each row sums to 1. The calculation procedure is as follows: 1) Calculate the Global Moran’s Index 2) Calculate the Local Moran’s Index In the above calculations, denotes the total number of spatial units (in this study, ); and are the coupling coordination degree values of the and province or municipality, respectively; is the overall sample mean; is the sample variance; and is the element of the row-standardized spatial weight matrix. is the Global Moran’s Index, with values ranging from : a significantly positive indicates that similar values tend to cluster spatially (positive spatial autocorrelation); a significantly negative indicates that dissimilar values tend to cluster (negative spatial autocorrelation); and approaching 0 indicates a random spatial distribution. is the Local Moran’s Index, which identifies four types of local spatial agglomeration — high–high (H-H), low–low (L-L), high–low (H-L), and low–high (L-H) — through comparison with neighboring units, thereby providing a foundation for the analysis of spatial disparities [36]. Spatial durbin model The Spatial Durbin Model simultaneously accounts for the spatial lag effects of both the dependent variable and the independent variables, enabling a comprehensive characterization of the spatial interaction mechanisms among variables. In this study, a spatiotemporal dual fixed-effects Spatial Durbin Model was constructed to investigate the key factors influencing the supply–demand coupling coordination degree of older adult care services and their associated spatial spillover effects [37]. The model specification is as follows: In the above model, denotes the coupling coordination degree of the province or municipality in year ; ρ is the spatial autoregressive coefficient, reflecting the influence of the dependent variable in neighboring regions on the local region. is the element of the spatial weight matrix; the benchmark model employs the geographic distance matrix, with the matrix specification consistent with that described in Section “Spatial autocorrelation analysis”. In the robustness checks, the benchmark matrix is replaced by the economic-geographic nested matrix, which further incorporates inter-provincial differences in economic development levels as a moderating factor on the basis of geographic distance, thereby jointly capturing the dual effects of geographic proximity and economic association, and enabling a more comprehensive characterization of the spatial dependence structure of inter-provincial supply–demand coupling coordination [38]. In the context of this study, both the fiscal capacity for older adult care service investment and the extent to which latent demand is activated across provinces and municipalities are profoundly constrained by the level of economic development; provinces with similar levels of economic development exhibit stronger coordinated linkage in their supply–demand coordination mechanisms, whereas the intensity of spatial interaction between provinces with substantial economic disparities is significantly attenuated even when they are geographically adjacent. The empirical results of this study are consistent with this logic: the coupling coordination degree exhibits a spatial differentiation pattern of “high in the east, low in the west,” and the Moran’s Index under the economic-geographic nested matrix is also higher than that under the geographic distance matrix, indicating that differences in economic development levels constitute an important dimension regulating inter-provincial spatial interaction intensity, which further validates the appropriateness of this matrix in the context of this study [39]. The specific construction procedure for the economic-geographic nested matrix is as follows. First, the unstandardized weights are calculated: where denotes the spherical distance between the provincial capital cities of province and province ; and are the time-series averages of per capita GDP (in 10,000 yuan per person) for the two provinces over the sample period (2014–2023), respectively; and the diagonal elements are set to . Row standardization is then applied: is the matrix of explanatory variables, comprising five dimensions of driving factors: economic conditions (PGD), policy regulation (GES), social structure (ODR), education expenditure (FEE), and consumption capacity (SDI). All explanatory variables are continuous, and a natural logarithmic transformation was applied to all of them to reduce heteroscedasticity in model estimation. Logarithmic transformation is a standard practice in spatial econometric research for handling continuous socioeconomic variables [40]; following transformation, the coefficients of all explanatory variables are expressed in comparable units, facilitating the identification of the relative magnitude of each driving factor’s effect on the coupling coordination degree. β is the regression coefficient of the explanatory variables, reflecting the direct effect of each factor; is the regression coefficient of the spatially lagged explanatory variables, reflecting the spatial spillover effects of factors in neighboring regions on the local region; and are the individual fixed effects and time fixed effects, respectively, used to control for unobservable province- or municipality-specific characteristics and time trends; and is the random error term. Based on the model estimation results, the direct effects, indirect effects, and total effects of each driving factor can be obtained through effect decomposition, thereby revealing their influence mechanisms and spatial transmission pathways [41]. Indicator selection The assessment of older adult care service resources involves two interacting subsystems: supply capacity and demand characteristics. In this study, the principles of systemic completeness, indicator availability, and cross-provincial comparability were followed in constructing separate indicator frameworks for the supply side and the demand side. The dimensional structure of the supply side follows the hierarchical logic of older adult care capacity building: the infrastructure dimension corresponds to the scale capacity and spatial accessibility of services; the workforce and social protection dimension corresponds to the professionalization and institutionalization of services; the medical-care integration dimension corresponds to composite care service capacity; and the smart older adult care dimension corresponds to emerging supply forms enabled by digital technology. Together, these four dimensions cover the full chain from physical infrastructure and human resources to specialized services and technological innovation. The dimensional structure of the demand side is organized according to the multi-tiered nature and structural heterogeneity of the needs of the older adult population: population structure determines the total scale and type of demand; health status determines demand intensity; economic capacity determines the degree to which demand is realized; digital participation determines the accessibility boundary of demand; and service utilization directly reflects the outcomes of demand realization. These five dimensions collectively constitute a complete analytical framework spanning from latent demand to actual consumption. On the supply side, the infrastructure dimension incorporates three indicators — the number of older adult care institutions and facilities, the number of beds per 1,000 older adults, and the number of universities for older adults — reflecting, respectively, the scale of institutional provision, the intensity of bed supply, and the level of cultural enrichment and spiritual service provision, thereby characterizing the capacity of the supply system to meet the basic care needs of older adults at the hardware level [42]. The workforce and social protection dimension incorporates three indicators — the certification rate of older adult care workers, the coverage rate of community-based older adult care service institutions, and basic pension insurance expenditure — comprehensively reflecting the degree of development of the supply-side protection mechanisms from the three perspectives of workforce standardization, service network accessibility, and fiscal funding assurance [43]. The medical-care integration dimension incorporates the number of medical-care integration institutions and the number of geriatric hospitals, measuring, respectively, the institutional scale of medical-care integration and the allocation level of specialized geriatric medical resources, and characterizing the responsiveness of the supply system to the chronic disease management and long-term care needs of older adults [44]. The smart older adult care dimension incorporates the coverage rate of older adult care informatization platforms and the number of smart older adult care pilot cities: the former measures the provincial-level penetration of digital service infrastructure [45]; the latter takes the nationally designated pilot cities jointly recognized by the Ministry of Industry and Information Technology and other relevant authorities as an objective basis — pilot cities form replicable service models by integrating technological, financial, and managerial resources, and their number objectively reflects both the demonstrative leadership capacity of provincial-level smart older adult care supply and the maturity of system development [46, 47]. Together, these two indicators characterize smart older adult care supply capacity from the dual perspectives of infrastructure coverage and demonstrative leadership. On the demand side, the population structure dimension incorporates three indicators — the population aged 60 and above, the proportion of older adults with disabilities or partial functional impairment, and the proportion of empty-nest older adult households — reflecting, respectively, the total scale of basic service demand, the intensity of demand for specialized care services, and the degree to which demand for socialized services has been released against the backdrop of the weakening of family-based care functions [48]. The health status dimension incorporates the prevalence of chronic diseases among older adults and the age-standardized incidence rate of Alzheimer’s disease: the former reflects the scale of the older adult population requiring continuous health management, while the latter characterizes the demand pressure that the cognitively impaired population places on specialized memory care services; together, they reflect the multi-tiered older adult care service demand driven by the accumulation of health risks [49]. The economic capacity and consumption dimension adopted effective demand as its core analytical perspective, evaluating the economic capacity and degree to which the older adult population is able to convert latent care needs into actual service consumption [50]. Latent care needs can only be converted into actual market consumption when accompanied by the corresponding payment capacity, and institutionally guaranteed income support constitutes the fundamental prerequisite for this conversion process [43]. Against the backdrop of China’s still-underdeveloped older adult care service market, pension income represents the primary source of funding through which the vast majority of older adults purchase formal care services; the scale of enrollment in urban basic pension insurance therefore serves as an important dimension for measuring the institutional foundation supporting effective regional demand [51]. At the same time, the supply-side workforce and social protection dimension has already incorporated “basic pension insurance expenditure” to measure the fiscal supply capacity of the system, while the demand-side enrollment figure focuses on characterizing the scale of the demand population with institutionalized payment capacity; the two differ fundamentally in analytical perspective and indicator content, and do not constitute conceptual overlap [52]. Per capita medical expenditure of older adults reflects the actual level of medical and health service consumption among the older adult population [53], while the urban–rural income disparity coefficient among older adults characterizes the structural constraints that income polarization places on the effective release of demand [54]. The digital participation dimension incorporates the internet usage rate among older adults, an indicator that serves both as a direct basis for assessing the accessibility of smart older adult care service demand and as a key measure of the degree to which the digital divide structurally excludes older adults from service utilization [55]. The service utilization dimension uses the total number of residents in older adult care institutions at year-end to measure the actual intensity of demand among older adults for institutional care services; as a direct observational indicator of the actual match between supply and demand, it authentically reflects the degree to which the demand side has absorbed the existing supply of care resources [56]. The indicator frameworks on both the supply and demand sides follow the theoretical logic of systems coupling: the supply side reflects older adult care service supply capacity from the dimensions of physical infrastructure, human resources, medical services, and technology, while the demand side characterizes the need profiles of the older adult population from the dimensions of demography, health, economic capacity, digital participation, and service consumption. The organic integration of the two frameworks enables an accurate assessment of the coordination level of supply and demand for older adult care service resources, providing a data foundation for the analysis of spatiotemporal characteristics and regional disparities (Table 2). Table 2Evaluation indicator framework for the supply and demand of older adult care service resources in ChinaSystemDimensionVariable (Unit)FormulaMeanSDMinMaxWeightSupplyInfrastructureNumber of older adult care institutions and facilities (units)-12140.79415721.1264769,8930.238Beds per 1,000 older adults (beds)Number of beds in older adult care institutions / Population aged 60 and above / 1,00027.4827.35312.96245.9170.047Number of universities for older adults (institutions)-141.40091.172134030.031Workforce and Social ProtectionCertification rate of older adult care workers (%)Number of certified older adult care workers / Total staff in older adult care institutions20.7955.7409.83639.1660.051Coverage rate of community-based older adult care service institutions (%)Number of communities with older adult care service facilities / Total number of urban and rural communities58.91311.53630.53591.2140.119Basic pension insurance expenditure (100 million yuan)-1206.3621149.03196.8816038.9080.144Medical-Care IntegrationNumber of medical-care integration institutions (units)-129.010101.203134790.083Number of geriatric hospitals (units)-85.78180.55273270.155Smart Older Adult CareCoverage rate of older adult care informatization platforms (%)Number of older adult care institutions connected to informatization platforms / Total number of older adult care institutions61.78513.30934.41998.1760.044Number of smart older adult care pilot cities (units)-2.0261.372060.088DemandPopulation StructurePopulation aged 60 and above (10,000 persons)-5453.4764146.973146.320434.70.181Proportion of older adults with disabilities or partial functional impairment (%)Number of older adults with disabilities or partial functional impairment / Population aged 60 and above13.8632.4348.15420.08450.062Proportion of empty-nest older adult households (%)Number of households with older adults living alone / Total number of households with older adults30.2824.75519.24542.8140.060Health StatusPrevalence of chronic diseases among older adults (%)-62.1077.38140.11185.1650.038Age-standardized incidence rate of Alzheimer’s disease (per 100,000 persons)-3.9320.5042.60645.2110.105Economic Capacity and ConsumptionPer capita medical expenditure of older adults (yuan)Total medical expenditure of older adults / Population aged 60 and above4055.7601419.5531650.0167846.2320.077Urban–rural income disparity coefficient among older adults (—)Per capita income of urban older adults / Per capita income of rural older adults2.2730.4681.3083.4080.047Number of urban basic pension insurance enrollees at year-end (10,000 persons)-1382.6861087.41815.2475392.4320.061Digital ParticipationInternet usage rate among older adults (%)43.46410.33522.14774.5260.175Service UtilizationTotal number of residents in older adult care institutions at year-end (persons)-93891.28773384.612545307,2290.195 Evaluation indicator framework for the supply and demand of older adult care service resources in China Data sources To ensure the representativeness and availability of research data, this study took the 31 provinces, autonomous regions, and municipalities directly under the central government of China over the period 2014–2023 as the units of analysis, constructing a panel dataset. Indicator data pertaining to older adult care institutions and facilities, bed supply, workforce allocation, medical-care integration, social insurance revenue and expenditure, the health status and medical consumption of older adults, and macroeconomic and fiscal investment were primarily sourced from the China Statistical Yearbook, the China Civil Affairs Statistical Yearbook, the China Human Resources and Social Security Yearbook, the China Health Statistics Yearbook, the China Social Insurance Yearbook, and the China Social Services Statistical Yearbook (editions spanning 2015–2024). Indicators relating to the demographic characteristics of the older adult population, internet usage rates, and the level of smart older adult care development were sourced from the Statistical Bulletin on the Development of China’s Cause for Older Persons, the Statistical Report on Internet Development in China, and the White Paper on the Development of the Smart Health and Older Adult Care Industry, respectively. All of the above data were officially published by authoritative sources and were characterized by a high degree of credibility and traceability. In addition, in accordance with the regional classification standards set out in the China Statistical Yearbook, the country was divided into three major regions — eastern, central, and western — to facilitate comparative regional analysis. Results Spatiotemporal analysis of the supply and demand of older adult care service resources in China The entropy weight method was applied to measure the supply and demand development levels of older adult care service resources across the 31 provinces and municipalities of China. Based on the regional classification set out in the China Statistical Yearbook, the supply and demand development levels of older adult care service resources were obtained for the country as a whole and for the three major regions — eastern, central, and western — as shown in Fig. 1. Temporal trends in the supply and demand development levels of older adult care service resources in China and its three major regions, 2014–2023 Throughout the study period, the supply level of older adult care service resources exhibited a gradual upward trend at both the national and regional levels. The eastern region maintained a leading position throughout, with its supply level rising from 0.339 in 2014 to 0.507 in 2023, significantly above the national average. The central region’s supply level increased from 0.193 to 0.334, remaining slightly below the national average while maintaining a steady catch-up trajectory. The western region’s supply level rose from 0.111 to 0.178, reflecting comparatively slower development with a notable gap relative to the eastern region, though the pace of catch-up has accelerated in recent years. Regional disparities have emerged from the convergence of multiple factors: the eastern region benefits from its higher level of economic development, well-established older adult care infrastructure, concentration of professional care workers, active policy innovation, and mature older adult care service market; whereas the central and western regions are characterized by relatively weaker economic foundations and resource allocation capacity, with older adult care institutions, bed supply, medical-care integration facilities, and smart care platform development lagging behind, resulting in a comparatively lower level of older adult care service resource supply. In sum, the supply of older adult care service resources in China exhibits a pronounced trend of regional differentiation, manifested as a pattern of “high in the east, low in the west, with a gradient decline.” The demand level of older adult care service resources at both the national and regional levels likewise demonstrated a continuous upward trend. The eastern region’s demand level rose from 0.393 in 2014 to 0.561 in 2023, consistently ranking first. The central region’s demand level increased from 0.316 to 0.462, broadly in line with the national average and maintaining steady growth. The western region’s demand level rose from 0.204 to 0.286, with a relatively moderate rate of increase but an overall trajectory of steady improvement. The regional differentiation pattern reflects the spatial heterogeneity of population structure, health status, and economic capacity. In the eastern region, the deep degree of population aging, the large scale of the older adult population, and the high proportion of individuals with disabilities or partial functional impairment, combined with strong economic payment capacity, high levels of medical consumption, and high rates of digital participation, have allowed older adult care service demand to be more fully released. In the central region, the aging process is accelerating and the proportion of empty-nest older adults is rising; however, demand levels remain moderate, constrained by urban–rural income gaps and limited basic pension insurance coverage. In the western region, the relatively slower pace of population aging, limited economic conditions, and the continued dominance of traditional family-based care models have meant that effective demand for older adult care services has yet to be fully expressed. In sum, the demand for older adult care service resources in China exhibits a spatial pattern of “high in the east, moderate in the central region, and low in the west,” with inter-regional demand disparities fundamentally reflecting the joint effect of the aging process, the accumulation of health risks, and economic payment capacity. Spatiotemporal analysis of the supply–demand coupling coordination of older adult care service resources in China Temporal evolution trends Figure 2 presents the trends in the supply–demand coupling coordination degree of older adult care service resources at the national level and across the three major regions from 2014 to 2023. Over the study period, at the national level, the supply–demand coupling coordination degree of older adult care service resources in China maintained a steady upward trajectory, rising from 0.479 in 2014 to 0.589 in 2023, with an average annual growth rate of 2.300%. In terms of evolutionary stages, the period from 2014 to 2016 was characterized by near dissonance (0.479–0.494), while the period from 2017 to 2023 entered the marginally coordinated stage (0.501–0.589), indicating that the alignment between older adult care service supply capacity and the needs of the older adult population has progressively strengthened. Nevertheless, the overall level of coordination remains low, and a positive, efficient interaction between the supply and demand subsystems has yet to be achieved. The underlying structural causes of this phenomenon lie in the fact that, while the number of older adult care institutions and the scale of bed supply have increased substantially, persistent structural imbalances — most notably a severe shortage of professional care workers and insufficient depth of medical-care integration — remain prominent. Concurrently, the rapid growth of the older adult population, the increasing number of older adults with disabilities or partial functional impairment, and the accelerating diversification of care needs have intensified demand-side pressure, and the mismatch in the pace of development between the supply and demand subsystems has constrained further improvement of the coupling coordination degree. Temporal trends in the supply–demand coupling coordination degree of older adult care service resources in China, 2014–2023 At the regional level, the coupling coordination degree exhibits significant spatial differentiation, consistently maintaining a gradient pattern of eastern > central > western regions. The coupling coordination degree of the eastern region rose from 0.589 in 2014 to 0.712 in 2023; from 2014 to 2015, it remained in the marginally coordinated stage (0.589–0.594), transitioning into the primarily coordinated stage from 2016 onwards (0.611–0.693), and entering the moderately coordinated stage from 2022 (0.702–0.712), consistently and substantially exceeding the national average. The central region increased from 0.493 to 0.621; from 2014 to 2015, it remained in the near dissonance stage (values of 0.493 and 0.490, respectively), transitioning into the marginally coordinated stage from 2016 (0.503–0.583), and advancing into the primarily coordinated stage from 2021 (0.603–0.621), demonstrating a sustained catch-up trajectory. The western region rose from 0.369 to 0.454; from 2014 to 2018, it remained in the mild dissonance stage (0.369–0.397), entering the near dissonance stage from 2019 (0.409–0.454). Although some improvement was observed, no further stage-level breakthrough was achieved, and the coordination level remained consistently below the national average. These regional disparities stem from the spatial heterogeneity of economic development levels, fiscal investment capacity, and prevailing social attitudes toward older adult care. The eastern region, benefiting from a robust economic foundation, has achieved relatively balanced supply–demand development; the central region has seen steady improvements in older adult care resource allocation efficiency, driven by preferential national policy support; whereas the western region, constrained by limited resource endowments and a weak development base, faces the dual challenge of insufficient effective supply and structurally impeded demand release. Over the study period, the average growth rates of the coupling coordination degree across the three major regions, ranked from highest to lowest, were as follows: central region (2.620%), western region (2.300%), and eastern region (2.090%). The higher growth rates recorded in the central and western regions relative to the eastern region reflect the channeling of resources toward these less-developed areas through fiscal transfer payments, paired-assistance programs, and other policy instruments, which has facilitated improvements in older adult care service capacity and enabled the gradual realization of their “late-mover advantages.” In the eastern region, by contrast, the construction of the older adult care service system has entered a comparatively mature phase, narrowing the scope for marginal improvement, such that the growth rate has stabilized while remaining on a steady upward trajectory. At the provincial level, the temporal evolution of the coupling coordination degree exhibits pronounced individual variation across provinces and municipalities (Fig. 3). Internal differentiation within the eastern region is particularly prominent: Guangdong, Shandong, and Jiangsu maintained a consistent lead, reaching values of 0.905, 0.895, and 0.859, respectively, by 2023, successively advancing into the well coordinated and even highly coordinated stages. By contrast, Hainan — also classified within the eastern region — remained throughout the entire study period within the range of 0.324–0.387, persistently in the mild dissonance stage, yielding an intra-regional gap of as much as 0.581. Hainan’s persistently low coordination level is closely related to its distinctive “migratory bird aging” phenomenon: large numbers of older adults from northern China relocate to Hainan during the winter months, generating pronounced seasonal fluctuations in local demand for older adult care services. The local supply system — planned on the basis of the permanent resident population — is ill-equipped to accommodate these cyclical surges in demand, thereby producing a structural supply–demand mismatch that persistently constrains the improvement of the coupling coordination level. Provinces in the central region generally exhibited a steady upward trend, with Henan recording the most pronounced increase, rising from 0.590 in 2014 to 0.723 in 2023. Marked inter-provincial disparities were also evident within the western region, with a substantial gap in coordination levels between Sichuan (0.680) and Tibet (0.190). Sichuan’s notably superior coupling coordination level relative to other western provinces is closely associated with its distinctive demographic structure as one of China’s largest labor-exporting provinces: sustained large-scale outmigration of working-age adults has resulted in an advanced degree of rural aging and a high prevalence of older adults living alone, and the accumulating care pressure has prompted local governments to prioritize the construction of older adult care service systems at a comparatively earlier stage, objectively driving the relative coordination of supply and demand. Tibet, Qinghai, and Ningxia remained in a state of severe or mild dissonance throughout the study period, with extremely limited improvements in their coordination levels, constituting the primary weak links constraining the overall improvement of national coupling coordination.The pronounced bipolarization in provincial-level coupling coordination — characterized by the persistent disparity between high-value eastern provinces, represented by Guangdong, Shandong, and Jiangsu, and low-value western provinces, represented by Tibet, Qinghai, and Ningxia — fundamentally reflects the spatial heterogeneity of multidimensional driving factors across regions, including levels of economic development and the extent of population aging, and thus constitutes an important descriptive basis for the subsequent Spatial Durbin Model in identifying the regionally heterogeneous effects of these drivers. Provincial distribution and temporal variation of the supply–demand coupling coordination degree of older adult care service resources across 31 provinces and municipalities in China, 2014–2023 Spatial evolution trends To elucidate the spatial differentiation characteristics of the supply–demand coupling coordination of older adult care service resources across China’s provinces and municipalities, this study characterizes the proportional distribution of coupling coordination types for the years 2015, 2017, 2019, 2021, and 2023 (Fig. 4). From a spatial perspective, the supply–demand coupling coordination level of older adult care service resources in China exhibits a distribution pattern of “high in the east, low in the west.” Proportional distribution of supply–demand coupling coordination types for older adult care service resources across China’s provinces and municipalities in selected years In 2015, the marginally coordinated type was predominant. Among the 31 provinces and municipalities, the marginally coordinated type accounted for 32.258%, the near dissonance type for 25.806%, and the mild dissonance type for 16.129%. Based on the spatial distribution of the coupling coordination degree, provinces and municipalities reaching the marginally coordinated level or above were concentrated primarily in the eastern and central regions — with Guangdong, Jiangsu, and Zhejiang, for instance, having already entered the moderately coordinated stage — whereas those in a state of coupling dissonance were located predominantly in the western region, with Tibet, Qinghai, and Ningxia remaining at the mild dissonance level or below. In 2017, the marginally coordinated and near dissonance types were jointly predominant, together accounting for 58.065%. Guangdong, Jiangsu, Zhejiang, and Fujian advanced into the moderately coordinated stage, with the proportion of this type increasing from 9.677% to 12.903%. Compared with 2015, the number of provinces and municipalities in the mild dissonance category increased from five to six, with the corresponding proportion rising from 16.129% to 19.355%, while the number of primarily coordinated provinces and municipalities declined from two to one. In 2019, the near dissonance type became predominant, accounting for 29.032% of all provinces and municipalities; the marginally coordinated type accounted for 22.581%; the primarily coordinated type accounted for 19.355%; and a well coordinated province was identified for the first time, as Guangdong entered this stage, accounting for 3.226%. In 2021, the marginally coordinated and primarily coordinated types were jointly predominant, together accounting for 51.613%. The number of provinces in the well coordinated stage increased from one to four, as Jiangsu, Zhejiang, Shandong, and Guangdong entered this stage, with the corresponding proportion rising from 3.226% to 12.903%. The number of mild dissonance provinces and municipalities declined from three to two, with the proportion decreasing from 9.677% to 6.452%, while the number of near dissonance provinces and municipalities fell from nine to five, with the proportion dropping from 29.032% to 16.129%. In 2023, the marginally coordinated and primarily coordinated types remained jointly predominant, together accounting for 48.387%. A highly coordinated province was identified for the first time, as Guangdong advanced into this stage, accounting for 3.226%. The number of well coordinated provinces and municipalities declined from four to three — namely Jiangsu, Zhejiang, and Shandong — with the proportion decreasing from 12.903% to 9.677%; the number of moderately coordinated provinces and municipalities increased from two to three, with the proportion rising from 6.452% to 9.677%; and the proportions of provinces in the severe dissonance and moderate dissonance categories remained consistently low throughout the study period, with a combined share of 6.452%. Overall, the predominant coupling coordination types in China’s supply–demand system for older adult care service resources have gradually shifted from near dissonance and marginally coordinated to marginally coordinated and primarily coordinated, with the coupling coordination level rising steadily. Provinces reaching the well coordinated stage or above have emerged and gradually increased in number. However, at the intra-regional level, Tibet, Qinghai, and Ningxia in the western region have long remained in a state of severe or mild dissonance, and significant disparities also persist within the eastern and central regions. The spatial heterogeneity in coupling coordination types indicates that there remains considerable room for improvement in the supply–demand coupling coordination level of older adult care service resources across China’s provinces and municipalities. Resolving the developmental bottlenecks faced by dissonant provinces and municipalities in areas such as older adult care resource allocation, the advancement of medical-care integration, and the cultivation of professional care workers — and thereby addressing the shortfalls in older adult care service supply — has become a central priority in the implementation of the national strategy for actively addressing population aging. Analysis of the spatial autocorrelation characteristics of the supply–demand coupling coordination degree of older adult care service resources in China Global spatial autocorrelation analysis As shown in Table 3, the Global Moran’s I indices for the supply–demand coupling coordination degree of older adult care service resources across China’s 31 provinces and municipalities from 2014 to 2023 were all positive and passed the 1% significance test in every year, indicating the presence of significant spatial autocorrelation in the supply–demand coupling coordination degree of older adult care service resources in China. Under the geographic distance matrix, the Global Moran’s I index ranged from 0.102 to 0.113, with p-values of 0.000 in all years; under the economic-geographic nested matrix, the Global Moran’s I index ranged from 0.133 to 0.146, with p-values at or below the 0.001 level in all years. The consistent results obtained under both spatial weight matrices provide robust mutual corroboration that the supply–demand coupling coordination degree of older adult care service resources exhibits significant spatial agglomeration characteristics. Overall, the Global Moran’s I index for the coupling coordination degree displayed a fluctuating upward trend from 2014 to 2023. Examining this by phase: under the geographic distance matrix, the Global Moran’s I index rose from 0.103 in 2014 to 0.113 in 2021–2022, before declining slightly to 0.110 in 2023; under the economic-geographic nested matrix, the index rose from 0.137 in 2014 to 0.146 in 2020, thereafter declining slightly to 0.140 in 2023. These findings indicate that the spatial agglomeration effect of the coupling coordination degree across provinces and municipalities has strengthened overall, and that spatial correlation and dependence remain at a relatively high level once economic linkages are incorporated. The Moran’s I indices under both spatial weight matrices are significantly positive and relatively stable, demonstrating that the supply–demand coupling coordination degree of older adult care service resources exhibits strong spatial associations among neighboring provinces and municipalities, reflecting a comparatively stable spatial distribution pattern. Table 3Global spatial autocorrelation Moran’s I indexYearGeographic Distance MatrixEconomic-Geographic Nested MatrixMoran’s IP-valueMoran’s IP-value20140.1030.0000.1370.00120150.1060.0000.1380.00120160.1020.0000.1330.00120170.1080.0000.1390.00020180.1070.0000.1390.00020190.1100.0000.1390.00020200.1120.0000.1460.00020210.1130.0000.1450.00020220.1130.0000.1450.00020230.1100.0000.1400.000 Global spatial autocorrelation Moran’s I index Local spatial autocorrelation analysis To provide a more detailed account of the spatial agglomeration patterns of the supply–demand coupling coordination degree of older adult care service resources across China’s provinces and municipalities, this study selected 2014, 2019, and 2023 as reference time points, calculated the Local Moran’s I index, and generated Moran scatterplots (Fig. 5), classifying the 31 provinces and municipalities into four spatial association types: high–high agglomeration (H-H), low–high agglomeration (L-H), low–low agglomeration (L-L), and high–low agglomeration (H-L), with the specific distribution presented in Table 4. Moran scatterplots of the supply–demand coupling coordination degree of older adult care service resources in China Table 4Spatial agglomeration type distribution of the local Moran’s i index for the supply–demand coupling coordination degree of older adult care service resources in ChinaYearH-HH-LL-LL-H2014Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Shandong, Henan, Hubei, HunanTianjin, Shanxi, Jilin, Jiangxi, HainanInner Mongolia, Heilongjiang, Guangxi, Guizhou, Yunnan, Tibet, Gansu, Qinghai, Ningxia, XinjiangLiaoning, Guangdong, Chongqing, Sichuan, Shaanxi2019Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Shandong, Henan, Hubei, HunanTianjin, Shanxi, Jilin, Jiangxi, HainanInner Mongolia, Heilongjiang, Guangxi, Guizhou, Yunnan, Tibet, Gansu, Qinghai, Ningxia, XinjiangLiaoning, Guangdong, Chongqing, Sichuan, Shaanxi2023Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, HunanTianjin, Shanxi, Jilin, HainanInner Mongolia, Heilongjiang, Guangxi, Chongqing, Guizhou, Yunnan, Tibet, Gansu, Qinghai, Ningxia, XinjiangLiaoning, Guangdong, Sichuan, Shaanxi Spatial agglomeration type distribution of the local Moran’s i index for the supply–demand coupling coordination degree of older adult care service resources in China Overall, the supply–demand coupling coordination degree of older adult care service resources in China exhibits pronounced spatial agglomeration characteristics, with the majority of provinces and municipalities distributed in the first and third quadrants, and high–high and low–low agglomeration as the dominant patterns. This indicates that the coupling coordination level across provinces and municipalities is not randomly distributed in space; rather, regions with similar levels of development tend to cluster together geographically. From a temporal perspective, the spatial distribution pattern remained relatively stable between 2014 and 2023, with only a small number of provinces and municipalities — such as Jiangxi and Chongqing — transitioning between different agglomeration types, reflecting the fact that most provinces and municipalities have developed strong spatial interdependencies with neighboring regions with respect to their coupling coordination level. In terms of spatial distribution characteristics, the high–high agglomeration zone is concentrated primarily in the eastern coastal and central regions, encompassing provinces and municipalities such as Beijing, Jiangsu, Zhejiang, Shandong, Henan, and Hubei. These areas are characterized by strong older adult care service resource supply capacity and a high degree of demand release among the older adult population, and neighboring provinces have established collaborative synergies in the inter-regional flow of care resources and the cross-provincial sharing of policy experience. The low–low agglomeration zone is distributed mainly across the western and northeastern regions, with representative provinces including Tibet, Qinghai, Ningxia, Gansu, and Heilongjiang; these areas not only exhibit low levels of older adult care resource allocation in their own right, but together with their neighboring regions form a spatial “depression” in terms of coupling coordination degree. The low–high agglomeration zone is scattered across a small number of provinces and municipalities, including Tianjin, Shanxi, and Jilin; although these areas have comparatively lower coupling coordination levels themselves, they are surrounded by high-coordination provinces, affording them the potential to benefit from regional spillover effects through inter-regional cooperation. The high–low agglomeration zone encompasses provinces such as Liaoning, Guangdong, and Sichuan; these provinces have relatively high coupling coordination levels of their own, yet their neighboring regions remain comparatively underdeveloped, giving rise to a characteristic pattern of “enclave-type” development. The formation of this spatial agglomeration pattern is rooted in the combined influence of regional economic development levels, the collaborative mechanisms of older adult care service policies, and population mobility patterns. The eastern and central regions, drawing on robust economic foundations and well-established regional cooperation mechanisms, have formed contiguous high-coordination agglomeration zones, whereas the western and northeastern regions — constrained by lagging economic development and the compounding structural pressures of population aging and labor outmigration — continue to experience a self-reinforcing spatial lock-in effect at low levels of coordination. This pattern of spatial heterogeneity reveals the spatial dependence mechanism underlying the inter-regional imbalance in the supply–demand coupling coordination development of older adult care service resources in China. Analysis of the driving factors of the supply–demand coupling coordination degree of older adult care service resources in China Selection of influencing factors Drawing on the theoretical framework of supply–demand coupling coordination for older adult care service resources and the practical realities of older adult care system development in China, this study constructs an analytical framework for influencing factors across five macro-level dimensions — economic conditions, policy regulation, demographic structure, human capital, and consumption capacity — to systematically identify the key driving factors affecting the level of coupling coordination. The following indicators are selected to examine the driving mechanisms underlying the supply–demand coupling coordination degree of older adult care service resources (Table 5). Table 5Influencing factors of the supply–demand coupling coordination degree of older adult care service resources in ChinaVariable NameVariable Description (Unit)SymbolMeanSDMinMaxCoupling coordination degreeCalculated coupling coordination degree of the two subsystemsD0.5290.1590.1290.905Economic FactorPer capita GDP (yuan/person)PGD67900.75532473.62326165.261200,278Policy factorGovernment fiscal expenditure on older adult care services (10,000 yuan)GES127579.495152166.493652.11006343.1Demographic factorOld-age dependency ratio (%)ODR16.9864.8877.0130.6Human capital factorFiscal expenditure on education (100 million yuan)FEE1010.957659.770122.684004.45Consumption factorPer capita disposable income of older adults (10,000 yuan)SDI3.1951.0631.4075.981 Influencing factors of the supply–demand coupling coordination degree of older adult care service resources in China Economic factor (PGD): The level of economic development constitutes the fundamental basis for both older adult care service resource supply capacity and demand release. Regional per capita GDP directly influences investment in older adult care infrastructure, the wage levels of professional care workers, and the payment capacity of the older adult population. Economically developed regions are better positioned to build a more comprehensive older adult care service supply system, while simultaneously facilitating the effective release of demand for older adult care services among the older adult population, thereby advancing the coordinated development of the supply–demand system. Policy factor (GES): Government fiscal expenditure on older adult care services reflects the degree of priority accorded by public finance to the construction of the older adult care service system. The scale of fiscal investment directly determines the number of older adult care institutions, the supply of beds, and the construction standards of medical-care integration facilities. Through a resource allocation mechanism oriented toward “securing the basics, providing a safety net, and promoting equity,” fiscal investment effectively compensates for market failures and narrows the inter-regional gap in older adult care service supply capacity. Demographic factor (ODR): The old-age dependency ratio captures the structural pressure that population aging exerts on the demand for older adult care services. Rising dependency ratios drive the transition from family-based care toward socialized older adult care services, while simultaneously generating rapid growth in the care needs of specific groups — including older adults with disabilities or partial functional impairment and empty-nest older adults — thereby placing higher demands on the quantity, quality, and structural composition of older adult care service resources. Human capital factor (FEE): Fiscal expenditure on education indirectly influences the efficiency of older adult care resource allocation through improvements in human capital quality. On one hand, it contributes to cultivating a professional care workforce and enhancing the level of specialization and quality of older adult care service supply; on the other, by raising the health literacy and digital participation capacity of the older adult population, it promotes the effective release of demand for older adult care services. Consumption factor (SDI): The per capita disposable income of older adults directly determines their payment capacity for older adult care services and the degree to which their demand is released. Older adults with higher income levels are more inclined to select diversified, high-quality care services, whereas those with lower incomes face economic constraints that prevent their care needs from being effectively converted into actual consumption. Income polarization therefore directly affects the supply–demand matching in the older adult care service market. Variable diagnostic analysis To ensure the statistical validity and reliability of the Spatial Durbin Model estimation results, this study conducted, in sequence, a correlation test, a multicollinearity test, and a panel unit root test on all explanatory variables prior to formal regression analysis. The correlation test was employed to provide an initial assessment of the degree of linear association among variables; the multicollinearity test further quantified the potential interference of inter-variable collinearity on the precision of model estimation; and the unit root test addressed the possible non-stationarity of macroeconomic variables in the panel data, fundamentally eliminating the risk of spurious regression and thereby establishing a sound statistical foundation for the subsequent spatial econometric estimation. As indicated by the correlation test results presented in Table 6, a certain degree of positive correlation exists among the explanatory variables, with all correlations significant at the 1% level. Among these, the correlation coefficient between the economic factor (PGD) and the consumption factor (SDI) is the highest, at 0.683; the correlation coefficient between the economic factor (PGD) and the policy factor (GES) is 0.680, with both reflecting the underlying role of economic development levels in driving the scale of fiscal expenditure and the income levels of the older adult population. The correlation coefficient between the human capital factor (FEE) and the consumption factor (SDI) is 0.611; that between the policy factor (GES) and the consumption factor (SDI) is 0.572; and that between the policy factor (GES) and the human capital factor (FEE) is 0.557. These associations indicate that regions with higher levels of economic development tend to exhibit simultaneously larger fiscal expenditure scales, higher incomes among the older adult population, and greater educational investment, reflecting an inherent structural interdependence among the driving factors. By contrast, the correlation coefficient between the demographic factor (ODR) and the human capital factor (FEE) is only 0.343 — the lowest among all variable pairs — suggesting that the explanatory variables possess a degree of conceptual and informational independence, which provides preliminary evidence that multicollinearity may fall within an acceptable range, pending confirmation through further diagnostic testing. Table 6Variable correlation testVariablePGDGESODRFEESDIPGD1.000GES0.680***1.000ODR0.373***0.471***1.000FEE0.348***0.557***0.343***1.000SDI0.683***0.572***0.350***0.611***1.000Note: *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively; values in parentheses are standard errors; the same applies to the tables below Variable correlation test Note: *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively; values in parentheses are standard errors; the same applies to the tables below As indicated by the multicollinearity test results presented in Table 7, the variance inflation factors (VIF) for all explanatory variables fall within an acceptable range. Specifically, the economic factor (PGD) has the highest VIF value at 2.76, followed by the consumption factor (SDI) at 2.68 and the policy factor (GES) at 2.54; the human capital factor (FEE) has a VIF value of 2.01; and the demographic factor (ODR) has the lowest VIF value at 1.31, with a mean VIF of 2.26 across all variables. It is generally accepted that a VIF value exceeding 10 indicates severe multicollinearity; all explanatory variables in this study are well below this threshold, and the tolerance values (1/VIF) range from 0.363 to 0.762, further confirming the absence of serious collinearity among the variables that would compromise the robustness of the model estimation, and demonstrating that the model specification possesses satisfactory statistical properties. Table 7Multicollinearity testVariableVIF1/VIFPGD2.760.363SDI2.680.373GES2.540.394FEE2.010.498ODR1.310.762Mean VIF2.26 Multicollinearity test As indicated by the panel unit root test results presented in Table 8, this study simultaneously applied three methods — the LLC test, the IPS test, and the ADF test — to verify the stationarity of both the dependent variable, the coupling coordination degree (D), and all five explanatory variables. The test results show that all variables reject the null hypothesis of a unit root under all three testing methods. Specifically, for the coupling coordination degree (D), the LLC statistic is − 7.388, the IPS statistic is − 5.181, and the ADF statistic is 183.528, all significant at the 1% level; the LLC statistic for the economic factor (PGD) is − 11.412, the largest in absolute value among all variables; the IPS statistic for the consumption factor (SDI) is − 6.550 and the ADF statistic is 345.696, both the largest in absolute value among all variables and both significant at the 1% level; the IPS test statistic for the human capital factor (FEE) is − 1.345, significant at the 10% level — the lowest significance level among all test results — however, its LLC and ADF test statistics are − 10.568 and 143.835, respectively, both significant at the 1% level, with consistent conclusions across all three testing methods. These results confirm that all variable series in this study constitute stationary panel data, that no spurious regression risk arising from non-stationary time series is present, and that the regression results possess full statistical validity. Table 8Panel unit root testVariableLLC TestIPS TestADF TestTest ResultD−7.388***−5.181***183.528***StationaryPGD−11.412***−2.856***118.215***StationaryGES−6.559***−4.377***139.323***StationaryODR−7.033***−3.572***251.986***StationaryFEE−10.568***−1.345*143.835***StationarySDI−10.779***−6.550***345.696***Stationary Panel unit root test Taken together, the results of the three diagnostic tests indicate that the correlation coefficients among the explanatory variables fall within an acceptable range, the mean VIF is only 2.26, and multicollinearity does not constitute a substantive concern; all variables pass the stationarity test and satisfy the fundamental statistical assumptions required for panel data regression. The results of these pre-estimation diagnostic tests confirm that the variable selection and data processing procedures adopted in this study are methodologically sound, providing a robust statistical foundation for the subsequent construction of the spatiotemporal dual fixed-effects Spatial Durbin Model and the decomposition of direct effects, indirect effects, and spatial spillover effects for each driving factor. Spatial econometric model selection Given the significant spatial autocorrelation present in the supply–demand coupling coordination degree of older adult care service resources in China, this study employs spatial econometric models to empirically examine the mechanisms through which each influencing factor operates. The spatial econometric models considered include the Spatial Lag Model (SLM), the Spatial Error Model (SEM), and the Spatial Durbin Model (SDM). To determine the optimal model specification, this study sequentially conducted the Lagrange Multiplier (LM) test, the Wald test, the Likelihood Ratio (LR) test, and the Hausman test, with results presented in Table 9. Table 9Spatial econometric model specification test resultsTest MethodGeographic Distance MatrixEconomic-Geographic Nested MatrixStatisticP-valueStatisticP-valueLM- error30.1240.00021.7970.000Robust LM- error42.2480.00033.7430.000LM- lag2.3710.12422.0030.000Robust LM- lag14.4950.00033.9490.000Wald- error20.140.00112.050.034Wald- lag20.9900.00113.050.023LR- error18.9400.00211.8100.037LR- lag19.8300.00112.7200.026Hausman10.3600.06698.2200.000LR_both_ind31.9200.00131.8800.001LR_both_time777.0000.000793.490.000 Spatial econometric model specification test results With respect to the test results under the geographic distance matrix, the statistics for LM-error and its robust form, Robust LM-error, are 30.124 and 42.248, respectively, both significant at the 1% level; the LM-lag statistic is 2.371 (P = 0.124), which does not pass the significance test, whereas the Robust LM-lag statistic is 14.495, significant at the 1% level. These results indicate that both spatial error autocorrelation and spatial lag effects are simultaneously present in the model. Furthermore, both the Wald test and the LR test significantly reject the null hypothesis that the Spatial Durbin Model can be reduced to either the Spatial Lag Model or the Spatial Error Model (Wald-error = 20.14, P = 0.001; Wald-lag = 20.990, P = 0.001; LR-error = 18.940, P = 0.002; LR-lag = 19.830, P = 0.001), indicating that the Spatial Durbin Model cannot be simplified into a nested model. This demonstrates that the supply–demand coupling coordination degree of older adult care service resources is subject to both the spatial spillover effects of the dependent variable in neighboring regions and the spatial interaction effects of the independent variables in neighboring regions; accordingly, the Spatial Durbin Model — which simultaneously captures the spatial effects of both the dependent and independent variables — constitutes the appropriate model choice. With respect to the test results under the economic-geographic nested matrix, the statistics for LM-error, Robust LM-error, LM-lag, and Robust LM-lag are all significant at the 1% level, indicating that spatial effects are even more pronounced under this specification. Both the Wald test and the LR test likewise reject the null hypothesis of model reduction (Wald-error = 12.05, P = 0.034; Wald-lag = 13.05, P = 0.023; LR-error = 11.810, P = 0.037; LR-lag = 12.720, P = 0.026), further confirming the appropriateness of the Spatial Durbin Model. Regarding the selection of model effects, the Hausman test results show that under the economic-geographic nested matrix, the test statistic is 98.220 (P = 0.000), significantly rejecting the random effects model at the 1% level and indicating that a fixed effects model should be adopted; under the geographic distance matrix, the statistic is 10.360 (P = 0.066). In addition, the LR_both_ind and LR_both_time test statistics are both significant at the 1% level (geographic distance matrix: 31.920 and 777.000; economic-geographic nested matrix: 31.880 and 793.49), indicating the presence of significant individual fixed effects and time fixed effects in the model, and confirming that a spatiotemporal dual fixed-effects model specification should be adopted. On the basis of the foregoing, this study ultimately selects the spatiotemporal dual fixed-effects Spatial Durbin Model (SDM) for empirical analysis. This model effectively controls for the influence of unobservable individual heterogeneity and time-varying factors on the estimation results, while simultaneously capturing the spatial dependence and spatial spillover effects of the supply–demand coupling coordination degree of older adult care service resources, thereby providing more robust estimation results for elucidating the mechanisms through which each influencing factor operates. Regression results analysis Based on the regression results presented in Table 10, this study conducts a comparative analysis of the estimation results of the spatial econometric models from two dimensions: the significance of spatial effects and the goodness of model fit. Table 10Regression results analysisVariableGeographic Distance MatrixSEMSLMSDMPGD0.008(0.022)0.030(0.021)0.016(0.022)GES-0.004**(0.002)-0.004**(0.002)-0.005**(0.002)ODR0.054**(0.028)0.063***(0.022)0.043*(0.026)FEE0.103***(0.029)0.069***(0.022)0.082***(0.027)SDI0.022(0.026)0.016(0.027)-0.007(0.029)W*Pgd0.147**(0.063)W*Ges0.007(0.006)W*Odr0.212***(0.080)W*Fee-0.125**(0.053)W*Sdi-0.279(0.195)Rho0.906***(0.027)0.654***(0.067)0.272*(0.154)Sigma2_e0.001***(0.000)0.001***(0.000)0.001***(0.000)Log-L582.470596.422605.658R20.8300.8730.893 Regression results analysis With respect to the spatial correlation test, the spatial autoregressive coefficients ρ for all three models (SEM, SLM, and SDM) are significantly positive: the ρ value for the SEM is 0.906, for the SLM is 0.654, and for the SDM is 0.272, significant at the 1%, 1%, and 10% levels, respectively. This result indicates that the supply–demand coupling coordination degree of older adult care service resources in China exhibits significant spatial dependence, with the coupling coordination levels of neighboring provinces mutually influencing one another through spatial association. This finding empirically corroborates the conclusion drawn from the earlier global spatial autocorrelation analysis — that Moran’s I is significantly positive — and confirms the objective existence of spatial agglomeration effects in the supply–demand coupling coordination degree of older adult care service resources. With respect to model fit, the Spatial Durbin Model (SDM) demonstrates clear superiority. In terms of the log-likelihood value (Log-L), the SDM reaches 605.658, substantially higher than the 596.422 of the SLM and the 582.470 of the SEM, indicating that the SDM more effectively captures the information structure of the data. In terms of goodness of fit (R²), the SDM achieves 0.893, likewise outperforming the SLM at 0.873 and the SEM at 0.830, demonstrating the SDM’s stronger explanatory power over the variation in the dependent variable. Both the Log-L and R² indicators point to the SDM as possessing the optimal statistical properties, further confirming at the level of empirical results the conclusions of the earlier Wald and LR tests, and validating the appropriateness and robustness of selecting the spatiotemporal dual fixed-effects Spatial Durbin Model for empirical analysis. Based on the effect decomposition results presented in Table 11, this study conducts an in-depth analysis of the direct effects, indirect effects, and total effects of each influencing factor, with a view to revealing the mechanisms through which each driving factor influences the supply–demand coupling coordination degree of older adult care service resources and its associated spatial spillover characteristics. Table 11Effects decomposition resultsVariableGeographic Distance MatrixDirect EffectIndirect EffectTotal EffectPGD0.019(0.022)0.213***(0.087)0.232***(0.090)GES-0.005**(0.002)0.008(0.008)0.003(0.008)ODR0.050**(0.024)0.294***(0.079)0.344***(0.079)FEE0.081***(0.028)-0.133**(0.069)-0.052(0.067)SDI-0.013(0.029)-0.408(0.299)-0.421(0.313) Effects decomposition results The direct effect of the economic factor (PGD) is 0.019, which does not pass the significance test; the indirect effect is 0.213, significantly positive at the 1% level; and the total effect is 0.232, significantly positive at the 1% level. The non-significance of the direct effect does not imply that economic development has no bearing on the coordination of older adult care service supply and demand; rather, it reflects the absence of an automatically operative direct transmission pathway between the two. The fiscal revenue expansion generated by economic growth must be channeled through government budgetary decision-making processes before it can be converted into older adult care service investment, and given that local governments face competing expenditure priorities, the dividends of economic development do not automatically accrue to the older adult care sector [57, 58]. At the same time, improvements in the supply–demand coupling coordination level depend on the systematic advancement of multiple interconnected processes — including institutional development, workforce training, and medical-care integration — and the impetus that economic development provides to these processes is subject to considerable time lags, making it difficult for statistically significant direct impacts to materialize in the short term [59, 60]. This is consistent with the theoretical proposition in the existing literature that economic development influences public service supply levels primarily through intermediate mechanisms. By contrast, the indirect effect of 0.213 substantially exceeds the direct effect, indicating that the influence of economic development on the coupling coordination degree operates primarily through cross-regional transmission mechanisms — economically developed regions, drawing on their mature older adult care service systems, accumulated policy experience, and advantages in factor mobility, exert significant demonstration effects and provide sustained developmental impetus to neighboring regions, thereby driving overall improvements in supply–demand coordination levels in adjacent areas [19, 61]. The direct effect of the policy factor (GES) is − 0.005, significantly negative at the 5% level; the indirect effect is 0.008, which does not pass the significance test; and the total effect is 0.003, likewise not significant. This result indicates that increases in government fiscal expenditure on older adult care services have exerted a significantly negative direct effect on the local supply–demand coupling coordination degree, contrary to theoretical expectations. This finding reflects, to a considerable extent, the structural misallocation of fiscal resources at the current stage of development — with investment heavily concentrated in hardware areas such as bed construction, while core service dimensions more directly relevant to supply–demand matching, including professional workforce development, medical-care integration operations, and community-based home care services, remain underfunded. As a result, the expansion of supply scale has not been effectively translated into substantive improvements in supply–demand coordination [62]. Furthermore, allocation efficiency problems — such as the concentration of fiscal resources in urban public institutions and the underdeveloped operating subsidy mechanism — have further undermined the capacity of fiscal investment to enhance the overall coupling coordination degree [63]. The non-significance of both the indirect effect and the total effect indicates, on one hand, that the misallocation problem in the current fiscal expenditure model is primarily concentrated at the local level, and on the other, that the high degree of homogeneity in the fiscal investment structures across provinces and municipalities has precluded the formation of differentiated inter-provincial spatial spillover effects. The direct effect of the demographic factor (ODR) is 0.050, significantly positive at the 5% level; the indirect effect is 0.294, significantly positive at the 1% level; and the total effect is 0.344, significantly positive at the 1% level. Rising old-age dependency ratios have not only significantly promoted improvements in the local supply–demand coupling coordination degree of older adult care service resources, but have also significantly advanced the coupling coordination level of neighboring regions through spatial spillover mechanisms, with the magnitude of the spillover effect substantially exceeding that of the local effect. Population aging pressure has become the core driving force behind the coordinated development of supply and demand for older adult care service resources, simultaneously stimulating the local uptake of older adult care services while generating cross-regional transmission effects through regional linkage mechanisms [64]. The direct effect of the human capital factor (FEE) is 0.081, significantly positive at the 1% level; the indirect effect is − 0.133, significantly negative at the 5% level; and the total effect is − 0.052, which does not pass the significance test. The increase in fiscal expenditure on education has significantly enhanced the local supply–demand coupling coordination degree of older adult care service resources, yet has generated significant negative spillover effects on neighboring regions, with the overall effect being non-significant. This phenomenon reveals the talent agglomeration effect of educational resource investment: while increases in local educational expenditure raise the quality of local human capital, they simultaneously undermine the capacity of neighboring regions to develop their professional older adult care workforce, producing a pronounced talent-drain effect [65]. The direct effect of the consumption factor (SDI) is − 0.013, the indirect effect is − 0.408, and the total effect is − 0.421; none of the three effects passes the significance test. The influence of per capita disposable income of older adults on the supply–demand coupling coordination degree of older adult care service resources is statistically non-significant, with neither local effects nor cross-regional spillover effects being evident. This may be related to the current level of development of the older adult care service market in China, whereby rising income levels among the older adult population have yet to be effectively converted into a full release of demand for older adult care service consumption [66]. Robustness checks To ensure the reliability and robustness of the empirical results, this study conducts robustness checks on the baseline regression model across five dimensions: replacing the spatial weight matrix, applying winsorization, substituting the dependent variable, introducing a one-period lag for the explanatory variables, and replacing selected explanatory variables. As shown in the left panel of Table 12,given that spatial econometric models are sensitive to the choice of spatial weight matrix, the economic-geographic nested matrix was constructed to re-estimate the benchmark model for the purpose of verifying the robustness of the research findings. The economic-geographic nested matrix incorporates the dual influence of both geographic proximity and economic association, enabling a more comprehensive characterization of the spatial dependence structure of the supply–demand coupling coordination degree of older adult care service resources across provinces and municipalities. The spatial effect decomposition results show that the direct effects, indirect effects, and total effects of each influencing factor are broadly consistent with those of the benchmark model in terms of the direction of regression coefficients, relative magnitude, and significance levels, indicating that the research conclusions are robust. Table 12Robustness check results for replacing the spatial weight matrix and winsorizationVariableEconomic-Geographic Nested MatrixWinsorizationDirect EffectIndirect EffectTotal EffectDirect EffectIndirect EffectTotal EffectPGD0.023(0.022)0.246***(0.074)0.268***(0.077)0.015(0.021)0.131*(0.079)0.146*(0.079)GES-0.005**(0.002)0.005(0.008)0.001(0.008)-0.005(0.003)0.004(0.009)-0.001(0.009)ODR0.063***(0.024)0.173***(0.067)0.236***(0.069)0.144***(0.023)0.227***(0.069)0.371***(0.067)FEE0.081***(0.028)-0.111*(0.066)-0.030(0.064)0.098***(0.024)-0.135**(0.064)-0.037(0.065)SDI-0.001(0.027)-0.219(0.188)-0.220(0.197)0.009(0.030)-0.326(0.282)-0.317(0.296) Robustness check results for replacing the spatial weight matrix and winsorization 0.023 (0.022) 0.246*** (0.074) 0.268*** (0.077) 0.015 (0.021) 0.131* (0.079) 0.146* (0.079) -0.005** (0.002) 0.005 (0.008) 0.001 (0.008) -0.005 (0.003) 0.004 (0.009) -0.001 (0.009) 0.063*** (0.024) 0.173*** (0.067) 0.236*** (0.069) 0.144*** (0.023) 0.227*** (0.069) 0.371*** (0.067) 0.081*** (0.028) -0.111* (0.066) -0.030 (0.064) 0.098*** (0.024) -0.135** (0.064) -0.037 (0.065) -0.001 (0.027) -0.219 (0.188) -0.220 (0.197) 0.009 (0.030) -0.326 (0.282) -0.317 (0.296) As shown in the right panel of Table 12, to eliminate the potential distortions introduced by extreme values, all continuous variables were winsorized at the 1st and 99th percentiles, and the Spatial Durbin Model was re-estimated accordingly. The regression results following winsorization show that the direction, relative magnitude, and significance levels of the direct effects, indirect effects, and total effects of each influencing factor are broadly consistent with those of the benchmark regression results, indicating that extreme values have not exerted any substantive influence on the empirical conclusions. As shown in the left panel of Table 13, the composite development gap index — defined as the ratio of the absolute difference between the supply and demand composite scores to their sum, serving as a measure of the relative degree of supply-demand imbalance — was used to replace the coupling coordination degree D as the dependent variable, thereby re-estimating the baseline model from the perspective of supply-demand imbalance. The results indicate that the indirect effect and total effect of the economic factor (PGD) remain significantly positive, exhibiting high consistency with the baseline regression results in both the direction of coefficients and the level of statistical significance, which further corroborates the core finding that economic development promotes improvements in supply-demand coordination through inter-regional transmission mechanisms. The direction of the direct effect of the policy factor (GES) is also consistent with the baseline model. Certain discrepancies in the direction or significance of the effects of other variables relative to the baseline model are primarily attributable to the fundamental conceptual differences between the two indicators: the coupling coordination degree D simultaneously captures both the degree of coupling and the overall development level of the supply-demand system, whereas the composite development gap index focuses on reflecting the relative imbalance between supply and demand, capturing a different dimension of systemic coordination. Consequently, the two indicators do not respond identically to the various driving factors. Overall, these results provide partial validation of the robustness of the baseline model. Table 13Robustness check results for substituting the dependent variable and introducing a one-period lag for explanatory variablesVariableSubstituting the Dependent VariableOne-Period Lag for Explanatory VariablesDirect EffectIndirect EffectTotal EffectDirect EffectIndirect EffectTotal EffectPGD0.157***(4.541)0.848***(3.427)1.004***(4.018)0.008(0.680)-0.116*(-1.900)-0.108*(-1.812)GES-0.000(-1.453)-0.006**(-2.909)-0,006**(-2.927)-0.000*(-1.897)-0.001**(-2.405)-0.001**(-2.517)ODR-0.012***(-7.310)-0.025(-1.270)-0.038(-1.864)0.012***(20.094)-0.009(-1.335)0.003(0.511)FEE-0.001(-0.842)0.037**(2.789)0.036*(2.550)0.012***(24.019)-0.005(-1.535)0.007*(1.930)SDI0.010(0.655)0.295*(1.990)0.305*(1.992)0.023***(4.429)0.070(1.586)0.092**(2.071) Robustness check results for substituting the dependent variable and introducing a one-period lag for explanatory variables 0.157*** (4.541) 0.848*** (3.427) 1.004*** (4.018) 0.008 (0.680) -0.116* (-1.900) -0.108* (-1.812) -0.000 (-1.453) -0.006** (-2.909) -0,006** (-2.927) -0.000* (-1.897) -0.001** (-2.405) -0.001** (-2.517) -0.012*** (-7.310) -0.025 (-1.270) -0.038 (-1.864) 0.012*** (20.094) -0.009 (-1.335) 0.003 (0.511) -0.001 (-0.842) 0.037** (2.789) 0.036* (2.550) 0.012*** (24.019) -0.005 (-1.535) 0.007* (1.930) 0.010 (0.655) 0.295* (1.990) 0.305* (1.992) 0.023*** (4.429) 0.070 (1.586) 0.092** (2.071) As shown in the right panel of Table 13, explanatory variables were lagged by one period prior to re-estimation in order to mitigate potential reverse causality between the explanatory variables and the dependent variable. The results show that the direction of the direct effect of the economic factor (PGD) remains consistent with the baseline model; the direct effect of the policy factor (GES) remains significantly negative, consistent with the baseline regression in both direction and significance; both the direction and significance of the direct effect of the demographic factor (ODR) are consistent with the baseline model, and the direction of the total effect likewise remains unchanged; the direction and significance of the total effect of the human capital factor (FEE), as well as the direction of its indirect effect, are consistent with the baseline results. Certain changes in the significance and magnitude of other effects are primarily attributable to the dynamic temporal adjustments introduced by the one-period lag: the influence of each driving factor on the coupling coordination degree is subject to a certain transmission lag, causing some statistical characteristics to shift accordingly, while the directional effects of the core variables remain substantively unchanged. Overall, these results provide supporting evidence for the robustness conclusion of the baseline model. As shown in Table 14, to further validate the robustness of the baseline regression results, this study replaces the policy factor variable government expenditure on care services for older adults (GES) with per capita government expenditure on care services for older adults (PGS = government expenditure on care services for older adults / population aged 60 and above), to eliminate potential confounding effects arising from interprovincial differences in population size on the measurement of fiscal expenditure intensity, and re-estimates the Spatial Durbin Model accordingly. The results indicate that the direction of the direct effect of the economic factor (PGD) remains consistent with the baseline regression; both the direction and significance of the direct effect of per capita government expenditure on care services for older adults (PGS) are consistent with the baseline model; the direction and significance of the direct effect of the demographic factor (ODR) are consistent with the baseline model, and the direction of the total effect likewise remains unchanged; the direction and significance of the direct effect of the human capital factor (FEE) are consistent with the baseline results, and the direction of the indirect effect also remains consistent. Overall, these results demonstrate that the baseline regression conclusions are not subject to distortion from interprovincial differences in population size, thereby confirming the strong robustness of the baseline regression results. Table 14Robustness check results for replacing explanatory variablesVariableReplacing Explanatory VariablesDirect EffectIndirect EffectTotal EffectPGD0.003(0.275)-0.118(-1.845)-0.114(-1.796)PGS-0.000*(-1.999)-0.001*(-2.328)-0.001*(2.505)ODR0.012***(19.817)-0.009(-1.367)0.003(0.527)FEE0.012***(23.404)-0.006(-1.676)0.006(1.709)SDI0.024***(4.621)0.098*(2.163)0.122**(2.621) Robustness check results for replacing explanatory variables Drawing on the combined findings of all five robustness checks — replacing the spatial weight matrix, applying winsorization, substituting the dependent variable, introducing a one-period lag for the explanatory variables, and replacing key explanatory variables — the directional effects and spatial effect characteristics of each driving factor on the coupling coordination degree of supply and demand for care services for older adults remain highly consistent with the baseline regression results across all specifications. This comprehensively validates the reliability of the empirical findings of this study and provides a solid empirical foundation for the subsequent policy recommendations. Regional heterogeneity analysis The three major regions of eastern, central, and western China exhibit pronounced differences in their levels of economic and social development, and the allocation of older adult care service resources likewise displays significant regional differentiation. It is therefore necessary to further examine the regional heterogeneity in the effects of each driving factor on the supply–demand coupling coordination degree of older adult care service resources. Table 15 presents the direct effects, indirect effects, and total effects of each influencing factor as estimated by the Spatial Durbin Model for the three major regions. Table 15Spatial durbin model estimation results for the coupling coordination level across the three major regionsVariableEastern RegionCentral RegionWestern RegionDirect EffectIndirect EffectTotal EffectDirect EffectIndirect EffectTotal EffectDirect EffectIndirect EffectTotal EffectPGD0.020***(0.010)0.036(0.051)0.056(0.060)-0.04(0.039)0.288**(0.129)0.248*(0.143)0.001(0.007)0.007(0.024)0.008(0.027)GES-0.006(0.004)0.024(0.023)0.018(0.026)-0.016**(0.008)0.022(0.019)0.006(0.022)-0.004**(0.002)-0.007(0.010)-0.011(0.010)ODR0.033(0.025)0.192***(0.055)0.225***(0.059)-0.010(0.053)0.195*(0.110)0.185*(0.106)0.005***(0.001)0.007***(0.001)0.012***(0.001)FEE-0.004(0.004)-0.071***(0.028)-0.075**(0.031)0.010**(0.005)-0.018(0.013)-0.008(0.015)0.001**(0.000)0.001(0.002)0.002(0.002)SDI0.079***(0.039)0.351**(0.155)0.430**(0.178)0.036(0.052)0.136(0.190)0.172(0.226)0.001(0.005)0.02*(0.012)0.020*(0.012) Spatial durbin model estimation results for the coupling coordination level across the three major regions 0.020*** (0.010) 0.036 (0.051) 0.056 (0.060) -0.04 (0.039) 0.288** (0.129) 0.248* (0.143) 0.001 (0.007) 0.007 (0.024) 0.008 (0.027) -0.006 (0.004) 0.024 (0.023) 0.018 (0.026) -0.016** (0.008) 0.022 (0.019) 0.006 (0.022) -0.004** (0.002) -0.007 (0.010) -0.011 (0.010) 0.033 (0.025) 0.192*** (0.055) 0.225*** (0.059) -0.010 (0.053) 0.195* (0.110) 0.185* (0.106) 0.005*** (0.001) 0.007*** (0.001) 0.012*** (0.001) -0.004 (0.004) -0.071*** (0.028) -0.075** (0.031) 0.010** (0.005) -0.018 (0.013) -0.008 (0.015) 0.001** (0.000) 0.001 (0.002) 0.002 (0.002) 0.079*** (0.039) 0.351** (0.155) 0.430** (0.178) 0.036 (0.052) 0.136 (0.190) 0.172 (0.226) 0.001 (0.005) 0.02* (0.012) 0.020* (0.012) The operating mechanisms of the economic factor (PGD) differ markedly across the three regions. In the eastern region, the direct effect is 0.020, significantly positive at the 1% level, while the indirect effect and total effect are both non-significant. In the central region, the direct effect is non-significant, whereas the indirect effect is 0.288, significantly positive at the 5% level, and the total effect is 0.248, significantly positive at the 10% level. In the western region, all three effects are non-significant. These divergences reflect deep-seated structural differences across the regions. In the eastern region, the higher level of economic development is associated with stronger payment capacity among the older adult population, enabling market-based demand for older adult care services to be more fully realized; the comparatively higher level of economic development also facilitates the attraction of private capital into the older adult care sector, promoting spontaneous market-level supply–demand matching and thereby generating a significant local direct effect. In the central region, the non-significant direct effect but prominent indirect effect reflects the tendency for the gains from central region economic growth to radiate outward to neighboring areas through factor mobility and policy diffusion, rather than translating rapidly into substantive improvements in local supply–demand matching efficiency, given that the local older adult care service market remains insufficiently developed for this purpose. In the western region, all three effects are non-significant, fundamentally because economic development has yet to cross the critical threshold required to drive the development of the older adult care service market; the limited economic increment must be directed preferentially toward more pressing basic livelihood needs, making it difficult in the short term to meaningfully advance supply–demand coordination. The policy factor (GES) exhibits a consistent directional effect across the three regions, but with marked differences in significance and magnitude. In the eastern region, the direct effect is − 0.006, which does not pass the significance test, and the indirect effect and total effect are likewise non-significant. In the central and western regions, the direct effects are − 0.016 and − 0.004, respectively, both significantly negative at the 5% level, while the spatial spillover effects are non-significant. This regional differentiation pattern is closely related to the fiscal implementation capacity and the pace of expenditure structure adjustment in each region. In the eastern region, certain provinces have taken the lead in optimizing the investment structure, progressively expanding the proportion of expenditure on operational subsidies and professional workforce development, thereby relatively narrowing the degree of mismatch between funding allocation and actual demand, such that the negative effect has tended toward non-significance. In the central region, the divergence between the rate at which fiscal resources expand following the receipt of central transfer payments and the pace of structural adjustment is the most pronounced among the three regions; large volumes of new funding have flowed into hardware construction, continuously amplifying the marginal misallocation effect and producing the largest negative effect in absolute value among all three regions. In the western region, the smallest coefficient is primarily attributable to the objective constraint imposed by limited local fiscal capacity, which has compressed the magnitude of the negative effect; nevertheless, structural constraints — including a severe shortage of professional care workers and high facility vacancy rates — remain prominent, and the supply–demand mismatch caused by the concentration of limited funding in hardware construction continues to constitute an important constraint on the persistently low coupling coordination level in the west. The indirect effects are non-significant across all three regions, indicating that the impact of fiscal expenditure structural misallocation is primarily concentrated at the local level and has yet to generate cross-regional spatial transmission effects. The demographic factor (ODR) exhibits pronounced differentiation in its effects across the three regions. In both the eastern and central regions, the direct effects are non-significant, yet both generate significant positive spatial spillover effects, significant at the 1% and 10% levels, respectively. In the western region, all three effects are significantly positive at the 1% level, with a direct effect of 0.005, an indirect effect of 0.007, and a total effect of 0.012 — the smallest coefficients among the three regions. Population aging proceeded at an earlier stage in the eastern and central regions, and following years of policy investment and market development, local older adult care service supply has developed a degree of responsiveness to aging-related demand; further aging pressure is therefore transmitted primarily through regional linkage mechanisms to neighboring areas, generating spillover effects. The spillover effect of the central region (0.195) is broadly comparable to that of the eastern region (0.192), indicating that the two regions possess similar levels of inter-regional transmission capacity. In the western region, where population aging has progressed more slowly and the supply system remains in a phase of addressing fundamental gaps, aging pressure has yet to reach the point where it can diffuse outward to neighboring regions and is therefore channeled primarily into direct local improvements in supply–demand coordination; however, all three coefficients are the smallest among the three regions, indicating that, constrained by the overall weakness of supply capacity, the capacity of aging pressure to stimulate commensurate supply-side responses remains highly constrained. The operating mechanisms of the human capital factor (FEE) exhibit clear differentiation across the three regions. In the eastern region, the direct effect is non-significant, whereas the indirect effect is − 0.071, significantly negative at the 1% level, and the total effect is − 0.075, significantly negative at the 5% level. In the central region, the direct effect is 0.010, significantly positive at the 5% level, while the indirect effect and total effect are non-significant. In the western region, the direct effect is 0.001, significantly positive at the 5% level but with the smallest coefficient, and the spatial spillover effect is non-significant. These regional differences are rooted in the spatial asymmetry of older adult care workforce mobility patterns. The non-significant direct effect in the eastern region indicates that educational investment has not significantly enhanced local supply–demand coordination; the significantly negative indirect effect, however, reveals a pronounced talent competition mechanism — the eastern region, drawing on its higher wage levels and well-developed career pathways, continuously attracts professional older adult care workers from neighboring regions, objectively undermining the workforce development capacity of adjacent provinces and producing a strong talent-drain effect. The positive direct effect of educational expenditure on local coupling coordination in the central region is associated with the relatively higher talent retention rates and the comparatively closer integration between vocational education and older adult care practice in the region; the non-significant indirect effect indicates that the central region has yet to develop an effective capacity to radiate professional talent to neighboring areas. In the western region, although the direct effect is statistically significant, the extremely small coefficient reflects the comparatively limited scale and quality of older adult care professional training programs in the region; compounded by the tendency for trained professionals to migrate toward the eastern region, educational investment has had limited substantive impact on improving local supply quality. The consumption factor (SDI) exhibits significant differences in both direction and magnitude across the three regions. In the eastern region, the direct effect is 0.079, the indirect effect is 0.351, and the total effect is 0.430, all significantly positive at the 1% or 5% level, with the largest coefficients among the three regions. In the central region, all three effects are non-significant. In the western region, the direct effect is non-significant, whereas the indirect effect is 0.020, significantly positive at the 10% level, and the total effect is likewise 0.020. This divergence is fundamentally rooted in differences in the maturity of the older adult care service market and the extent to which income gains translate into actual consumption across the three regions. In the eastern region, the diverse and multi-tiered supply of older adult care services enables income growth among the older adult population to be relatively smoothly converted into actual service consumption, generating significant spillover effects on neighboring provinces through consumption demonstration effects and the diffusion of service standards; the indirect effect (0.351) substantially exceeding the direct effect (0.079) further indicates the particularly prominent role of the eastern region’s mature market in shaping regional older adult care consumption expectations. In the central region, all three effects are non-significant: on one hand, the older adult care service market remains in an early stage of development, with effective service supply still unable to fully accommodate diversified consumption preferences; on the other, traditional family-based care values continue to exert a degree of constraining influence in the central region, limiting to some extent the willingness of older adults to convert income into formal service consumption and thereby preventing income growth from meaningfully advancing the coupling coordination degree. In the western region, the non-significant direct effect reflects both the low degree of local market development and the generally low income levels among the older adult population, such that an effective income-to-consumption transmission channel has yet to be established; the modest positive indirect effect may be related to weak consumption linkages between certain provinces and adjacent more developed regions through population mobility, though the overall effect remains very limited in magnitude. On the basis of the foregoing, each driving factor exhibits pronounced regional heterogeneity, the underlying causes of which are rooted in the systematic differences among the three major regions in terms of market maturity, workforce mobility patterns, the stage of population aging, and fiscal implementation capacity. In the eastern region, where market mechanisms are comparatively well-established, economic and consumption drivers are more fully realized through mature supply–demand matching systems, with strong spillover effects on neighboring areas through talent competition and consumption demonstration. The central region is in a transitional phase from policy-driven to market-driven development, with the local human capital effects of educational investment and the regional stimulative effects of economic growth being relatively prominent, though the supply–demand linkage mechanisms are still in need of optimization. In the western region, aging pressure constitutes the most direct local driving force, yet a weak economic base, limited capacity to convert income gains into care service consumption, and a severe shortage of professional care workers have collectively constrained the pace of improvement in supply–demand coordination; breakthroughs in this regard will require strengthened fiscal support and the targeted development of complementary institutional frameworks. This pattern of heterogeneity reveals the stage-specific regularities governing the development of China’s older adult care service system, and provides empirical support for the formulation of regionally differentiated, precisely targeted older adult care policies. As shown in Table 16, to further validate the robustness of the regional heterogeneity conclusions, this study adopts the eastern region as the reference group. By introducing interaction terms between regional dummy variables and each explanatory variable within the full-sample Spatial Durbin Model framework, a joint test is conducted to verify the regional heterogeneity characteristics identified in the subsample regressions. The results indicate that the interaction term coefficients of core variables — including PGD, GES, and SDI — are broadly consistent with the conclusions of the subsample regional heterogeneity analysis in terms of both direction and statistical significance, thereby providing partial support for the robustness of the regional heterogeneity conclusions. The observed divergence in the inter-regional effects of FEE and ODR from the corresponding subsample results is primarily attributable to the methodological distinction between the dummy variable approach — which imposes a uniform spatial structure constraint within the full-sample framework — and independent subsample estimation; the regional heterogeneity characteristics of these variables warrant further investigation. Overall, the full-sample dummy variable estimation results provide partial support for the core conclusions of the subsample regional heterogeneity analysis. Table 16Robustness check results for the introduction of regional dummy variablesVariableEastern Region (Reference)Central-EasternWestern-EasternDirect EffectIndirect EffectTotal EffectDirect EffectIndirect EffectTotal EffectDirect EffectIndirect EffectTotal EffectPGD-0.049***(0.018)-0.266***(0.081)-0.315***(0.089)0.017(0.025)1.012***(0.201)1.029***(0.205)0.262***(0.022)-0.031(0.140)0.231(0.145)GES-0.000*(0.000)0.000(0.000)-0.000(0.000)0.000**(0.000)-0.001**(0.001)-0.001*(0.001)0.000(0.000)-0.009***(0.003)-0.008***0.003ODR0.013***(0.001)0.008*(0.005)0.021***(0.005)-0.001(0.001)-0.028***(0.009)-0.029***(0.009)-0.002***(0.001)0.000(0.007)-0.002(0.007)FEE0.012***(0.001)-0.008**(0.004)0.004(0.004)0.001(0.001)-0.004(0.007)-0.004(0.008)-0.004**(0.002)-0.017*(0.009)-0.021**(0.010)SDI0.026***(0.008)0.009(0.034)0.035(0.036)-0.006(0.011)-0.221***(0.081)-0.227***(0.084)-0.027**(0.011)0.117**(0.059)0.090(0.063)Note: The eastern region serves as the reference group. The coefficients in the Central-Eastern and Western-Eastern columns represent the differential effects of each region relative to the eastern region Robustness check results for the introduction of regional dummy variables -0.049*** (0.018) -0.266*** (0.081) -0.315*** (0.089) 0.017 (0.025) 1.012*** (0.201) 1.029*** (0.205) 0.262*** (0.022) -0.031 (0.140) 0.231 (0.145) -0.000* (0.000) 0.000 (0.000) -0.000 (0.000) 0.000** (0.000) -0.001** (0.001) -0.001* (0.001) 0.000 (0.000) -0.009*** (0.003) -0.008*** 0.003 0.013*** (0.001) 0.008* (0.005) 0.021*** (0.005) -0.001 (0.001) -0.028*** (0.009) -0.029*** (0.009) -0.002*** (0.001) 0.000 (0.007) -0.002 (0.007) 0.012*** (0.001) -0.008** (0.004) 0.004 (0.004) -0.004 (0.007) -0.004 (0.008) -0.004** (0.002) -0.017* (0.009) -0.021** (0.010) 0.026*** (0.008) 0.009 (0.034) 0.035 (0.036) -0.006 (0.011) -0.221*** (0.081) -0.227*** (0.084) -0.027** (0.011) 0.117** (0.059) 0.090 (0.063) Note: The eastern region serves as the reference group. The coefficients in the Central-Eastern and Western-Eastern columns represent the differential effects of each region relative to the eastern region Conclusions and recommendations Discussion Institutional transmission gaps between economic growth and supply–demand coordination The non-significant direct effect of per capita GDP does not imply that economic development lacks substantive influence on the construction of the older adult care service system; rather, it reveals the presence of institutional transmission gaps that cannot be automatically bridged. Under the prevailing fiscal framework, the fiscal space generated by economic growth must pass through local government budgetary decision-making processes before it can be converted into older adult care service investment — a conversion process constrained by multiple structural impediments: the temporal mismatch between short-term performance orientation and the long-cycle returns characteristic of older adult care services; the crowding-out pressure exerted by competing expenditure demands in areas such as education and infrastructure; and the inherent difficulty of incorporating the composite output of supply–demand coordination in older adult care services into conventional performance evaluation systems [67]. These constraints mean that even when economic increments are objectively present, they are unlikely to flow reliably into the older adult care sector. The high significance of the indirect effect, by contrast, points to an effective alternative pathway from a different analytical dimension: mature regions, drawing on institutionally replicable experience and mobile human capital accumulated over the long term, generate demonstration spillover effects on neighboring provinces — conditions that cannot be directly produced by short-term economic growth alone [68]. The findings of this study indicate that designing institutional mechanisms to effectively channel the dividends of economic growth into the older adult care sector constitutes an important policy priority for improving the overall level of supply–demand coordination. Three mechanisms of efficiency loss through which fiscal investment suppresses coordination The finding that increased fiscal investment actually suppresses the supply–demand coupling coordination degree represents the most policy-relevant discovery of this study. The underlying mechanism does not lie in any inherent problem with fiscal expenditure per se, but rather in a systemic misalignment between the structure of fund allocation and the genuine service needs of older adults — a misalignment that can be understood at three distinct levels. First, public investment may generate crowding-out effects on private capital. Public institutions, drawing on fiscal subsidies to establish significant price competitive advantages, occupy a dominant position in the institutional care market, potentially deterring private capital entry due to compressed profit margins; if this mechanism holds, it would narrow the diversity of the supply structure — causing service types and quality tiers to converge — and would systematically undermine the capacity of supply to match the full spectrum of heterogeneous demand [69]. Second, performance evaluation incentives generate a hardware bias. The prevailing performance assessment system for older adult care governance centers on readily quantifiable hardware metrics — such as the number of care institutions and bed capacity — creating institutional incentives that direct local governments to prioritize visible facility construction, while relegating to secondary importance the soft-side investments most directly relevant to supply–demand matching quality, including professional care worker training, home-based care operational subsidies, and the advancement of medical-care integration [70]. Third, distorted subsidy benchmarks produce perverse incentives. A fiscal subsidy mechanism centered on bed counts decouples institutional profitability from service quality — capacity expansion automatically generates additional subsidies, without any prerequisite of service utilization rates or demand-matching outcomes. This perverse incentive structure produces the simultaneous coexistence of continuously expanding bed supply and stagnant fulfillment of actual care needs, with the marginal utility of fiscal resources declining steadily, ultimately generating a policy paradox in which growing investment fails to improve — and indeed suppresses — the level of supply–demand coordination [71]. Path lock-in mechanisms through which spatial polarization becomes self-reinforcing The “high east, low west” spatial pattern remained highly stable and continued to intensify throughout the study period, underpinned by the mutually nested and cyclically self-reinforcing operation of three mechanisms — capital agglomeration, talent mobility, and institutional barriers — that market forces alone are insufficient to resolve spontaneously. At the capital level, eastern provinces, drawing on the scale of their mature older adult care service markets and comparatively well-developed policy environments, continuously attract secondary agglomeration of social capital and policy innovation resources, thereby sustaining a virtuous cycle on the supply side; western provinces, by contrast, face a chronic structural deficit of external capital inflows driven by limited market scale and subdued profitability expectations, creating a mutually reinforcing dynamic of low supply capacity and low market attractiveness. At the talent level, the negative spatial spillover effect of educational investment reveals an important capacity-building paradox: even when central and western provinces sustain substantial investment in older adult care professional education, the graduates produced are continuously drawn away by the superior wages and career development opportunities available in the eastern region, severing the link between investment and local capacity accumulation and producing a talent-drain trap in which greater investment yields greater outflows. At the institutional level, the long-standing absence of cross-provincial administrative coordination and inter-provincial cooperation mechanisms has left low–low agglomeration zones unable to achieve exogenous breakthroughs through regional resource spillovers, nor to realize endogenous advancement through their own accumulated capacities, causing the spatial lock-in effect to become increasingly entrenched over time [72]. Recommendations Establishing institutionalized channels for converting economic growth into older adult care service investment In addressing the institutional transmission gap between economic growth and older adult care investment, the policy priority should be to leverage institutional design to mandate the establishment of resource channeling conduits, thereby transforming the latent association between economic increments and older adult care service investment into explicit institutional constraints. It is recommended that the central government’s general transfer payment allocation formula incorporate the size of the older adult population and the deviation value of the supply–demand coupling coordination degree as weighting factors, enabling the fiscal allocation mechanism itself to serve a transmission function and thereby reducing the dependence of older adult care investment on discretionary local decision-making [73]. Concurrently, the degree of improvement in the supply–demand coordination of older adult care services should be incorporated into the performance evaluation indicators of provincial governments, using accountability mechanisms to compel localities to prioritize older adult care budget allocations. With respect to regional differentiation, eastern provinces should prioritize product innovation in commercial long-term care insurance and health management insurance for older adults, effectively linking residents’ income growth to older adult care consumption demand; central and western provinces are advised, within the framework of paired-assistance programs, to establish industry guidance funds for the older adult care service sector, using policy-backed guarantees and risk-sharing mechanisms to lower barriers to social capital entry, thereby guiding economic increments toward the generation of effective supply through institutional safeguards and progressively nurturing the foundations for local market development. Restructuring fiscal resource allocation with an efficiency-oriented approach The systemic misalignment between fiscal resource allocation structures and the genuine service needs of older adults constitutes the fundamental cause by which fiscal investment suppresses the supply–demand coupling coordination degree. Reform efforts should therefore target the root causes of three distinct efficiency loss mechanisms — crowding-out effects, hardware bias, and perverse incentives — and restructure the investment framework with efficiency and demand-matching as its core orientations. With respect to the crowding-out effect mechanism, the functional boundaries of public institutions should be explicitly defined around their role as a safety-net provider, with their target population delimited to particularly vulnerable groups such as functionally impaired and low-income older adults. Competitive market access mechanisms should be introduced in the mid-to-high-end care service market, replacing government monopoly with public–private collaboration so as to create sustainable profit margins for private capital. With respect to the hardware bias mechanism, a tiered evaluation and accreditation system centered on service quality should be established, directly linking institutional operating subsidies to service quality ratings; fiscal expenditure structures should simultaneously be reoriented toward soft-side investments — including professional care worker training, home-based care operational support, and the advancement of medical-care integration — thereby correcting the chronic structural distortion of over-concentrating resources in visible infrastructure construction. With respect to the perverse incentive mechanism, subsidy benchmarks should be shifted from bed counts to service utilization rates and demand-matching quality; a binding threshold on bed vacancy rates should be set as a prerequisite condition for approving new facility construction, leveraging utilization efficiency to activate existing assets and curb the inertia of low-efficiency capacity expansion. As the applicability of these three reform pathways varies across regions, the three major regional groupings should calibrate their priorities according to their respective stages of development: eastern regions should take the lead in transitioning subsidy mechanisms; central regions should prioritize the establishment of county-level demand assessment databases to enable precision-targeted resource allocation; and western regions should concentrate their limited fiscal resources on improving rural basic pension benefit levels and expanding long-term care insurance coverage, thereby consolidating the institutional foundations of basic social protection [74]. Establishing targeted compensation and cooperative mechanisms to break the spatial lock-in of low–low agglomeration zones Breaking the spatial lock-in of low–low agglomeration zones in the western region requires simultaneous interventions across three dimensions — targeted compensation, talent flow management, and cooperative mechanisms — to construct a regional equalization framework that combines external intervention with endogenous activation. With respect to targeted fiscal compensation, it is recommended that a special transfer payment fund for the regional equalization of older adult care services be established, with low–low agglomeration provinces as the primary beneficiaries. Compensation intensity should be linked to each province’s coupling coordination degree deviation and its older adult population size, and a five-year dynamic adjustment mechanism should be instituted, automatically reducing the compensation ratio as the coordination degree improves, thereby preventing the formation of dependency-driven fiscal inertia. With respect to talent flow management, to address the talent-drain trap associated with educational investment, a targeted commissioned training cooperation program for older adult care and nursing professionals between eastern and western regions should be launched — with eastern vocational colleges accepting training commissions from western provinces — supplemented by full-subsidy incentives conditional on a three-year local employment commitment. Simultaneously, leading eastern care institutions should be encouraged to establish professional staff rotation and secondment systems in western provinces, transforming the current unidirectional talent extraction into an organized bidirectional flow and progressively rebuilding sustainable pathways for local professional talent accumulation in the west. With respect to cooperative mechanism development, a Western Regional Older Adult Care Service Cooperation Alliance should be established, anchored by provinces with comparatively advanced coordination levels such as Sichuan, integrating the functions of experience-sharing, mutual recognition of standards, and cross-provincial service referral, and thereby forming an east–central–west cascading policy diffusion channel. At the national level, the outcomes of cross-regional cooperation should be incorporated into the annual performance assessments of participating provinces, with dedicated performance incentives provided, ensuring through institutionalized evaluation that cooperative mechanisms remain consistently operational rather than perfunctory [75]. Research limitations and future directions This study has undertaken a systematic exploration of both the theoretical framework and empirical methodology; however, several limitations remain that warrant attention in subsequent research. First, although the spatiotemporal dual fixed-effects Spatial Durbin Model effectively controls for spatial autocorrelation and individual heterogeneity, the endogeneity between the supply and demand subsystems has not been fully addressed within the existing framework: supply-side resource expansion may activate latent service consumption through an “induced demand” mechanism, while the sustained release of demand-side pressure may in turn drive optimization of the supply system. Future research should introduce instrumental variable two-stage least squares estimation (IV-2SLS) or simultaneous equation models to address this limitation. Second, the robustness verification framework has room for further expansion: subsequent studies could select alternative indicators — such as older adult care resource allocation efficiency — for cross-validation of the core findings, and could supplement the analysis by reporting the complete decomposition of direct, indirect, and total effects within the Spatial Lag Model (SLM) framework, thereby further consolidating the reliability of the empirical conclusions through multi-model comparison. Furthermore, the use of the three broad regional divisions — eastern, central, and western China — as the primary analytical units in this study partially obscures the spatial heterogeneity within each region. Future research may consider reconstructing spatial econometric models at a finer spatial scale to more thoroughly identify differences in the underlying driving mechanisms within these regions. Looking ahead, the research team will continue to deepen the investigation of supply–demand coordination mechanisms in older adult care services, with particular emphasis on three directions: refining the characterization of spatial differentiation patterns at the county level, evaluating the policy effects of the long-term care insurance system, and examining the mechanisms through which digital technology enables the equilibration of older adult care service supply and demand — thereby providing sustained empirical support for evidence-based decision-making in the implementation of the national strategy for actively addressing population aging. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We thank all the authors for their participation and all the editors and reviewers for their helpful comments on this article. Abbreviations Coupling coordination degree Spatial Durbin Model Spatial Lag Model Spatial Error Model Lagrange Multiplier Likelihood Ratio Variance inflation factor Levin-Lin-Chu test Im-Pesaran-Shin test Augmented Dickey-Fuller test High–High agglomeration Low–Low agglomeration High–Low agglomeration Low–High agglomeration Per capita GDP (Economic Factor) Government fiscal expenditure on older adult care services (Policy Factor) Per capita government expenditure on older adult care services Old-age dependency ratio (Demographic Factor) Fiscal expenditure on education (Human Capital Factor) Per capita disposable income of older adults (Consumption Factor) Composite development gap index Author contributions LG and YW have made equally significant contributions to this study and share equal responsibility and accountability for it. The symbol “#” denotes co-first authors. Their specific contributions are as follows:“L.G. was responsible for conceptualization, formal analysis, methodology, software, and data curation. Y.W., X.Z., and K.C. contributed to data curation and investigation. H.W. secured funding, supervised the project, provided resources, validated the work, contributed to visualization, and participated in writing the original draft as well as review & editing. Q.T. oversaw project administration, provided resources, supervised the work, validated the findings, contributed to visualization, and participated in writing the original draft and review & editing. All authors read and approved the final manuscript.” Funding The authors declare that financial support was received for the conduct of the research, the writing of the manuscript, and/or the publication of this article. This study was supported by the following projects: Key Laboratory of Philosophy and Social Sciences of Anhui Province – Data Science and Traditional Chinese Medicine Innovation and Development Laboratory (Project No.: GXXT-2022-095); Research on the Effect Evaluation and Improvement Suggestions of Medical Insurance Supporting the Development of Traditional Chinese Medicine (Project No.: YB2025A007); Construction of Traditional Chinese Medicine Industry and Comprehensive Reform Demonstration Zone for Traditional Chinese Medicine (Project No.: RZ2300001174). Data availability All data generated or analysed during this study are included in this published article [and its supplementary information files]. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Lingxiao Guo and Yiwen Wei contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement All data generated or analysed during this study are included in this published article [and its supplementary information files].