Introduction

Investing in Urban Cultural Facilities (UCFs) is an invaluable strategy for promoting social equity and creating inclusive and sustainable cities (Borowiecki and Castiglione, 2014; Ferrucci et al., 2022). UCFs serve as material and spatial carriers, meeting the spiritual and cultural needs of residents while providing them with a sense of belonging and identity (Tomasello et al., 2005). For instance, libraries and museums act as material carriers, offering gathering spaces for residents to exchange culture and satisfy their knowledge requirements. Preserving local culture and natural landscapes is crucial for enhancing a city’s core competitiveness and addressing the dilemma of convergence in urban development (Komasi et al., 2023). The significance of UCFs for city inclusiveness and sustainability has been continuously increasing.

In theoretical research, the theory of public goods provision focuses on the public attributes of UCFs and their contributions to social welfare (Andrade-Agudelo et al., 2023). Cultural capital theory examines how these facilities enhance the cultural capital of individuals and communities (Blessi et al., 2012). Maslow’s hierarchy of needs theory evaluates the functionality of UCFs from the perspective of fulfilling the multi-tiered needs of society members (Talebi et al., 2023), while location value theory investigates how the geographical placement of UCFs influences their value and utility (DeVerteuil, 2000). Furthermore, supply and demand theory serves as a foundational theoretical framework for the study of UCFs (Wei et al., 2022). Researchers typically focus on the spatial clustering of these facilities, utilizing GIS and multivariate analysis techniques to explore the impacts of urban development policies and geographical environments (de Graaff et al., 2009; Mahmoudzadeh and Noori, 2018). Despite numerous scholarly contributions, research on UCFs still exhibits significant gaps, such as the lack of regional studies transcending administrative boundaries (Shorabeh et al., 2020), an overemphasis on the facility layouts of single periods (Brida et al., 2012), limited use of open-source Point of Interest (POI) data, and insufficient consideration of systemic driving factors.

In China, under the guidance of policies like the ‘National Cultural Development Planning Outline,’ significant strides have been made in the construction of UCFs. Ye (2013) highlights that these policy interventions have substantially fostered the growth of UCFs. However, as noted by Macintyre et al. (2008) and Shoham et al. (1990), challenges persist in the development of UCFs, especially concerning regional disparities and uneven spatial distribution within cities. Existing research on UCFs predominantly focuses on their spatial distribution, aiming to uncover the unevenness in cultural spaces. For instance, Chang et al. (2021) revealed less spatial inequality in museum numbers between provinces in China than within provinces, highlighting the importance of spatial scale in research. Additionally, the spatial clustering of UCFs is particularly pronounced within individual cities. A case is He et al. (2021), which investigated the spatial clustering effects of various cultural and entertainment facilities in Beijing.

These case studies underscore the critical interaction between space and culture in the context of rapid urbanization and economic transformation. They show how cultural facilities reflect and influence social structures and urban development, revealing cultural disparities and promoting equalization of urban cultural services. Regional-scale studies, though less emphasized, are crucial for a comprehensive understanding. The Yangtze River Economic Belt (YREB), embodying a fusion of economic and cultural exchanges across China and spanning diverse geographical and cultural landscapes from inland mountainous areas to coastal metropolises, stands as an exemplary economic region for such research. Investigating the role of UCFs within the YREB can significantly contribute to understanding the distribution of regional cultural facilities and the advancement of equitable public services. This approach is instrumental in broadening the cultural dimensions of sustainable development theories, particularly within varied regional contexts.

Therefore, this study aims to investigate the spatiotemporal dynamics and influencing factors of UCFs from a regional perspective. Two research questions guide the study:

How do the spatiotemporal dynamics of UCFs, including their subcategories Non-profit Urban Cultural Facilities (NPUCFs) and Profitable Urban Cultural Facilities (PUCFs), vary across the YREB from 2012 to 2020, and what are the influencing factors?

What is the spatial heterogeneity of the impact effects of key influencing factors on UCFs, and how do these effects differ between NPUCFs and PUCFs in the YREB?

Data from 2012 to 2020 were collected through multiple sources, including yearbooks and POI data. By employing the GeoDetector method, we calculate the impact effects of influencing factors and their interactions on UCFs. The Multi-Scale Geographically Weighted Regression (MGWR) model is then used to investigate the spatial heterogeneity of these effects across different types of UCFs. The study contributes to offering new insights into the spatiotemporal changes and driving forces of basin-scale UCFs.

Literature review

Regional disparities in cultural facility distribution

UCFs play a pivotal role in promoting social equity and advancing urban development. These facilities, encompassing a variety of public and private spaces and activities, aim to strengthen cultural and creative expressions, appreciation, and consumption in urban areas. Research by Woronkowicz (2013) and Terzi et al. (2015) underscores the critical importance of the equitable distribution of UCFs in enhancing urban cultural development, particularly in achieving equality in public services and fostering cultural diversity. Scientific assessment and optimization of UCFs layouts can significantly enhance the fairness of public services (Yue et al., 2023). Then, UCFs have a profound impact on the multidimensional development of cities. Economically, UCFs not only generate direct revenues through ticket sales and the revitalization of obsolete buildings (Eom and An, 2018) but also indirectly boost tourism, retail, and other related economic activities through attracting both tourists and local residents (Lak et al., 2020), thereby substantially increasing employment opportunities. Socially, UCFs provide foundational support for civic participation and satisfaction, enhancing community cohesion through interactions and dialogues among people from diverse backgrounds (Carrà, 2016). Culturally, UCFs are tasked with preserving and showcasing historical heritage and providing platforms for artistic creation and expression, thus fostering cultural innovation and enhancing public cultural literacy (Kagan et al., 2018).

Existing research not only highlights the importance of UCFs but also focuses increasingly on the issue of their inequity. These studies are primarily categorized into three areas: Firstly, addressing the needs of specific populations, Park and Lee (2015) evaluated the current state of supply and demand match for cultural facilities, while Cui et al. (2022) assessed the distribution equity and accessibility of rural cultural facilities based on surveys of individualized needs among farmers. Secondly, regarding the overall needs of urban populations, Wei et al. (2022) and Jiang (2023) explored the spatial equity of cultural facilities, whereas Dadashpoor et al. (2016) examined the spatial mismatch between cultural facilities distribution and population distribution in Hamadan city. Thirdly, considering market demands such as tourism and commerce, Boudiaf (2016) and Doling and Gibson (1979) analyzed the balance between supply and preservation of UCFs using census data. These studies underscore the necessity of optimizing the provision of public services and provide empirical and theoretical support for broader UCFs research.

Simultaneously, research on UCFs explores issues of public service equalization at different scales. At the national level, studies in countries like the USA, China, and the UK predominantly analyze how national policies shape the overall layout of cultural facilities and their connection with urbanization processes (Ballatore and Candlin, 2023; Donnelly, 2014). However, such research often overlooks in-depth exploration of local characteristics and dynamics. Provincial-scale studies highlight the impact of economic development and cultural policies (such as integrated tourism) on the distribution of cultural facilities within provinces (Tytelova et al., 2021), but often lack consideration of inter-provincial cultural facility cooperation and impact. Urban-scale research concentrates on meeting the needs of different demographic groups and the supply of UCFs within the city, such as UCFs studies in cities like Tshwane, South Africa, and Beijing, China, focusing on policy assessment, facility accessibility, and community participation (He et al., 2021; Nawa, 2018). However, urban research often neglects the impact of interregional linkage effects (like a watershed area). These studies aim to elucidate the contribution of cultural facility configuration to social equity and cultural development, yet require further exploration, particularly in considering regional differences and effects.

Determinants influencing urban cultural facility development

UCFs are crucial in shaping urban development while also being influenced by a multitude of factors within cities. Although some theories explain UCF growth or changes (Arnaboldi and Spiller, 2011; Yang et al., 2016), the supply-demand theory provides a more precise understanding. According to Green et al. (2005), the supply-demand theory explains UCF growth or changes by examining how the balance between the availability of cultural facilities (supply) and the needs or desires of the population (demand) influences their development.

On the demand side, as urban populations grow or cultural interests evolve, the demand for various types of UCFs (such as museums, theaters, and galleries) increases. High demand in densely populated or economically prosperous areas can drive the expansion and development of these facilities. Population, income and consumption, and education are significant drivers affecting UCF demand. Specifically, factors like population growth, density, and targeted groups such as college students are critical in driving demand for UCFs (Mouate and Travers, 2024; Smith and Bond, 2019). As societies progress and urban lifestyles become more prevalent, cultural services become integral to residents’ daily lives and long-term development plans (Zhao et al., 2022). Moreover, shifts in consumer attitudes and income disparities are key factors affecting UCFs demand (Tubadji et al., 2022; Wanyan and Wang, 2022). Urban residents are increasingly willing to invest more time and money in high-quality and diverse cultural activities, with higher-income groups typically having greater expectations and demands for cultural facilities. Lastly, educational advancement is a significant driver of UCFs demand (Mouate and Travers, 2024), as overall improvements in educational levels directly increase the demand for knowledge acquisition and cultural experience institutions. Pricing and diversity in cultural facilities also play roles in shaping demand (Borowiecki and Castiglione, 2014; Fincher et al., 2014).

From the supply side, the availability of UCFs is influenced by factors such as government funding, innovation, and urban status. Areas with better financial resources or supportive policies are likely to see more UCFs. Factors like the adoption of innovative technologies, financial investments, and urban administrative rankings are crucial in shaping UCF provision (Grodach, 2017; Woronkowicz, 2013). Urban planners and policymakers anticipate and evaluate specific cultural needs, aligning them with broader strategies for industrial innovation, such as incorporating new technologies and creating new job opportunities in the cultural sector (Grodach and Loukaitou‐Sideris, 2007). Additionally, cities with different administrative levels vary in their cultural transmission roles, with higher-level cities carrying greater responsibility for cultural promotion and preservation, necessitating additional cultural facilities for adequate support. Concurrently, supply costs such as land prices, construction expenses, and operational costs can also disrupt the provision of UCFs (Anelli and Tajani, 2022).

Empirically, various scholars have investigated the factors influencing UCFs from a supply-demand view. For instance, Sung and Ki (2023) found that socioeconomic factors significantly influenced the spatial distribution of art galleries in Seoul, while Kim (2007) emphasized the role of the creative economy in promoting Seoul’s art gallery clusters. Altunkasa et al. (2017) explored the relationship between cultural facility distribution and green spaces in the Adana community in Turkey. Wang et al. (2010) analyzed how innovation influenced the development of urban cultural facilities in Los Angeles. While single-factor analysis plays a crucial role in revealing specific pathways of influence, it often fails to capture the comprehensive effects of the entire system. Therefore, in addition to focusing on individual factors, research also delves into the complex interactions between multiple factors influencing spatial changes in UCFs (Woronkowicz, 2013). For example, Lu (2022) showed that the spatial layout of cultural facilities in Zhengzhou was shaped by both resident factors, such as consumption habits and cultural levels, and external factors, such as population, transportation, and economic conditions. Zhang et al. (2017) in the Greater London area found that museum numbers were linked not only to the business environment but also to composite demand factors, including resident and tourist demands. These findings highlight the importance of a comprehensive understanding of the multidimensional factors affecting spatial changes in UCFs (Shorabeh et al., 2020).

However, current research in exploring these issues has certain limitations. Studies on complete regions crossing administrative boundaries are relatively scarce, particularly evident in research on economic zones and watersheds, limiting our understanding of complex interactions within regions and their impact on urban cultural facility layouts. Additionally, long-term spatiotemporal dynamic studies are lacking, failing to fully reflect the changing trends of UCFs. The over-reliance on traditional statistical data also obscures the potential of POI big data in UCF distribution research. Lastly, the impact of multidimensional factors such as population demand and innovation on the spatiotemporal dynamics of UCFs is often overlooked.

Towards a conceptual framework

This study focuses on the YREB as the research area to explore the dynamic changes and influencing factors of UCFs. As a key area in China’s major strategy, with its rich economic and population resources, diverse cultures, and varying city types, YREB serves as a prime example for studying UCFs from a regional view.

A theoretical framework has been developed to analyze the changes in UCFs and their influencing factors (refer to Fig. 1). This framework identifies four primary dimensions influencing UCFs dynamics: economic development, population demand, policy support, and innovation. Each dimension includes specific indicators. Specifically, the economic development dimension includes indicators such as per capita GDP, per capita consumption expenditure of urban residents, and domestic tourism revenue. The population demand dimension covers indicators like population size, urbanization rate, and the number of undergraduate and junior college students on campus. Policy support includes urban hierarchy level, education expenditure in the general public budget, and the scientific and technological investment in the general public budget. The innovativeness dimension involves the amount of invention patents granted, employment of scientific research and technical services industry, and general institutes of higher education. These factors are hypothesized to influence the changes in UCFs. The following sections will utilize this framework to further analyze the spatiotemporal dynamics of UCFs and their specific influencing factors.

Fig. 1
figure 1

A conceptual framework.

Full size image

Materials and methodology

Study area

The YREB evolved from the “industrial intensive belt of the Yangtze River” and currently covers nine provinces and two municipalities directly under the Central Government, including 130 administrative units. It spans China’s east, central, and western regions, occupying 21% of China’s land area and gathering over 40% of China’s population and economic aggregate (Bi et al., 2023; Yang et al., 2021). As a region with high potential for cultural and economic development, this study selected 110 prefecture-level cities in the YREB as study areas, after excluding areas lacking data (See Fig. 2).

Fig. 2
figure 2

Study area.

Full size image

Data source and variables

Cultural facilities data

UCFs, vital carriers and disseminators of cultural functions, not only undertake cultural education and outreach tasks but also provide leisure and entertainment services (Tomasello et al., 2005). These facilities are categorized into NPUCFs and PUCFs, based on their operational models (Chirieleison and Montrone, 2013). In this study, we synthesized definitions of UCFs, relevant literature, and the availability of data from Amap (Point of Interest, POI) to identify 15 types of UCFs.

NPUCFs include museums (MUS), archives (ARC), memorial halls (MH), churches (CHU), science and technology museums (STM), art galleries (AG), aquariums (AQU), temples (TEM), planetariums (PLAN), libraries (LIB), cultural centers (CM), and exhibition halls (EH). These 12 facilities primarily serve cultural promotion and educational purposes (Chatterjee and Camic, 2015; García-Muiña et al., 2019; Mura, 2009). Conversely, cinemas (CIN), theaters (THE), and concert halls (CH) are categorized as PUCFs, primarily satisfying residents’ cultural and entertainment needs (Wang and Zou, 2021; Zhao et al., 2022).

For this research, POI data for these 15 types of UCFs were obtained for the years 2012 and 2020 through Amap (http://ditu.amap.com/). After meticulous verification, deduplication, and spatial matching processes, we successfully compiled an exhaustive dataset of 102,671 UCFs across the study area. Between 2012 and 2020, there was an increase of 57,557 UCFs, as detailed in Table 1.

Table 1 The quantity and distribution of POI of UCFs.
Full size table

Influencing variables

The determinants of UCFs are determined by a combination of the influencing factors and the correlations between them. Therefore, the selection of the most appropriate influencing factors is crucial to ensure the rationality of the results (Zhang et al., 2017). The selection of influencing factors is a complex and challenging process, and we referred to several key relevant studies (Cheng et al., 2021; Guo et al., 2017), most of which followed the principles of systematicity, representativeness, timeliness, and regional specificity in constructing the system of influencing factors. Accordingly, this study identifies four principal categories of key drivers: economic development, population demand, policy support, and innovativeness. To ensure a comprehensive and precise analysis, twelve representative indicators have been carefully selected from these categories, as detailed in Table 2.

Table 2 Variable definitions and computation methods.
Full size table

Firstly, economic elements play an indispensable role in public goods provision. Specifically, per capita GDP is a key indicator of overall economic levels, reflecting average economic capacity and influencing cultural consumption capabilities and the demand for high-quality UCFs (Scott, 2004). Domestic tourism revenue directly relates to the attractiveness of cultural facilities; the development of tourism often relies on rich cultural resources, such as museums and historical sites, to attract visitors (Bowitz and Ibenholt, 2009). Additionally, the per capita consumption expenditure of urban residents indicates purchasing power, where higher income levels can enhance participation in cultural activities and consumption of cultural products, particularly for PUCFs (Bille and Schulze, 2006).

Secondly, population support is another crucial determinant of UCFs. Key indicators selected include population size, urbanization rate, and the number of undergraduate and junior college students on campus. Population size directly determines the overall demand for UCFs (Woronkowicz, 2013)—the larger the population, the greater the demand for both NPUCFs and PUCFs. The urbanization rate reflects the level of urban development and the modernization of residents’ lifestyles, with higher rates typically indicating stronger dependence on and demand for UCFs (Chang et al., 2023). The number of students in higher education indicates the city’s educational level and the size of its youthful population, who are major participants and drivers of cultural activities and are particularly sensitive to the demand for UCFs (Çelik and Akyol, 2015).

Thirdly, policy support is a critical dimension, with indicators such as urban hierarchy level and public expenditures on education and technology. Higher city levels often mean stronger policy support and richer resource allocation, enabling cities, due to their political and economic status, to attract and implement more investments in cultural facilities (Kong, 2007). In China, cities with higher administrative levels, such as municipalities directly under the central government, sub-provincial cities, and provincial capitals, generally provide more substantial government support for cultural entities like museums and theaters. For example, in 2020, Zhengzhou, a provincial capital, allocated 18.8 billion RMB in fixed assets for culture and entertainment, more than double the investment of Xinyang, a prefecture-level city in the same province. The proportion of educational expenditure in the public budget reflects the government’s financial commitment to cultural and educational infrastructure, indirectly fostering the development of non-profit UCFs like science museums and museums, often in collaboration with educational institutions to provide learning and research platforms (Mosaferi et al., 2021). Technology expenditure highlights the government’s emphasis on investment in innovation and technological development, directly impacting the modernization and technical content of facilities like science museums and planetariums.

Lastly, innovativeness is identified as a key dimension, focusing on indicators such as the amount of invention patents granted, employment in scientific research and technical services industry, and general institutes of higher education. The amount of invention patents granted reflects the region’s innovative capacity and technological development, directly influencing the modernization of cultural facilities (Romanchukov and Berestneva, 2018). Employment in scientific research and technical services industry indicates the demand for highly skilled research personnel, crucial for enhancing the scientific displays and interactive experiences in facilities like science museums and planetariums (Bakhshi and Throsby, 2012). The general institutes of higher education serve not only as an indicator of educational resources but also reflect the scale of potential cultural consumers and innovative talent in the region, crucial for the demand and development of UCFs (Altintas and Kutlu, 2021).

Research methods

Exploratory spatial data analysis

Exploratory Spatial Data Analysis mainly includes two aspects: Global Moran’s I and Local-Moran’s I (Griffith, 2018). Global Moran’s I can describe the overall spatial correlation and difference characteristics of UCFs in the YREB, while Local Moran’s I can describe the Local aggregation characteristics of UCFs in the YREB. In this study, Global-Moran’s I and Local Moran’s I were used to analyze the overall and Local spatial characteristics of UCFs’ dynamic changes in the YREB. This study utilized GeoDa software (version 1.20) and ArcGIS software (version 10.8) for model computations and the visualization of results.

GeoDetector

The GeoDetector is employed to detect spatial heterogeneity and identify influencing factors. Specifically, the factor detection function analyzes the spatial differentiation of the dependent variable Y and explores the influence of the independent variable X on Y, quantifying the extent to which X explains the spatial variation of Y (measured by the q value). Additionally, the interaction detection feature is used to identify interactions between different risk factors Xs, assessing the increase or decrease in explanatory power for Y when multiple factors interact.

The focus of this study is to explore the spatial heterogeneity characteristics and determining factors of changes in UCFs, making it a suitable application scenario for GeoDetector. We utilized ArcGIS software (version 10.8) and the “GeoDetector_2015_Example (NDVI Dataset).xlsm” tool for data classification transformation and model computations. Furthermore, we assessed the variance inflation factor (VIF) of the independent variables, post-classification transformation, using STATA 17.0, which yielded a VIF of 3.65—significantly lower than the threshold of 10—indicating that multicollinearity is not an issue. By applying the factor and interaction detection features of the GeoDetector, this study identifies the principal factors and their interactions affecting the spatial distribution and dynamic changes of UCFs in the YREB.

$$q=1-\left(\mathop{\sum }\limits_{h=1}^{L}{N}_{h}{\sigma }_{h}^{2}\right)/N{\sigma }^{2}$$
(1)

Where L is the number of layers; N is the total number of urban samples, which is 115. σ2 is the total discrete variance of UCFs of cities in the YREB; \({\sigma }_{h}^{2}\) is the variance of UCFs at layer h. Interactive exploration can explore the effects of the combination of two different factors, including nonlinear weakening, unilinear weakening, double factor strengthening, nonlinear strengthening and mutual independence (Wang et al., 2016). The relevant judgment formulas are as follows:

$${\rm{Non}}-{\rm{linear}}\; {\rm{attenuation}}:{\rm{q}}\left({\rm{X}}1\cap {\rm{X}}2\right) \,< \,{\rm{Min}}\left({\rm{q}}\left({\rm{X}}1\right),{\rm{q}}\left({\rm{X}}2\right)\right)$$
(2)
$${\rm{Single}}-{\rm{factor}}\; {\rm{non}}-{\rm{linear}}\; {\rm{attenuation}}:{\rm{Min}}\left({\rm{q}}\left({\rm{X}}1\right),{\rm{q}}\left({\rm{X}}2\right)\right)\, < \,{\rm{q}}\left({\rm{X}}1\cap {\rm{X}}2\right)\, < \,{\rm{Max}}\left({\rm{q}}\left({\rm{X}}1\right),{\rm{q}}\left({\rm{X}}2\right)\right)$$
(3)
$${\rm{Dual}}-{\rm{factor}}\; {\rm{enhancement}}:{\rm{q}}\left({\rm{X}}1\cap {\rm{X}}2\right) \,>\, {\rm{Max}}\left({\rm{q}}\left({\rm{X}}1\right),{\rm{q}}\left({\rm{X}}2\right)\right)$$
(4)
$${\rm{Independence}}:{\rm{q}}\left({\rm{X}}1\cap {\rm{X}}2\right)={\rm{q}}\left({\rm{X}}1\right)+{\rm{q}}\left({\rm{X}}2\right)$$
(5)
$${\rm{Non}}-{\rm{linear}}\; {\rm{enhancement}}:{\rm{q}}\left({\rm{X}}1\cap {\rm{X}}2\right) \,> \,{\rm{q}}\left({\rm{X}}1\right)+{\rm{q}}\left({\rm{X}}2\right)$$
(6)

Multiscale geographically weighted regression

The factors influencing the changes in UCFs exhibit typical scale effects and heterogeneity in geographical space. Consequently, this study employs the MGWR model for further analysis. The MGWR model, by relaxing the assumption of a single bandwidth for all variables, identifies the roles of various covariates at local, regional, and global scales, thus more accurately depicting the actual spatial processes (Yu et al., 2020). The fundamental regression equation of the MGWR model can be expressed as:

$${{\rm{y}}}_{{\rm{i}}}=\mathop{\sum }\limits_{{\rm{j}}=1}^{{\rm{k}}}{{\rm{\beta }}}_{{{\rm{b}}}_{{\rm{w}}}{\rm{j}}}({{\rm{u}}}_{{\rm{i}}},{{\rm{v}}}_{{\rm{i}}}){{\rm{x}}}_{{\rm{ij}}}+{\rm{\varepsilon }}$$
(7)

Where \({{\rm{y}}}_{{\rm{i}}}\) represents the dependent variable, \({x}_{{ij}}\) the covariates; \({{\rm{\beta }}}_{{{\rm{b}}}_{{\rm{w}}}{\rm{j}}}\) denotes the local regression coefficient for bandwidth \({{\rm{b}}}_{{\rm{w}}}\) at location \({\rm{j}}\)；\(({{\rm{u}}}_{{\rm{i}}},{{\rm{v}}}_{{\rm{i}}})\) epresents the spatial geographic location of the subject \({\rm{i}}\)；\({\rm{\varepsilon }}\) is the model regression residual. Each regression coefficient \({{\rm{\beta }}}_{{{\rm{b}}}_{{\rm{w}}}{\rm{j}}}\) in MGWR is derived from local regression, and the optimal bandwidth is determined.

Compared to the traditional GWR model, the MGWR model demonstrates lower AICc values and higher adjusted R2 values, thus confirming its superior model fitting performance (Roy et al., 2024) (see Table 3). First, we used STATA 17.0 to measure the VIF of the independent variables (original values), which was found to be 2.85 (far less than 10), avoiding the influence of multicollinearity. Using MGWR software (version 4.2) and ArcGIS software (version 10.8), we identified the core driving factors for each city and conducted interpolation analysis and visualization of the significance (P values) of these factors. These analytical results provide crucial references for proposing specific optimization strategies.

Table 3 Comparison of GWR and MGWR fitting results.
Full size table

Results

Spatial Differentiation of UCFs from 2012 to 2020

The spatial dynamics and clustering of UCFs

Figure 3 illustrates significant regional disparities in the dynamic changes of UCFs and distinct growth patterns among various sub-categories of UCFs. Overall, from 2012 to 2020, the number of UCFs increased by 2.6 times, with NPUCFs growing at a much faster rate than PUCFs. Spatial characteristics of UCFs and their non-profit and profit-oriented branches tended to solidify, with major growth concentrated in the eastern part of the YREB, as well as in western cities like Chengdu and Chongqing, displaying a higher concentration compared to the central region. Specifically, significant growth and concentration were observed in provincial capitals of Sichuan, Hebei, Hunan, Zhejiang, Anhui, Jiangsu, cities along the eastern coast of the YREB, and places like Chongqing. These cities typically possess higher administrative levels and stages of economic development, whereas other prefecture-level cities showed lower growth. From the perspective of sub-facilities, cities with lower consumer spending on cultural facilities such as libraries, cultural museums, and cinemas experienced rapid growth, while facilities catering to higher-income groups, such as art galleries and concert halls, grew more slowly. Additionally, faith-related facilities like temples and churches saw the fastest growth, followed by comprehensive facilities, whereas specialized institutions like science museums, aquariums, and planetariums had the slowest growth.

Fig. 3: Spatial analysis of UCFs from 2012 to 2020.
figure 3

a Spatial overlap histogram of UCFs, b Spatial overlap histogram of NPUCFs, c Spatial overlap histogram of PUCFs, d Spatial clustering of UCFs, e Spatial clustering of NPUCFs, f Spatial clustering of PUCFs.

Full size image

The spatial auto-correlation analysis of dynamics of UCFs

Figure 4 demonstrates that the spatial patterns of UCFs and NPUCFs exhibit similarities, whereas the spatial distribution of PUCFs is more dispersed. However, a closer examination reveals that PUCFs tend to cluster in larger cities. A global analysis (see Table 4) shows significant spatial clustering of UCFs and NPUCFs, in contrast to the less pronounced spatial clustering of PUCFs. From a local perspective (see Fig. 4), UCFs and NPUCFs are concentrated in specific areas such as the YREB cities, including the Suzhou and Jinhua areas, along with other scattered cities like Zunyi, a city known for red tourism. In contrast, PUCFs are primarily clustered around major cities in the YREB, particularly around Shanghai and Suzhou. Additionally, ‘star’ regions such as Chongqing, Chengdu, Kunming, and Hefei show significant clustering of PUCFs. Surrounding areas of Chongqing exhibit significantly lower growth of UCFs compared to neighboring cities, becoming major ‘collapse’ areas.

Fig. 4: LISA maps of UCFs from 2012 to 2020.
figure 4

a UCFs, b NPUCFs, c PUCFs.

Full size image
Table 4 Global Moran’s I of UCFs change in YREB.
Full size table

Factors influencing the dynamics of UCFs in the YREB

Single factor detection of UCFs, NPUCFs, and PUCFs

Table 5 represents the detection power(q-value) of each studied factor for the distribution and change of UCFs from 2012 to 2020, especially highlighting the top three factors in terms of explanatory power. All single-factor effects are positive and passing the 5% significant test, except per capita GDP(X1) and urbanization rate(X5) in certain detection. Regarding spatial heterogeneity, the three variables in the policy dimension (X7, X8, and X9) occur the top three drivers for each UCFs multiple times, all passing the 1% significance test, highlighting the dominant role of policy in UCF distribution and change. According to the result of the change of all-type UCFs between 2012 and 2020, the policy support dimension wins the strongest explanatory power, followed by population demand and innovativeness, with economic development being the least crucial.

Table 5 Factor detection results of the dynamic change of UCFs in the YREB.
Full size table

Factors affect the inter-city distribution of UCFs have changed over time. In 2012, X8, X2, and X9 were the main influences on UCF distribution, indicating urban economic development and related policy support for the early construction of cultural facilities in the YREB. By 2020, X8, X4, and X10 became the key factors, reflecting a shift towards marketization and technology in UCFs. Education expenditure in the general public budget (X8) consistently remained the core factor for its highest q value.

The high driving factor combinations for NPUCFs were similar to those for all UCFs, with education expenditure in the general public budget (X8) still being the most significant, although with a slightly lower q value. For PUCFs, X8, X9, and X10 were the core drivers both in 2012 and 2020, suggesting a trend to invest cultural industry towards cities with active policy in science, technology, and innovation. Therefore, cultivating regional scientific and cultural potential may draw PUCFs and provide extra urban cultural well-being for the YREB cities.

The factor analysis of UCF changes from 2012 to 2020 reveals that X8, X7, and X4 are the main drivers of changes in all UCFs. The strong individual effects of urban hierarchy status and education expenditure in the general public budget highlight the importance of urban government decisions. The core drivers for NPUCFs changes align with those for UCFs, while for PUCFs, X7, X8, and X9 are the main drivers with positive effects. These detection results point out that investments in science and technology education not only enhance public cultural facilities but also encourage the development of cultural venues. Furthermore, the number of undergraduate and junior college students on campus (X6) has a significant impact on UCFs, NPUCFs, and particularly PUCFs, for which it demonstrates stronger explanatory power. This is because university students are one of the primary consumer groups of public cultural facilities due to their accessibility and affordability. In the cultural consumption patterns of university students, venues such as cinemas and theaters—which are PUCFs—constitute the main types of facilities they frequent.

Single factor detection of urban cultural sub-facilities

In order to explore the inner difference of UCFs changing precisely, the explanatory power of X1-X12 on urban cultural sub-facilities is categorized based on the single-factor detection of each UCFs’ type. Figure 5 shows two main external factors influencing modes. For most urban cultural sub-facilities, the changes in their numbers from 2012 to 2020 are predominantly driven by X7, X8, X9, and X4, with these factors having a stronger impact than others. Policy support (X7-9) and population size(X4) generally boost the growth of NPUCFs and PUCFs, while per capita GDP(X1), per capita consumption expenditure of urban residents(X2), urbanization rate(X5), and general institutes of higher education(X12) are the least influential. However, temples and planetariums are less affected by these factors, showing relatively low q values. It’s likely due to the stronger influence of local environmental restrictions and historical constraints on these types of UCFs.

Fig. 5: Factor detection results of the dynamic change of urban cultural sub-facilities.
figure 5

a Factor detection results of MUS, ARC, MH, CHU, STM, b Factor detection results of AG, AQU, TEM, PLAN, LIB, c Factor detection results of CM, EH, CIN, THE, CH.

Full size image

Among the 12 sub-cultural facilities with clear factor-driven effects, each has its core drivers. Education expenditure in the general public budget (X8) strongly encourages NPUCFs’ boost. NPUCFs include a series of public education sites, for instance, aquariums as a natural study place and MH as a history classroom, so NPUCFs gain investment from the education department. The proportion of education expenditure also connects regional talent and cultural development strategy. Urban hierarchy status(X7), reflecting a city’s regional ranking, shows that higher-ranked cities, with more cultural resources and advanced cultural services, draw high-end cultural experience consumption. Thus, facilities providing drama, fine art, and other performance (museums, archives, exhibition Halls, and theaters) were gathered. The scientific and technological investment in the general public budget (X9) significantly explains changes in science and technology museums. Moreover, population size (X4), representing city size, has a more noticeable impact on PUCFs such as cinema than on NPUCFs.

Interaction factor detection of the dynamic change of UCFs、NPUCFs, and PUCFs

The interactive effects analysis revealed that among the three groups of interactions studied, UCFs, NPUCFs, and PUCFs showed both two-factor enhancement effects and nonlinear enhancement effects. This suggests that the interactions among driving factors have improved the explanatory power regarding changes in various types of UCFs from 2012 to 2020. Nonlinear effects are particularly prevalent between auxiliary factors(0<q ≤ 0.3) and other factors. And the strong interactions(q > 0.8) emerge mostly driven by core factors(q > 0.6).

The growth of UCFs is not the result of one factor alone but depends on the combined effect of the policy support, economic development, population demand, innovativeness, and so on. For UCFs and NPUCFs (Fig. 6a, b), nonlinear enhancement effects are generally stronger than two-factor ones. Similar interactions between domestic tourism revenue, the amount of invention patents granted, and education expenditure in the general public budget (X3 ∩ X8, X10 ∩ X8) significantly drive the increase of UCFs and NPUCFs. For PUCFs (Fig. 6c), numerous effective two-factor interactions centered around factors of the dimension of policy support were noted, underscoring the central role of policies in PUCF development. The significant nonlinear effects from per capita GDP(X1), domestic tourism revenue(X3), and the number of undergraduate and junior college students on campus (X6) highlight the importance of considering how economic development and urbanization assist urban innovation policies to attract PUCFs.

Fig. 6: The detection paths and intensities of dynamic change from 2012 to 2020.
figure 6

a UCFs, b NPUCFs, c PUCFs.

Full size image

Spatial differences of core factors influencing the change of UCFs, NPUCFs, and PUCFs

Figure 7 shows the core factors with highest regression coefficients of each city in the YREB analysed by MGWR model. Results reveal that core factors for UCFs show provincial and spatial differences between upstream and downstream areas along the YREB. For all UCFs, the core factor in the middle and lower reaches is education expenditure in the general public budget (X8), while in the upper reaches, it is the amount of invention patents granted (X10). This spatial difference suggests that from 2012 to 2020, the upper regions relied more on technological innovations to boost regional cultural welfare, while the lower regions benefited more from educational constructions and policies in developing cultural facilities. NPUCFs in the middle and lower Yangtze River areas have been influenced by population size(X4). Cities where NPUCFs boosts are sensitive to population size are concentrated in Zhejiang Province, Shanghai, and nearby areas at the eastern end of the YREB. Additional NPUCFs may be invested to meet urban living needs under the context of urban expansion. For PUCFs, the spatial distribution of core factors shows regional patterns: education expenditure in the general public budget (X8) is the main driver for downstream cities, the amount of invention patents granted (X10) for northern cities in the middle and upper reaches, and the employment of scientific research and technical service industry(X11) for southern cities in those areas. The link between technology service employment and the growth of cultural facilities highlights the positive impact of innovation in central Yangtze River cities on cultural welfare.

Fig. 7: Spatial heterogeneity of core driving factors and their p values of dynamic changes from 2012 to 2020.
figure 7

a UCFs, b NPUCFs, c PUCFs.

Full size image

Meanwhile, the MWGR results also exhibit regional patterns in significance. While all results passed the 5% significance level, there is a noticeable decrease in significance in the middle and lower reaches area. For all UCFs, cities in the northern end of basin divider are less significant, so does plenty lower reaches cities for NPUCFs and few southern middle reaches cities for PUCFs. This trend may reflect the influence of changes in urban cultural development strategies in the downstream regions during 2012–2020. Less significant areas may experience more complex effects on NPUCFs’ development, as this phenomenon happened in lower reaches, the economically vibrant and culturally diverse areas of the YREB.

Discussion

The investigation into spatial processes of UCFs at a regional scale and their respective impacts remains notably under-explored in academic research. In response to this, the present study synthesizes the primary research findings on the spatio-temporal characteristics of the dynamics of UCFs in the YREB and their influencing factors. We further reflect on the findings.

First, this study has established an analytical framework to explore the spatiotemporal variations of UCFs within the YREB, offering new perspectives on UCFs at the regional-scale UCFs. Previous studies have largely focused on static factors influencing UCFs (Guo et al., 2017; He et al., 2021; Zhang et al., 2017) or on isolated or specific categories of UCFs (Andrade-Agudelo et al., 2023; Chang et al., 2021; Wang and Zou, 2021). In contrast, our research adopts a dynamic perspective, thoroughly assessing the factors influencing 15 types of both profitable and non-profit UCFs in the YREB, and elucidating the spatial differentiation of the impact effect of key factors. This holistic and dynamic approach provides vital theoretical insights and actionable guidelines for other nations or regions to develop diverse cultural policies and to refine the scale and efficiency of UCFs in stages. Then, the discourse on equity of UCFs at the watershed or regional scale, particularly in emerging economies, has been limited. Our research contributes to filling this gap and suggests policy recommendations for enhancing public service equity at a regional level.

Second, the study reveals key factors influencing the dynamic changes of UCFs, emphasizing the enhancing effects of variable interactions. It confirms significant impacts from resident consumption (Lu, 2022), tourism development (Ballatore and Candlin, 2023), population demand, and educational levels (Woronkowicz, 2013) on UCFs. Education expenditure, for instance, consistently stands out as a strong core factor affecting UCFs, including both NPUCFs and PUCFs. According to demand-pull theory, highly educated populations have a greater demand for UCFs and their services (Mouate and Travers, 2024). Additionally, human capital theory suggests that investing in education enhances workforce skills and productivity (Annabi, 2017), providing a sustained human capital reserve for cultural industries and institutions, thereby promoting a balanced distribution of UCFs.

Then, there are significant differences between NPUCFs and PUCFs in terms of core influencing factors. For NPUCFs, the key factors are science and technology expenditure and the number of granted invention patents. For PUCF, the core influencing factor is the administrative level of the city. Based on public goods theory, NPUCFs are primarily supplied by the government and are considered public goods (Bille, 2024). The Chinese government’s “innovation-driven” strategies introduced in 2012 and 2016 have increased science and technology expenditure and patent activities while enhancing public awareness and demand for scientific culture, necessitating a substantial number of NPUCFs to meet and sustain this demand. Conversely, central place theory and institutional economics posit that PUCFs are mainly market-supplied (Hesmondhalgh, 2010). High-level cities exert a siphoning effect on the populations, industries, and resources of surrounding areas (Guan et al., 2018) and benefit from more comprehensive institutional environments and policy support (Marton and Wu, 2006). Their vast market potential attracts significant investment and establishment of PUCFs.

Besides, the study identifies differences in the core influencing factors among 15 subtypes of facilities. Unlike other sub-facilities, temples and planetariums are strongly affected by per capita consumption expenditure, underscoring the close link between residents’ spending and their spiritual and cultural needs. Cultural consumption theory posits that increased consumer capacity leads to higher demand for quality cultural products and services (Assadourian, 2010). By offering unique cultural experiences (such as religious and tourism experiences) and educational experiences (like high-tech services), temples and planetariums cater to the needs of high-spending groups. Understanding this mechanism aids in formulating cultural policies that optimize the cultural needs of residents across different levels.

In addition, this study extends existing research on the impacts of population mobility and tourism on UCFs in countries such as Turkey and the United Kingdom (Ballatore and Candlin, 2023). Through extensive testing of the interactions between tourism revenue and factors such as economics, population, policy, and innovation, it has been revealed that tourism revenue often exhibits non-linear and dual-factor enhancement effects when interacting with these variables. This finding confirms the inherent synergy between culture and tourism (Boudiaf, 2016). Culture enriches tourism with content and meaning, while tourism provides a platform for cultural dissemination and economic support. Furthermore, tourism maximizes its impact on UCFs through coordinated development with other factors, such as enhancing service industry GDP, attracting investment and visitors, and driving innovation in products and talent (Liu and Wu, 2019).

Third, the study reveals the dynamic spatial variations of UCFs and the spatial heterogeneity of the impact effect of influencing factors across different cities, laying a foundation for more targeted spatial policy strategies. In high-ranking cities (e.g., those with higher administrative levels or more developed economies), UCFs generally grow faster than in less developed cities. This distribution pattern is evident both in developed and developing countries. For instance, in developed nations, the northeastern United States and its major metropolitan areas possess an abundance of library resources (Donnelly, 2014), while the per capita number of museums in England’s regions significantly exceeds that in other areas (Ballatore and Candlin, 2023). In developing countries, such as South Africa’s Tshwane and China’s eastern coastal cities and major urban agglomerations, UCFs are primarily located in economically more developed areas (Nawa, 2018; Zhao et al., 2023). These regions become hubs for UCFs due to high demand and their attractiveness, further exacerbating spatial supply disparities. Additionally, this study identifies the spatial heterogeneity of the impact effect of influencing factors across cities. This study reveals that the impact of influencing factors on UCFs varies significantly between cities, highlighting the need for tailored policies that address these spatial differences.

Finally, policy recommendations are proposed. Initially, the research emphasizes optimizing facility categories by increasing PUCFs in major cities and boosting NPUCFs investments in smaller cities. Given the slow growth of PUCFs, policy incentives and industry partnerships can stimulate economic activity and meet higher entertainment demands in large cities. In smaller cities, increasing NPUCF investment addresses service gaps effectively. Then, the study also proposes a stratified implementation strategy based on the significance of core influencing factors (strong, medium, weak) and their impact across cities (fast, medium, slow). This approach allows for tailored short-term, medium-term, and long-term projects and policies specific to each urban area. Additionally, the research examines non-linear interactions among factors such as educational investment, patent grants, and tourism revenue. By focusing on these interactions, the study aims to enhance overall effectiveness through better coordination between different departments.

Concluding remarks and limitations

The equalization of UCFs is a crucial aspect of achieving urban equity. This study employed ESDA and GeoDetector methods to systematically analyze the spatiotemporal dynamics and influencing factors of UCFs and their subcategories in the YREB from 2012 to 2020, providing a theoretical foundation for future urban cultural facility planning and optimization. The study highlights the importance of planning and policy interventions in ensuring the equitable distribution of cultural facilities, particularly by supporting underdeveloped regions to foster the development and operation of cultural infrastructure. Additionally, the study reveals the distinct cultural demand patterns in different cities, offering valuable insights for optimizing facility layout, improving utilization rates, and enhancing market efficiency, especially in core cities with high cultural demand. These findings contribute to the sustainable utilization of cultural resources and promote greater equity in public services at the regional level.

The key findings are as follows:

First, UCFs, including both NPUCFs and PUCFs, show considerable spatial variation and clustering. NPUCFs exhibit stronger global clustering, while PUCFs demonstrate higher local clustering in several cities (i.e., municipalities, provincial capitals, and the Yangtze River Delta region). Second, policy support and population demand are major factors influencing the spatiotemporal dynamics of UCFs, with important variables being population size, urban hierarchy status, and education spending in the public budget. The interactions among these variables further strengthen their effects. Third, mapping the P-value of key factors shows significant spatial differentiation, particularly between upstream and downstream areas, with UCFs and NPUCFs decreasing from southwest to northeast, while PUCFs show no significant trend.

However, this study also acknowledges several limitations. Firstly, the inability to comprehensively collect specific cultural policies from various provinces and cities limited the assessment of these policies’ impacts. Moreover, the failure to consider the visitor volumes of different facility types made it challenging to accurately evaluate their effectiveness in meeting residents’ cultural needs. Thirdly, although GeoDetector was utilized to reveal linear and nonlinear relationships, this tool may not have fully captured all nonlinear trends. To address these limitations, future research will aim to obtain more comprehensive policy data from government agencies, integrate user access data from mobile platforms like Weibo, and employ machine learning to delve deeper into the nonlinear relationship between UCFs and influencing factors.