Introduction

Land is an essential resource that plays an important role in socioeconomic development and conservation of the ecological environment, and that provides a critical space that serves as a carrier for urbanization (Bazame et al. 2018; Liu et al. 2022). With the rapid development of urbanization around the world, the scale of urbanized land has expanded continuously (Seto et al. 2012; Li et al. 2018). From 1985 to 2015, the global area of urban land increased by 80%, and urban land became an increasingly important support for human lives and production activities (Liu et al. 2020). However, the rapid growth of urban land was not based on the efficiency of land use (Xue et al. 2022). The disorderly spread of urban land led to land and ecosystem degradation, seriously hindering sustainable development (Chakraborty et al. 2022). In September 2015, the United Nations General Assembly adopted the 2030 Agenda for Sustainable Development with 17 Sustainable Development Goals, which provided a blueprint for sustainable development (UN 2015; Zhang et al. 2022). In terms of urban land, the 11th sustainable development goal was “Sustainable Cities and Communities” and proposed to “make cities and human settlements inclusive, safe, resilient and sustainable” (Jiang et al. 2021). As an important component of this goal, the sustainability of urban land use has received widespread attention (Zhang et al. 2011; Lu and Ke 2018). In this paper, we focused on one specific aspect of urban sustainability: the growth in the area of urbanized land, which represents land that provides no ecological services and instead requires complete support from surrounding ecosystems. As a result, creation of this land (e.g., by replacing green space or cultivated land with buildings) decreases a city’s sustainability unless urban planners compensate for this loss by increasing the urban ecosystem’s capacity to provide this support.

China has experienced unusually rapid urbanization, leading to the largest flow in history of rural residents into urban areas, causing a proportionally rapid growth of the area of urbanized land (Friedmann 2006; Cai et al. 2021). In contrast with other countries and regions, China has found a peaceful and efficient road to urbanization (Cai et al. 2019). China’s land finance model has been an important tool for local governments to obtain large amounts of funds by selling land-use rights to citizens or enterprises, thereby supporting large-scale infrastructure construction and ultimately contributing to urbanization (Li 2011; Liu et al. 2018). However, this approach also created a serious problem of disordered growth of urbanized land, as local governments over-relied on the land finance model to meet their increasing financial needs (Huang and Chan 2018; Cheng et al. 2022). As a result, the disordered growth of urban land has caused urbanization of land to proceed much faster than urbanization of the population, leading to contradictions in human land use, such as building more apartments than there are people who can afford to live in them (Li et al. 2020). Moreover, when land is urbanized, it decreases the area of land that can sustain the city; the reduction of ecological or green space to make room for urban land threatens ecosystem security, and the reduction of cultivated land threatens food security, all of which pose a threat to sustainable development (Liu 2018; Yao et al. 2022). Therefore, understanding the factors driving the growth of urban land is a key challenge to effort to promote the sustainability of urban land use in China (Cui et al. 2019).

Many scholars have studied the sustainability of urban land use from different angles (Zhang et al. 2011; Jiang et al. 2021). First, scholars extended the concept of sustainable development to land use in the 1990s and proposed the concept of sustainable land use. For example, in 1993, the Food and Agriculture Organization chose land productivity, safety, conservation, economic feasibility, and social acceptability as their evaluation criteria for sustainable land use (Smyth and Dumanski 1993; Jiang et al. 2021). Second, in terms of evaluation methods, scholars developed many indicator systems and evaluation tools that accounted for nature, the environment, the economy, and society (Walter and Stützel 2009; Shen et al. 2010; Lu et al. 2022). Third, the evaluation results revealed that sustainable urban land use was influenced by human socioeconomic activities and the ecological environment, and that unsustainable land use exacerbated the damage to resources and the environment (Lu and Ke 2018; Pan et al. 2020). What is more, researchers demonstrated that the growth of urban land is an inevitable trend during rapid urbanization, and that this growth was closely related to the impacts for sustainable development (Pandey and Seto 2015; Cui et al. 2019). Effective management of urban land use should be based on constraining the growth of urban land to promote sustainable urban land use (Musakwa and Niekerk 2013; Li et al. 2020). However, focusing only on land-use problems is not enough to understand what factors drive the growth in the area of urbanized land (Koroso et al. 2020). It is necessary to identify these factors to provide effective guidance for urban land-use sustainability.

To identify the factors that influence growth of urban land in China, and that therefore decrease sustainability, we studied Chinese prefecture-level cities using spatial regression analysis models based on panel data from 2007 to 2019. We also used a geographical and temporal weighted regression (GTWR) model to explore the spatial differences among regions in the most significant driving factors, with the goal of proposing appropriate development strategies that account for these differences. Two innovations of our study are the combination of spatially explicit regression analysis with contribution analysis to identify the most important driving factors, as well as the goal of identifying differences among a large number of cities in a large number of key factors to reveal how the key factors varied among China’s regions. Our goal was to provide a valuable reference to support future plans to increase the sustainability of urban land use and to improve the likelihood of achieving sustainable urbanization.

Materials and methods

Data sources

We selected 295 prefecture-level cities in China as our research subjects (Fig. 1). The selection was based on the availability of comparable data for each city, and included prefecture-level cities, leagues in Inner Mongolia, and autonomous prefectures. This let us comprehensively reflect the current overall level of urbanization in China using the most recent data that are available (from 2007 to 2019). The data we used were obtained from the China City Statistical Yearbook, and from the Statistical Communique on the National Economy and Social Development of China for each city (NBS 2020; http://www.stats.gov.cn/).

Fig. 1
figure 1

Spatial distribution of the 295 prefecture-level Chinese cities used in our study

Full size image

Indicator selection

We selected the growth of urban land (the urban land area divided by the total land area) as the dependent variable. For the independent variables, we selected 11 indicators (defined in Table 1) from categories that described the economy, society, ecological environment, and geographical conditions that were shown to affect the growth of urban land in previous research: industrialization (Gollin et al. 2016), marketization (Liao and Wei 2015), industrial upgrading (Liang and Yang 2019), urban green space (Cai et al. 2021), fiscal decentralization (Gu and Wall 2007), education (Zhou and Zhao 2019), the unemployment rate (Wang and Wan 2014), information technology (Tan et al. 2021), infrastructure (Cai et al. 2019), terrain conditions (Sarker 2021), and environmental degradation (Amin et al. 2020). To ensure comparability of the data, which were obtained in different units of measurement and had different magnitudes, we standardized the data using Z-scores. We then used the variance inflation factor (VIF) to test for multicollinearity. The VIF results for all variables were less than 2 (Table 1). These criteria ensured that there was no significant multicollinearity among the chosen variables and that the model had been defined reasonably (Wang and Mu 2018; Wang et al. 2018). Note that our approach can be expanded to include a larger number of indicators or different indicators that have been shown to be important in a specific study area. In future research, our approach could also be used to focus more narrowly on sub-indicators of an indicator that is shown to be significant in the present study. This would provide more precision in designing policies to attack specific problems.

Table 1 Results of the indicator selection and multicollinearity test to select the key factors that have influenced the growth in the area of urbanized land in 295 Chinese cities from 2007 to 2019. TOL, tolerance factor; VIF, variance inflation factor
Full size table

Spatial autocorrelation analysis

First, we used spatial statistical analysis based on the Moran index (I) to test for spatial autocorrelation and heterogeneity of growth of urban land (Xu and Lin 2017). We employed the spatial autocorrelation analysis module of version 1.14 of the GeoDA software (https://geodacenter.github.io/) to calculate Moran’s I using the Queen matrix method:

$$I = \left( {\frac{N}{{\mathop \sum \nolimits_i \mathop \sum \nolimits_j {W_{ij}}}}} \right)\left( {\frac{{\mathop \sum \nolimits_i \mathop \sum \nolimits_j {W_{ij}}\left( {{x_i} - \bar x} \right)}}{{\mathop \sum \nolimits_i {{\left( {{x_i} - \bar x} \right)}^2}}}} \right),$$
(1)

where N is the number of spatial units (here, 295 cities); xi is the value of tested variable i for i = 1 to 11 variables; \(\overline {x}\) is the average value of the tested variable; and Wij is the corresponding value in the spatial weight matrix for variable i in city j. The global value of Moran’s I for the growth of urban land was positive and significant (p < 0.05) from 2007 to 2019, which showed that there was significant spatial autocorrelation.

Regression analysis

To ensure the validity of the results of our regression analyses and better reveal the factors that influenced the growth of urban land among the cities, we constructed a spatial panel model for empirical analysis. This model can overcome the spatial correlation and spatial heterogeneity caused by factors related to a city’s geographical location. The spatial panel models that have been used in previous research mainly include spatial autoregressive models (SAR), spatial error models (SEM), and spatial Durbin models (SDM) (Wang et al. 2022a). The formula for calculating an SDM is

$${Y_{it}} = \rho \mathop \sum \limits_{i,j = 1}^n {W_{ij}}{Y_{jt}} + \beta {X_{it}} + \gamma \mathop \sum \limits_{i,j = 1}^n {W_{ij}}{X_{jt}} + {\alpha_i} + {v_t} + {\varepsilon_{it}},$$
(2)
$${\varepsilon_{it}} = \lambda \mathop \sum \limits_{i,j = 1}^n {W_{ij}}{\varepsilon_{jt}} + \varphi ;i = 1, \ldots ,n;t = 1, \ldots ,T$$
(3)

where i and j represent different city indicators (295 cities), t is the research period (2007 to 2019), W is the spatial weight (Queen matrix), wij is the weight of variable i in city j, Y is the explained variable (growth of the urbanized land area from 2007 to 2019), and X is the explanatory variable (the 11 influencing factors), ρ is the spatial regression coefficient, ε is the random error term, and λ is the spatial autocorrelation coefficient of the error. When λ = 0, θ = 0, the equation becomes the calculation formula of an SAR model and when ρ = 0, θ = 0 and the equation becomes the calculation formula of an SEM model.

In the actual modeling process, we began by using an ordinary least-squares (OLS) model to diagnose the spatial dependence Lagrange multiplier results and judge which spatial regression model to choose (Anselin 1995; Table 2). Next, according to the likelihood ratio test results, we judged whether the SDM model can be degraded into the SEM and SAR models; if not, we would not construct the SEM and SAR models. Next, based on the results of the Hausman test, we determined whether fixed effects or random effects are needed in the model. Finally, we used a joint significance test to judge which of the individual effects, time effects, or combined individual and time effects was the best for the model to select the most suitable and reliable spatial regression model for this study. Based on the results of this analysis, the SEM model with fixed time was the most effective model to explore the factors influencing China’s growth of urban land from 2007 to 2019.

Table 2 Model test results for the factors that influence the comprehensive regional development level. Data used in the analysis were obtained from 2007 to 2019. DF, degrees of freedom; LM, Lagrange multiplier test; R-LM, robust-Lagrange multiplier
Full size table

Next, we used the absolute value of the regression coefficients in β to calculate the strength of the influence of the independent variables on the growth in the area of urban land (Cai et al. 2021). We defined this strength as the contribution, which was calculated as follows:

$$Co{n_K} = \frac{{A{C_K}}}{{\mathop \sum \nolimits_{K = 1}^{11} A{C_K}}} \times 100\%$$
(4)

where Conk is the contribution of variable k to the dependent variable (Y), and ACk is the absolute value of the coefficient βk.

Finally, we used a GTWR model to describe the spatial effect of the growth of urban land in China so as to more clearly reflect the spatial imbalance and spatial trends for the driving mechanisms (Fotheringham et al. 1998; Wang et al. 2022b). We selected the significance index in the GTWR model and carried out the VIF test by using version 10.7 of the ArcGIS software (www.esri.com). The calculation formula is

$${Y_i} = {\beta_0}\left( {{\mu_i},{v_i},{t_i}} \right) + \mathop \sum \limits_{j = 1}^n {\beta_{ji}}\left( {{\mu_i},{v_i},{t_i}} \right){x_{ij}} + {\varepsilon_i}$$
(5)

where β0 is an assumed constant, βji is the regression parameter for variable j in region i, and (μi, vi, ti) is the spatial and temporal coordinate for region i, and εi is a residual error term that follows a normal distribution, which we calculated using the GTWR function provided by ArcGIS. We then mapped the values of the key driving factors to display their spatial distribution intuitively.

Results

The global value of Moran’s I for growth of urban land was positive and significant (p < 0.05) from 2007 to 2019, with I values ranging between 0.21 and 0.41, which indicated that there was significant spatial autocorrelation in the growth of urban land. This means that areas with similar levels of growth of urban land tend to be spatially agglomerated in China. For example, areas with high growth of urban land are adjacent to each other and form strong–strong agglomerations (in which both parts of the agglomeration show strong growth); areas with low growth of urban land are also adjacent to each other to form weak–weak agglomerations (in which both parts of the agglomeration show weak growth). In addition, the spatial Moran’s I (error) residual statistical value was significant (p < 0.1) and its value reached 1.7, which indicated that the influence of various factors on growth of urban land had obvious spatial correlation, and the use of spatial panel models would improve the validity of the empirical analysis results.

The overall spatial analysis results showed that 10 of the 11 indexes were statistically significant in the SEM (environmental degradation was not significant), and four indexes had a contribution greater than 10% (Table 3). The index with the strongest impact on the growth of urban land was fiscal decentralization (with a contribution of 20.7%), followed by infrastructure (19.0%), industrialization (16.7%), and industrial upgrading (14.0%). In addition, marketization (7.3%), information technology (3.9%), education (3.4%), and urban green space (2.2%) had important positive impacts on the growth of urban land. However, it should be noted that the unemployment rate and terrain conditions had a negative impact, with contributions of 6.0 and 5.7%, respectively.

Table 3 Statistical significance and contributions to the proportion of urban land use in China for the 11 driving factors from 2007 to 2019. βk regression coefficient for variable k, Conk the contribution of variable k (defined in Eq. 4), OLS ordinary least-square regression, SEM spatial error model
Full size table

The GTWR model can further explore the spatial variation of the effects of the factors that significantly affected the growth of urban land based on statistical analysis of the coefficient values. Specifically (Fig. 2), the effects of industrialization, marketization, industrial upgrading, and fiscal decentralization on growth of the area of urban land decreased from south to north and from east to west, but the effects of the unemployment rate and terrain conditions showed the opposite trends. The influence of information technology, infrastructure, education, and environmental degradation showed a trend of “concentric circles,” with high effects in the central and northeastern regions, but with low effects along the southeastern coast and in the western region. Urban greening was lowest in central regions and highest in the surrounding areas, and the areas with the greatest effect on the growth of urban land were mainly concentrated in the northeastern and western regions.

Fig. 2
figure 2figure 2

Spatial distributions of the coefficients of geographical and temporal weighted regression (GTWR) models for the proportion of urban land for the 11 factors that drove the growth of urban land in China from 2007 to 2019

Full size image

Discussion

As the world’s second-largest economy, China’s socioeconomic development has attracted worldwide attention, especially given China’s rapid urbanization process (Lu et al. 2019). China’s reform of its economic institutions has created an impetus for urbanization. China has evolved away from the centrally planned economic system, and local governments have obtained more autonomy of urban land use, including the right to control regional urbanization. On the other hand, China has gradually implemented a market economy system, and this mode of resource allocation has driven a large amount of social investment in urban infrastructure construction. The fiscal decentralization and marketization have increased initiative and the efficiency of growth of urban land (Shang et al. 2016). However, as the reforms were first implemented in the southern coastal areas and gradually implemented in northern and western regions, the difference in implementation rates has created regional differences in the progress of the reform process, with economic institutions more advanced in southern China (Yueh 2010). Therefore, the contribution of fiscal decentralization and marketization both decrease from south to north.

The reform of economic institutions also promoted industrialization, which has been a main driver of the growth of urban land (Gollin et al. 2016). The development of industrialization has promoted an economic transformation from agriculture to industry, which has promoted a huge population migration from rural areas to urban areas and rapid growth of urban land. Influenced by China’s regionally unbalanced development, industrialization in the southern region has developed faster than that in the northern and western regions, which has led to regional differences in the impact of industrialization on the growth of urban land (Fig. 2a). More importantly, industrialization has inevitably damaged the urban ecological environment, thereby creating huge challenges to the land’s ecological carrying capacity (Zhang et al. 2020). Therefore, it is necessary not only to accelerate the development of industrialization, but also to strengthen government management of green space and the discharge of industrial pollution to improve the urban ecological environment, ultimately achieving a win–win situation between industrial development and preservation or restoration of the ecological environment.

Upgrading a city’s industrial structure will be an important way to optimize growth of urban land in China (Xu et al. 2018; Liang and Yang 2019). This will not only improve how industrialization occurs and mitigate the problem of environmental degradation, but will also improve the urban land-use efficiency by increasing the economic output per unit area of land. Affected by the development process of industrialization, industrial upgrading has shown the same spatial trend as industrialization (Fig. 2a, c). Meanwhile, the impact of information technology cannot be ignored (Fig. 2h). Information technology has supported industrial upgrading and has become a new engine that is driving the growth of urban land after industrialization (Ye et al. 2018). Information technologies have promoted the rise of smart cities and innovative planning, exploitation, and utilization of urban land, leading to more intensive use of urban land and improved quality of urbanization (Zhao et al. 2018). Therefore, different regions can use information technology to promote industrial upgrading and improve education levels to match the needs of emerging jobs, so as to optimize the growth of urban land.

Infrastructure is an important factor that affected the growth of urban land (Zhou et al. 2015). Since 1978, the Chinese government has continued to expand and support urban infrastructure construction, while also adopting a land finance model (in which the government sells land to developers) to raise funds for urbanization (Cai et al. 2019). However, due to differences in the available funding for infrastructure construction in different regions, infrastructure construction has not proceeded equally fast everywhere, and this has led to regional differences in the impact on the growth of urban land (Li and Fang 2014; Fig. 2i).

We also found that terrain conditions and urban green space significantly affected the growth of urban land. This is because terrain conditions affect the living and production environments, and because urban green space affects the urban ecological environment and the quality of life (Xu et al. 2022). Therefore, it is necessary to not only overcome limitations of the natural environment such as terrain conditions, but to also protect and restore the urban ecological environment to promote the sustainable development of a city.

Conclusion

The rapid development of China’s urbanization has relied on the support provided by the unique land finance model, which has raised large amounts of funds for urban construction, thereby supporting the initial urbanization. However, over-reliance on the land finance model has led to disordered growth of the area of urban land, thereby threatening the sustainable development of cities by threatening the ecological environment and the economy. The sustainability of urban land use will become an important path to promote sustainable urbanization in China. In this study, we used panel data from 295 Chinese cities to explore the factors that are driving the growth of urban land using spatial econometric methods. We demonstrated that fiscal decentralization, infrastructure, industrialization, and industrial upgrading were the major factors responsible for the growth of urban land, but that their degrees of influence differed between regions (Table 3; Fig. 2). Based on our findings, potential solutions include further deepening of the reforms of China’s economic institutions to give local governments more autonomy, improving the urbanization infrastructure, and promoting industrial upgrading by encouraging innovative use of information technology to improve industrial efficiency. Alternatively, the national government could provide sustainability targets with the goal of bringing urbanized land into a better balance with the ecosystems that support that land but leave the choice of solutions that will achieve that balance to be defined by urban governments.

In future research on the sustainability of urban land use, various gaps in our research need to be filled. First, the development of urban agglomerations has become a new trend that will give full play to the roles of an agglomeration’s central cities to drive the development of urbanization in their surrounding areas. Thus, the mutual promotion (or sometimes restriction) of urban land use among the different cities in an urban agglomeration must be more fully considered. Second, the national government’s macro-scale land-use policy needs to be combined with micro-scale land-use spatial planning at the scale of individual cities to better allocate the spatial distribution of various functional areas and to promote sustainable urban land use more accurately. Overall, the development path of China's urbanization suggests that improving the sustainability of urban land use (and particularly restraining the growth of the area of urbanized land) will be the key to sustainable urbanization. Our study provides important insights into this sustainability by exploring the main factors that influence the growth of urban land from a spatial perspective. What is more, as the world’s largest developing country, China can provide a rich reference to guide other countries and regions around the world to promote their own sustainable urbanization.