Spatial Interaction and Driving Factors between Urban Land Expansion and Population Change in China

Abstract

The rational matching of urban land and population has become an important prerequisite for sustainable urban development. In this paper, the traditional urban land scale elasticity model was improved, and combined with the gravity model, the spatial interaction between land expansion and population change in 618 cities in China during the period 2006–2021 was investigated. The geographical detector method was used to reveal what drives them. The main results were as follows: (1) China’s urban land expansion rate was 1.83 times faster than the population growth rate during 2006–2021. After the implementation of the New-type Urbanisation Plan in 2014, the ratio of land expansion rate to population growth rate dropped from 2.46 to 1.12. (2) Among the six interaction types identified, land rapid expansion is the most significant, accounting for 41.59% of urban samples. (3) The geographical detector method found that the indicators of urban development rights such as the level of administrative hierarchy and the ratio of fiscal revenue to fiscal expenditure were the main factors affecting land expansion and that economic indicators such as gross domestic product and employment opportunities dominated population change. Fortunately, the intervention role of urban development rights has declined, and the constraints of market mechanisms, resources and environment have gradually become the dominant factors in urban land expansion and population change. These findings provide a theoretical basis for alleviating the human–land contradiction and achieving sustainable urban development.

1. Introduction

The United Nations laid out 17 Sustainable Development Goals (SDGs) on 25 September 2015. Among them, SDG 11 is targeted at ‘making cities and human settlements inclusive, safe, resilient and sustainable’, which requires coordinated development in land use and land cover, demographics and other sectors [1].

However, rapid urban land expansion due to population growth presents serious challenges to sustainable development. Over the next 30 years, the pace of world urbanisation will continue to accelerate, with the world urbanisation rate expected to increase from 56% in 2021 to 68% by 2050, which means 2.2 billion new urban residents [2]. These challenges pose significant obstacles to achieving SDG 11 [3]. China is experiencing rapid urbanisation, which is ‘perhaps the greatest human resettlement experiment in history’ [4]. According to the Statistical Yearbook of China’s Urban Construction, the urbanisation rate of China’s permanent resident population has increased rapidly from 17.9% in 1978 to 65.22% in 2022, while the number of cities has also increased rapidly from 193 to 687, and it is responsible for air pollution and resource depletion on both local and global scales [5,6]. These challenges pose a major obstacle to achieving SDG 11, especially in China, where explosive population growth and increased mobility may lead to a misallocation of urban infrastructure supplies.

In the long-term process of urbanisation, population and land are the core elements of the urban system [7]. Improving urban areas is crucial to addressing these challenges and achieving SDGs. Along with the growth of urban land, the formulation of a sustainable development plan is even more important to achieve the trade-off between land expansion and population changes. Such a trade-off strategy could help to avoid wasted resources, which has aroused widespread attention from the government and academia [8].

Therefore, what is the spatial interaction relationship between urban land expansion and population change in China and its spatial and temporal patterns? What are the basic determinants of their spatial interaction? These are very pressing issues. Although the existing literature answers the above questions to a certain extent, there are still some shortcomings.

It is well-known that urban land expansion is faster than population growth in the global urban development experience, also known as urban sprawl [9]. Even as the urban population stops growing in post-urbanised Europe, urban land continues to expand [10]. Marshall found that the land expansion rate of cities in the United States during the period 1950–2000 was 10% faster than the population growth rate [11]. Most empirical results show that China’s urbanisation process presents a state of great leap forward and spatial disorder spread [12], and urban land expansion is much faster than population growth [13]. The New-type Urbanisation Plan of China considers that ‘Urban land expansion is obviously faster than population growth, and urban land use is inefficient’ is an important threat to sustainable urban development [14]. In contrast, some studies believed that it was normal for urban land to expand faster than population growth in a certain period [15]. In the early stage of urbanisation, China’s per capita urban land was less. In 2000, global urban land per capita was 214.1 m², developed countries 351.7 m² and even developing countries reached 153.0 m² [16]. However, China’s per capita urban land area was only 102.9 m², ranking 179th in 2000. It can be seen that there is still some debate about the spatial interaction between urban land expansion and population change in China. It can be seen that there is still some debate about the spatial interaction between urban land expansion and population change in China. In addition, the objects of research are mainly focused on the overall [17] or individual city level in China [18], or the interactive relationship between population and land in local areas [19]. There are few studies on the differences between cities at different stages of urbanisation.

Elasticity coefficient [20], Tapio Model [21], coupled coordination coefficient [20] and other methods are the most widely used in this research field. In the UN 2030 Agenda for Sustainable Development SDG 11.3.1 goals, the elasticity coefficient of urban land scale is an important measure of the interactive relationship between land expansion and population change [22]. However, it only considers the ratio of land to the population change rate, which does not easily reflect the status quo of urban per capita land use and geographical location, etc. Some empirical studies have designed the ideal value of urban per capita land use and modified the traditional elasticity coefficient of the urban land-use scale [23,24], but even so, due to the wide climate difference between cities in China and the different building standards in different climate zones, these ideal values of per capita land use still lack detail on regional differences. Therefore, it is easy to fall into the cycle of playing games with data, and reach completely different conclusions, which need to be further developed.

In terms of driving factors, the existing studies mainly explain the reasons for the interaction between urban land expansion and population change from two aspects. On the one hand, land rent, transportation cost and environmental cost [25] are the dominant urban development mechanisms, which tend to be the objective laws of economic development. The economic law of urban development is revealed by the theory of decreasing return to scale and rent and price of land. On the other hand, China’s unique land system and fiscal and taxation system promote the urban development mode of land revenue and land mortgage financing by local governments, which intensifies the extensive use of land [26]. At the same time, the urban development mode lacking industrial support is prone to population loss, which hinders the further development of the city. Combined with China’s hukou system, it is difficult for the floating population to truly integrate into the city [27]. Compared with Western cities, China’s urbanisation problem is more complex. As Friedmann said, China’s urbanisation is the result of endogenous forces and is evolving in its own way [28]. The traditional urbanisation model in China emphasizes scale and speed, often making economic development the primary evaluation criterion for local governments. This focus has led to disparities between land expansion and population growth in most regions, threatening the sustainable development of cities [29,30]. Existing studies lack a comprehensive consideration of institutional, economic, resource and environmental constraints. However, the analysis of these reasons is also important, because it will help formulate sustainable urban development policies.

In short, there is much literature that answers the above two questions; that is, the spatiotemporal pattern and driving factors of the spatial interaction between urban land expansion and population change in China. However, three gaps remain. First, there is still some debate about the spatial interaction between urban land expansion and population change in China, especially the comparison between cities at different stages of urbanisation. Second, in terms of methodology, there is a lack of methods to reflect regional differences. Third, in terms of driving factors, there is a lack of comprehensive consideration of institutional, economic, resource and environmental constraints.

Based on this, the study makes an original contribution to several important areas. First of all, it is different from previous empirical studies that focus on large geographic scales, or individual small geographic scales. Using the panel data of 618 cities in China from 2006 to 2021, this paper explores the general rules and results of the interaction between urban land expansion and population change. Second, this paper revises the traditional method of using the urban land scale elasticity coefficient, considers the heterogeneity of urban scale and geographical location comprehensively, and constructs a spatial interaction coefficient model. In particular, we attempt to explain the driving mechanism of urban land expansion and population change using the geographical detector method (GDM). This has been mentioned in existing studies but has not been thoroughly tested.

2. Materials and Methodology

2.1. Gravity Model

The concept of a centre of gravity comes from physics, which refers to the point of action of the gravitational force exerted on the parts of an object [31]. It was first used to analyse the spatial distribution of population in the U.S. in 1872 [32]. Over the years, extensive research on barycentric models has focused on population [33], economics [34], urbanisation [20], land use [19] and environmental pollution [35].

In this study, the centre of gravity theory is applied to analyse the centre of gravity of urban land and population, and the combined centre of gravity trajectory of the two is determined [36], to reflect the overall combination of urban land expansion and population growth in China. The centre of gravity model consists of 618 cities. The centre of gravity models of urban land and population are as follows:

$$G_L(x,y)={\displaystyle \frac{\sum _{i=1}^n(L_i·Q_i(x_i,y_i))}{\sum _{i=1}^n{L}_i}}$$

(1)

$$G_P(x,y)={\displaystyle \frac{\sum _{i=1}^n(P_i·Q_i(x_i,y_i))}{\sum _{i=1}^n{P}_i}}$$

(2)

where $G_L$ and $G_P$ are construction land and population centre of gravity, respectively; $L_i$ and $P_i$ are the land and urban population size of city $i$, respectively. $Q_i(x_i,y_i)$ is the longitude and dimensional coordinates of the geographical centre of gravity of city $i$.

To further explore the spatial interaction between urban land expansion and population change, the overlap index of population and land gravity distribution was introduced [37]. The overlap index of the population centre of gravity and land centre of gravity in spatial distribution is used to measure their spatial interaction. The calculation formula is as follows:

$$D=d\left(G_L,G_P\right)=\sqrt{{\left(x_L-x_P\right)}^2+{\left(y_L-y_P\right)}^2}$$

(3)

where $D$ is the distance between urban land and the population centre of gravity, and the closer the distance, the higher the overlap; $G_L$ and $G_P$ are the construction land and the population centre, respectively. $x_L$, $y_L$ are the longitude and dimension coordinates of the centre of gravity of urban construction land, respectively. $x_P$, $y_P$ are the longitude and dimensional coordinates of the centre of gravity of the urban population, respectively.

2.2. Defining a New Interaction Coefficient

Different building standards were found in each climate zone. The ideal value of per capita construction land in cities located in climate zones III, IV and V is 5 m² less than that in cities located in climate zones I, II, VI and VII [38].

Due to the small sunshine spacing in southern China, the per capita residential land scale is 5 m² less than that of cities in northern China. We presented the ideal values of urban land use per capita for different building climate zones and different grades of cities (Table 1). Through the ideal value of per capita urban land use, the traditional elasticity coefficient of urban land use is corrected accordingly. We then measured the interactive relationship between urban land expansion and population change based on the interaction coefficient constructed by per capita constraint.

We evaluated the two abovementioned documents designed by the Chinese government comprehensively and found that the maximum value of urban construction land per capita in cities located in climate zones III, IV and V is 5 m² less than that in cities located in climate zones I, II, VI and VII. Due to the smaller daylight spacing of buildings in the southern region, the per capita planning land standard is lower than that of cities in the northern region [39,40]. Based on this, the ideal values of urban land use per capita in different building climate zones and sizes are proposed (Table 1). Through the new ideal value of per capita urban land, the interaction coefficient between urban land expansion and population change was constructed. The specific formula is as follows:

$$CLPI={\displaystyle \frac{L{R}_i}{P{R}_i}}\times R$$

(4)

$$R={\displaystyle \frac{{LP}_t}{{LPI}_t}}/{\displaystyle \frac{{LP}_0}{{LPI}_0}}$$

(5)

$$L{R}_i=\sqrt[t]{{\displaystyle \raisebox{1ex}{$L_t$}\!\left/ \!\raisebox{-1ex}{$L_0$}\right.}}-1$$

(6)

$$P{R}_i=\sqrt[t]{{\displaystyle \raisebox{1ex}{$P_t$}\!\left/ \!\raisebox{-1ex}{$P_0$}\right.}}-1$$

(7)

Table 1. The ideal land use per capita for cities of different sizes.

Urban Hierarchy	Urban Population Scale (Ten Thousands of People)	Ideal Value of per Capita Land Use (m2/person)
		Previous Research	I, II, VI and VII Climatic Zone for Architecture	III, IV and V Climatic Zone for Architecture
Megacities	≥1000	95	95	90
Supercities	(500–1000)	95	95	90
Large cities	(100–500)	100	100	95
Medium cities	(50–100)	105	105	100
Small cities	<50	110	110	105

Notes: The previous ideal value of urban per capita land use mainly refers to the research of [23,41].

Table 1. The ideal land use per capita for cities of different sizes.

Urban Hierarchy	Urban Population Scale (Ten Thousands of People)	Ideal Value of per Capita Land Use (m2/person)
		Previous Research	I, II, VI and VII Climatic Zone for Architecture	III, IV and V Climatic Zone for Architecture
Megacities	≥1000	95	95	90
Supercities	(500–1000)	95	95	90
Large cities	(100–500)	100	100	95
Medium cities	(50–100)	105	105	100
Small cities	<50	110	110	105

Notes: The previous ideal value of urban per capita land use mainly refers to the research of [23,41].

Based on the standards of CLPI > 1.3, 0.9 < CLPI ≤ 1.3, 0 < CLPI ≤ 0.9 and CLPI < 0 [18,23], the interaction between urban land expansion and population change in China can be divided into four types (Figure 1). The urban shrinkage type can be subdivided into three sub-types: land shrinkage and population growth, land shrinkage and population shrinkage and land expansion and population shrinkage.

2.3. The Geographical Detector Method

As a statistical method to detect spatial change and reveal its driving force, the GDM is widely used in the study of the formation mechanism of spatial distribution of geographical phenomena [42]. Its principle assumes that if the independent variable has a significant effect on the dependent variable, its spatial distribution should be similar [43,44]. The most outstanding advantage of the GDM over other methods is that it can detect the relationship between drivers and geographical phenomena without linear assumptions, so its calculation process and results will not be affected by multi-variate collinearity [45]. The GDM consists of four modules—the factor detector, the interaction detector, the risk detector and the ecological detector [46]. Among these, the factor detector and interaction detector were used to detect the influence of various factors on urban land expansion and population change and the intensity and type of multi-factor interaction.

2.3.1. Factor Detector

The factor detector uses a $q$ value to quantify the influences of variable X on Y; q is expressed mathematically as follows:

$$q=1-{\displaystyle \frac{\sum _{h=1}^L{N}_h{\sigma }_h^2}{N{\sigma }^2}}=1-{\displaystyle \frac{SSW}{SST}}$$

(8)

$$SSW={\sum }_{h=1}^L{N}_h{\sigma }_h^2$$

(9)

$$SST=N{\sigma }^2$$

(10)

where q represents the explanatory power of factor X to attribute Y, the range of q values is between [0, 1]. The greater the q, the greater the explanatory power of the factor. $N\mathrm{a}\mathrm{n}\mathrm{d}$ $N_h$ are the number of sample units in the entire region and the subregion, respectively; $h$ = 1, 2, …, L represents the number of subregion. ${\sigma }^2$ is the global variance of Y over the entire study region, and ${\sigma }_h^2$ is the samples’ variance in subregion $h$. $SSW$ and $SST$ are the within the sum of squares and the total sum of squares, respectively.

2.3.2. Interaction Detector

This is used to test the effect of different factors after interaction. After the interaction of factors $X_1$ and $X_2$, it is judged whether their explanatory power to the dependent variable $Y$ is enhanced or weakened and how much it is enhanced or weakened or whether the influence of each factor on $Y$ is independent of the other. The evaluation method is to first calculate the q values of two factors $X_1$ and $X_2$ for Y: $q\left(X_1\right)$ and $q\left(X_2\right)$. Then, the two factors interact and the q value is obtained: $q\left(X_1\cap X_2\right)$; Finally, $q\left(X_1\right)$, $q\left(X_2\right)$ and $q\left(X_1\cap X_2\right)$ are compared [47]. The strength and type of interaction between the two factors can be divided into five categories (

Table 2. The possible interaction between two variables and their interactive impacts.

Graphical Representation	Description	Interaction Categories
	$q\left(X_1\cap X_2\right)<Min(q\left(X_1\right),q\left(X_2\right))$	Non-linearly weaken
	$Min(q\left(X_1\right),q\left(X_2\right))<q\left(X_1\cap X_2\right)<Max(q\left(X_1\right),q\left(X_2\right))$	Un-weaken
	$q\left(X_1\cap X_2\right)>Max(q\left(X_1\right),q\left(X_2\right))$	Bi-enhance
	$q\left(X_1\cap X_2\right)=q\left(X_1\right)+q\left(X_2\right)$	Independent
	$q\left(X_1\cap X_2\right)>q\left(X_1\right)+q\left(X_2\right)$	Non-linearly enhance

Legend: : $Min(q\left(X_1\right),q\left(X_2\right))$, : $Max(q\left(X_1\right),q\left(X_2\right))$, : $q\left(X_1\right)+q\left(X_2\right)$, : $q\left(X_1\cap X_2\right)$.

).

The analysis method of the GDM requires that the independent variable should be a type variable [48]. In this work, when calculating driving factors’ impacts on urban land expansion and population changes, the natural breaks were used to classify the driving factor value into five levels in ArcGIS software (ArcGIS 10.5).

2.4. Data Sources and Pre-Processing

In this study, the city’s land area and population size data from the China Urban Construction Statistical Yearbook (https://www.mohurd.gov.cn/gongkai/fdzdgknr/sjfb/tjxx/jstjnj/index.html, 5 July 2024). It is a statistic released by the Ministry of Housing and Urban–Rural Development. There are generally two statistical standards for urban land in China: the construction land area and the built-up area [8]. The area of urban construction land refers to the horizontally projected area of land enclosed by the boundaries of various types of construction land determined by the administrative department of urban planning. The built-up area is more extensive, including the area of construction land and the area of land expropriated [49]. However, the land expropriated may not be built on or put into use, so the construction land area reflects the urban function relatively more accurately.

In our study, urban population is the sum of permanent and temporary residents in urban areas. Due to the change in the regional demographic standard in the China Urban Construction Statistical Yearbook, the statistical standard of China’s urban population has changed from registered population to permanent population since 2006. To ensure the horizontal comparability of urban population change characteristics, the study period is set for 2006–2021.

In addition, the driving factors selected seven indicators in the categories of urban development rights, economic and social development and natural resources, namely, the urban administrative hierarchy level (HIERARCHY), the total quantity of water supply (WATER), the gross domestic product (GDP), the ratio of fiscal revenue to fiscal expenditure (FISCAL), the proportion of employment in the third industry (EMPLOYMENT), green space rate of built district (GREEN), and the road surface area per capita (ROAD).

The data in the paper are derived from the China Urban Statistical Yearbook and China Urban Construction Statistical Yearbook, except for HIERARCHY, which is a virtual index. The missing data in some years were estimated by the trend extrapolation method.

China has adopted a multi-level administration division to regulate large populations and areas [50]. The indicators of HIERARCHY are a directly controlled municipality, municipalities directly under the central government, vice-ministerial-level cities, separately planned cities, provincial capitals, prefecture-level cities and county-level cities. Some county-level cities’ data are derived from the supplement and improvement of the China County Statistical Yearbook.

During the period 2006–2021, the administrative divisions of some cities have been adjusted, such as Chaohu City in Anhui Province, which was absorbed into neighbouring cities, Laiwu City of Shandong Province, which was merged into Jinan City, and Tongzhou City, Wujiang City, Jiangsu Province, Nankang City, Jiangxi Province, which were disestablished and set up as districts. To ensure the continuity of data, in 2021, 692 cities were screened one by one, and the final sample covered 618 cities in 31 provincial-level administrative regions (excluding Hong Kong, Macao and Taiwan) (Figure 2). Among them, 289 are prefecture-level cities and above, and 329 are county-level cities.

3. Results

3.1. Overall Situation of Interaction between Urban Land Expansion and Population Change

3.1.1. Staged Transformation of Urban Land Expansion and Population Change

In the period 2006–2021, the average annual rate of urban land expansion and population growth in China was 3.93% and 2.14%, respectively, the former being 1.83 times that of the latter. At the same time, the urban per capita land area has also risen from 103 m² to 130 m², exceeding the upper limit of 105 m² per capita in the code for classification of urban land use and planning standards of development land [38]. It can be seen that land expansion was significantly faster than population growth.

Through the superposition analysis of urban land expansion and population change curves (Figure 3), the interaction between urban land expansion and population change in China has a special staged transformation feature, and it can be divided into two periods with obvious differences in characteristics. From 2006 to 2015, urban land expansion was significantly faster than population growth. During this period, urban land expanded rapidly, at an average annual rate of 4.68%. However, the average annual population growth rate was only 1.90%. Urban land expansion was significantly faster than population growth, with a rate ratio of 2.46. From 2015 to 2021, the trend of urban land expansion becoming faster than population growth was initially curbed. During this period, the rate of urban land expansion decreased to 2.80%, while the urban population growth accelerated, and the growth rate increased to 2.50%, which was 1.12 times that of the previous period.

In addition, an interesting phenomenon was found. After the New−type Urbanisation Plan had been implemented in 2015. The curves of urban land expansion and population change tend to be similar, but the peaks appear in different years. For example, 2017 and 2019 are the peaks of land expansion, while population change is the trough. The time curve of urban land expansion lags behind that of population change by 2 years. Surprisingly, this phenomenon was not evident before 2015.

3.1.2. Spatial Mismatch between Urban Land and Population

Using formulas 1, 2 and 3, the centre of gravity coordinates of urban land and population from 2006 to 2021 were calculated (Figure 4). The spatial characteristics of the interactive relationship between urban land expansion and population change were determined by the spatial superposition and distance of the two centres of gravity.

The centre of gravity coordinates of urban land in China in 2006 are 115°22′48″ E, 33°8′24″ N, located in Fuyang City, Anhui Province. During the period 2006–2021, the centre of gravity moved 76.13 km from northeast to southwest, entering Zhumadian City, Henan Province. The centre of gravity of the urban population, from 114°41′02″ E, 32°54′34″ N in 2006, moved to 114°41′02″ E, 32°54′34″ N in 2021, from Zhumadian City in Henan Province into Xinyang City. The centre of gravity of the urban population moved 107.03 km from northwest to southeast.

From the perspective of the distance between heavy centres, the spatial characteristics of the interaction between urban land expansion and population growth still have clear stage differences, which are highly consistent with the division of time evolution stages mentioned above. (1) From 2006 to 2015, the distance between the centre of gravity of urban land and population was shortened from 76.51 km to 23.59 km. The spatial coupling degree between urban population and construction land was enhanced. At this stage, the dominant direction of the centre of gravity movement was east–west. (2) From 2015 to 2021, the centre of gravity moved away rapidly at this stage. The distance between the centre of gravity of urban land and population was expanding rapidly from 23.59 km to 106.60 km. Although the distance between population centres and land centres did not change in the east-west direction during this stage, the centres of urban development were shifting from ‘east–west’ to ‘north–south’.

3.2. Spatiotemporal Interaction Types between Urban Land Expansion and Population Change

Based on Formula (4) of the relationship model between urban land expansion and population change, we evaluated the interaction between urban land expansion and population change in China from 2006 to 2021.

3.2.1. Spatial Pattern Characteristics of Interaction Types

During 2006–2021, the interaction between urban land expansion and population change in China was dominated by the rapid land expansion type, accounting for 41.59% of urban samples. The proportion of the urban shrinkage type and the coordinated development type is similar, at 25.08% and 24.92%, respectively. The proportion of the population growth type is the lowest at only 8.41% (Figure 5).

(1)

Rapid land expansion type. This includes 257 cities, accounting for 41.83% of the land and 35.67% of the population in China’s cities. Both the urban land and population are increasing, but the urban land expansion is 2.44 times the population growth rate, and the per capita construction land will increase from 96.04 m² in 2006 to 157.79 m² in 2021. They are mostly distributed in the periphery of urban agglomerations such as the Beijing–Tianjin–Hebei region and the Yangtze River Delta region, and in rapidly developing urban agglomerations such as the Central Plains City cluster and the middle reaches of the Yangtze River.

(2)

Coordinated development type. This type accounts for 24.92% of the number of cities in China, and the ICLP is between 0.9 and 1.3, which shows the coordinated growth of urban land and population. The per capita land area increased from 114.35 m² in 2006 to 135.04 m² in 2021. Such cities are mainly distributed in eastern and central China, represented by Hangzhou, Ningbo, Suzhou and Xuzhou in the Yangtze River Delta, and Tianjin, Shijiazhuang and Langfang in North China.

(3)

Rapid population growth type. This type has only 52 cities and an ICLP between 0 and 0.9. Such cities are mainly exemplified by Beijing, Shenzhen, Zhuhai and other big cities. Population growth in these cities is significantly faster than the rate of land expansion, with the per capita land falling from 116.21 m² in 2006 to 91.96 m² in 2021.

(4)

Shrinking city type. This city type manifests as urban population shrinkage or land shrinkage, which can be divided into three subcategories: the land shrinkage and population growth type, the land shrinkage and population shrinkage type and the land expansion and population shrinkage type. These are also known as shrinking cities. Cities with land shrinkage and population growth have a declining land consumption level, belonging to the category of actively shrinkage cities. However, there are only 18 cities of this type, such as Shanghai, Fuzhou and Haikou. The land shrinkage and population shrinkage type, and land expansion and population shrinkage type belong to negatively shrinkage cities. There are 19 cities with land shrinkage and population shrinkage, mainly distributed in Northeast China, Gansu, Inner Mongolia and Hubei, most of which are county-level cities or resource-based cities. There is relatively more land expansion and population shrinkage in cities, numbering up to 118. These cities are characterised by an uncoordinated state where the population decreases while the land increases. This type of city has obvious spatial agglomeration characteristics, with 51 cities located in Northeast China, and the rest mainly located on the periphery of urban agglomerations such as Beijing–Tianjin–Hebei, the Yangtze River Delta, the middle reaches of the Yangtze River and Chengdu–Chongqing.

3.2.2. Different Scale Cities’ Interaction Types

Based on the Notification on Adjusting the Standards for Classifying Urban Scales issued by the State Council of China in 2014, urban scales were classified using the urban population size data of 2021, and an analysis of the human–land interaction types of cities of different scales was conducted (

Table 3. Different scale cities’ interaction coefficient between urban land expansion and population change.

Interaction Types	Megacities	Supercities	Large Cities	Medium Cities	Small Cities
Rapid land expansion	/	0.49%	5.34%	6.96%	28.80%
Coordinated development	0.32%	0.65%	5.02%	5.18%	13.75%
Rapid population growth	0.32%	/	0.49%	0.32%	7.28%
Land shrinkage and population growth	0.16%	/	0.32%	0.32%	2.10%
Land shrinkage and population shrinkage	/	/	0.16%	0.32%	2.59%
Land expansion and population shrinkage	/	/	0.81%	3.07%	15.21%

).

• There are five megacities, namely, Shanghai (land shrinkage and population growth type), Beijing, Shenzhen (rapid population growth type), Chongqing and Tianjin (coordinated development type). In China’s urbanisation strategy, the size of megacities is strictly controlled. Under the dual reinforcement of fewer new construction land indicators and greater population attraction, the land consumption level of the megacities has declined further.
• There are 15 megacities, mainly of the coordinated development type, supplemented by the land rapid expansion type. Four cities, Hangzhou, Changsha, Chengdu and Xi’an exhibit coordinated urban land and population, with the per capita land increasing from 94.11 m² in 2006 to 108.57 m² in 2021. Nanjing, Wuhan and Guangzhou exhibit land expansion, with the per capita land increasing from 73.29 m² in 2006 to 165.19 m² in 2021, a growth rate of 2.25 times, indicating a significant land expansion characteristic.
• There are 75 large cities, accounting for 35.16% of the urban land nationwide and supporting 32.01% of the population. The per capita land is 147.81 m², higher than the national average of 134.54 m². These cities mainly exhibit two types of rapid land expansion and coordinated development. In 2006, the per capita land of the rapid land expansion type of cities was 90.75 m², with a small base that was only 73.36% of the coordinated development type of cities. However, the growth rate of land was 2.48 times higher than the latter. In 2021, the per capita land of the two types of cities was 142.79 m² and 152.36 m², respectively, and therefore almost at the same level.
• Small and medium-sized cities. Small and medium-sized cities are primarily of the rapid land expansion type, accounting for 41.62%. It is worth noting that 21.28% of these cities, while experiencing population outflow, have increased rather than decreased their urban land, exhibiting a low-density development trend. This has exacerbated the already acute contradiction between farmland protection and urban land expansion.

3.2.3. Comparison of IULP before and after the New-Type Urbanisation

In 2014, China implemented the New-type Urbanisation Plan [14]. Combined with the overall situation of interaction between urban land expansion and population change in Section 3.1. (Figure 3 and Figure 4). The period from 2006 to 2021 was divided into two stages. The first stage was before the implementation of the New-type Urbanisation Plan from 2006 to 2015, and the second stage was after its implementation from 2015 to 2021. By comparing the two stages, it can be found that after the implementation of the new strategy, both the quantity and spatial distribution of the interaction between urban land expansion and population change have changed (Figure 6 and Figure 7).

In the interactive relationship between urban land expansion and population change in China, the number of land-expanding cities is greatly reduced, and the spatial distribution shows a convergence trend. During the period 2006–2015, the interaction between urban land expansion and population change in China was mainly dominated by land expansion, accounting for 50.16%. After the implementation of the new strategy, the land expansion type city is still the highest, but the proportion drops to 29.61%. The cities that were originally distributed in urban agglomerations such as the middle reaches of the Yangtze River, the Central Plains, Harbin–Dalian–Qiqihar and Tianshan have transformed from rapid land expansion to urban shrinkage. Cities that have consistently exhibited the feature of rapid land expansion were mainly distributed in the periphery of urban agglomerations such as the Yangtze River Delta, Beijing–Tianjin–Hebei, and Chengdu–Chongqing (Figure 6).

At the same time, the interaction between urban land expansion and population change in China has created new problems. The number of cities with a shrinking population has increased significantly, from 17.15% during the period 2006–2015 to 33.33% during the period 2015–2021. These types of cities are mainly located in the periphery of urban agglomerations, and the urban shrinkage characteristics are more obvious than those in the central area of urban agglomerations, especially in the Beijing–Tianjin–Hebei and the middle reaches of the Yangtze River. They present a divergent trend in spatial distribution, spreading from the eastern region represented by Fushun, Benxi, Jinzhou and Jilin to the south margin of the Beijing–Tianjin–Hebei region represented by Xingtai, Zhuozhou and Renqiu, and to the small and medium-sized cities of the Yangtze River Delta urban agglomeration represented by Zhenjiang, Danyang, Yangzhou, Jingjiang and Huzhou and to the northwest region of the urban agglomeration in the middle reaches of the Yangtze River represented by Tianmen, Suizhou, Hanchuan, Yicheng and Zhongxiang.

3.3. Driving Factors of Interaction between Urban Land Expansion and Population Change

3.3.1. Factor Detector Results

The Pearson correlation coefficient method was used to analyse the linear relationship between urban land expansion, urban population change and independent variables. WATER, HIERARCHY and GDP are important influencing factors, and the correlation coefficients between them and urban land expansion or population change are all greater than 0.6 (Figure 8).

However, the Pearson correlation coefficient only analyses the linear correlation between variables, ignoring the heterogeneity of spatial stratification. Therefore, a new statistical analysis method of spatial differentiation perspective is adopted, namely, the GDM. The influencing factors of urban land expansion and population change in China are thereby further analysed. The results of factor detection analysis show that these seven independent variables all pass the significance test of 1%; in other words, these seven variables are important driving factors affecting urban land expansion and population change (Figure 8). In the q value of the GDM, after considering spatial heterogeneity, the factor detector results were all lower than the Pearson correlation coefficient score, indicating that the degree of correlation between variables was corrected after fully considering heterogeneity.

The results show that HIERARCHY, WATER and GDP are the main factors driving urban land expansion and population change. In terms of the urban development right, the coefficient of HIERARCHY on land expansion is 0.788, which is higher than that of population change (0.708). The same phenomenon is also reflected in FISCAL. The higher the HIERARCHY of the city and the richer the financial resources, the greater the power of development it has but it is more likely to lead to an imbalance in the interactive relationship between land expansion and population change. In terms of urban economy and society, GDP is the factor with the highest population change coefficient (0.745). Surprisingly, its coefficient of influence on land expansion is only 0.560. The coefficient of population change of EMPLOYMENT is also significantly higher than that of land expansion. It can be seen that the higher the level of economic and social development, the easier it is to promote the reduction in the urban land consumption level. In terms of urban resources and environment, the influencing factors of urban WATER and GREEN are higher than those of population change. Water shortage is a major factor restricting urban development in China. The scale of urban development is limited by water resources, and urban land development cannot exceed the carrying capacity of water resources. If the city has rich water and plant resources, its land expansion motivation is significantly higher.

The factor detection results of urban land expansion and population change in China before and after the implementation of the New-type Urbanisation are further compared (Figure 9).

It was found that the q value of urban HIERARCHY, GDP and WATER affecting land expansion were large during the periods 2006–2015 and 2015–2021, but the ranking changes. The influence of urban HIERARCHY on land expansion dropped from first to third place, by 25.79%. The influence of WATER and GDP on land expansion increased by 37.53% and 11.57%, respectively. In particular, the influence of WATER rose from third to first place.

In terms of urban population change, the leading factors affecting urban population change were ranked as HIERARCHY > GDP > WATER during the period 2006–2015. With the implementation of the New-type Urbanisation, GDP has become the most important factor affecting urban population change, while the influence of WATER has increased by 17.49%, becoming the second-largest influence. However, the effect of HIERARCHY, FISCAL, GREEN and other factors on urban population change has weakened.

3.3.2. Interaction Detector Results

The interactive detection method of the GDM is used to detect the interaction of different factors affecting urban land expansion and population change. There are 21 pairs of results of interactive detection among the 7 independent variables (Figure 10).

After the interaction, factor enhancement occurred in all of them, in which non-linear enhancement was the main one and cooperative enhancement was the auxiliary one. In other words, the explanatory power of factors increases significantly after the bivariate interaction. It can be seen that the influence of any two independent variables after interaction has stronger explanatory power than the original single factor, indicating that urban land expansion and population change are the result of multiple factors.

From the interactive detection results of urban land expansion from 2006 to 2021, 52.38% of the interaction results are factor non-linear enhancement, and 47.62% are collaborative enhancement. According to the q value, HIERARCHY has the strongest interaction with WATER (0.912), followed by WATER and GDP (0.882), WATER and EMPLOYMENT (0.877) and WATER and FISCAL (0.834). This shows that the combination of HIERARCHY, WATER and GDP are the most important influencing factors for urban land expansion.

In the interactive detection results of urban population change, it was found that 61.90% of the interaction results are factor non-linear enhancement and 38.10% are cooperative enhancement. The interaction between WATER and HIERARCHY is the strongest (0.936), followed by GDP and HIERARCHY (0.890) and GDP and WATER (0.875). It is further confirmed that HIERARCHY, GDP and WATER are the most important influences on urban population change.

4. Discussion

4.1. The Spatial Interaction between Urban Land Expansion and Population Change

We constructed the interaction coefficient for evaluating urban land expansion and population change. The ideal value standard of per capita urban land is proposed for cities with different building climate zones and different scale levels (Table 1). Compared with the elasticity coefficient in the UN 2030 Agenda for Sustainable Development of SDG 11.3.1 goals [22], our method can make up for the lack of consideration of the difference in city size and geographical location. We provide a reproducible method for the interaction between urban land expansion and population change, which is one of the peripheral contributions of this paper.

During the period 2006–2015, China’s urban land expansion rate was 1.83 times that of the population growth rate, indicating that urban land expansion was significantly faster than population growth, which also verified the empirical evidence of some scholars [24,51]. However, this conclusion has to be seen in stages. The urban land expansion rate/population growth rate changed from 2.46 in 2006–2015 to 1.12 in 2015–2021. Therefore, the conclusion that China’s urban land expansion is faster than population growth is only valid during the period 2006–2015, and this conclusion is not valid during the period 2015–2021 since 1.12 is just the ideal value of the elasticity coefficient of urban land-use growth [18]. Since China implemented the New-type Urbanisation Plan in 2015, the proportion of cities with rapid land expansion has dropped from 50.16% to 29.61%. At the same time, the proportion of cities experiencing population shrinkage increased from 17.15% in 2006–2015 to 33.33% in 2015–2021. It can be seen that the interaction between urban land expansion and population change in China is facing new structural problems. To the best of our knowledge, there is no available research in previous empirical studies in China.

From the perspective of the spatial pattern of interactive relationship types, the structure of the centre–periphery is presented (Figure 5). In the Yangtze River Delta, Pearl River Delta, Beijing–Tianjin–Hebei and other mature urban agglomerations, the core cities are mostly of the rapid population growth type, and the peripheral cities are mostly of the rapid land expansion type. The outer cities in the middle reaches of the Yangtze River, Chengdu–Chongqing, Harbin–Dalian–Changchun and other rapidly developing urban agglomerations show the characteristics of land expansion and population shrinkage. This may be due to the relatively saturated land market in the central cities of urban agglomerations, which still has strong population attraction, while the land price in the peripheral areas of urban agglomerations is relatively low and land expansion is, therefore, easier [52].

In urban development, due to the lag of the population introduction mechanism and the foresight of land development [53,54], the time curve leading to urban land expansion is 2 years behind the curve of population change. According to land management law in China, if the actor who buys the land does not develop it within 2 years, the government will take back the land forcibly [49]. This 2-year requirement coincides with the lag time of urban land expansion and population change curves. It can be seen that the interactive relationship between urban land expansion and population change needs to be viewed from the perspective of dynamic evolution.

4.2. The Driving Mechanism of Interaction between Urban Land Expansion and Population Change

The results of the GDM show that urban land expansion and population change in China and their interaction are the result of multiple factors, and the driving mechanism is different in different stages. In general, urban development rights, economic and social development, resources and environment have varying degrees of impact on urban land expansion and population change (Figure 7 and Figure 8).

China’s urban construction land supply is an indicator allocation system from the central government to the local government [8]. Therefore, the higher the administrative level of a city, the more indicators of new construction land will be obtained, and the easier urban land expansion will be [55]. In 1994, China implemented a tax-sharing reform [56], and the land transfer income accounted for 70% of the total urban income [4]. The excessive dependence on land finance accelerated the expansion of urban land. At the same time of rapid urban land expansion, some cities do not have enough population, elements and economic activities to coordinate with them and, hence, become ghost cities [54].

Urban population growth is mainly due to migration [57]. Therefore, the higher the level of GDP and the more job opportunities in the city, the faster the urban population growth. The population attraction of the eastern coastal cities is significantly higher than that of the cities in the central and western regions, resulting in the shift of the urban population centre from northwest to southeast [58]. The centres of gravity of urban land and population move in opposite directions (Figure 4). The hidden law is the spatial mismatch of urban construction land index against population growth, which leads to the coexistence of urban land going through disordered expansion and inefficient use.

Fortunately, since the release of the New-type Urbanisation Plan, the government has placed greater emphasis on human-oriented urbanisation [59]. It issued a policy description on the demarcation of the urban growth boundary and addressed the dilemma of the local government’s financial dependence on land effectively [49]. The driving force of development right factors such as HIERARCHY and FISCAL on urban land expansion and population change declined. The indicators of GDP and employment opportunities, which represent the level of economic development, and the driving force of water on urban land expansion and population change are enhanced. It can be seen that China’s urban development is gradually eliminating inefficient policy intervention. The market mechanism and the constraints of resources and the environment have gradually become the dominant factors of urban land expansion and population change.

4.3. Policy Implications

These insights can be valuable in designing land-use policies that incorporate data-based solutions and address the challenges of sustainable urban development while also accommodating population change.

Urbanisation is not only the accumulation of material wealth represented by the urbanisation of land but also the change in behaviour mode and the improvement in living standards represented by the urbanisation of people. Only with the coordinated development of the two can the sustainable development of cities be realised.

The cities identified with population growth, such as Shanghai, Guangzhou and Shenzhen, are included in this study. This shows the characteristics of population inflow, a good economy and high resource and environment carrying capacity. These cities need to moderately expand the scale of land supply to improve the carrying capacity of the population and economic development. These cities should be encouraged to provide equitable basic public services to urban and rural residents based on policies such as liberalising restrictions on household registration and promoting national pooling of pension insurance, and using market forces to promote the concentration of population in cities.

Therefore, urban land expansion and population change do not always change in the same direction. China should reform the allocation system of new urban construction land quotas, increase the flexibility of land supply and promote the overall allocation efficiency and use efficiency of land resources in the country. For cities in different regions and of different sizes, differentiated policies should be adopted to balance the demand for urban land, consider the change in the population size and reduce the scale of urban land appropriately in areas where people move, to improve land-use efficiency and promote the sustainable and healthy development of cities.

5. Conclusions

To explore the spatial interaction between urban land expansion and population change, and based on the elasticity coefficient of urban land scale, the interaction coefficient is constructed by integrating urban scale, climate and ideal value of land use. Combined with the centre of gravity model, the spatial interaction characteristics of land expansion and population change in 618 Chinese cities from 2006 to 2021 were investigated. The GDM was used to reveal the drivers of interactions.

(1)

From 2006 to 2021, China’s urban land expansion rate was 1.83 times that of population growth, confirming the traditional consensus that land expansion is significantly faster than population growth. The rising level of land consumption may be an obstacle to the sustainable development of Chinese cities. However, the stages of this consensus are obvious; especially before and after China implemented the New-type Urbanisation Plan in 2014, the rate of urban land expansion/population growth dropped from 2.46 to 1.12. The urban character of rapid land expansion has been curbed.

(2)

It was found that the peripheral cities of urban agglomerations such as Beijing–Tianjin–Hebei and the Yangtze River Delta show the characteristics of rapid land expansion. The rapid development of urban agglomerations in the middle reaches of the Yangtze River, Chengdu–Chongqing and Harbin–Dalian–Changchun are mostly characterised by land expansion and population shrinkage.

(3)

The GDM results showed that multiple factors such as urban development rights, economic aspects, resources and the environment work together, leading to an increase in the land consumption level. Fortunately, after the implementation of the New-type Urbanisation Plan, the development of Chinese cities gradually eliminated inefficient policy intervention, and market mechanisms and resource and environmental constraints have gradually become the dominant factors in urban land expansion and population change.

This paper analyses the spatial interaction and driving factors between urban land expansion and population change in China, and provides a valuable reference for regulating the allocation of human–land resources in different regions, but there are still some limitations to this approach.

The population data in this study are the sum of the urban population and urban temporary resident population. They come from the Chinese Ministry of Housing and Urban–Rural Development provided by the China Urban Construction Statistical Yearbook. Temporary population statistics based on the application of temporary residence permits are the standard, so the calibre is narrower than the permanent resident migrant population. Therefore, it underestimates the permanent population of big cities; however, the population of small and medium-sized cities is relatively large, so there is a certain overestimation of the permanent population of small and medium-sized cities [60]. This is not only consistent with the definition of the resident population who have lived locally for more than 6 months but also with the population that the city serves [61]. Therefore, the quality of this data is controversial [57,62].

However, population data provided by the National Bureau of Statistics are collected every 10 years, and 1% of the surveys are sample surveys every 5 years [63], but the quality of the Ministry of Housing and Urban–Rural Development’s urban population data is questionable. At present, they are the only long-term continuous data that can be specific to the level of urban entities.

In addition, our study did not explore the cross-coupling relationship between different types of land and population change. In future studies, more refined research can be carried out through remote sensing, global positioning systems and location-sharing services [64]. We can try to consider whether only the New-type Urbanisation Plan will affect the phenomena studied, such as the level of public services, land prices, suburbanisation, transportation accessibility, and growth poles. These indicators will be used to pass spatial econometric models such as the geographically weighted regression tool to better represent the spatial characteristics of urban expansion and urban population growth.