Introduction

Since the initiation of economic reforms in 1978, China has experienced profound urban sprawl accompanying its accelerated urbanization process (Qiao and Huang, 2024). This unprecedented spatial expansion has precipitated inefficient urban land utilization patterns, generating significant challenges to the sustainable management of urban spaces and ecological spaces (Koprowska et al. 2020; You, 2016; Zhang et al. 2025). Urban growth boundary (UGB), which is mainly used to control the urban sprawl of low-density, discontinuous sub-urbanized expansion, has been the most representative management tool for urban spatial growth in the United States since the 1970s and 1980s (Bengston et al. 2004). China’s spatial governance framework has historically incorporated UGB-like controls through multiple regulatory instruments: The “Three Functional Areas and Four Lines” system in urban planning, “Three Boundaries and Four Zones” demarcations in land use planning, and the red line of basic farmland protection collectively constituted the proto-UGB mechanisms prior to the establishment of unified territorial spatial planning (Zhang et al. 2013; He et al. 2018; Chen et al. 2021; Deng et al. 2024). The current institutional innovation in China’s spatial planning system involves synthesizing these legacy regulatory tools into an integrated framework. In July 2014, the Ministry of Housing and Urban-Rural Development of the People’s Republic of China and the Ministry of Land and Resources of the People’s Republic of China selected 14 trial cities for trial UGB delineation. The trial UGB is also known as the “urban development boundary” (cheng shi kai fa bian jie). The trial UGB delineation represents a strategic localization of western UGB concepts, adaptively incorporating China’s distinctive planning paradigms and land governance policies to develop an operational framework suited to its urbanization context.

UGB is a strategic land-use planning tool aimed at regulating urban expansion by legally establishing a geographic demarcation that separates areas designated for urban development from protected rural or natural lands. The primary objectives of implementing a UGB are multifaceted: to curb urban sprawl by concentrating development within a defined boundary, to safeguard farmland, forests, and ecosystems from conversion to urban uses, and to promote efficient infrastructure utilization while reducing public service costs. The effectiveness of UGB has been argued. Ding et al. (1999) provided a theoretical framework to analyze the impacts of UGB. They developed a dynamic urban model incorporating a congestible public good, a fixed stock of urban infrastructure, and average cost pricing for public goods. Their findings indicated that, within a specified planning horizon, the implementation of a UGB could enhance social welfare. Cho et al. (2006) approached the issue from a political economy perspective, suggesting that the reduction in land development within UGB areas might be partially attributed to the city government’s authority to annex land parcels within the UGB boundary without requiring landowner consent. This insight highlights the complex interplay between policy mechanisms and landowner rights in the context of urban growth management. UGB has been widely implemented as a tool for managing urban growth, with Portland, Oregon, serving as a prominent example. By adopting a high-density development model facilitated by UGB, Portland has achieved significant urban planning outcomes. Studies have shown that this approach not only increased land value and improved land use efficiency (Jun, 2004; Nelson and Moore, 1993; Kline and Alig, 1999) but also effectively alleviated the challenges typically associated with high-density development. Through strategic expansion of green spaces, public areas, and enhanced accessibility to service facilities (Jun, 2004), the UGB framework has played a pivotal role in strengthening Portland’s land carrying capacity, demonstrating its effectiveness as a sustainable urban growth management strategy. However, the implementation of UGB is not without its challenges. One significant issue is the difficulty in effectively curbing improper urban expansion, as highlighted by Mubarak (2004). Furthermore, while UGB has the potential to improve urban transportation systems, they can also introduce distortions in the land market, as noted by Knaap (1985) and Zhang and Kockelman (2016). These challenges underscore the complexities and trade-offs inherent in UGB policies, necessitating careful consideration and adaptive management strategies to balance urban growth with sustainable development goals. The effectiveness of UGB suggests that the trial implementation of UGB delineation in China has the potential to curb urban sprawl and foster more intensive land use. In China, the trial UGB function by demarcating boundaries between urban construction lands and non-construction lands. The primary aim of the UGB is to control urban sprawl and promote intensive land use. They are designed to protect farmland and ecological lands from the rapid encroachment of urban expansion (Huang et al. 2019; Tan and Xu, 2023). This strategic delineation helps in preserving essential non-urban spaces while accommodating urban growth.

Land resources play a critical role in supporting human survival and development, with land carrying capacity (LCC) being a key indicator closely tied to sustainable development (Xu and Yang, 2024). The importance of LCC has been underscored by the United Nations’ Sustainable Development Goals (Zhou et al. 2021). Initially, LCC was defined as the maximum population that an area’s farmland can sustain indefinitely without significant depletion or degradation of resources (Shi et al. 2019; Lane et al. 2014). However, as cities have evolved into primary hubs for population, economic activities, public services, and social interactions (Shi et al. 2019), the concept of LCC has expanded. The ultimate objective of urban land use is to harmonize human survival, social development, economic growth, and ecological preservation. To address the challenges of sustainable land use, researchers have categorized LCC into three interconnected systems: social, economic, and ecological (Luo et al. 2020; Tehrani and Makhdoum, 2013; Shen et al. 2022; Tsou et al. 2017). Thus, LCC reflects the capacity of land resources to support economic activities, meet social demands, and maintain ecological balance under natural resource constraints. LCC is deeply intertwined with urban changes. For instance, Qian et al. (2015) found that the decline in farmland and energy consumption in Xiamen significantly reduced its LCC. Similarly, Li et al. (2020) highlighted that unchecked urban sprawl exerts substantial pressure on LCC. Conversely, Shen et al. (2022) demonstrated that urban agglomeration can create a synergistic “1 + 1 > 2 effect” on LCC, as observed in the Pearl River Delta Urban Agglomeration in China. This suggests that trial UGB delineation, which influences the quantity and spatial distribution of land, may also impact LCC. While existing research has made significant contributions to understanding the role of UGB in controlling urban construction land boundaries and shaping urban form—such as limiting urban sprawl and promoting compact development—it has largely overlooked the broader implications of UGB on the interplay between economic, social, and ecological systems. This narrow focus represents a critical gap in the literature, as it fails to address whether UGB can effectively balance competing demands for economic growth, social equity, and ecological sustainability within constrained land resources. Additionally, UGB separates urban areas from surrounding natural and agricultural lands, potentially altering ecological conditions. It remains uncertain whether constrained land space can effectively support economic activities, social needs, and ecological sustainability. By addressing these gaps, this study seeks to provide a more holistic understanding of the relationship between UGBs and sustainable development. Specifically, it investigates how trial UGB delineation in 14 cities influences changes in multi-dimensional LCC, encompassing economic, social, and ecological dimensions. In doing so, it aims to shed light on the broader implications of UGB implementation and contribute to the development of more balanced and effective urban growth management strategies.

To evaluate the impact of trial UGB delineation on LCC, it is essential to compare changes in LCC before and after the implementation of UGB over an extended period. This study treats the trial UGB delineation introduced in 14 pilot cities in 2014 as a quasi-natural experiment. And 15 cities including municipalities, provincial capitals, and cities with a population of more than 5 million are selected as the control group. The Difference-in-Differences (DID) model is employed to assess the effects of trial UGB delineation on LCC, encompassing its economic, social, and ecological dimensions. Additionally, the study examines how the heterogeneity of urban locations influences the outcomes of UGB delineation. Beyond the introduction, this paper is structured into five sections. The first section establishes a conceptual framework to analyze the effects of UGB on LCC. The second section details the selection of indicators, the methodology for calculating LCC, and the application of the DID model. The third section applies the DID model to uncover the effects of trial UGB delineation on LCC and tests for urban heterogeneity in these effects. The fourth section explores the underlying causes of the relationship between UGB delineation and LCC, providing insights into the urban heterogeneity observed in the results. Finally, the fifth section summarizes the key findings and conclusions of the study.

Conceptual framework

The technical approaches employed in UGB delineation demonstrate a significant influence on LCC. These approaches primarily encompass two methodologies: the positive guidance method and the negative control method. The positive guidance method conceptualizes urban construction space as a continuously expanding spatial entity. It establishes land use boundaries necessary for urban spatial expansion through a comprehensive analysis of projected population growth, economic trends, and resource availability. This method delineates future-oriented urban development boundaries that integrate economic, social, and ecological considerations, reflecting the city’s developmental aspirations. Conversely, the negative control method adopts a precautionary approach by considering worst-case scenarios in urban growth, recognizing the inherent limitations to urban expansion. This method identifies ecologically sensitive areas on the urban periphery that are unsuitable for development, establishing them as inviolable limits to urban spatial growth. By employing a “subtraction approach,” this method serves as a regulatory force in UGB determination. The negative control method effectively safeguards against potential ecological degradation and resource depletion in future urban development. Existing research consistently demonstrates the relationship between UGB delineation and LCC. Li et al. (2020) highlighted that uncontrolled urban expansion exerts substantial pressure on LCC. Ding et al. (1999) observed that local governments frequently adjust UGB to enhance economic efficiency. Empirical studies by Wassmer (2002) revealed that UGB implementation can mitigate retail decentralization. Furthermore, Frenkel (2004) demonstrated that growth-management policies, including UGB implementation, can counteract biodiversity loss resulting from ecosystem fragmentation due to the depletion of open spaces and agricultural land.

Figure 1 presents a conceptual framework illustrating the relationship between UGB and LCC. The UGB delineation process incorporates existing planning control lines, implying that the influence of these control lines on LCC is inherently reflected in the effects of UGB on LCC (Zhang et al. 2013; He et al. 2018; Chen et al. 2021). The impact of established UGBs on LCC is subsequently mediated through specific land policies, as UGB achieves their objectives via various control mechanisms. The potential pathways of the effects of UGB on LCC include the control mean selection and typical land policies in China. Lastly, the outcomes of UGB implementation significantly influence four key land-related dimensions: land use structure, land quantity, spatial layout, and functional utilization (Zheng et al. 2017; Ma et al. 2020; Cho et al. 2006). These land transformations are fundamentally interconnected with and directly affect the overall LCC.

Fig. 1
figure 1

Conceptual framework for effects of UGB on LCC.

Full size image

Owing to China’s unique land ownership system characterized by public ownership, the country has developed a distinctive land management framework. Consequently, the mediating role of China’s land management system is crucial in understanding the relationship between UGB delineation and LCC. Box A in Fig. 1 illustrates the UGB delineation processes within this context. China has implemented specific spatial control measures to address rapid urban expansion, including the establishment of a permanent basic farmland control boundary. This boundary serves to protect high-quality agricultural land from conversion to non-agricultural uses (Cheng et al. 2017). Additionally, the ecological conservation redlines have been delineated to restrict human activities in areas critical for maintaining national ecological security and providing essential ecosystem services (Gao et al. 2020). These redlines play a vital role in safeguarding urban ecological security and preventing uncontrolled urban sprawl (Yue et al. 2024). The UGB delineation process comprehensively considers the interrelationships among these spatial control measures. Specifically, the negative control method in UGB delineation prioritizes maintaining the integrity of both the permanent basic farmland control boundary and ecological conservation redlines. The trial UGB delineation process emphasizes the coordination and integration of these three spatial control measures, which forms the fundamental basis for establishing a practical and effective UGB.

Box B in Fig. 1 is the potential pathways of the effects of UGB on LCC. The delineation and implementation of UGB necessitates comprehensive land space control, which is primarily executed through three interconnected mechanisms: control target setting, control incentives and punishments, and control effect checking. These mechanisms collectively aim to achieve effective land use regulation and promote sustainable land utilization. China’s unique land policies play a crucial role in facilitating land space control. For instance, the policy of restricting incremental construction land expansion while optimizing existing land stock serves to control urban sprawl and mitigate arable land loss (Zhou et al. 2023). This approach is particularly evident in major Chinese cities, where “zero growth” targets for construction land have been implemented alongside initiatives to revitalize underutilized urban areas (Shi et al. 2024). Furthermore, the protection of arable land and permanent basic farmland represents a critical strategy for preserving both the quantity and quality of agricultural land amidst rapid urbanization (Wu et al. 2017). This is achieved through stringent territorial and spatial use regulations that prevent the unauthorized conversion of farmland to construction land. Additionally, China’s ecological security strategy involves the construction of comprehensive ecological security patterns (Wang et al. 2024). This includes the identification and strict protection of zones characterized by high biodiversity, critical ecological functions, or significant ecological sensitivity. The effectiveness of control target setting, control incentives and punishments, and control effect checking is fundamentally dependent on the implementation of these specific land policies, which collectively shape the relationship between UGB implementation and LCC enhancement.

Box C in Fig. 1 presents the implementation results of UGB, focusing on their effects on land utilization. As the fundamental resource for production and urban development, land undergoes significant modifications following UGB delineation through three primary mechanisms: control target setting, control incentives and punishments, and control effect checking. These mechanisms collectively induce transformations in four critical dimensions of land utilization: land use structure, land quantity, land spatial layout, and land use functions. These land use transformations demonstrate UGB’s capacity to influence three crucial urban development aspects: economic activities, social needs, and ecological environment. The urban status, representing the human production and living activities supported by land resources, encompasses these three aspects and can be quantitatively expressed through LCC measurements. The conceptual framework not only illustrates the domestic implications but also highlights the international relevance of China’s empirical study, particularly through the mediating role of its unique land management system. China’s land management, operating under public ownership, is predominantly administered through administrative mechanisms. This stands in contrast to market-driven land management systems prevalent in countries with private land ownership. Consequently, this study provides valuable insights for comparative analysis of UGB’s differential impacts on LCC under two distinct paradigms: market-led versus administrative-led land management systems.

Data and methods

This study focuses on addressing two fundamental research questions: (1) How can LCC be effectively measured? (2) What are the impacts of trial UGB delineation on LCC? To address the first question, this study employs the Mean Square Error-Technique for Order Preference by Similarity to an Ideal Solution (MSE-TOPSIS) model for LCC measurement. This comprehensive model evaluates LCC through three distinct dimensions: economic carrying capacity, social carrying capacity, and ecological carrying capacity of land. Regarding the second research question, the study utilizes the trial UGB delineation as a quasi-natural experiment and applies the Difference-in-Differences (DID) model to rigorously analyze the effects of trial UGB implementation on LCC.

Indicators selection and calculation for LCC

In the early conceptualization of LCC, it was widely assumed that LCC had an insurmountable upper limit. However, with technological advancements, this notion has been challenged, sparking debate over whether LCC is constrained by a static and unchangeable threshold (Lindberg et al. 1997). Despite these debates, it remains undeniable that the finite nature of land resources fundamentally limits their ability to meet the material and spatial demands of human production and living activities. Consequently, LCC has been conceptualized as a metric to describe the pressures exerted on land due to socioeconomic development (Shen et al. 2022). To operationalize this concept, previous research has categorized LCC into several dimensions and proposed specific indicators for its assessment, as LCC cannot be measured directly (Table 1). These indicators provide a framework for evaluating the capacity of land to sustain economic, social, and ecological activities under the constraints of resource availability and environmental conditions.

Table 1 Indicators of LCC in previous research.
Full size table

Building on previous research, this study categorizes LCC into three distinct components: economic carrying capacity of land, social carrying capacity of land, and ecological carrying capacity of land. However, the selection of indicators for LCC varies across studies due to differences in local contexts and data availability. In this study, the selection of specific indicators was guided by established literature (Table 1) and constrained by the availability of publicly accessible data (Luo et al. 2020; Tehrani and Makhdoum, 2013; Shen et al. 2022; Tsou et al. 2017). A total of 21 indicators were initially identified across the three domains: economic, social, and ecological carrying capacity of land. To refine these indicators, an expert panel evaluation was conducted, following methodologies employed in prior studies (You et al. 2018, 2020). The panel comprised seven experts specializing in land spatial planning and land use policy. The evaluation process was structured around four criteria: measurability, discrimination, scientific validation, and suitability (Su et al. 2019; Li et al. 2019). Each indicator was assessed by the experts and assigned a score based on its alignment with these criteria, using a four-point scale: 1 (very unfit), 2 (unfit), 3 (fit), and 4 (very fit). Equal weight was assigned to each criterion to calculate the average scores for the indicators. Only those indicators receiving consistently high scores were retained. Following the experts’ recommendations, the final set of 15 indicators for LCC was selected (Table 2). The study utilizes original data spanning from 2005 to 2019, sourced from a variety of authoritative databases and publications, including www.cei.cn, statistical yearbooks of the selected cities, statistical communiqués on the environment and national economic and social development of the selected cities, China City Statistical Yearbooks, and China Statistical Yearbooks on Environment.

Table 2 Selected indicators for LCC and weights.
Full size table

The MSE-TOPSIS model is based on the concept that the chosen alternative should have the shortest geometric distance from the positive ideal solution and the longest geometric distance from the negative ideal solution (Yoon and Hwang, 1995). The MSE-TOPSIS model uses a simple computation process. It is suitable for different data types, and can solve a variety of decision-making problems, including comprehensive assessment, sorting, and selection. Therefore MSE-TOPSIS model was applied to assess LCC. The Min-Max normalization method was used to obtain dimensionless original data. The normalization equations are as follows:

$${\rm{Positive}}\,{\rm{indicator}}:{x}_{ij}=({v}_{ij}-\,\min {v}_{ij})/(\max {v}_{ij}-\,\min {v}_{ij})$$
(1)
$${\rm{Negative}}\,{\rm{indicator}}:{x}_{ij}=(\max {v}_{ij}-{v}_{ij})/(\max {v}_{ij}-\,\min {v}_{ij})$$
(2)

where vij is the original value of the ith indicator in year j, max vij is the maximum value of the corresponding indicator, min vij is the minimum value of the corresponding indicator, and xij is the normalized dimensionless value.

The mean Square Error model was used to assign weights to the indicators, and the indicator weight values wi are shown in Table 1. Technique for Order Preference by Similarity to an Ideal Solution model was utilized to calculate the LCC. The normalized standard matrix X is multiplied by the indicator weight value wi to get the weighted normalized matrix Y = (yij)n×m = (xij -wi)n×m (i = 1,2,…,m; j = 1,2,…,n), and the positive and negative ideal solutions and distances are calculated with the following equation:

$${Y}^{+}=\{\mathop{\max }\limits_{1\le i\le m}{y}_{ij}|i=1,2,\cdots ,m\}=\{{y}_{1}^{+},{y}_{2}^{+},\cdots ,{y}_{m}^{+}\}$$
(3)
$${Y}^{-}=\{\mathop{\min }\limits_{1\le i\le m}{y}_{ij}|i=1,2,\cdots ,m\}=\{{y}_{1}^{-},{y}_{2}^{-},\cdots ,{y}_{m}^{-}\}$$
(4)
$${D}_{j}^{+}=\sqrt{\mathop{\sum }\limits_{i=1}^{m}{({y}_{i}^{+}-{y}_{ij})}^{2}}$$
(5)
$${D}_{j}^{-}=\sqrt{\mathop{\sum }\limits_{i=1}^{m}{({y}_{i}^{-}-{y}_{ij})}^{2}}$$
(6)

where yi+ is the maximum value of the ith indicator in the weighted normalized matrix, namely the positive ideal solution; yi- is the minimum value of the ith indicator in the weighted normalized matrix, namely the negative ideal solution; and Dj+ and Dj- denote the Euclidean distances between the ith indicator and the positive and negative ideal values, respectively.

The nearness degree Tjk is selected to describe the state of the LCC and the three components composed of LCC. A greater Tj means a better state of the LCC and three components composed of LCC in the jth year. The nearness degree Tjk is calculated as follows:

$${T}_{jk}=\frac{{D}_{jk}^{-}}{{D}_{jk}^{+}+{D}_{jk}^{-}}$$
(7)

DID model

The DID model is a widely used quasi-experimental method for evaluating the causal impact of a treatment or intervention (Lechner, 2011). It estimates the change in the mean outcome for a group that undergoes a treatment or intervention (the experiment group) before and after the treatment, and compares this change with the corresponding change for a similar group that does not receive the treatment (the control group). The DID model has been extensively applied to assess the outcomes of policies or natural events, particularly in the field of urban planning and development. For instance, Dempsey and Plantinga (2013) employed a DID estimator to evaluate the effect of Urban Growth Boundaries (UGBs) on the probability of development. Similarly, Chen and Dagestani (2023) utilized the Chinese smart city pilot policy—a novel urban planning initiative aimed at promoting sustainable urban development—as a natural experiment and applied the DID model to examine its impact on clean energy development. The DID model offers several advantages: (1) it can establish causal relationships using observational data, provided that key assumptions are met; (2) it can effectively isolate the effect of a specific treatment or intervention, as long as other concurrent treatments or interventions do not significantly influence the outcome; and (3) its results can be rigorously validated through a variety of robustness tests, such as placebo tests, to ensure reliability. Given these strengths, the DID model is particularly well-suited for assessing the effects of trial UGB delineation on LCC.

The trial implementation of UGB is treated as a quasi-natural experiment in this study. The selection of experiment and control groups should follow three principles: (1) clearly identify the individuals or entities directly affected by the policy or intervention as the experiment group; (2) construct a control group that closely mirrors the experiment group in terms of pre-intervention characteristics and trends, while ensuring it remains entirely unaffected by the intervention; and (3) ensure the availability of adequate longitudinal data for both groups, covering multiple time points both before and after the implementation of the intervention. In DID model, there is no absolute rules for the sample sizes of experiment and control groups. However, the following principles should be followed to ensure reliable and valid estimates: (1) the number of cities in control group should not be too small: The control group must be large enough to support the common trends assumption. If the control group is much smaller than the experiment group, validating this critical assumption becomes challenging. (2) Extreme imbalance of sample sizes between experiment group and control group. It may introduce bias due to high heterogeneity or matching difficulties. Try to keep the sample sizes of the experiment and control groups close, for example, a ratio between 1:1 and 1:3 is common.

Therefore 14 pilot cities in trial UGB delineation—Beijing, Shanghai, Guangzhou, Shenzhen, Nanjing, Suzhou, Hangzhou, Xiamen, Shenyang, Wuhan, Zhengzhou, Xi’an, Chengdu, and Guiyang—were selected as the experiment group (Fig. 2). Prior to the implementation of the policy or intervention, the control group should be as similar as possible to the experimental group in all relevant aspects, such as economic characteristics, demographic composition, historical trends, and other key dimensions. Therefore The selection of the control group was guided by four key criteria in this study: population size, data availability, geographic distribution, and city representativeness. In China, urban hierarchy is closely tied to population size, with cities exceeding 5 million inhabitants classified as megacities. Since most of the pilot cities fall into this megacity category, the control group was carefully chosen to include cities with comparable population sizes. Consequently, 15 cities—comprising municipalities directly under the central government, provincial capitals, or other cities with populations of 5 million or more—were identified as the control group. These cities include Tianjin, Dalian, Changchun, Harbin, Jinan, Qingdao, Nanchang, Hefei, Changsha, Fuzhou, Chongqing, Shijiazhuang, Ningbo, Taiyuan, and Kunming (Fig. 2). This approach ensures that the experiment and control groups are comparable in terms of urban scale and demographic characteristics, thereby enhancing the validity of the quasi-experimental design.

Fig. 2
figure 2

Experiment group and control group in the DID model.

Full size image

The study period was 2005–2019, of which 2014 was the year when the Ministry of Housing and Urban-Rural Development of the People’s Republic of China and the Ministry of Land and Resources of the People’s Republic of China jointly identified 14 megacities with populations of more than 5 million to start the trial work of delineating UGB. A difference-in-difference model is constructed to analyze the impact of the trial UGB delineation on the LCC with the following equation:

$$\begin{array}{l}{Z}_{it,c}={\beta }_{0,c}+{\beta }_{1,c}\cdot DID+{\beta }_{2,c}\cdot YEA{R}_{it,c}\\\qquad\quad+\,{\beta }_{3,c}\cdot TREA{T}_{it,c}+{X}_{it,c}+{\mu }_{i,c}+{\varepsilon }_{it,c}\end{array}$$
(8)

where µi,c is the fixed effect of the city, εit,c is the random disturbance term. Zit,c is the explained variables that describe the state of the LCC in tth year of the ith city. Zit,c includes the economic carrying capacity, the social carrying capacity, the ecological carrying capacity of land, and LCC. DID is the explanatory variable that denotes the trial UGB delineation policy. The trial UDB delineation is considered as a policy shock in the DID model. DID is the cross-product term of the time dummy variable YEAR and the area dummy variable TREAT. Xit,c are the control variables, and the selection of control variables varies according to the differences of the three explanatory variables: the economic carrying capacity of land, the social carrying capacity of land, and the ecological carrying capacity of land. Based on the studies by Sun et al. (2020), 11 control variables were selected. The control variables and their explanations are shown in Table 3. Statistical description of control variables is shown in Table 4. To reduce the influence of different units of indicators and improve regression convergence in regressions, the control variables of the economic carrying capacity of land and the social carrying capacity of land are normalized by Eq. 1, and the control variables of the ecological carrying capacity of land are normalized by Eq. 2.

Table 3 Selected control variables and definitions.
Full size table
Table 4 Statistical description of selected control variables.
Full size table

Results

Basic regressions

Table 5 shows the basic regression results. Model 1.1-Model 1.8 reveal the effects of trial UGB delineation on the LCC. The coefficients of DID are significantly negative in Model 1.1 and Model 1.2. Meanwhile, R2 value is improved in Model 1.2 after adding the control variables. These regression results denote that the trial UGB delineation reduces the economic carrying capacity of land. The coefficients of DID are significantly positive in Model 1.3 and Model 1.4. Meanwhile, R2 value is also improved in Model 1.4 after adding the control variables. These regression results denote that the trial UGB delineation reduces the social carrying capacity of land. The coefficients of DID are significantly positive in Model 1.5 and Model 1.6. Meanwhile, R2 value is improved in Model 1.6 after adding the control variables. These regression results denote that the trial UGB delineation reduces the ecological carrying capacity of land. The coefficients of DID are significant in Model 1.7 and Model 1.8. But the coefficient of DID in Model 1.7 is positive, and negative in Model 1.8. The reason may be that the control variables in Model 1.8 include all three explanatory variables: economic carrying capacity, social carrying capacity, and ecological carrying capacity of land. Model 8 does not delete the “bad variables” among them. Therefore, this study no longer analyzes the LCC as an explained variable in the subsequent analysis, and focuses on the three explanatory variables of economic carrying capacity, social carrying capacity, and ecological carrying capacity of land.

Table 5 Basic regressions of DID models.
Full size table

Robustness tests

Robustness test includes parallel trend test and placebo test. The parallel trend test verifies whether there is a significant difference between trial cities (experiment group) and non-trial cities (control group) before policy implementation, and only those non-trial cities with similar characteristics can be used as the control group. The parallel trend test generates the time dummy variables, and then the cross-product term of the generated time dummy variables and policy dummy variable is added in regression analysis. In the parallel trend test, the cross-product term for 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, and 2019 with the dummy variables of the experiment group are represented as before8, before7, before6, before5, before4, before3, before2, current, after1, after2, after3, after4, and after5, respectively. To avoid collinearity, the first period of 2013 before policy implementation (before 1) is dropped as the reference group in the test (Dragone et al. 2019). The coefficients of before3 and before2 are not significant in Models 2.1–2.3 in Table 6. Therefore, before the trial UGB delineation, economic carrying capacity of land, social carrying capacity of land, and ecological carrying capacity of land in the cities in the experiment group and control group shows the same trend of change without significant differences. The results justify the robustness of our model, and the parallel trend test satisfies the requirements of the DID model.

Table 6 Parallel trend test.
Full size table

Placebo tests diagnose problems with research designs in observational studies. Placebo tests fictitious policy implementation time in the robustness test. In the placebo test, the sample from 2005 to 2013 without the trial UGB delineation was selected. Xu and Li (2020) designed a selection of the lead time for the occurrence of fictitious processing effects. According to Xu and Li (2020), 2008, 2009, and 2010 are selected as the hypothetical years for UGB delineation. As shown in Table 6, the estimated coefficients of DID for the fictitious policy variables are not significant during the fictitious policy implementation time in Models 3.1–3.9 in Table 7. It indicates that excluding the impacts of the UGB delineation on trial cities, there are no systematic differences in the trend of the economic, social, and ecological carrying capacities of land between trial cities in the experiment group and non-trial cities in the control group.

Table 7 Placebo Test.
Full size table

Urban heterogeneity tests

Urban heterogeneity tests are conducted to examine how the effects of trial UGB delineation vary across different urban locations. In the DID model, there are no strict rules governing the sample sizes of the experimental and control groups. This study utilizes original data spanning from 2005 to 2019, ensuring that the sample sizes of both the experimental and control groups are adequate for conducting subgroup analyses in urban heterogeneity tests. For instance, Zhang et al. (2023) employed a DID model to analyze panel data from 57 counties between 2002 and 2019, investigating the relationship between industrial transfer parks and county-level economic growth. In their heterogeneity analysis, they categorized the data into three regions: the coastal area of eastern Guangdong (EG), the coastal area of western Guangdong (WG), and the mountainous area of northern Guangdong (NG). Notably, the experimental group in the EG region comprised only 3 cities. Furthermore, DID model analyses using province-level samples are common in China, with typical sample sizes around 30 provinces (Xu et al. 2022; Gao et al. 2020). Based on these precedents, this study classifies urban locations into three regions—eastern region, central region, and western region—for the purpose of urban heterogeneity testing. This regional classification allows for a more nuanced understanding of how UGB delineation impacts different areas.

Regional dummy variable LOCATION was added in basic regressions for urban heterogeneity tests. The dummy variables LOCATION were assigned as follows: concerning the eastern dummy variables, LOCATION = 1 for the eastern cities, while LOCATION = 0 for the central and western cities; concerning the central dummy variables, LOCATION = 1 for the central cities, while LOCATION = 0 for the eastern and western cities; concerning the western dummy variables, LOCATION = 1 for the western cities, while LOCATION = 0 for the eastern and central cities. Then the cross-product term DID × LOCATION was added to establish the triple difference-in-differences models for urban ___location heterogeneity tests. As shown in Table 8, trial UGB delineation has heterogeneous effects on the economic carrying capacity, social carrying capacity, and ecological carrying capacity of land in cities with different regions. Regarding the economic carrying capacity of land, the coefficient of DID × LOCATION is significantly negative after adding the dummy variables of the eastern region and western region. However, the coefficient of DID × LOCATION is negative but insignificant after adding the dummy variable of the central region. Regarding the social carrying capacity of land, the coefficient of DID × LOCATION is significantly positive after adding the dummy variables of the east region. The coefficient of DID × LOCATION is significantly negative after adding the dummy variables of the central region and western region. Regarding the ecological carrying capacity of land, the coefficient of DID × LOCATION is significantly positive after adding the dummy variable of the eastern region. The coefficient of DID × LOCATION is significantly negative after adding the dummy variable of the central region. The coefficient of DID × LOCATION is negative but insignificant after adding the dummy variable of the western region.

Table 8 Urban ___location heterogeneity tests.
Full size table

Then the urban heterogeneity tests analyze the influence of the difference of urban hierarchy on the effect of trial UGB delineation. According to the National Urban System Plan (2010–2020), the Chengdu−Chongqing Urban Agglomeration Development Plan, the National Development and Reform Commission Reply on Supporting to Build Wuhan as a National Central City, the National Development and Reform Commission Reply on Supporting to Build Zhengzhou as a National Central City, and the Guanzhong Plain Urban Agglomeration Development Plan, Beijing, Tianjin, Shanghai, Guangzhou, Chongqing, Chengdu, Wuhan, Zhengzhou, and Xi’an are set as nine national central cities in China. The national central cities are the highest hierarchy cities in China. Therefore the sample cities were divided into two groups, namely national central city and non-national central city, to test the effects of trial UGB delineation on the LCC in national and non-national central cities. Dummy variable of urban hierarchy (HIERARCHY) was added in basic regressions for urban heterogeneity tests. The dummy variables HIERARCHY were assigned as follows: if the city is a national central city, then HIERARCHY = 1; otherwise HIERARCHY = 0. Then the cross-product term DID × HIERARCHY was added to establish the triple difference-in-differences models for urban hierarchy heterogeneity tests. As shown in Table 9, the differences in the urban hierarchy also have an impact on the effect of trial UGB delineation. Regarding the economic carrying capacity of land, the coefficient of DID × HIERARCHY is significantly negative after adding dummy variables of urban hierarchy. Regarding the social carrying capacity and ecological carrying capacity of land, the coefficient of DID × HIERARCHY is significantly positive after adding dummy variables of urban hierarchy.

Table 9 Urban hierarchy heterogeneity tests.
Full size table

Discussion

Hidden causes of relationships between UGB delineation and LCC

The trial UGB delineation exerts a significant negative impact on the economic carrying capacity of land, as evidenced by the DID coefficients of −0.057 and −0.059 (Table 5). This reflects a deliberate policy trade-off, wherein UGB delineation restricts the spatial expansion of urban construction land, particularly in trial cities that serve as China’s primary economic hubs. The primary objective of UGB delineation is to manage the boundaries between construction and non-construction areas (Tan and Xu, 2023), with the concentrated and contiguous urban construction land within the UGB carrying the majority of the city’s economic activities. By concentrating development within the UGB, the policy limits the availability of land for new economic activities, directly constraining short-term economic output. This aligns with the DID results, which quantify the economic cost of prioritizing boundary control over unchecked urban sprawl. These findings are consistent with the study by Long et al. (2015), which identified that Beijing’s UGBs were highly effective in containing human activities, as most areas outside the UGBs were not efficiently utilized.

The trial UGB delineation demonstrates significant positive impacts on both social and ecological carrying capacities of land, as evidenced by DID coefficients of 0.051, 0.031 and 0.030, 0.027 respectively (Table 5). The social capacity of land is enhanced since the policy systematically increases accessibility to public amenities by mandating reserved spaces for healthcare, education, and transportation infrastructure within UGB. Regression results confirm this causal pathway, with improved social carrying capacity reflecting spatial prioritization of livelihood-related services—a core objective of China’s “People-Centric Urbanization” strategy. The ecological capacity of land is enhanced since ecological capacity gains. While cities adopt varied boundary configurations in UGB trials, all implementations consistently protect restricted development zones, including green spaces, farmland, and permanent basic farmland (Yang et al. 2021). These measures enhance ecosystem service provision through landscape connectivity preservation and carbon emission reduction (Chen et al. 2023), while coordinated infrastructure expansions within UGBs mitigate environmental pressures from rapid urbanization. Besides, the effect of trial UGB delineation on land economic carrying capacity is opposite to its effect on social carrying capacity and ecological carrying capacity of land, so there is a dilemma in balancing land economic carrying capacity with land social and ecological carrying capacities. It can be seen from the regression results that the current UGB is more likely to improve the social and ecological carrying capacities of land at the cost of sacrificing a certain amount of economic carrying capacity.

Social factors constitute drivers underlying the relationship between UGB delineation and LCC. As a complex socio-spatial intervention, UGB implementation responds to China’s dual challenges of overinvestment-driven growth and urban livability crises. By containing urban sprawl, UGB mitigates excessive capital allocation to land development, thereby reducing economic carrying capacity as a necessary trade-off for sustainable urbanization. The Chinese government’s strategic intentions play a pivotal role in shaping UGB policies. Confronted with pervasive “urban diseases”—including traffic congestion, environmental pollution, and healthcare accessibility deficits—the state has prioritized building “People-Centric Cities” through enhanced public service provision. Simultaneously, the “Beautiful China” initiative drives ecological modernization, emphasizing ecosystem diversity, stability, and sustainability through green development pathways. UGB delineation serves as a spatial instrument to operationalize these national priorities.

Understanding urban heterogeneity in effects of UGB delineation on LCC

Regarding the economic carrying capacity of land, the coefficient of DID × LOCATION is negative but insignificant after adding the dummy variable of the central region. The possible reason is that the land development of trial cities in central China, such as Zhengzhou and Wuhan, faces stronger economic development needs than those in the eastern and western regions (Zhang et al. 2018). To meet the uncertain demand for urban development in this economic development, central cities such as Zhengzhou and Wuhan have established an open-ended elastic adjustment mechanism in the trial UGB delineation (Tan and Xu, 2023). It allows the adjustment of UGB morphology at the premise of satisfying land indicators. The negative impact of UGB delineation on the economic carrying capacity of land is limited, thus resulting in insignificant regression results. Regarding the social carrying capacity of land, the coefficient of DID × LOCATION is significantly positive after adding the dummy variables of the east region. The coefficient of DID × LOCATION is significantly negative after adding the dummy variables of the central region and western region. The possible reason is that the major trial cities in the eastern region, such as Beijing, Shanghai, Guangzhou, Shenzhen, Nanjing, and Xiamen, have adopted a citywide boundary delineation method for their UGB. These cities provide more space to carry public service facilities for ensuring people’s livelihoods. It is also in line with the characteristics of the social development stage of the trial cities in the eastern region. The majority of trial cities in the central and western regions adopt the two methods for UGB delineation: delineating UGB in the planning areas involved in the central urban area, and delineating UGB in the central urban area and key planning control areas. So it is difficult to increase the carrying capacity of land for infrastructures in trial cities in the central and western regions since existed high-density infrastructures in urban centers. Regarding the ecological carrying capacity of land, there are contradictions between the UGB delineation and the protection redlines of permanent basic farmland and ecology (Wang et al. 2020). The trial cities in the eastern region coordinate the urban development space and protective space, which is conducive to improving the ecological environment. Therefore the coefficient of DID × LOCATION is significantly positive after adding the dummy variable of the eastern region. The trial cities in the central region suffer the strong demand for urban development, thus resulting in conflicts between urban development space and protective space as well as the reduction in the supply of ecological products and services. Therefore the coefficient of DID × LOCATION is significantly negative after adding the dummy variable of the central region. Besides, the expansion of development space in the western trial cities is likely weaker than that in the central trial cities, and the damage to protective spaces is also smaller. Therefore the coefficient of DID × LOCATION is negative but insignificant after adding the dummy variable of the western region.

Urban hierarchy heterogeneity tests reveal the heterogeneous effects of UGB delineation on LCC adding dummy variables of urban hierarchy. The social factors relate to the heterogeneous effects of UGB delineation on LCC. The selection of a national central city is an administrative act. The different cities try to become national central cities based on their competitive advantage. The national central city is positioned not only by its economic influence in the country, but also by social factors, such as its ability to participate in international competition and affect the national politics, economy, and culture (Feng et al. 2021). Therefore, to realize the function of the national central city under the limited resource and environment-carrying conditions, the UGB delineation mainly aims at satisfying the demand that land can carry the increasing public service facilities, ecological products, and ecological services for people’s livelihood.

Policy implications

This study’s empirical findings on the effects of UGB delineation on LCC provide a foundation for the following targeted policy recommendations.

Regionally differentiated UGB adjustment mechanisms should be adopted since the heterogeneous effects of UGB delineation on LCC. Regionally differentiated UGB adjustment mechanisms should prioritize three factors: regional development stages, land-use conflicts, and urban functional positioning. For eastern megacities facing growth pressures, “flexible UGB + floor area ratio incentives” can be adopted, allowing high-density zones to compensate for land constraints through vertical expansion. In areas with land-use conflicts, participatory zoning should resolve competing demands through a priority hierarchy. Finally, UGB must align with cities’ strategic roles, differentiating between industrial hubs and ecological conservation zones to avoid uniform rigidity. These mechanisms, grounded in spatial equity and risk mitigation, and balance growth needs across diverse regional contexts.

The dynamic resource-carrying capacity evaluation should be integrated into UGB delineation since the tightly relationship between UGB delineation and LCC. To effectively integrate dynamic resource-carrying capacity evaluation into UGB delineation, it is critical to develop a predictive resource-carrying capacity assessment framework that anticipates future social, economic, and ecological demands in alignment with planning objectives. This requires embedding advanced spatial analysis tools—such as GIS-based decision support systems and scenario modeling—into territorial spatial planning to strengthen the linkages between resource-carrying capacity evaluations and land management practices. Specifically, China should institutionalize their mandatory integration into municipal-level UGB delineation processes, as outlined in the Guidelines for Municipal Territorial Spatial Planning (2020). Such reforms would ensure that UGB dynamically adapts to evolving resource constraints while balancing development needs across spatial and temporal scales.

Implementing full-cycle management of UGB should be established as a critical institutional mechanism, given the profound impacts of UGB delineation on economic growth, social equity, and ecological sustainability. To ensure UGBs fulfill their intended role as a “growth container” rather than a symbolic line, a robust framework must integrate implementation, supervision, evaluation, assessment, and enforcement across the entire lifecycle of UGB. This process should begin with multi-stakeholder coordination during implementation, ensuring alignment between land use, infrastructure development, and environmental conservation. Besides, it is imperative to promote the specialized legislation of the “Territorial Spatial Planning Law” to clarify the legal status of UGB. A dual control mechanism of “negative list + positive guidance” should be established.

Limitations of the work

However, it is important to acknowledge several limitations inherent in our study. First, while we have examined the impact of trial UGB delineation on LCC, it is crucial to recognize that UGB implementation is a multifaceted social issue. Our research does not account for the potential interactions between UGB delineation and other social factors that may influence LCC. Second, the indicators we selected to characterize LCC are specifically tailored to the Chinese context. Their applicability to other regions requires further validation, as different areas may have distinct social, economic, and ecological conditions that could affect the relevance of these indicators. Third, the DID model relies on a sufficient volume of sample data for accurate regression estimation. Insufficient data can compromise the reliability of the results. In this study, the analysis covers the period from 2005 to 2019. However, inconsistencies in statistical indicators across cities and missing data for certain years may have affected the quality of the DID model, suggesting room for further improvement in data collection and standardization.

The limitations of our work imply that future studies should adopt other regions and combine more diverse methods to deepen the knowledge of the relationship between UGB delineation on LCC. Specifically, more control variables regarding social influencing factors need to be added to the models. Meanwhile, the indicators for describing the LCC should be adjusted according to the local social, economic, and ecological conditions. This study also reveals the directions for future practice and policy. The empty institutions emerge because land users and public administrations are unwilling or unable to implement planning institutions (You et al. 2022). The UGB delineation needs to combine with regional development needs. And policy makers should pay attention to the feasibility of planning policies.

Conclusion

China’s rapid urbanization has precipitated significant urban sprawl, necessitating effective spatial governance tools like UGB. This study evaluates the impact of China’s trial UGB delineation (initiated in 14 pilot cities in 2014) on LCC, decomposed into economic, social, and ecological dimensions. Results reveal that UGB delineation significantly reduces economic carrying capacity but enhances social and ecological capacities, reflecting trade-offs between growth constraints and sustainability. Urban heterogeneity tests highlight divergent outcomes: eastern cities prioritize social-ecological gains, western cities focus on ecological benefits, while central cities face unresolved tensions between development needs and spatial constraints. Additionally, national central cities exhibit unique hierarchical impacts, balancing public service demands with ecological preservation.

This study advances our understanding of UGB by establishing two critical contributions to sustainable land governance. First, we reveal how China’s unique land management system, characterized by centralized control and public ownership, acts as both a mediator and amplifier of UGB impacts on LCC. Unlike market-driven systems, China’s administrative-led framework enables rapid spatial reconfiguration but intensifies trade-offs between economic, social, and ecological dimensions of LCC, as quantified by DID coefficients. Second, we demonstrate that UGB effectiveness is inherently place-specific: eastern cities face steeper economic costs due to high development pressure, western cities prioritize ecological gains, while central cities balance social equity demands. The results were validated through regional heterogeneity analysis.

These findings in this study offer two actionable insights for global implementation. First, the “economic-social-ecological” triple squeeze effect of LCC necessitates establishing an adaptive UGB framework. Cities in eastern regions should alleviate growth constraints through resilient boundary design (e.g., floor area ratio incentive mechanisms), while less-developed areas ought to implement ecological compensation systems to balance development rights in protected areas. Second, developing countries must construct risk early-warning systems when implementing UGB, preemptively assessing chain risks such as economic stagnation from excessive boundary constraints, social imbalances caused by public service gaps, and ecological degradation triggered by regulatory failures.