Distribution dynamics, regional differences, and convergence of sustainable development of cities and communities in China

Abstract

China is one of the most populated and rapidly urbanizing countries worldwide and was among the earliest countries to integrate sustainable development into urban construction. To achieve high-quality development and implement the objectives of "Transforming Our World: The 2030 Agenda for Sustainable Development", it is crucial to measure and analyze the current level of sustainable development of cities in China. Following the principles of relevance, scientific rigor, universality, reliability, and timeliness, this study constructs an assessment index system for sustainable development, covering seven themes corresponding to the UN Sustainable Development Goal 11. Through detailed calculations, we obtained sustainable development indices for 139 Chinese cities from 2016 to 2022 and analyzed them in three dimensions: distribution dynamics, regional differences, and convergence. The key findings are as follows. First, the level of sustainable development showed improvement, with the average score of included cities increasing by 11.88% from 2016 to 2022. Second, the level of sustainable development was relatively balanced, maintaining low Gini coefficients between 0.05 and 0.06. Third, a weak overall σ convergence feature existed, with increased differentiation in 2021. From a regional perspective, a σ convergence feature was observed in the northeastern but not in the western region. Fourth, both overall absolute β-convergence and conditional β effects were significant. Regional absolute β-convergence and conditional β-convergence were also significant. This study contributes to the literature by providing evidence of China' s urban sustainable development, offering policy insights for deepening the implementation of development goals in the future, and providing experiential reference for other developing countries to achieve sustainable development.

Headnote

ARTICLE INFO

Keywords:

Urban sustainability

Sustainable Development Goal 11

Dynamic evolution of distribution

Regional differences

Spatial convergence

ABSTRACT

China is one of the most populated and rapidly urbanizing countries worldwide and was among the earliest countries to integrate sustainable development into urban construction. To achieve high-quality development and implement the objectives of "Transforming Our World: The 2030 Agenda for Sustainable Development", it is crucial to measure and analyze the current level of sustainable development of cities in China. Following the principles of relevance, scientific rigor, universality, reliability, and timeliness, this study constructs an assessment index system for sustainable development, covering seven themes corresponding to the UN Sustainable Development Goal 11. Through detailed calculations, we obtained sustainable development indices for 139 Chinese cities from 2016 to 2022 and analyzed them in three dimensions: distribution dynamics, regional differences, and convergence. The key findings are as follows. First, the level of sustainable development showed improvement, with the average score of included cities increasing by 11.88% from 2016 to 2022. Second, the level of sustainable development was relatively balanced, maintaining low Gini coefficients between 0.05 and 0.06. Third, a weak overall σ convergence feature existed, with increased differentiation in 2021. From a regional perspective, a σ convergence feature was observed in the northeastern but not in the western region. Fourth, both overall absolute β-convergence and conditional β effects were significant. Regional absolute β-convergence and conditional β-convergence were also significant. This study contributes to the literature by providing evidence of China' s urban sustainable development, offering policy insights for deepening the implementation of development goals in the future, and providing experiential reference for other developing countries to achieve sustainable development.

1. Introduction

China is one of the most populated and rapidly urbanizing countries worldwide and was among the earliest nations to incorporate sustainable development into urban construction. Urban and rural development in China has profoundly influenced sustainable development worldwide. Since China signed the 2015 Agenda for Sustainable Development, called "Transforming Our World: the 2030 Agenda for Sustainable Development" (hereinafter, the 2030 Agenda), along with over a hundred global leaders at the United Nations General Assembly in 2015, China has consistently highlighted sustainable development in significant international activities. Hence, sustainable development as a national strategy, together with global advancement of the 2030 Agenda, has increasingly gained public attention.

The construction of inclusive, safe, resilient, and sustainable cities and communities is a critical part of the 2030 Agenda (United Nations, 2015) not only because it represents the 11th of 17 sustainable development goals (SDGs), but also because of its close association with other goals. Most SDG indicators involve cities and human settlements and approximately one-third are measured at the local level (UN-Habitat, 2018). The reason for attaching such importance to urban sustainable development is that cities have large populations and economic activities. Cities' development patterns directly affect the efficiency of resource utilization and environmental conditions. In addition, sustainable development of cities and communities plays a vital role in addressing the social and environmental challenges faced by global development. The coronavirus disease pandemic and other crises have posed momentous challenges to sustainable development, particularly in urban areas. The experience of recovering from the pandemic demonstrates the resilience and adaptability of urban areas in dealing with emergencies and establishing "new-normal" mechanisms. Some cities have become important engines for economic and social recovery and innovation, with significant consideration of, and envisioning opportunities for, urban areas as centers for sustainable and inclusive growth. China's urban population is substantial. In 2022, it accounted for 19.86% of the world's urban population and 11.29% of the world's total population". Therefore, measuring and analyzing China's current level of urban sustainable development is crucial for achieving high-quality development and implementing the 2030 Agenda.

In 2023, the United Nations updated the global indicator framework for the 17 SDGs and 169 specific targets set by the 2030 Agenda (United Nations, 2016). Among them, there are 247 suggested monitoring indicators. However, most indicators mentioned in the 2030 Agenda focus on the national level and diverge from the objectives and means of urban sustainable development. Therefore, the construction of an indicator system for sustainable development must follow two principles. First, it should adapt to local conditions by selecting available data that can reflect the actual situation in China; and second, it must focus on cities by choosing city-level data that cover the majority of cities. Indeed, aside from being crucial for assessing the sustainable development of Chinese cities, these principles also serve as reliable references when assessing urban sustainable development in other countries.

2. Liferature review

Several studies focusing on various countries have assessed urban sustainable development. Extensive research was conducted on sustainable development assessments prior to the official release of the 2030 Agenda. Sala et al. (2015) provided a methodological framework for sustainability assessment from three perspectives: ontological, methodological, and epistemological. Castelnovo et al. (2015) proposed a comprehensive method for assessing governance and policy decisions in the context of smart and sustainable cities. Yigitcanlar et al. (2015) introduced a multiscalar urban sustainability approach, analyzing sustainability performance at the micro and meso levels in Gold Coast City, Australia, and generating multiscalar outcomes at the macro level. Braulio-Gonzalo et al. (2015) summarized 13 representative indicator systems and constructed a sustainable development assessment indicator system considering the characteristics of a Mediterranean city in Spain. International organizations such as Arcadis, Siemens, the Economist Intelligence Unit, United Nations Human Settlements Programme, and U.S. Department of Housing and Urban Development have conducted in-depth research on urban sustainability indicators.

Since the formal release of the 2030 Agenda, many scholars and institutions have aligned previous studies with the SDGs, further analyzing the implications of sustainable development. They actively explored localized methods for achieving SDGs, established indicators tailored to specific countries or regions based on SDGs, and monitored and analyzed the implementation of SDGs. Klarin (2018) comprehensively introduced the historical origins, development processes, and current status of the concept and theory of sustainable development, referencing the 2030 Agenda from economic, social, and resource perspectives. Coscieme et al. (2020) concluded that enhancing policy coherence for sustainable development in the 2030 Agenda will provide advantages by ensuring interconnectedness among goals via targets and indicators, recognizing global sustainability dimensions, fostering a unified narrative of shared objectives across diverse nations, and intensifying efforts to develop consistent indicators aligned with sustainability principles and advancements in sustainability science and governance. Giles-Corti et al. (2020) examined the extent to which SDGs help cities assess their efforts toward achieving sustainable development and health outcomes. They recommended a more comprehensive approach toward setting benchmarks for policies aimed at achieving healthy and sustainable cities, monitoring and evaluating policies, and assessing spatial inequality. Koch and Krellenberg (2018) focused on the specific role of cities in promoting sustainable development and analyzed how Germany linked SDGs with urban-level sustainability through different measures, emphasizing the relationship between the national and local levels. Steiniger et al. (2020) assessed six cities in Chile using a method involving expert consultations and selecting five sustainable development categories and 29 indicators. These 29 indicators may be allocated to specific SDGs. They proposed that prioritizing an effective set of indicators is crucial for addressing urban development challenges highlighted in the 2030 Agenda and the "New Urban Agenda", and emphasized reviewing the changing levels of urban sustainable development according to the situation over time. Bartniczak and Raszkowski (2022) studied the implementation of SDG 11 in EU countries and analyzed and ranked the sustainable development levels of 28 countries from 2005 to 2020.

Additionally, some studies have examined the relationship between different types of cities and industries and the achievement of SDGs. Pittman et al. (2019) discussed the connection between coastal cities and the SDGs, highlighting the multiple SDGs associated with these cities. Swain and Karimu (2020) explored the close relationship between the electricity industry and sustainable development. Monteiro et al. (2019) investigated the alignment between mining activities and the SDGs and identified numerous possibilities for applying the SDGs to the mining industry to contribute toward their achievements.

Abundant research has focused on assessing the sustainable development of urban areas in China. Before the formal release of the 2030 Agenda, many extensive studies were conducted regarding the fundamental concepts of sustainable development, construction of key indicator systems for sustainable development, assessment methodologies for sustainable development capabilities or levels, policies, and pathways. Niu (2012) analyzed the foundation and theoretical practices of China's sustainable development by extracting the fundamental concepts of sustainable development, three-dimensional mapping, mathematical analysis, and threshold judgment for sustainable development. Regarding the research on indicator systems for sustainable development assessment, Yang et al. (2011) conducted a systematic interpretation of the quantitative assessment of urban sustainability through thematic literature analysis.

From a theoretical perspective, this study begins with various dimensions of urban areas, such as population growth, production input, consumption behavior, and construction operations, to summarize and refine various fundamental models for the quantitative evaluation, comparison, and analysis of their strengths and weaknesses. Practically, this study reviews and evaluates the methods applied by different international and local institutions in this field.

After the formal release of the 2030 Agenda, China established a national-level coordination mechanism and formulated specific national plans for the 2030 Sustainable Development Strategy. The implementation and realization of SDGs were also integrated into the "13th Five-Year Plan". Simultaneously, the achievement of the SDGs on a larger scale has been actively promoted. Specifically, domestic scholars and major research institutions have actively promoted localized research on SDGs, measuring and monitoring indicators for each SDG target. Furthermore, the China International Economic Exchange Center and the Earth Institute at Columbia University released the China Sustainable Development Assessment Report. This report established a city-level sustainable development evaluation indicator system consisting of primary indicators of economic development, social well-being, resources and environment, consumption and emissions, and environmental governance. This study conducted an exploratory evaluation of sustainable development levels in 100 major Chinese cit- ies.

Domestic scholars have also advanced localized SDGs research. Sun et al. (2016) developed a system comprising 24 sustainable development indicators categorized into economic development, social progress, and ecological environment for 277 differently sized prefecture-level cities in China. To advance sustainable regional development policies, pathways, and mechanisms, Yao et al. (2019) established a people-industry-space-institutions research framework and proposed pathways for urban sustainable development. Wang et al. (2018) constructed an open framework for urban sustainability assessment indicators based on a study of international urban sustainable development assessment indicators with SDG 11 as the research focus. Building on the implementation of the 2030 Agenda, Zhu et al. (2018) established a comprehensive method and indicator system applicable at the national level in China. It emphasized coordinated development among the economy, society, resources, and the environment, aligning with the emphasized principles of sustainable development strategies, thus serving as a theoretical foundation. Ma and Ai (2019) constructed an evaluation indicator system for the sustainable development of urbanization based on the SDGs, researching the current state of sustainable development of urbanization in Jilin Province from both provincial and external perspectives. Yang et al. (2021) used a system dynamic assessment model for urban sustainability, selecting 13 cities in the Beijing - Tianjin - Hebei region as samples to assess and simulate spatiotemporal changes in the sustainable development levels of these cities from 2005 to 2035.

The extensive research conducted from various perspectives on urban sustainable development has revealed certain limitations. To address these limitations, the indicator assessment system constructed in this study adheres to the following five principles for selecting the indicators.

(1) Relevance. The selected indicators are linked to the implementation of SDG 11 and consider specific goals and indicators related to urban sustainable development within other targets. However, it excludes indicators primarily at the national level or involving international cooperation, such as "11. b.1-the number of countries that have national and local disaster risk reduction strategies in line with the Sendai Framework for Disaster Risk Reduction 2015-2030".

(2) Scientificity. The indicator system established in this study is based on scientific principles. It objectively reflects the status of urban development and uses research methods rooted in scientific principles to depict interconnections within sustainable development.

(3) Universality. The indicator data used in this study generally covers more than 70% of the selected cities.

(4) Reliability. Data collection and processing in this study are based on effective and reliable statistical methods. Regularly updated datasets are prioritized to monitor progress until 2030.

(5) Timeliness. The selected indicators are current and are released according to a reasonable schedule.

The marginal contributions of this study are threefold. First, in terms of indicator construction, this study devises an evaluation framework for urban sustainable development at the city level, measuring Chinas urban sustainability from multiple dimensions. Second, in processing data, scientific methods are employed to standardize the data and establish rules for determining the optimal values. Third, concerning research content, this study comprehensively demonstrates the structural characteristics of urban sustainable development such as distribution dynamics, regional differences, structural disparities, and convergence; and diagnoses inhibiting factors.

The remainder of this paper is organized as follows. Section 2 introduces the construction of the urban sustainable development evaluation system. Section 3 analyzes the distribution dynamics of urban sustainable development using the kernel density estimation. Section 4 presents an analysis of regional disparities in urban sustainable development using the Dagum Gini coefficient subgroup decomposition and variance decomposition methods. Section 5 separately explores the o-convergence and /-convergence characteristics of sustainable development.

3. Sustainable development: materials, methods, and overall analysis

3.1. Establishment of an evaluation indicator system

Referring to the SDGs evaluation indicator system developed by Shao et al. (2021), the construction of the urban sustainable development index (USDI) in this study involved thematic areas and specific indicators, presented in Table 1.

The primary data sources consisted of official statistical reports such as national and local statistical yearbooks, local statistical communiques on national economic and social development, ecological and environmental status bulletins, reports on the work of local governments, official statistical reports such as the implementation of the local budget, and reports on the local draft budget. For official indicators with missing data or other issues, reliable alternative data sources and measurement methods were used for enhancement such as Anjuke and Creprice; domestic databases such as the China Economic and Social Big Data Research Platform, China Economic and Social Development Statistical Database, and Qianzhan Database; and other official databases such as published journal articles and household surveys.

...

Where, subscript / represents the i-th thematic area; j represents the j-th indicator within the i-th thematic area; x, denotes the raw numerical value; max and min respectively indicate the extremes of the best and worst performances for all data within the same indicator, and /; denotes the computed standardized value, representing the final score for the indicator.

After obtaining scores for the indicators, an aggregation process was performed in two steps: (D Arithmetic averaging using standardized indicator values to derive scores for each thematic area. (2) Employment arithmetic averaging to aggregate each thematic area, resulting in the USDI score. Specifically, this is expressed by the following two equations.

...

Where, 7; represents the score of the i-th thematic area; N; signifies the number of indicators within the i-th thematic area; / denotes the score of the USDI, and N represents the number of thematic areas (for most cities and years, this value is 7, although missing thematic areas for specific samples may result in a value less than 7.

3.2. Overall analysis

Figure 1 shows the average and trend of the USDIs from 2016 to 2022 for China as a whole and its eastern, central, western, and northeastern regions. Overall, the total score had an upward trend, rising from 55.30 in 2016 to 61.86 in 2022, indicating a steady improvement in the sustainable development levels of cities nationwide. Specifically, until 2020, the sustainable development levels in the central and western regions aligned closely with the national average, and that in the eastern region exhibited the highest level. The northeastern region displayed the lowest sustainable development level, trail- ing the other three regions significantly. This outcome may be correlated with economic development, as effective infrastructure construction is crucial for urban sustainable development. The betterdeveloped eastern region is capable of supporting more advanced infrastructure construction, hence maintaining a leading position until 2020 compared with the other regions. However, cities with higher economic development had lower housing security scores, indicating the need to emphasize housing issues in urban sustainable development efforts. After 2020, the development trends among regions changed. Notably, by 2021, the central region experienced rapid development, surpassing the eastern region. By 2022, the central region achieved the highest level, while the eastern and western regions exhibited similar sustainable development levels. The northeastern region, although still lagging behind the other three regions, showed a reduction in gap.

4. Distribution dynamics and evolution of the urban sustainable development index

4.1. Kernel density method

Kernel density estimation uses a smoothed kernel function to fit the sample data and thereby captures the distribution shape of a random variable through a continuous density curve. It possesses advantages such as weak model dependency and strong robustness. This method assumes the density function of the random variable X as follows.

...

Where, n represents the number of observations; Y, represents independently and identically distributed observations; x signifies the mean value; K denotes the kernel density, and 7 represents the bandwidth. A relatively small bandwidth is generally chosen to ensure a higher accuracy. The kernel density, which functions as a weighting or smoothing transformation function, must typically satisfy the following conditions.

...

The kernel density can be divided into uniform, triangular, and Gaussian densities. This study used the widely applied Gaussian kernel density method for estimation and analysis. Generally, one can observe the distributional position, pattern, spread, and polarization phenomena related to the level of urban sustainable development from the curve obtained through kernel density estimation. The distributional position reflects the high or low levels of urban sustainable development. The distribution shape (height and width of peaks) is used to analyze the magnitude of regional differences. The spread characterizes the differences between cities with the highest or lowest levels of urban sustainable development and other cities, with longer tails indicating larger differences. Lastly, polarization refers to the number of peaks, which reflects the degree of multipolarity (Quah, 1993).

4.2. Results of the kernel density estimation

By employing the kernel density estimation method with a Gaussian kernel density function and optimal bandwidth, this study derived the urban sustainable development levels for the entire country and for the eastern, central, western, and northeastern regions. The distribution dynamics were analyzed based on four aspects of distribution: position, pattern, spread, and polarization.

4.2.1. Overall estimation

Figure 2 displays the dynamic evolution trend of the overall USDI across the country. In terms of position, the distribution shifted rightward over time, reflecting an overall increase in sustainable development, consistent with the information presented in Figure 1. Regarding shape, the distribution's form remained relatively unchanged, indicating a stable level of inequality in sustainable development, with the distribution centered around the median in 2020. In terms of spread, there a pronounced leftward tail occurred in 2021, indicating that some cities scored far below the average level for that year. All distributions were unimodal, with no signs of polarization. Additionally, the graph shows small changes between 2016 and 2019 as well as between 2021 and 2022, with a significant rightward shift in the latter years. The distributions for 2021 and 2022 were situated to the right of the 2016-2019 distributions. Among scores below 56, the probability density was the lowest for 2022, whereas among scores above 65, the probability density was the highest also for 2022. This indicates a significant improvement in cities' sustainable development levels in 2022 compared with previous years.

4.2.2. Regional estimation

Next, we examine the distribution dynamics of sustainable development levels in the eastern, central, western, and northeastern regions. Figure 3 shows the dynamic evolution trend of the USDI distribution in the eastern region. In terms of position, from 2016 to 2020, the distribution position fluctuated around the median of 60 without a significant trend developing. However, in 2021 and 2022, the distribution shifted to the right. In terms of shape, compared with the overall national distribution, the curve in the east was broader, with relatively stable changes over time. Regarding extensiveness, although the median of the distribution changed, those in the tail distribution were small each year, displaying a certain degree of tailing, which was especially pronounced in 2016. This indicates considerable differences among cities in the eastern region, and that cities with lower sustainable development levels found it challenging to make progress. Regarding polarization, a small peak in the tail occurred in 2016, indicating that some cities in the eastern region had lower levels of sustainable development. In addition, the distributions in the other years were unimodal, without evidence of multimodality. Furthermore, in 2022, the median increased compared with those in previous years, but the probability density of moderate scores also increased, suggesting a decline in some cities that previously had higher scores.

Figure 4 illustrates the dynamic evolution trend of the USDI distribution in the central region. Regarding the position, the distribution remained almost unchanged from 2016 to 2019. However, from 2020 to 2022, a noticeable rightward shift occurred in the distribution, indicating a general improvement in the sustainable development level of cities in the central region during these three years. In terms of shape, the distributions for each year were nearly identical except for 2020, during which the distribution tended to be more concentrated around the median than in the other years. In terms of extensiveness, no apparent tailing occurred in any year. Regarding polarization, each year displayed a unimodal distribution. This indicates a relatively balanced situation among the scores for cities in the central region, without severe development differentiation.

Figure 5 depicts the dynamic trend of the USDI distribution in the western region. Concerning position, the distribution did not move significantly from 2016 to 2020. In 2021, the distribution noticeable shifted rightward, and its position in 2022 remained consistent with that in 2021, indicating a general improvement in the sustainable development level of cities in the western region during these two years. Regarding shape, the distributions in 2016-2019 and 2022 were broader. The distribution was the narrowest in 2020 and the second narrowest in 2021. This implies that except for 2020, the scores of cities in the western region were relatively dispersed. In terms of extensiveness, left tailing from 2016 to 2019 and right tailing in 2022 occurred, indicating that some cities in the western region had a relatively lower sustainable development level in the earlier period, but improved recently, with some cities significantly enhancing their sustainable development level. Regarding polarization, 2020 exhibited a bimodal distribution with a smaller peak on the left, whereas the other years displayed unimodal distributions.

Figure 6 displays the dynamic trend of the USDI distribution in the northeastern region. In terms of position, the distribution from 2016 to 2020 did not vary much. However, from 2021 to 2022, there was a noticeable rightward shift, indicating an improvement in the sustainable development level of cities in the northeastern region during these two years. Regarding shape, the distribution in 2022 was narrower, with most city scores concentrated around 60. However, the distribution from 2016 to 2021 was broader and more dispersed. Concerning extensiveness, no apparent tailing in the distributions occurred for each year, partly because the distributions for most years were broad and the score distribution itself was relatively dispersed, thus not elongating the tail, even with larger or smaller scores. In terms of polarization, 2022 exhibited a significant bimodal distribution with a secondary peak on the left side of the main peak. A slight bimodal distribution occurred in 2021, and a pronounced trimodal shape with a secondary peak on each side of the main peak occurred in 2018, indicating that the polarization phenomenon in the distribution of the northeastern region was relatively common. This suggests a clear gradient effect on the sustainable development level among cities in this region.

5. Regional differences of the urban sustainable development index

5.1. Dagum Gini coefficient method

This study applied the Dagum Gini coefficient method to analyze regional differences and their sources in the sustainable development of Chinese cities. Based on the traditional Gini coefficient, Dagum (1997) decomposed the overall differences measured by the Gini coefficient into three parts: within-group difference, between-group net difference, and between-group transvariation density. Specifically, we defined the between-region Gini coefficient as follows.

...

Where, j and h represent any two of the four regions; nj and nh denote the number of cities in their respective regions; yji and yhr represent the sustainable development index of the i-th city in region j and the r-th city in region h, respectively, whereas yj and yh denote the mean of the USDIs in the corresponding regions. If the two regions are the same (j = h), the resulting value would be the within-group Gini coefficient Gjj for region j. Furthermore, the Dagum Gini coefficient can be decomposed into three parts.

...

Where, pj = nj /n represents the proportion of the number of cities nj in region j to the total number of cities n, and sh = nh yh / (ny) represents the proportion of the sum of the USDIs of cities in region h to the total sum of sustainable development indices of all cities. The equation ... indicates that the overall Dagum Gini coefficient is the weighted average of between-group (within-group) Gini coefficients Gjh for all pairwise combinations of regions, with respective weights pj sh. Gw representing the total contribution of withinregion differences to the overall difference. Meanwhile, Ggb =Gnb +Gt represents the total contribution of the differences among all regions, and Djh represents the relative influence between regions j and h, calculated as follows.

...

Before computing djh and pjh, the numbering of the two regions must be adjusted such that yj 3 yh. Fj (x) and Fh(x) represent the cumulative distribution functions of the USDI for regions j and h, respectively. djh represents the total influence between regions j and h, calculated as the mathematical expectation of all yji - yhr > 0 for j and h. The expectation is estimated using a weighted average of the sample differences, where the weight for each difference is 1/njnh. pjh represents the first-order moment of transvariation between regions j and h, calculated as the mathematical expectation of all yhr - yji > 0 for j and h. The expectation is estimated using a weighted average of the sample with the same weights as djh. Djh actually represents the proportion of net influence between regions, calculated as djh - pjh over its maximum possible value djh + pjh, and Djh =Dhj with its value constrained within the range of [0, 1]. It equals 0 when yji = yhr and equals 1 when regions j and h do not overlap. Thus, Gnb (Gt ) represents the contribution of the between-region net difference (transvariation density).

Intuitively, if the average score of regions with higher USDIs decreases, while those with lower USDIs increases, the overall Gini coefficient would be reduced by narrowing the between-region differences. However, an overlap between subsamples, (i.e., when some cities in regions with lower-average USDIs have scores higher than those in regions with higher-average USDIs)-which increases the scores of high-scoring cities in regions with a lower average USDI and decreases the scores of low-scoring cities in regions with a higher average USDI-might increase the within-region difference, reduce the between-region net differences, and exacerbate the inequality in the overlapping parts between regions, leading to an overall increase in the Gini coefficient. This portion of the Gini coefficient caused by intergroup overlap is referred to as the between-region transvariation density, which equals zero when regions do not overlap.

5.2. Decomposition of the Dagum Gini coefficient

Figure 7 illustrates the Gini coefficient of sustainability and its decomposition. In summary, the Gini coefficient was relatively stable. It maintained levels between 0.05 to 0.06 until 2021 and decreased to 0.048 5 in 2022. This suggests a relatively stable level of urban sustainable development, with low inequality. After decomposing the Gini coefficient, we obtained within-group differences, betweengroup differences, and the transvariation density. Within-group and between-group differences accounted for a relatively small proportion of the overall Gini coefficient, with minimal differences between them. The transvariation density contributed the most to the Gini coefficient. Small within-group differences imply relatively equal sustainable development levels among cities within each region, indicating a minor degree of imbalanced development within the regions. Similarly, small between-group differences suggest that inequality in sustainable development is less strongly associated with heterogeneous development conditions among regions. The significant transvariation density indicates a substantial overlap in the sustainable development levels of cities in different regions, implying that the proportions of cities with high and low sustainable development levels within each region were relatively close and that these cities, which were quite similar in terms of development levels, contributed significantly to the overall imbalance.

6. Convergence of urban sustainable development indices

6.1. Convergence model specification

The study of convergence originated from neoclassical growth theory. The neoclassical growth model explains the disparity in economic growth levels among countries, suggesting that economies eventually reach a steady state owing to diminishing marginal returns on capital. This phenomenon, called "convergence" in economic growth theory, has found widespread application beyond economic growth research, notably in significant advancements concerning resource allocation efficiency. The USDI falls under the category of efficiency indicators; therefore, it is highly plausible that as cities progress steadily, their level of sustainable development may converge toward a steady state. Hence, employing convergence models to study the long-term variation characteristics of China' s USDI is quite fitting. Common convergence models include σ-convergence and β-convergence. σ-convergence refers to the tendency of the deviation in USDIs among regions to gradually decrease over time. This study uses the coefficient of variation to measure the σ-convergence trend, as shown in the following equation.

... Where, X, represents the USDI of the i-th city within the region; Y denotes the mean score of the cities in the region, and N denotes the number of cities in the region.

β-convergence refers to a scenario where cities with lower scores experience higher growth rates over time to catch up with cities having higher score, thereby gradually reducing the gap between them, eventually reaching the same steady-state level. β-convergence can be further categorized into absolute β-convergence and conditional β-convergence. Absolute β-convergence refers to the trend where the USDIs of cities converge, irrespective of several influential factors affecting sustainable development. The setup of the absolute β-convergence model is defined as follows.

it

Where, Xit represents the USDI of the i-th city in period t; β denotes the convergence coefficient of our concern, where β < 0 indicates a convergence trend in the region' s USDI, while the opposite suggests a divergence trend; α is the intercept term; µi represents the region-specific effect; 7), stands for the time-specific effect; and e, denotes the random disturbance term.

The conditional $ -convergence model, building on the absolute convergence model, incorporates a series of factors that importantly impact sustainable development as control variables. This incorporation aims to provide more convincing and robust results regarding the nature of convergence. The setup for the conditional B-convergence model is as follows.

...

Where, Control, „ represents the j-th control variable. The descriptive statistics of the model variables are presented in Table 2. d_In sust is the dependent variable, representing the growth rate of the USDI; Insust is the independent variable, denoting the logarithm of the USDI; and In gdp and Ingove are the control variables that represent the logarithm of regional gross domestic product (GDP) and local general public budget expenditure (GOVE), respectively. Regional GDP measures the economic development level of a region, whereas the USDI reflects a city's development under sustainable development principles. Hence, the level of economic development, as the foundation of comprehensive development capability, may influence the convergence of the USDI. Local general public budget expenditure often reflects a city's governmental capacity to provide public goods. As urban sustainable development relies on public goods such as infrastructure construction, this control variable may also impact the convergence of sustainable development levels. Given that the construction of the USDI already considered multiple factors affecting urban development, we controlled for only these two critical variables when evaluating convergence.

6.2. Results of o-convergence

Figure 8 illustrates the o convergence, and Table 3 presents the specific values of the coefficient of variation.

Overall, the σ convergence effect was weak. From 2016 to 2020, the coefficient of variation slightly decreased, but it increased in 2021, reaching its peak. In 2022, it significantly dropped, hitting a historical low. Therefore, at the national level, the rebound in 2021 led to a substantial decrease in the σ convergence effect during the sample period. This resulted in the coefficient of variation exhibiting more fluctuating characteristics than a downward trend.

Looking at different regions, the coefficient of variation in the eastern region was the smallest, showing a "decline - increase - decline" trend that displays clear fluctuating characteristics. In terms of both value and evolution, the coefficient of variation in the central region was consistent with that in the overall national level. In the western region, except for 2020, the coefficient of variation was higher than that in the overall level in all years. A substantial increase in 2021 drove the overall coefficient of variation in the western region. The σ convergence situation in the northeast region significantly differed from the other three regions. From 2016 to 2020, the coefficient of variation in the northeastern region was higher than that in the other three regions. In 2021 and 2022, it was only higher than that in the eastern region but lower than that in the central and western regions and the overall level. Both figure and table demonstrate a clear declining trend in the coefficient of variation in the northeast region.

Overall, apart from the northeast region, σ convergence in other areas and the overall national urban sustainable development levels was generally non-existent. Except for the western region, the coefficient of variation in all regions by 2022 remained at its lowest during the sample period, with year-end values lower than the initial values.

6.3. Results of f-convergence

Table 4 presents the absolute β-convergence results for the whole sample. The first three columns show results for models with no fixed effects, with city and year double fixed effects, and with city and year double fixed effects and robust standard errors clustered at the regional level. The results of the absolute convergence test show that the β coefficients are -0.17, -0.68, and -0.68, all significant at the 1% level, indicating a significant absolute β -convergence effect under these three model settings. The last two columns consider robustness. To eliminate the impact of outliers on the regression model, we winsorized and trimmed the independent variable (ln sust). The results presented in Columns (4) and (5) show that after winsorizing and trimming, the β coefficients are -0.73 and -0.65, both significant at the 1% level, demonstrating the robustness of a significant absolute β- convergence effect even after removing the influence of extreme values.

Table 5 illustrates the results of conditional A-convergence for the whole sample. Unlike the absolute convergence test, this model incorporates two control variables: regional gross domestic product and local general public budget expenditure. The first three columns are for models with no fixed effects, with city and year double-fixed effects, and with city and year double-fixed effects and robust standard errors clustered at the regional level. The results of the conditional convergence test reveal β coefficients of -0.17, -0.67, and -0.67, all significant at the 1% level. This indicates a significant conditional β- convergence effect under these three model settings. The last two columns consider robustness. Columns (4) and (5) reveal that after winsorising and trimming, the β coefficients are -0.72 and -0.65, both significant at the 1% level. This demonstrates that even after eliminating the influence of extreme values, a significant conditional β-convergence effect persists.

Table 6 presents the absolute β-convergence test results with double fixed effects for the various regions. The β coefficients for the eastern, central, western, and northeastern regions are -0.64, -0.74, -0.70, and -0.69, respectively, all significant at the 1% level. The findings indicate the existence of significant absolute β-convergence across these regions. Comparing these regional results to that of the overall level (Column (2) of Table 4), the absolute values of the β coefficients for the central, western, and northeastern regions are larger, suggesting a faster convergence rate. Conversely, the absolute value of the β coefficient for the eastern region is smaller, indicating a slower convergence rate than at the national level.

Table 7 presents the conditional β-convergence test results with double fixed effects for the different regions. The β coefficients for the eastern, central, western, and northeastern regions are -0.67, -0.74, -0.70, and -0.69, respectively, all significant at the 1% level. The results indicate a significant conditional β-convergence across these regions. Comparing these regional results to the overall level (Column (2) of Table 5), the absolute values of the β coefficients for the central, western, and northeastern regions are larger, suggesting a faster convergence rate. Conversely, the absolute value of the β coefficient for the eastern region is smaller, indicating a slower convergence rate than at the overall level.

Table 8 presents the β-convergence test grouped by the initial USDI. Columns (1) and (2) respectively represent the absolute β-convergence test for samples with USDI scores below ("low" group) and above ("high" group) the median in 2016. The table shows that the β coefficients for the high-initial-score and low-initial-score groups are -0.72 and -0.74, respectively, significant at the 1% level. Columns (3) and (4) represent the conditional β-convergence test for samples with USDI scores below ("low" group) and above ("high" group) the median in 2016. The table indicates that the β coefficients for the high-initial-score and low-initial-score groups are -0.71 and -0.74, respectively, significant at the 1% level. The results from both absolute and conditional β-convergence tests suggest that the absolute value of the β coefficient is higher for the high-initial-score group than that for the low-initial-score group, indicating a faster convergence rate. This may be because cities with higher initial sustainable development index scores face more challenges and have fewer conditions for subsequent score improvement, resulting in smaller score increments and a greater degree of growth slowdown, thereby leading to a faster convergence rate.

Table 9 displays the β-convergence test grouped by the initial regional gross domestic product. Columns (1) and (2) represent the absolute β-convergence test for samples in 2016 with GDP below ("low" group) and above ("high" group) the median. The table indicates that the β coefficients for the high-initial-GDP and low-initial-GDP groups are -0.71 and -0.65, respectively, significant at the 1% level. Columns (3) and (4) represent the conditional β-convergence test for samples in 2016 with GDP below ("low" group) and above ("high" group) the median. The table shows that the β coefficients for the high-initial- GDP and low-initial-GDP groups are -0.71 and -0.66, respectively, significant at the 1% level. The results from both the absolute β-convergence and conditional β-convergence tests indicate that the absolute value of the β coefficient is larger for the low-initial-GDP group than that for the high-initial-GDP group, suggesting a faster convergence rate. This could be attributed to cities with lower initial outputs generally having fewer resources and smaller populations, thus having less momentum for sustainable development, resulting in a faster convergence speed.

Table 10 presents the β-convergence test grouped by the initial local general public budget expenditure (GOVE). Columns (1) and (2) respectively represent the absolute β-convergence test for samples in 2016 with GOVE below ("low" group) and above ("high" group) the median. The table shows that the β coefficients for the high-initialscore and low-initial-score groups are -0.68 and -0.69, respectively, significant at the 1% level. Columns (3) and (4) represent the conditional β-convergence test for samples in 2016 with GOVE below ("low" group) and above ("high" group) the median. The table indicates that the β coefficients for the high-initial-score and low-initialscore groups are -0.68 and -0.69, respectively, significant at the 1% level. The results from both the absolute β-convergence test and the conditional β-convergence test suggest little difference in the convergence speed between the high-initial-score and low-initial-score groups, indicating minimal impact of the GOVE level on the convergence of urban sustainable development.

7. Conclusion

In summary, to support the implementation of the 2030 Agenda in China, this study assesses the practical effects of local urban development processes on SDG 11, that is, "Make cities and human settlements inclusive, safe, resilient, and sustainable". Specifically, this study captures the developmental essence, scientifically measures the level of urban sustainable development, delineates its distribution dynamics and regional disparities in both time and spatial dimensions, and further diagnoses obstacles to the urban sustainable development process. A comprehensive understanding of these aspects contributes to an objective recognition of the true situation and distribution characteristics of China's high-quality economic development. It also provides insightful strategies to promote comprehensive improvement in high-quality economic development across different regions, with significant theoretical and practical significance.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

The paper is supported by the National Key Research and Development Program of China under the theme "Research on urban sustainable development evaluation data fusion management technology" [Grant No. 2022YFC3802903].

Sidebar

Received 21 April 2024; Accepted 12 August 2024

Footnote

* Corresponding author. E-mail address: [email protected] (M. Li)

References

References

Bartniczak, B., Raszkowski, A., 2022. Implementation of the sustainable cities and communities sustainable development goal (SDG) in the European union. Sustainability 14, 16808.

Braulio-Gonzalo, M., Bovea, M.D., Rud, M.J., 2015. Sustainability on the urban scale: proposal of a structure of indicators for the Spanish context. Environmental Impact Assessment Review 53, 16-30.

Castelnovo, W., Misuraca, G., Savoldelli, A., 2015. Citizen's engagement and value coproduction in smart and sustainable cities. International Conference on Public Policy, 1-16.

Coscieme, L., Mortensen, L.F., Anderson, S., Ward, J., Donohue, I., Sutton, P.C., 2020. Going beyond Gross Domestic Product as an indicator to bring coherence to the Sustainable Development Goals. Journal of Cleaner Production 248, 119232.

Dagum, C., 1997. Decomposition and interpretation of Gini and the generalized entropy inequality measures. Statistica 57(3), 295-308.

Giles-Corti, B., Lowe, M., Arundel, J., 2020. Achieving the SDGs: Evaluating indicators to be used to benchmark and monitor progress towards creating healthy and sustainable cities. Health Policy 124, 581-590.

Klarin, T., 2018. The concept of sustainable development: from its beginning to the contemporary issues. Zagreb International Review of Economics and Business 21, 67- 94.

Koch, F., Krellenberg, K., 2018. How to contextualize SDG 11? Looking at indicators for sustainable urban development in Germany. ISPRS International Journal of GeoInformation 7, 464.

Ma, Y.J., Ai, X.P., 2019. Evaluation of sustainable urbanization development in Jilin Province based on the 2030 sustainable development goals. Scientia Geographica Sinica 39(3), 487-495 (in Chinese).

Monteiro, N.B.R., da Silva, E.A., Moita Neto, J.M., 2019. Sustainable development goals in mining. Journal of Cleaner Production 228, 509-520.

Niu, W.Y., 2012. Theory and practice of China's sustainable development. Bulletin of Chinese Academy of Sciences 27(3), 280-289 (in Chinese).

Pittman, S.J., Rodwell, L. D., Shellock, R.J., Williams, M., Attrill, M. J., Bedford, J., Curry, K., Fletcher, S., Gall, S.C., Lowther, J., McQuatters-Gollop, A., Moseley, K. L., Rees, S.E., 2019. Marine parks for coastal cities: a concept for enhanced community well-being, prosperity and sustainable city living. Marine Policy 103, 160-171.

Quah, D., 1993. Galton's fallacy and tests of the convergence hypothesis. The Scandinavian Journal of Economics 95, 427.

Sala, S., Ciuffo, B., Nijkamp, P., 2015. A systemic framework for sustainability assessment. Ecological Economics 119, 314-325.

Shao, C., Chen, S., Gao J., He, Y., Zhou, H., 2021. Design of China's sustainable development evaluation index system based on the SDGs. China Population, Resources and Environment 31(4), 1-12 (in Chinese).

Steiniger, S., Wagemann, E., de la Barrera, F., Molinos-Senante, M., Villegas, R., de la Fuente, H., Vives, A., Arce, G., Herrera, J.C., Carrasco, J.A., Pastén, P.A., Muñoz, J.C., Barton, J.R., 2020. Localising urban sustainability indicators: The CEDEUS indicator set, and lessons from an expert-driven process. Cities 101, 102683.

Sun, X., Liu, X., Li, F., Tao, Y., 2016. Comprehensive evaluation of sustainable development for different scale cities in China. Acta Ecologica Sinica 36(17), 5590-5600 (in Chinese).

Swain, B. B., Karimu, A., 2020. Renewable electricity and sustainable development goals in the EU. World Development 125, 104693.

UN-Habitat, 2018. Tracking Progress towards Inclusive, Safe, Resilient and Sustainable Cities and Human Settlements: SDG 11 Synthesis Report - High Level Political Forum 2018. Nairobi, Kenya: UN-Habitat. https://doi.org/10.18356/36ff830e-en.

United Nations, 2015. Transforming our world: the 2030 Agenda for Sustainable Development. https://sdgs.un.org/2030agenda.

United Nations, 2016. Global indicator framework for the Sustainable Development Goals and targets of the 2030 Agenda for Sustainable Development. https://unstats.un.org/sdgs/indicators/indicators-list/.

Wang, P.L., Gao, F., Huang, C.L., Song, X.Y., Wang, B., Wei, Y.Q., Niu, Y.B., 2018. Progresson sustainable city assessment index system for SDGs. Remote Sensing Technology and Application 33(5), 784-792 (in Chinese).

Yang, D.F., Yin, C.Z., Long, Y., 2011. Quantitative assessment of urban sustainability: methodological and practical review. Urban Planning Forum (3), 58-65.

Yang, Z., Yang, Hang, Sun, D., 2021. Simulation of urban sustainability based on system dynamics: a case study of Beijing-Tianjin-Hebei region. Scientia Geographica Sinica 41(9), 1515-1524 (in Chinese).

Yao, Q., L, K., 1, Q., 2019. Research on sustainable development path for resource-type small cities: a case study of Daliuta-town-level Cities in Shaanxi Province. Journal of Ningxia University (Natural Science Edition) 40(3), 286-290 (in Chinese).

Yigitcanlar, T., Dur, F., Dizdaroglu, D., 2015. Towards prosperous sustainable cities: a multiscalar urban sustainability assessment approach. Habitat International 45, 36-46.

Zhu, J., Sun, X.Z., He, Z., 2018. Research on China's sustainable development evaluation indicators in the framework of SDGs. China Population, Resources and Environment 28(12), 9-18 (in Chinese).