Research on the Identification and Spatiotemporal Evolution of China’s Urban Life Cycle: From the Perspective of Organic Entities

Abstract

Based on the characteristics of cities as organic entities, this paper constructs a five-dimensional evaluation framework encompassing economy, industry, society, population, and space. A three-stage process of “fuzzy comprehensive evaluation—bi-level K-means clustering—state stability correction” is adopted to identify the development stages and spatiotemporal evolution of 286 Chinese cities from 2008 to 2023. The study finds that China’s urban development has shifted from “universal growth” to “divergent evolution,” exhibiting multiple characteristics such as the decline in the initial-stage cities and differentiation in the growth stage. Significant regional spatial differentiation is observed, with notable development gaps among the eastern, central, western, and northeastern regions, as well as between the northern and southern regions. Furthermore, most urban agglomerations exhibit a “mature center–lagging periphery” structure.

1. Introduction

As the fundamental unit for promoting coordinated regional development in China, the accurate identification of urban development stages is essential for promoting regionally coordinated development through differentiated policy interventions. The 2025 Central Urban Work Conference of the Communist Party of China (CPC) pointed out that “China’s urbanization is shifting from a period of rapid growth to one of stable development, and urban development is transitioning from large-scale incremental expansion to a stage centered on stock revitalization, quality improvement, and efficiency enhancement” [1]—herein referred to as the “two transformations.” At the same time, owing to long-standing disparities in historical foundations, locational conditions, and policy orientations, significant development gaps persist both between and within different regions of China [2,3]. Confronted with this new stage of urban development characterized by the “two transformations” and the objective reality of unbalanced regional development, a critical question for scientifically promoting coordinated regional development during the 15th Five-Year Plan period is how to accurately grasp the development stages of cities in different regions, thereby providing an objective basis for “adapting measures to local conditions and offering classified guidance”.

Cities, as organic entities, do not develop along a linear growth trajectory; rather, their development constitutes a natural evolutionary process characterized by inherent life cycle dynamics—including emergence, growth, maturation, adjustment, and renewal—typically encompassing five stages: initial, growth, maturity, decline, and transition [4,5]. The 2015 Central Urban Work Conference already emphasized the need to “respect the laws of urban development,” and the 2025 Conference further stressed that cities must be “understood, planned, and built as organic entities” [1]. Just as organic entities exhibit distinct growth stages with corresponding developmental mechanisms, urban development likewise displays differentiated life cycle characteristics. In particular, China’s regions differ considerably in natural resource endowments and socio-economic development; consequently, cities in different regions may occupy different life cycle stages, and the growth mechanisms suited to each cannot be applied in a “one-size-fits-all” manner. Therefore, grounded in the objective reality of unbalanced regional development, the essential prerequisite for promoting coordinated regional development through targeted and classified policies is to focus on the organic-entity characteristics of cities, and thereby objectively and scientifically grasp the life cycle evolution stages of cities across different regions. As Kurokawa (1991) argues, urban development embodies the vitalistic principles of metabolism and cyclical renewal, through which cities continuously achieve iterative regeneration [6].

Regrettably, current academic research on coordinated regional development predominantly focuses on narrowing objective regional development gaps, while studies on urban development stages have largely concentrated on evaluating the life cycles of resource-based cities [7] or identifying specific city types such as shrinking cities [8]. To date, no scholars have systematically assessed and identified the development stages of cities across different regions of China. In reality, cities of all sizes in all regions follow the objective evolutionary laws of urban development. Against this backdrop, the present paper, grounded in the characteristics of cities as organic entities, constructs an urban development stage evaluation framework to scientifically identify the development stages and spatiotemporal evolution characteristics of 286 prefecture-level and above cities in China from 2008 to 2023, thereby providing scientific evidence and decision-making support for improving the institutional mechanisms of coordinated regional development and promoting high-quality urban development.

The marginal contributions of this paper are threefold. First, departing from prior studies that treat urban development as a linear growth process, this paper conceptualizes urban development stage evolution through the lens of cities as organic entities, dividing the process into five stages—initial, growth, maturity, decline, and transition—thereby fully accommodating the nonlinear evolutionary characteristics of urban organic entities. Second, unlike previous research that largely equates development quality with development stage identification and employs methods such as quantile-based slicing, natural breakpoint classification, or cluster analysis [9,10] to directly categorize city types, this paper first conducts a fuzzy comprehensive evaluation of urban development stages, then accurately identifies nonlinear stage transitions through bi-level K-means clustering combined with a state stability correction principle. This approach offers a new framework for understanding the transition of Chinese cities from incremental expansion to stock-based development. Third, whereas prior research has predominantly focused on identifying the life cycle stages of specific city types, this paper systematically examines the evolutionary characteristics of all 286 prefecture-level and above cities in China from 2008 to 2023, offering a new perspective for promoting coordinated regional development through “differentiated measures tailored to local conditions” during China’s 15th Five-Year Plan period (2026–2030). This not only contributes a fresh analytical lens for Chinese urban studies but also provides a methodological reference for the vast majority of developing—and even developed—countries seeking to scientifically identify urban development stages and adjust their urban development strategies accordingly.

The remainder of this paper is structured as follows: Section 2 reviews the literature, systematically examining research on urban development stage identification and coordinated regional development. Section 3 elaborates the theoretical connotations of the urban life cycle and constructs a stage identification indicator system covering five dimensions: economic vitality, industrial momentum, social services, population development, and spatial carrying capacity. Section 4 introduces the stage identification methodology based on Fuzzy Comprehensive Evaluation, bi-level K-means clustering, and state-stability correction. Section 5 analyzes the spatiotemporal evolution characteristics and regional differences of urban development stages across 286 prefecture-level and above cities in China from 2008 to 2023. Section 6 discusses the findings in dialogue with the existing literature, and Section 7 summarizes the research conclusions and proposes differentiated policy recommendations.

2. Literature Review

At present, a considerable body of scholarship addresses the assessment of urban development stages. On the one hand, some studies equate the stratification of urban development quality with development stages, defining stages according to development quality gaps. On the other hand, a separate stream of research focuses on evaluating the life cycle evolution of resource-based cities and the systematic identification of special types of cities such as shrinking cities. However, these two streams of research share a common limitation: neither has treated the “development stage” as a theoretical category requiring independent conceptual construction and systematic identification. The former substitutes cross-sectional ranking for stage diagnosis, while the latter replaces a universal stage-division logic with diagnostic frameworks tailored to specific city types.

In research on urban development quality stratification, scholars employ development quality as the basis for stage division and classify cities according to various classification rules. Some scholars classify cities by the mean or quantile of measured values. For instance, Zhu (2021) constructed an index system based on the five development concepts, applied the entropy-weighted TOPSIS model to evaluate the development quality of eight cities in the Wanjiang City Belt, and identified three city types—leading, catching-up, and lagging—according to evaluation score quantiles [11]. Yang et al. (2021) evaluated 41 cities in the Yangtze River Delta across five dimensions—including economic vitality, innovation efficiency, and green development—using factor analysis and the entropy method, and divided them into high, medium, and low levels using the mean ± 0.5 standard deviations as boundaries [12]. Li (2023) assessed Yangtze River Delta cities using the entropy-weight TOPSIS method with five-dimensional indicators and classified them into three echelons based on mean comprehensive scores [13]. Other scholars employ methods such as natural breakpoint classification. Liu (2021) was among the first to integrate remote sensing data with Artificial Neural Network (ANN) algorithms, measuring China’s top ten urban agglomerations across four dimensions—economy, infrastructure, resource environment, and social development—and classifying them into leading, transformation, and lagging groups using the natural breakpoint method [14]. Qin & Qin (2025) constructed a 37-indicator system based on big data, evaluated 151 prefecture-level cities, and categorized them into five types—initial development, catching-up, developing, booming, and leading—using the natural breakpoint method [15]. A related study using the same big data evaluation framework further found that between 2017 and 2021, China’s urban development quality improved consistently and regional disparities gradually narrowed, with innovation emerging as the primary driver for enhancing urban development quality. Additionally, some scholars conduct cluster analysis following quality evaluation. Fan & Wang (2022) systematically constructed a statistical measurement system for high-quality urban development covering economy, society, ecology, and people’s livelihood, generated comparable comprehensive scores using the entropy-weight TOPSIS method, and divided Yangtze River Delta cities into high, medium, and low development echelons through cluster analysis [16]. Wen et al. (2022) also measured the development quality of 13 cities in the Beijing–Tianjin–Hebei region using four-dimensional indicators, but innovatively introduced SOM neural network clustering, identifying four levels: the capital (Beijing), the municipality directly under the central government (Tianjin), core industrial cities (Tangshan), and ordinary prefecture-level cities [17]. From a megacity perspective, recent research employing system dynamics simulation of China’s seven megacities revealed an upward trend in both high-quality development index and coupling coordination degree among subsystems, with notable stratification across three echelons—Beijing, Shanghai and Shenzhen forming the top tier; Guangzhou and Chongqing the second; and Chengdu and Tianjin the third—highlighting that even among the highest-development-level cities, distinct stage differentiation exists [18]. In summary, the common limitation of the above studies is that they directly equate development quality scores with development stages, essentially performing a cross-sectional ranking of urban development levels rather than a stage-based identification of the urban maturation process. This conflation of level with stage fails to answer the core questions of which developmental stage a city currently occupies and how far it remains from a stage transition.

In studies on the identification of special city types, the life cycle evaluation of resource-based cities and the identification of shrinking cities are the most extensively developed. Research on resource-based cities takes life cycle evolution as its core logic. In 2013, the State Council’s National Plan for Sustainable Development of Resource-based Cities (2013–2020) established, for the first time at the national level, a four-stage classification standard—growth, maturity, decline, and regeneration—based on two dimensions: resource security capacity and sustainable development capacity. A total of 262 Chinese cities were evaluated under this framework [19], providing an important reference for subsequent studies. Thereafter, Yu et al. (2018) applied this classification using two indicators: mining function intensity and resource contribution degree [20]. Lu et al. (2020) further identified two key thresholds—1.9% and 31.0%—for the employment share of the mining sector, quantitatively characterizing the transition mechanism of development stages [7]. Meanwhile, some studies have sought to further enrich the connotation of development stages. Lyu & Lu (2016) divided resource-based cities into four development states based on green growth capacity [21]. Zhao (2020) proposed a five-stage green evolution pathway—non-transformation, primary transformation, intermediate transformation, advanced transformation, and successful transformation—from the perspective of green transformation [22]. On this basis, recent research has deepened the understanding of industrial resilience restructuring in resource-based regions from the perspective of “innovation ecosystem organicity”. Taking the “π-shaped Curve Area” in the Yellow River Basin as a case, a three-dimensional “quality–structure–function” analytical framework was constructed, and fuzzy-set qualitative comparative analysis (fsQCA) revealed multiple equivalent pathways from resource dependence to resilience restructuring. The study found that the organicity of innovation ecosystems in this region exhibits a clear gradient pattern, with provincial capitals as core nodes, growing resource-based cities as highlands, and declining resource-based cities as depressed areas, and industrial resilience evolves along three trajectories: “fluctuating upward”, “fluctuating reversal”, and “fluctuating downward” [23]. This provides a new analytical tool for understanding the transition of resource-based cities from the “resource curse” to “resilience restructuring”. Complementing this perspective, an examination of the social–ecological systems adaptive cycling mechanism in Shanxi Province’s resource-based cities showed that industrial transformation policies significantly enhance system potential and resilience, while investment in social and human capital, technological innovation, and social infrastructure promotes advancement in the adaptive cycle—leading to a classification of resource-based cities into development-stopping, normal development, and leapfrog development categories [24]. In addition, the role of industrial collaborative agglomeration in accelerating green and low-carbon transformation of resource-based cities has been empirically established, with regional heterogeneity in these effects underscoring the importance of place-based transition strategies [25].

Research on shrinking cities is oriented toward problem diagnosis and causal attribution. Since most scholars consider urban shrinkage to be a compound phenomenon resulting from multi-dimensional decline in population, economy, and society, they have attempted scientific identification from multiple perspectives. Regarding the degree of shrinkage, Kabisch et al. (2012) identified three evolutionary pathways—continuous decline, stability, and recovery—through population trajectory clustering [26]. Sun et al. (2021) classified cities in China’s three northeastern provinces into five levels: pre-shrinkage, early shrinkage, mid-shrinkage, late shrinkage, and decline stage [8]. In terms of shrinkage types, Wiechmann & Pallagst (2012) proposed two spatial modes—perforated shrinkage in Europe and doughnut-type shrinkage in North America [27]. Zhang et al. (2020) identified four shrinkage modes: comprehensive, population, economic, and social [28]. Zhang (2019) further classified shrinking cities by causation into four types: those induced by harsh geographical environments, resource exhaustion, lagging industrial transformation, and locational marginalization [29]. More recently, the multidimensional nature of shrinkage has been further revealed through integrated morphological and functional approaches. Using a multi-model framework that integrates 12 morphological indicators, a study of 277 Chinese cities found that functional shrinkage is widespread and exhibits a “low in the east, high in the west” spatial pattern, with governance implications for morphological optimization [30]. From the perspective of “natural cities”, another study focusing on Northeast China found that more than 60% of cities in the region are shrinking, highly concentrated in “land marginal” areas and “peripheral urban circles”, arguing that shrinkage governance should shift from “resistance” to “adaptation” [31]. Empirical evidence also confirms the negative impact of urban shrinkage on innovation and entrepreneurial activities, with the effect being particularly pronounced in old industrial cities and economically weak cities with low levels of science and technology investment [32]. Moreover, the relationship between urban shrinkage and urban resilience in Northeast China reveals that shrinking cities tend to be less resilient than growing cities, yet urban shrinkage may paradoxically offer opportunities to build resilience through influencing fiscal self-sufficiency, human capital, and environmental quality—suggesting a nuanced perspective that decline and resilience are not simply opposing forces [33]. Advanced methodological innovations using time-series machine learning—specifically Dynamic Time Warping clustering—have classified 285 Chinese cities into four distinct groups based on shrinkage/growth trajectories, identifying group-specific drivers such as resource depletion, manufacturing recession, industrial structure shifts, and population migration [34]. Taken together, although the above studies provide important insights into the development trajectories of specific city types, their identification frameworks are tightly anchored to the proprietary attributes of specific city groups—resource-based cities relying on resource dependence as the core criterion, and shrinking cities requiring negative population growth as a precondition—and therefore lack the universality and commensurability needed for application to the national urban system as a whole.

Overall, existing research has made notable progress in the identification of urban development stages. However, because the two aforementioned research streams are respectively confined to cross-sectional ranking and the diagnosis of specific groups, three systematic gaps remain. First, some studies conflate development stage assessment with development quality stratification, equating composite score rankings with stage evolution while neglecting inter-city differences in developmental genes. As a result, certain cities are frequently misclassified as being in a primary or lagging stage merely because of their low absolute scale. In fact, development stage assessment should focus on characterizing a city’s developmental state and its transition thresholds. Second, although some studies have attempted to identify the life cycles of resource-based or shrinking cities, they rely excessively on single-dimensional indicators such as resource dependence or negative population growth. These frameworks essentially offer a fine-grained depiction of specific groups’ development rather than a universally applicable and generalizable identification framework. Third, when stratifying key identification indicators or assessment results, most studies linearly partition threshold intervals based on means or quantiles. This approach neither scientifically captures the structural breakthroughs of key indicators nor adequately accounts for the nonlinear manifestations of development stages, thereby making it difficult to precisely identify stage transition nodes along urban development pathways. In response to these methodological limitations, unsupervised machine learning methods have recently been applied to urban development stage identification, offering a new technical path to break through the “linear partitioning” dilemma. Taking England and Wales as the study area, a study used K-means clustering to analyze multi-scale real estate transaction data and census data, revealing how neighborhood dynamics affect spatial inequality and environmental sustainability across the urban system [35]. This methodological turn is part of a broader trend: machine learning applications in urban morphology have been categorized into two streams—the first revealing nonlinear relationships between urban form and ecosystem services, and the second employing deep learning-based representation learning methods for urban typology clustering, which significantly enhance the accuracy and reliability of urban classification [36]. Context-specific urban optimizations through data-driven classification have also been proposed as a computational framework to support targeted interventions for climate neutrality and livability, reinforcing the notion that algorithmic clustering provides a structured approach to understanding diverse urban contexts [37]. These methodological innovations demonstrate the significant potential of clustering algorithms in exploring the spatiotemporal dynamics of urban systems, although their general applicability still needs to be tested in developing country contexts and urban–rural difference scenarios.

These three gaps indicate that existing research lacks a systematic diagnostic framework for urban development stages—one that is grounded in the characteristics of cities as organic entities, possesses nationwide applicability, and is capable of capturing nonlinear stage transitions. To address these gaps, this paper constructs a comprehensive evaluation system covering five dimensions—economic vitality, industrial momentum, social services, population characteristics, and spatial carrying capacity—based on the organic-entity characteristics of cities. Through a three-stage progressive evaluation process of “fuzzy comprehensive evaluation → bi-level K-means clustering → state stability correction,” this paper identifies the development stages and spatiotemporal evolution characteristics of 286 prefecture-level and above cities in China from 2008 to 2023.

3. Theoretical Connotation and Construction of Evaluation System

3.1. Theoretical Connotation

Urban life cycle theory posits that cities, like organic entities, undergo an evolutionary process from birth through growth and maturity to possible decline or renewal. Within the classical theoretical tradition, Mumford (1961), through a long-term examination of Western urban development, delineated six evolutionary stages and forms—primitive city, polis, metropolis, megalopolis, tyrannical city, and ruined city—highlighting the cyclical and stage-based nature of urban development [4]. Hall (1971), approaching from the dimensions of industrial migration and population movement, proposed four stages of urban evolution: centralized urbanization, suburbanization, counter-urbanization, and re-urbanization [5]. Northam (1979), drawing on the century-long urbanization trajectories of developed Western countries such as the United Kingdom and the United States, advanced the classic “S-shaped” curve theory of urbanization, dividing cities into three stages: initial development (urbanization rate below 30%), rapid development (urbanization rate of 30–70%), and stable development (urbanization rate above 70%) [38]. The renowned Japanese architect Kisho Kurokawa likewise emphasized that urban development embodies the vitalistic principles of metabolism and cyclical renewal, through which cities continuously achieve iterative regeneration [6]. General Secretary Xi Jinping has also observed in this regard that “some towns will grow rapidly, some will grow slowly, and some may not grow or even shrink—this is natural selection” [39]. At the same time, urban development cannot be fully equated with organic entities in the ordinary sense. As a product of human social practice, the urban development process not only follows objective laws but is also profoundly shaped by human subjective agency and creativity. Human society is capable of making forward-looking assessments of urban development trajectories and, through institutional design, planning regulation, and policy intervention, guiding, correcting, or even reversing the urban life cycle—thereby mitigating the risk of decline to a certain extent and fostering the long-term maintenance of a relatively mature developmental state.

In light of existing research and practical applications, urban life cycle theory has been extensively applied in the evolutionary analysis of specific city types. Resource-based cities, which exhibit a clear “rise–boom–decline–transformation” trajectory shaped by resource exploitation cycles, represent one of the most typical applications of life cycle theory. Similarly, shrinking cities—characterized by population loss, spatial constraints, and functional degradation—have been widely incorporated into diagnostic frameworks for cities in decline or transformation. Urban development, accordingly, is not an instantaneous achievement but a staged, nonlinear evolutionary process that can generally be divided into five stages: initial stage, growth stage, maturity stage, decline stage, and Transition stage (see Figure 1). It is important to emphasize that cities in the growth stage do not necessarily progress to maturity, and cities in the Transition stage do not inevitably move toward decline. When the urban development path is misaligned with the stage, a city in the growth stage may bypass maturity and enter directly into transformation or even decline; conversely, a city undergoing transformation may, if it achieves high-quality restructuring, re-enter the growth or even maturity trajectory.

It is essential to clarify that the five stages delineated in this paper constitute, in essence, a nonlinear state classification system based on the combined characteristics of scale, structure, and speed, rather than a strictly unidirectional sequential typology. The aim is to capture the variations in development dynamics among different cities at specific temporal cross-sections, without presupposing that cities inevitably evolve along a single “initial-growth–maturity–decline” trajectory. In fact, the empirical results presented below demonstrate that some cities move directly from the initial stage to the decline stage, while others re-enter the growth or maturity stage after undergoing transformation—further corroborating that the paradigm adopted here is not a linear sequential one.

Furthermore, it must be stressed that the urban development stage is merely an objective description of a city’s position within the life cycle; it does not directly reflect its development quality, and the two must not be conflated. Just as the stages of childhood, youth, prime, and old age in the human life cycle cannot be ranked by quality simply according to their sequence—and different stages are not directly comparable—a well-rounded youth cohort with balanced moral, intellectual, physical, aesthetic, and labor development cannot be simplistically judged as inferior to a prime-age cohort with imbalanced physical and mental conditions. By the same logic, cities at different development stages should not be subjected to direct horizontal comparison; rather, development quality comparisons should be conducted within the same stage to ensure objectivity and scientific validity.

As an endogenous evolutionary process of organic entities, the stage transition of urban development is not a linear accumulation of a single indicator, but a systemic evolution across three dimensions: scale, structure, and speed. The dominant dimensions of change vary significantly across different life cycle stages. Cities in the initial stage are characterized primarily by scale expansion: population and total economic output agglomerate rapidly, but the structural configuration has not yet taken shape, and growth rates fluctuate sharply. Cities in the growth stage are driven by the dual engines of continuing scale expansion and accelerating structural transformation, with industrialization and urbanization advancing in tandem and growth rates remaining high. Once a city enters the maturity stage, its scale stabilizes, structural optimization (e.g., service-sector dominance, innovation-driven development) becomes the core marker, and the growth rate naturally moderates to a medium-low level. When a city enters the decline or Transition stage, however, it is often accompanied by scale contraction or stagnation, pronounced structural imbalance, and persistently low or highly volatile growth. Thus, relying on a single type of evolutionary indicator—such as scale or speed alone—risks misclassifying cities on different evolutionary trajectories as being in the same stage. Only by simultaneously observing the combined configuration of scale, structure, and speed can the systemic transformation from quantitative change to qualitative change in urban development be scientifically captured.

Drawing on China’s urban development practice since the founding of the People’s Republic and relevant international experience, and proceeding from the “scale–structure–speed” evolutionary characteristics, the five stages can be characterized as follows.

Initial stage: The initial accumulation phase of the urban life cycle. Economic, demographic, and spatial scales are all relatively small; the industrial base and public service system have not yet fully formed; the structure remains unstable; and the growth rate is generally low with considerable volatility. Its core feature is that the urban functional system is still in the process of formation and has not yet entered a trajectory of sustained expansion. The urbanization rate is typically below 30%, spatial carrying capacities are characterized by point-like growth with vague functional zoning, and the central developmental question is whether production factors can rapidly agglomerate to generate initial momentum.

Growth stage: The phase in which urban scale expansion and momentum release are most pronounced. The city has broken through the basic constraints of the initial stage; economic, demographic, and spatial scales expand continuously; the industrial system gradually takes shape; the urbanization rate steadily improves (entering the 30–70% rapid-growth range); and the speed of development is the most prominent among the five stages. It can be summarized as an expansion phase characterized by “moderate scale, fastest speed, and continuously optimizing structure.” This is a critical period for cities transitioning from quantitative accumulation to structural optimization.

Maturity stage: The phase in which urban development enters a period of relative stability and structural optimization. Cities in this stage typically possess the largest economic, demographic, and spatial scales among the five stages; public services, industrial systems, and urban functions are relatively well-developed; and the level of structural sophistication is the highest. The growth rate shifts from high-speed expansion to stable medium-low growth. The industrial structure has completed the “tertiary–secondary–primary” advanced transformation, and spatial development has shifted from incremental expansion to stock revitalization and smart growth, emphasizing “urban refinement” rather than “urban enlargement.” Its typical characteristics are “largest scale, most advanced structure, and decelerating speed”.

Decline stage: The phase in which urban development momentum weakens markedly or even contracts. Cities in this stage usually possess a certain scale base, often larger than that of the initial stage, but due to factors such as insufficient industrial momentum, population outflow, resource-dependent path lock-in, or waning growth drivers, they experience stagnation or even negative growth. The core characteristic is not a small scale, but growth stagnation or negative growth on an already established scale base. The economy languishes in prolonged downturn, traditional industries shrink while emerging industries remain absent, fiscal pressures lead to declining public service quality, the population undergoes sustained net outflow and accelerating aging, and spatial carrying capacities exhibit hollowing-out and functional degradation—the city as a whole falls into a structural impasse.

Transition stage: The phase in which a city, once constrained on its original development path, enters a period of adjustment, restructuring, and reorganization. In general, cities in the Transition stage maintain positive growth, but at a speed lower than that of the growth stage; their scale is smaller than that of the maturity stage and is comparable to or slightly larger than that of the growth stage. Its essence lies in the transformation of the developmental dynamic mechanism. There are two typical pathways into the Transition stage. First, a city that has experienced decline or sluggish growth may, through industrial restructuring, functional renewal, and the re-agglomeration of production factors, regenerate positive growth momentum—manifesting as an upward restorative restructuring from decline toward renewed growth. Second, a city that has not yet reached the scale and structural levels of maturity but whose original high-speed growth model—relying on input expansion, policy stimulus, or a single-industry base—has become unsustainable may prematurely enter a state of low-speed adjustment and growth driver transition from the initial or growth stage. The future trajectory of cities in this stage depends on the effectiveness of the transition from old to new growth drivers and structural reshaping: successful transformation leads back to the growth or even maturity stage, while failed transformation may result in sliding into decline.

3.2. Construction of the Evaluation System

Based on the theoretical connotation of urban development stage evolution from the perspective of organic entities, this paper proceeds from five core dimensions—economic vitality, industrial momentum, social service capacity, population development, and spatial carrying capacity—and follows the principles of scientific rigor, objectivity, and data availability to construct an indicator system for evaluating the evolution of urban development stages, as presented in

Table 1. Indicator Framework for Identifying the Evolution Stages of Urban Development.

Target Layer	Primary Indicator	Secondary Indicator	Unit	Core Connotation	Impact Direction
Urban Development Stage	Spatial Carrying Capacity	Population Density	Person/square kilometer	Measure the intensity of land use and urban carrying pressure	+
		Proportion of Built-up Area	%	Reflect the speed of urban expansion and intensity of spatial development	+
	Population Development	Urbanization Rate	%	Reflect the degree of population agglomeration to cities, an important symbol of modernization	+
		Permanent Resident Population	10,000 persons	Characterize the urban scale and population carrying capacity	+
		Aging Degree	%	Reflect the change of population structure, posing challenges to labor supply and social security	−
	Social Service Capacity	Number of Hospital Beds	Unit	Measure the supply capacity of medical resources	+
		Number of Urban Centers	Unit	Reflect the improvement and balance of urban public service functions	+
		Highway Passenger Volume	10,000 person-times	Reflect the level of transportation infrastructure and travel convenience of residents	+
	Industrial Momentum	Proportion of Secondary Industry	%	Measure the level of industrialization and the status of manufacturing industry	+
		Proportion of Tertiary Industry	%	Reflect the development level of the service industry and the direction of transformation and upgrading	+
		Number of Patent Authorizations	Unit	Characterize the innovation capacity and technological activity of the industry	+
		Number of Industrial Enterprises above Designated Size	Unit	Reflect the industrial scale and agglomeration degree of enterprises	+
		Total Profits of Industrial Enterprises above Designated Size	100 million yuan	Measure industrial benefits and market competitiveness	+
	Economic Vitality	GDP Scale	100 million yuan	Measure the total urban economic output, the core basic indicator of development stage	+
		GDP Growth Rate	%	Reflect the speed and potential of economic growth, reveal the development momentum	+
		Per Capita GDP	10,000 yuan	Reflect the development quality and per capita wealth level of residents	+
		Total Imports and Exports	100 million yuan	Measure the level of opening-up and international competitiveness	+
		Total Retail Sales of Consumer Goods	100 million yuan	Reflect the scale of the consumer market and the pulling power of domestic demand	+
		Local General Public Budget Expenditure	100 million yuan	Demonstrate the financial strength and public service guarantee capacity	+

.

Economic vitality, as the foundational dimension for identifying development stages, comprehensively reflects the city’s economic scale, growth momentum, development quality, openness, domestic demand support, and fiscal capacity through indicators such as gross domestic product (GDP), GDP growth rate, per capita GDP, total retail sales of consumer goods, total foreign trade volume, and local general public budget expenditure. Industrial momentum, as the key cross-stage driving force, captures the evolution of industrial structure through the share of value added by secondary and tertiary industries, and incorporates the number of patent grants, the number of industrial enterprises above designated size, and their total profits to gauge technological innovation activity, industrial base scale, and market competitiveness—thereby identifying transformation trends or decline risks. Social service capacity reflects the level of public service provision and livelihood security; the number of hospital beds, the number of urban centers, and highway passenger volume are selected to characterize medical resource allocation, public service accessibility, and transportation infrastructure carrying capacity, respectively. Population development, as the principal agent variable, systematically reveals trends in population agglomeration or loss, the urbanization stage, and the evolution of the population age structure through permanent resident population size, urbanization rate, and the degree of aging; notably, population aging in the maturity stage is largely a natural demographic outcome, whereas in the decline stage it is frequently compounded by the outflow of young and middle-aged workers. Spatial carrying capacity, as the spatial manifestation of the development stage, employs population density and the proportion of built-up area to reflect land-use intensity, the pace of urban development, and carrying pressure; its evolution typically exhibits a transition from initial outward expansion to intensive, regulated layouts in the maturity stage.

Several indicators in the index system may jointly reflect urban size, income, density, and fiscal capacity. This feature does not mean that these variables are treated as mutually independent causal factors, nor does it imply that the evaluation mechanically amplifies the role of absolute scale. In the perspective of cities as organic entities, scale-related variables such as GDP, population, fiscal expenditure, retail sales, and built-up area describe different manifestations of the size and carrying capacity of the urban organism. Their correlation is therefore partly an objective feature of urban development rather than a methodological defect. To prevent this correlation from turning the evaluation into a simple size ranking, the study first normalizes and grades all secondary indicators on a common scale, then aggregates them into five primary dimensions, and finally uses the second-layer clustering to distinguish the combined configuration of scale, structure, and speed.

In other words, final stage identification is not determined by any single scale variable. A city with a relatively small absolute scale may still be identified as being in the growth stage if it shows fast growth and an improving structural profile, whereas a city with a larger historical scale may be classified as being in the decline stage if its growth momentum is weak and its population or industrial structure has deteriorated. Supplementary diagnostic checks also show that while scale and carrying-capacity indicators are strongly associated with stage differentiation, speed and structural indicators such as GDP growth rate, industrial structure, urbanization rate, aging level, innovation output, and public service indicators also exhibit significant stage differences. This supports the interpretation that the classification is jointly shaped by scale, structure, and speed, rather than being dominated by urban size alone.

The temporal coverage of the above indicator data is 2008–2023, and the research objects are 286 prefecture-level and higher in China. Data sources include the China City Statistical Yearbook, China Urban–Rural Construction Statistical Yearbook, statistical yearbooks of individual prefecture-level cities, the China Customs Database, and various local statistical yearbooks. The selection of research objects is based on two considerations. First, comprehensiveness and representativeness: the 286 prefecture-level and above cities cover all regions of China, encompassing municipalities directly under the central government, sub-provincial cities and other national central cities, as well as ordinary prefecture-level cities and resource-based cities, thereby systematically reflecting the overall evolutionary characteristics of China’s urban system. Second, data availability and continuity: compared with county-level and lower administrative units, prefecture-level city data feature greater public accessibility, relatively consistent statistical coverage, and sound time-series continuity from 2008 to 2023, providing a reliable data foundation for subsequent multi-period indicator analysis, fuzzy comprehensive evaluation, and cluster assessment.

Regarding data processing, several measures were adopted to ensure data quality and longitudinal comparability. First, for sporadic missing values in the indicator time series, linear interpolation using adjacent years was applied to fill the gaps, ensuring complete and continuous series for each city. Second, to maintain consistent spatial units across the entire 2008–2023 period, the administrative boundaries of the most recent year were adopted as the uniform benchmark. For earlier years in which administrative adjustments occurred (e.g., conversion of counties into districts, mergers of districts, or redefinition of municipal boundaries), the original statistical data of the affected sub-units were retrospectively aggregated according to the latest boundary definitions. This approach eliminates artificial breakpoints in the statistical series caused by boundary changes and ensures that the panel reflects the same geographic scope throughout the study period. Through these steps, all indicator series for all 286 cities are aligned with a consistent set of geographical units, and year-on-year changes capture genuine socioeconomic dynamics rather than artifacts of administrative reclassification.

4. Evaluation Process of Urban Development Stages

To scientifically evaluate the evolution stages of urban development, this paper constructs a progressive three-stage identification process: “fuzzy comprehensive evaluation → bi-level K-means clustering → state stability correction.” First, based on the annual indicator data distribution of all 286 cities, the quartile method is used to determine the grade thresholds for each indicator. Using these thresholds, a fuzzy comprehensive evaluation matrix is synthesized via the weighted average operator, from which the composite scores for the five primary dimensions are calculated. Second, the first round of K-means clustering divides cities into three development level groups—high, medium, and low—to eliminate potential bias arising from cross-level comparisons. Within each group, a second round of clustering is then performed to identify the specific evolutionary development stage. Finally, the development stage of each city is confirmed based on the stability of its stage classification over three consecutive years, ensuring that the final results are objective, comparable, and robust. The overall assessment framework is illustrated in Figure 2.

4.1. Initial Identification of Development Stages Based on Fuzzy Comprehensive Evaluation

The identification of urban development stages is inherently multidimensional and fuzzy; a single threshold or linear classification cannot adequately capture the life cycle evolutionary state of a city. Accordingly, this paper introduces the fuzzy comprehensive evaluation method, which maps continuous indicator values to the possibility space of each stage via membership functions, thereby maximizing the flexibility and inclusiveness of stage assessment. To objectively identify the evolutionary stage of urban development, the quartile method is employed to determine the grade thresholds of each indicator, drawing on the historical evolution data of China’s urban development. Specifically, using the distribution of indicator values of the sample cities from 2008 to 2023, the first quartile (Q1, 25th percentile) and the third quartile (Q3, 75th percentile) are taken as the division boundaries. An indicator value below Q1 is classified as “low,” a value between Q1 and Q3 as “medium,” and a value above Q3 as “high.” This method, grounded in the natural quantiles of the data distribution, effectively identifies cities at the developmental tails while using the middle 50% of the sample to reflect the typical state of urban development, thereby balancing objectivity and robustness in stage assessment. It is important to emphasize that the quartile method here serves a data normalization function—eliminating inter-city scale differences—rather than directly determining development stages at this step. This is fundamentally distinct from the approach, criticized in the literature review, of directly stratifying cities by quantile-based thresholds. The specific threshold intervals for each indicator are presented in

Table 2. Threshold Interval of Each Indicator for Urban Development Stage Evaluation.

Primary Indicator	Indicator	Low	Medium	High
Spatial Carrying Capacity	Population Density (Person/square kilometer)	<173.54	173.54~512.25	>512.25
	Proportion of Built-up Area (%)	<0.93	0.93~5	>5
Population Development	Urbanization Rate (%)	<34.2	34.2~76.1	>76.1
	Permanent Resident Population (10,000 persons)	<236	236~662.67	>662.67
	Aging Degree (%)	>6.69	6.69~15.71	<15.71
Social Service Capacity	Number of Hospital Beds (Unit)	<10,722	10,722~29,608	>29,608
	Number of Urban Centers (Unit)	<3	3~7	>7
	Highway Passenger Volume (10,000 person-times)	<2143.45	2143.45~10,028.2	>10,028.2
Industrial Momentum	Proportion of Secondary Industry (%)	Tertiary share < Secondary share and Secondary share < 45%	Secondary share ≥ 45%	Tertiary share > Secondary share
	Proportion of Tertiary Industry (%)
	Number of Patent Authorizations (Unit)	<3357	3357~10,033	>10,033
	Number of Industrial Enterprises above Designated Size (Unit)	<690.02	690.02~2030.04	>2030.04
	Total Profits of Industrial Enterprises above Designated Size (100 million yuan)	<17.92	17.92~427.13	>427.13
Economic Vitality	GDP (100 million yuan)	<1354.37	1354.37~10,000	>10,000
	GDP Growth Rate (%)	<4	4~15	>15
	Per Capita GDP (10,000 yuan)	<2.67	2.67~10.93	>10.93
	Total Imports and Exports (100 million yuan)	<261.28	261.28~1554.83	>1554.83
	Total Retail Sales of Consumer Goods (100 million yuan)	<521.13	521.13~1527.38	>1527.38
	Local General Public Budget Expenditure (100 million yuan)	<132	132~212.12	>212.12

.

Before the quartile-based classification, Min-Max normalization was applied to the raw data of all 19 secondary indicators to eliminate differences in measurement scales and orders of magnitude. For positive indicators (where a larger value indicates a higher level of development, such as GDP and urbanization rate), the normalization formula is x′ij = [xij − min(xj)]/[max(xj) − min(xj)]. For inverse indicators (where a smaller value indicates a higher level of development), the formula is x′ij = [max(xj) − xij]/[max(xj) − min(xj)]. Through this transformation, all indicator values are mapped to the [0, 1] interval, with the directions of positive and inverse indicators unified—meaning that a value closer to 1 signifies a higher level of development.

On the basis of the above threshold divisions, this paper synthesizes the fuzzy comprehensive evaluation matrix through the weighted average operator and calculates the composite score of each city on the five primary dimensions. The specific steps of the fuzzy comprehensive evaluation are as follows:

(1) Establish the factor set. The factors for evaluating urban development stages are the evaluation indicators. The primary evaluation factor set comprises the five composite indicators, denoted as U = {U1, U2, U3, U4, U5}. The secondary evaluation factor set consists of the 19 secondary indicators, denoted as Ui = {Ui1, …, Uik}.

(2) Determine the comment set V = {V1, V2, …, Vn}. The comment set is the collection of possible evaluation outcomes. In this study, the preliminary evaluation results are divided into three grades: V = {V1, V2, V3} = {low, medium, high}.

(3) Establish the weight set. Since the relative importance of each indicator is difficult to be clearly divided in the development stage, the equal weight method is adopted for the weight design in this paper. The primary indicator weight set is W = {W1, W2, W3, W4, W5}, and the weight set of secondary indicator factors is wi = {wi1, wi2, …, wik}.

(4) Construct the membership function and obtain the fuzzy evaluation matrix R.

(5) Synthesize the fuzzy comprehensive evaluation result vector. The weighted average operator—which accounts for the influence of each factor on the comprehensive evaluation while preserving all information from the original data—is selected to synthesize the weight vector wi = {wi1, wi2, …, wik} of the urban development stage evaluation indicators with the fuzzy matrix R, yielding the evaluation result vector B of the urban development state.

Because the quartile method is used as a preprocessing and grading tool, rather than as the direct basis for assigning life-cycle stages, this paper further tests whether the final identification results are sensitive to the choice of discretization rule. Specifically, two alternative rules were adopted: the tercile method, which changes the breakpoint positions by dividing the sample into three equal portions, and the Jenks natural breaks method, which uses a data-driven classification to relax the fixed-proportion assumption. The quartile-based results are used as the benchmark, and the consistency of the alternative results is reported in

Table 3. Sensitivity Test of Alternative Discretization Methods.

Test Indicator	Tercile Method	Jenks Natural Breaks
Adjacent-stage agreement rate (difference <= 1)	92.99%	91.39%
Kappa coefficient	0.741	0.559
Weighted Kappa (linear weights)	0.799	0.695

.

The adjacent-stage agreement rates under both alternative methods exceed 91%, and the weighted Kappa values fall within the range of substantial agreement. This indicates that replacing the quartile rule with either an equal-proportion or data-driven discretization scheme does not lead to systematic changes in the identified urban development stages. The result supports the methodological interpretation that the quartile method performs a normalized grading function, while the substantive stage identification is determined by the subsequent bi-level K-means clustering and state-stability correction.

4.2. Calibration of Development Stages via Bi-Level K-Means Clustering

There are significant differences in the development level of cities in different regions. If direct global clustering is performed on 286 cities, it is very easy to cause high-development-level cities to dominate the clustering center and low-development-level cities to be marginalized, thus covering up the real state of objective evolution of urban development. Therefore, this paper designs a “Bi-level K-means clustering” method. First, clustering grouping is carried out according to the comprehensive state of urban objective development, and then clustering identification is carried out within the group according to the three dimensions of scale, structure and speed, so as to ensure that the nonlinear evolution characteristics of China’s urban development in the five stages of initial, growth, maturity, transformation and decline can be objectively and scientifically depicted.

In the process of urban development stage identification, the first clustering is K-means clustering on the scores of five primary indicators from the fuzzy comprehensive evaluation of 286 prefecture-level and above cities from 2008 to 2023. First, according to the research objectives and preliminary exploratory analysis, set the number of clusters k = 3 to obtain three categories of high, medium and low urban development states. Second, randomly select 3 samples as the initial clustering centers, then calculate the squared Euclidean distance from each sample point to each center according to Formula (1), and divide it into the nearest category according to the minimum distance principle. Finally, the clustering centers are further re-determined according to the mean value of all samples in each category on the five dimensions. The above process is iterated repeatedly until the clustering results converge. The convergence standard is that the Within-Cluster Sum of Squares (WSS) reaches the minimum (as shown in Formula (2)), that is, the total distance between all samples and their category centers no longer changes significantly, thus forming the initial classification for the identification of urban development stage evolution. The second clustering is the re-classification of the results of the first clustering, focusing on the evolution characteristics of the three dimensions of urban development: scale, structure and speed, and repeating the first clustering steps, so as to realize the correspondence between the “urban development state” and the “five stages of the life cycle”. Because high-development-level cities may be in the maturity stage with large scale but slow growth rate, or in the growth stage with medium scale but fast growth rate, low-development-level cities may also be in the initial stage or growth stage with rapid growth.

$$\mathrm{d}(x_i,x_j)={{\displaystyle \sum _{k=1}^m}\left(x_{ki}-x_{kj}\right)}^2={{\displaystyle \sum _{k=1}^m}‖x_{ki}-x_{kj}‖}^2$$

(1)

$$\mathrm{W}\mathrm{S}\mathrm{S}={\displaystyle \sum _{i=1}^k}\sum _{C(l)=i}‖x_i-{\overline{x}}_i‖$$

(2)

In the above formulas: k is the number of categories, m represents the dimension of clustering indicators, x_i represents the center of the i-th category, and C(i) represents the sample set of the i-th category.

Since K = 3 is adopted in both the first-layer clustering and the within-group second-layer clustering, the parameter setting was further evaluated before reporting the cluster results. This validation combines cluster-number diagnostics and stability checks: the Elbow method and Gap Statistic assess whether K = 3 provides sufficient information gain relative to simpler or more fragmented alternatives, while repeated random initialization and subsample tests examine whether the results depend on K-means randomness or sample perturbation. The key evidence is summarized in

Table 4. Validation of K = 3 and Clustering Stability.

Validation Item	Key Result	Interpretation
Elbow method	SSE reduction remains large from K = 2 to K = 3 (24.3%) and slows from K = 3 to K = 4 (16.3%).	K = 3 retains major structure while avoiding unnecessary fragmentation.
Silhouette coefficient	K = 3 reaches 0.2896; K >= 4 does not improve compactness.	Although K = 2 is coarser and slightly higher, K = 3 better captures intermediate stage differences.
Gap Statistic	Gap increases from 0.7029 at K = 2 to 0.8435 at K = 3.	K = 3 is supported as the minimum reasonable cluster number under the study’s theoretical design.
Cluster-size distribution	At K = 3, cluster sizes remain balanced; at K = 5, the smallest cluster contains only 8.4% of cities.	K = 3 improves interpretability and policy operability.
Initial-seed sensitivity	For K = 3, mean ARI = 0.8571 across 100 random seeds.	The clustering result is relatively stable under random initialization.
Subsample stability	For K = 3, the 90% random-subsample test yields mean ARI = 0.9177 (SD = 0.0683).	The identified cluster structure is robust to moderate sample perturbation.

.

Taken together, these tests show that K = 3 is not used as an unexamined prior assumption. Rather, it is a parsimonious and statistically defensible setting that aligns with the theoretical design of high-, medium-, and low-development-level grouping, while preserving the second-layer capacity to distinguish different scale-structure-speed configurations within each level. To further reduce algorithmic contingency, the actual implementation selected the optimal solution across repeated random initializations and then applied the three-year state-stability correction described in Section 4.3.

(1) Results of the First Clustering of Primary Indicators

Table 5. Average Values of Primary Indicators for Different City Clusters in the First Cluster Analysis.

Category	Economic Vitality	Industrial Momentum	Social Service Capacity	Population Development	Spatial Carrying Capacity
1	1.755	1.890	2.051	1.974	1.470
2	1.366	1.519	1.540	1.673	1.187
3	2.304	2.638	2.328	2.373	2.326

presents the mean values of the primary indicators for the three city categories obtained from the first cluster analysis. The cluster sizes are 1671, 2027, and 878. The first category exhibits intermediate comprehensive development levels and indicator values, corresponding to medium-development-level cities that may include growth-stage and Transition-stage cities. The second category records the lowest composite values, with all five primary indicator scores falling below those of the other two categories, indicating low-development-level cities that may encompass growth-stage, decline-stage, and initial-stage cities. The third category achieves the highest composite scores across all primary indicators, corresponding to high-development-level cities that may include maturity-stage and growth-stage cities.

(2) Re-clustering Analysis under Core Indicators

On the basis of the previous cluster analysis results of primary indicators, this paper further selects core indicators from three aspects, scale, growth rate and structure, to carry out the second clustering on the three types of cities, so as to further clarify the nonlinear distribution of urban development in the five life cycle stages. Among them, 3 indicators including GDP scale, per capita GDP and permanent resident population are selected as scale indicators, 1 core indicator of GDP growth rate is selected as growth rate indicator, and 3 indicators including proportion of tertiary industry, urbanization rate and aging degree are selected as structural indicators. Before the second cluster analysis, this paper standardized the above three types of indicators to eliminate the impact of different indicator data dimensions, and carried out the second K-means cluster analysis according to the steps of the first clustering.

Table 6. Average Value of Three Types of Indicators under the Second Clustering.

Development Level	Category	Scale			Structure			Speed
		GDP (100 Million)	Per Capita GDP (10,000)	Permanent Resident Population (10,000 Persons)	Proportion of Tertiary Industry (%)	Urbanization Rate (%)	Aging Degree (%)	GDP Growth Rate
High	1	10487.34	37.91	947.30	55.15	94.38	5.20	10.09
	2	5845.34	8.47	741.84	48.54	69.01	11.94	8.20
	3	22,308.52	13.44	1817.05	64.47	84.38	11.95	6.16
Medium	1	1657.71	3.33	537.83	35.52	45.83	10.17	11.20
	2	2724.22	8.63	338.40	45.06	67.26	12.15	7.36
	3	2257.95	5.08	460.73	46.60	54.11	14.82	5.64
Low	1	1034.87	8.22	143.38	36.51	70.72	10.89	8.45
	2	900.78	4.28	222.55	46.52	54.22	14.12	4.42
	3	664.76	2.47	296.46	34.24	41.02	9.10	12.44

shows the results of the second cluster analysis. A total of 9 types of urban groups with distinct characteristics are identified through the three cluster analyses, which not only reflects the diversity of urban development path differentiation under the same level, but also shows that China’s urban development indeed has the evolution characteristic of “same level-different stages”.

The second clustering further obtained 3 types of urban subgroups through separate clustering in the high-, medium- and low-development-level groups. Among them, in the high-development-level group, the sample sizes of the three types of cities are 2, 771 and 105 respectively. The first and third types of cities are leading in scale, structure and speed indicators, while the second type of cities are at the intermediate level in the three dimensions and show typical growth characteristics. Therefore, the first and third types of cities belong to the maturity stage, and the second type belongs to the growth stage. In the medium-development-level group, the sample sizes of the three types of cities are 757, 308 and 606. The first type of cities show typical characteristics of accelerated industrialization and rapid spatial expansion in the three dimensions, and the second type of cities show steady growth characteristics, thus belonging to typical growth stage cities. However, the third type of cities have a relatively lowest growth rate and the highest aging degree, with weakened growth momentum coexisting with the aging of population structure, which is a typical development trend of the Transition stage. In the low development level group, the sample sizes of the three types of cities are 315, 721 and 991. The first type of cities are relatively leading in scale and speed indicators, and have relatively better structural indicators, which are typical characteristics of growth stage cities. The second type of cities have weak growth superimposed with population aging, with obvious characteristics of decline stage cities. The third type of cities show the characteristics of small scale but fast growth rate, thus belonging to typical initial stage cities.

It should be emphasized that the stage names assigned above are not the direct output of the K-means algorithm. The clustering algorithm only produces statistically similar urban subgroups based on the observed configuration of scale, structure, and speed; it does not automatically generate conceptual labels such as maturity stage, growth stage, decline stage, or transition stage. The assignment of life cycle stage names is therefore a theory-driven interpretation step. Specifically, the centroid characteristics of each subgroup are compared with the theoretical expectations defined in Section 3.1. For example, a subgroup with large scale, advanced structure, and decelerating growth is interpreted as the maturity stage, whereas a subgroup with an existing scale base, weak growth, and more pronounced aging is interpreted as the decline stage.

Thus, the identification procedure consists of two analytically distinct steps. The first step is data-driven grouping, in which the bi-level K-means algorithm classifies cities into subgroups with similar multidimensional profiles. The second step is theory-driven stage labeling, in which these statistical subgroups are mapped onto the five life cycle stages according to the theoretical framework of urban organic evolution. This distinction avoids presenting the five stages as purely automatic algorithmic products and makes explicit the interpretive bridge from statistical clusters to substantive urban development stages.

4.3. Steady-State Correction of Development Stages

Finally, it must be emphasized that the evolution of urban development stages is a gradual and continuous process. Short-term economic fluctuations, statistical anomalies, or temporary policy shocks may cause a given year’s indicators to deviate from the city’s true developmental stage, potentially distorting the assessment. To mitigate the risk that abrupt shifts in core indicators—triggered by idiosyncratic events—produce spurious stage transitions, enhancing the robustness of the identification results is an inherent requirement. Accordingly, this paper introduces a “state stability correction” mechanism. By examining the dynamic evolution of the stage classifications of all 286 cities across three consecutive years from 2008 to 2023, a city is formally assigned to a particular development stage only when it has remained stably in that life cycle stage for three consecutive years. This correction mechanism effectively filters out “pseudo-changes” caused by transient factors, ensuring that the identification results reflect the long-term trend and genuine trajectory of urban development stage evolution rather than short-term fluctuations.

The complete identification chain can therefore be summarized as follows. First, the 19 secondary indicators are normalized, graded, and synthesized into five primary fuzzy comprehensive evaluation scores. Second, these five scores are used in the first-layer K-means clustering to form high-, medium-, and low-development-level groups, which controls for cross-city scale heterogeneity. Third, each group is further clustered according to scale, structure, and speed, generating nine statistically distinct subgroups. Fourth, the centroid characteristics of these subgroups are matched with the theoretical stage profiles discussed in Section 3.1 and mapped onto the five life cycle stages. Finally, the three-year state stability correction is applied to filter out temporary fluctuations and identify a city’s relatively stable development stage.

This logic also clarifies how the descriptive results reported later should be read. Table 5 and Table 6 show the mean characteristics of the first- and second-layer clusters, revealing how fuzzy scores and core scale–structure–speed indicators enter the grouping process. Figure 3 then presents the year-by-year distribution of final stage results from 2008 to 2023, while the stability correction described in this subsection explains how unstable or borderline annual classifications are handled. Accordingly, the final stage assigned to a city is not a single-year statistical label, but the outcome of a progressive procedure linking fuzzy scores, cluster hierarchy, theoretical stage interpretation, and temporal stability verification.

5. Evaluation Results and Their Spatiotemporal Differences

Based on the panel data of 286 prefecture-level and above cities in China from 2008 to 2023, this paper constructs a life cycle stage identification system covering five dimensions—economic vitality, industrial momentum, social services, population characteristics, and spatial carrying capacity—and systematically delineates the development stages of Chinese cities and their spatiotemporal evolutionary patterns. The specific evaluation results are presented in Figure 3.

5.1. National Level

At the national level, the evolution system of China’s urban development stages from 2008 to 2023 is undergoing a transition from “universal growth” to “divergent evolution,” exhibiting four overarching evolutionary features: the rapid recession of the initial stage, high-level differentiation within the growth stage, steady expansion of the maturity stage, and the accelerated emergence of the decline and Transition stages. The evolution of urban development stages presents a complex picture of pronounced differentiation and structural reconfiguration. At the same time, urban development pathways do not strictly conform to a unilinear “life cycle theory”; rather, they follow diverse nonlinear trajectories. Some cities transition directly from the initial stage to the decline stage, or from the growth stage to the decline or Transition stages, while others re-enter the growth stage after passing through decline or transformation. Figure 4 illustrates the changes in the number of cities in each of the five development stages from 2008 to 2023.

Cities in the initial stage have largely withdrawn from the historical stage. In 2008, there were 187 cities in the initial stage nationwide, predominantly concentrated in the underdeveloped areas of central and western China. By 2023, this number had plummeted to only six cities—Hezhou, Hechi, Laibin, Chongzuo, Baoshan, and Lincang.

Of the 2008 initial-stage cities, 109 advanced to the growth stage, nine transitioned to the Transition stage, and 63 slid into the decline stage. This shift indicates that the primary accumulation model—centered on infrastructure construction, industrialization, and urbanization—has been largely completed across the vast majority of Chinese regions over the past decade and a half, yielding a substantial overall improvement in urban development levels.

The number of cities in the growth stage followed a nonlinear trajectory of “expansion–differentiation–contraction.” The count peaked at 175 in 2014, accounting for 61.2% of all cities, before declining continuously to 116 in 2023. Notably, this decline does not signify a simple large-scale deterioration in urban development, but rather reflects the internal differentiation of development structures. Among these cities, 12 successfully ascended to the maturity stage, 80 entered the Transition stage due to lagging driver conversion, and 25 fell into the decline stage—marking a shift in China’s urban development stage evolution from a “bimodal” pattern to “multi-polar differentiation.” Cities in the maturity stage have expanded spatially from “coastal polar cores” to “multi-point diffusion.” In 2008, only four first-tier cities—Beijing, Shanghai, Guangzhou, and Shenzhen—had reached the maturity stage; by 2023, this number had grown to 16. New entrants include national central cities and strong provincial capitals such as Chongqing, Chengdu, Wuhan, Hangzhou, Nanjing, Suzhou, Ningbo, Tianjin, Qingdao, Xi’an, Zhengzhou, and Changsha. This expansion exhibits clear periodicity: prior to 2015, the average annual increase was fewer than 0.5 cities, whereas from 2015 to 2023, the pace accelerated to an average of 1.5 new maturity-stage cities per year, with maturity-stage cities beginning to emerge in central and western regions.

Spatially, in 2023, maturity-stage and growth-stage cities are highly concentrated in China’s eastern coastal areas and the core cities of major urban agglomerations, forming a contiguous high-quality development core zone. The boundary of this zone extends roughly along the line east of the Beijing–Guangzhou Railway and southward through the middle and lower reaches of the Yangtze River Plain, encompassing national-level urban agglomerations such as the Yangtze River Delta, Pearl River Delta, Beijing–Tianjin–Hebei, and Chengdu–Chongqing, which collectively constitute the strategic fulcrum for coordinated regional development. These regions not only lead in economic scale, industrial structure, and public services, but have also achieved a systemic leap from “quantitative expansion” to “qualitative enhancement” through efficient factor allocation and innovation-driven development.

Meanwhile, the number of cities in the Transition and decline stages has continued to rise, reaching a combined total of 148 in 2023—accounting for 51.7% of all cities nationwide—surpassing the number of growth-stage cities and becoming the dominant mode of urban development. Transition-stage cities are largely medium-development-level cities actively pursuing structural adjustment, scale management, and urban renewal, whereas decline-stage cities are predominantly characterized by chronically low growth, net population outflow, and industrial hollowing-out. Together, they constitute the “stock majority” of China’s current urban system, underscoring the shift from large-scale incremental expansion to stock-based quality improvement and efficiency enhancement. Spatially, Transition-stage cities are widely dispersed between the periphery of mature zones and decline belts, forming a transitional “buffer layer” found mostly in central provinces, the southwestern hilly regions, and the environs of certain provincial capitals. Decline-stage cities exhibit a pronounced “belt-plus-patch” distribution pattern nationwide, concentrated predominantly in the vast areas north and west of the Hu Huanyong Line. In particular, a large contiguous decline zone has formed in the northeast, extending into resource-based and old industrial areas such as Shanxi, Inner Mongolia, Gansu, and Ningxia, creating a “structural decline belt” stretching across northern China. This spatial configuration is not coincidental; it results from the long-term superimposition of resource dependency, heavy-industry path lock-in, and sustained population outflow, signaling that some cities have entered the downward channel of the life cycle.

5.2. Regional Level

Figure 5 and

Table 7. Proportion of the Number of Cities in Each Development Stage in Different Regions in 2023.

	Initial Stage	Growth Stage	Maturity Stage	Decline Stage	Transition Stage
Eastern Region	0.0%	61.6%	11.6%	5.8%	20.9%
Central Region	0.0%	30.0%	3.8%	20.0%	46.3%
Western Region	7.0%	39.5%	3.5%	27.9%	22.1%
Northeast Region	0.0%	14.7%	0.0%	73.5%	11.8%
Southern Region	3.9%	43.2%	7.1%	13.5%	32.3%
Northern Region	0.0%	37.4%	3.8%	37.4%	21.4%

jointly present the 2023 stage composition of the four major regions. The figure is added to make the regional differentiation more visually interpretable: the eastern region is dominated by growth-stage cities and contains the largest share of maturity-stage cities, indicating a maturity-growth dual engine; the central region is characterized by the highest proportion of Transition-stage cities, reflecting concentrated restructuring pressure; the western region displays a polarized structure in which growth and decline coexist; and the northeastern region is dominated by decline-stage cities, highlighting the urgency of revitalization. Table 7 reports the corresponding exact proportions and provides the quantitative basis for the following regional interpretation.

Examining the east–west axis, the four major regions in 2023 display a clear evolutionary gradient: “the eastern region leads with maturity, the central region bears the pressure of transformation, the western region exhibits pronounced internal polarization, and the northeastern region is dominated by decline.” The eastern region has remained at the forefront of national development throughout the study period, experiencing an orderly evolution from “growth and expansion” through “maturity and stability” to “intensive transformation.” In 2008, nearly 30 eastern cities were still in the initial stage, but after 2014, a large number completed their industrialization accumulation, and the number of maturity-stage cities rose rapidly. By 2023, the eastern region was home to 10 maturity-stage cities—62.5% of the national total—while growth-stage cities still accounted for 61.6% and decline-stage cities for merely 5.8%, forming an evolutionary structure driven primarily by a “maturity–growth” dual engine and supplemented by transformation.

The central region underwent a pronounced boom–bust adjustment. After 2010, initial-stage cities withdrew on a large scale and growth-stage cities expanded rapidly; however, after 2015, development momentum weakened considerably and transformation pressures were unleashed in concentrated fashion. By 2023, the proportion of Transition-stage cities in the central region was the highest in the country at 46.3%, with decline-stage cities accounting for an additional 20.0%. This reflects the dual predicament facing central cities after the dividends of industrial relocation have faded: the transformation of traditional industries lags while emerging drivers remain insufficient, and internal development imbalances have become increasingly acute. The western region’s development trajectory exhibits pronounced “bipolarization” that has intensified over time. In the early years, the overwhelming majority of western cities were in the initial stage, but with the rise of urban agglomerations such as Chengdu–Chongqing and Guanzhong, national central cities like Chongqing, Chengdu, and Xi’an were the first to enter the maturity stage. Meanwhile, cities in peripheral areas have progressed slowly: the western region still contains all six of the country’s remaining initial-stage cities, and the proportion of decline-stage cities has climbed to 27.9%. The internal regional structure has shifted from “generalized backwardness” to a widening gradient of “core leapfrogging, periphery stagnation”.

The northeastern region presents the most severe picture. Its structural predicament has deepened continuously over the 16-year study period. In 2008, some northeastern cities were still in the growth or initial stage, but their capacity to transition to maturity was limited. After 2015, growth momentum further contracted and the decline trend accelerated markedly. By 2023, 25 of the 34 sample cities (73.5%) were in the decline stage, none had reached the maturity stage, and only four cities (14.7%) remained in the growth stage. The region as a whole has fallen into a systemic impasse characterized by “growth stagnation, transformation inertia, and spreading decline,” with a critical deficit of development momentum.

Beyond the east–central–west–northeast comparison, Figure 6 further visualizes the north–south divide in 2023. The south is characterized by a growth–Transition combination, with 43.137% of cities in the growth stage and 30.719% in the Transition stage, while decline-stage cities account for only 15.033%. By contrast, the north shows a more fragile structure: growth-stage cities account for 37.594%, but decline-stage cities also reach 35.338%, and maturity-stage cities account for only 3.759%. This visual comparison helps clarify that the regional gap has shifted from a simple level difference to a stage-based fault line between a more dynamic southern urban system and a more structurally constrained northern urban system.

Turning to the north–south axis, the gap in urban development stage evolution between the two regions widened continuously from 2008 to 2023, evolving from “similar starting points” to “divergent trajectories.” In 2008, there were 102 initial-stage cities in the south and 85 in the north—reflecting a modest initial difference. However, with the deepening of regional development strategies, industrial relocation, and market-oriented reforms, the southern urban system accelerated its upward transition, while the north gradually became mired in structural adjustment difficulties. By 2023, southern growth-stage and maturity-stage cities totaled 78, accounting for 50.3% of the southern total, indicating that more than half of southern cities remained in expansion or high-quality stable development; decline-stage cities accounted for only 13.5%. In stark contrast, although northern initial-stage cities had disappeared by 2021, the north had only 49 growth-stage cities and a mere five maturity-stage cities (3.8%). More gravely, the number of decline-stage cities in the north stood at 49—equal to the number of growth-stage cities—with an additional 28 Transition-stage cities. This evolutionary pattern is not the product of short-term fluctuations but the cumulative outcome of long-run urban development dynamics. The number of northern decline-stage cities rose relentlessly from 2 in 2012 to 49 in 2023, while maturity-stage cities increased by only 4 over the same 16-year period—a pace significantly lagging behind the south (which grew from 1 to 11). It is particularly alarming that the proportion of decline-stage cities in the north (37.4%) is nearly 2.8 times that in the south (13.5%), whereas the number of maturity-stage cities in the north is merely 45.5% of that in the south. The north–south development gap has thus evolved from an early-stage speed differential into a contemporary stage fault line and kinetic energy chasm.

Table 8. Proportion of Each Stage in the Five Major Urban Agglomerations in 2023.

Urban Agglomeration	Initial Stage	Growth Stage	Maturity Stage	Decline Stage	Transition Stage
Yangtze River Delta Urban Agglomeration	0.0%	63.0%	18.5%	7.4%	11.1%
Beijing–Tianjin–Hebei Urban Agglomeration	0.0%	38.5%	15.4%	7.7%	38.5%
Pearl River Delta Urban Agglomeration	0.0%	77.8%	22.2%	0.0%	0.0%
Middle Reaches of the Yangtze River Urban Agglomeration	0.0%	40.7%	7.4%	11.1%	40.7%
Chengdu–Chongqing Economic Circle	0.0%	12.5%	12.5%	6.3%	68.8%

presents the stage composition of China’s five major urban agglomerations. The Yangtze River Delta and Pearl River Delta urban agglomerations exhibit relatively strong internal coordination capacity, while the remaining agglomerations display a pronounced “core–periphery” structural divide. The Yangtze River Delta is the most concentrated zone of maturity-stage cities in the country, evincing a high degree of economic maturity and structural stability. In 2023, Shanghai, Hangzhou, Ningbo, Nanjing, and Suzhou—five cities in total—were at the maturity stage, accounting for 31% of all maturity-stage cities nationwide. Growth-stage cities constituted 63% of the Delta’s urban total, second only to the Pearl River Delta. It is noteworthy, however, that internal structural differentiation persists within the Delta: some cities in Anhui have entered the Transition or decline stage, indicating residual imbalances in the region’s coordinated development process. The Pearl River Delta presents a “dual-core leadership with region-wide growth” configuration: Guangzhou and Shenzhen are maturity-stage cities, while all seven remaining cities are in the growth stage, reflecting a significant synergistic spillover effect. By contrast, although the Beijing–Tianjin–Hebei, Chengdu–Chongqing, and Middle Reaches of the Yangtze River urban agglomerations each possess two maturity-stage core cities, the majority of their surrounding cities are in the Transition or decline stage, indicating that inter-city coordinated development capacity in these agglomerations still requires considerable enhancement.

6. Discussion

The findings of this study both corroborate and extend the existing literature on regional inequality in China. The overarching pattern of regional inequality has been extensively documented. Kanbur and Zhang (2005) systematically reviewed the evolution of regional inequality since the founding of the People’s Republic, noting that the coastal–inland gap widened sharply after the reform and opening-up [40]; Démurger (2001) attributed this disparity to the spatial differentiation of infrastructure and locational advantages [41]; and Wei (2015) further revealed the spatial characteristics of regional inequality from a multi-scalar perspective [42]. Our finding that 85.7% of cities in the eastern region are in the maturity or growth stage is highly consistent with the coastal agglomeration advantage identified in the aforementioned literature. However, our incremental contribution lies in measuring the gap not merely by the level of development, but by utilizing the concept of “development stages” to further disaggregate the advantages of eastern coastal cities into the structural characteristics of “steady expansion in the maturity stage” and “high-level differentiation in the growth stage.” This reveals that the eastern region is not a monolithic block, but a multi-stage mosaic.

The widespread decline in Northeast China has likewise attracted substantial scholarly attention. Zhang (2008) systematically analyzed the trajectory and challenges of revitalization policies for the Northeast Old Industrial Base [43]; He et al. (2017) and Yang and Dunford (2018) focused on shrinking cities in Northeast China, identifying population outflow, industrial path lock-in, and institutional inertia as the core drivers of decline [44,45]; and Xie et al. (2020) further delineated the spatial carrying capacity of urban shrinkage in Northeast China from a multi-dimensional perspective [46]. Our finding that 73.5% of northeastern cities are in the decline stage—with none having reached maturity—strongly corroborates this literature. However, unlike previous studies that concentrated on specific city types (e.g., resource-based or shrinking cities), the present study situates northeastern cities within a unified life cycle framework encompassing 286 prefecture-level and above cities nationwide for comparative analysis. This more clearly reveals that the Northeast is not an isolated case of “local distress,” but rather the regional block lying closest to the systemic decline end of the national urban development stage spectrum. This perspective helps elevate the “Northeast Phenomenon” from a narrative of localized plight to a structural proposition within the evolution of national urban development stages.

Furthermore, the five-stage classification system constructed in this paper (initial, growth, maturity, decline, and transition) offers a novel analytical tool for understanding regional inequality in China. Unlike the existing literature, which predominantly employs continuous inequality measures such as per capita GDP or composite development indices, our stage-based classification method simultaneously captures the configurational features of cities across three dimensions—scale, structure, and speed. This makes it possible to reveal the nonlinear dynamics in which the same development level interval may correspond to different life cycle stages, thereby providing a more structurally nuanced picture of regional disparities.

Several limitations of this study should be acknowledged. First, while the bi-level K-means clustering approach effectively identifies stage configurations, it does not establish causal relationships between specific policies and stage transitions; future research could incorporate quasi-experimental designs to evaluate the effects of place-based policies on stage evolution. Second, the indicator system, though comprehensive across five dimensions, could be further enriched—for instance, by incorporating environmental sustainability and digital economy indicators—as data availability improves. Third, the analytical framework established here could be extended to international comparative studies to examine whether the stage evolution patterns observed in China hold in other developing and developed country contexts.

7. Conclusions and Policy Recommendations

7.1. Research Conclusions

Grounded in the characteristics of cities as organic entities, this paper constructs an evaluation system spanning five dimensions—economic vitality, industrial momentum, social services, population characteristics, and spatial carrying capacity—and, through a three-stage progressive evaluation process of “fuzzy comprehensive evaluation → bi-level K-means clustering → state stability correction,” systematically identifies the development stages and spatiotemporal evolution characteristics of 286 Chinese cities from 2008 to 2023. The principal findings are as follows.

At the national level, China’s urban development stage evolution system is transitioning from “universal growth” to “divergent evolution,” exhibiting four overarching features: the rapid recession of the initial stage, high-level differentiation within the growth stage, steady expansion of the maturity stage, and accelerated emergence of the decline and Transition stages. Urban development pathways do not strictly follow a unilinear “life cycle theory,” but display diverse nonlinear trajectories. The number of initial-stage cities plummeted from 187 in 2008 to 6 in 2023; growth-stage cities declined from a peak of 175 to 116; the evolutionary pattern shifted from “bimodal” to “multi-polar differentiation”; and maturity-stage cities expanded steadily from the original four (Beijing, Shanghai, Guangzhou, and Shenzhen) to 16. Meanwhile, the combined proportion of Transition- and decline-stage cities reached 52%, surpassing the number of growth-stage cities and becoming the dominant mode of China’s urban development.

At the regional level, in the east–west direction, the four major regions exhibit a clear evolutionary gradient of “eastern maturity leadership, central transformation pressure, western polarization, and northeastern decline dominance.” The eastern region hosts 62.5% of the nation’s maturity-stage cities and 61.6% of its growth-stage cities. The central region saw its development momentum weaken significantly after 2015, with the highest national proportion of Transition-stage cities (46.3%) in 2023. The western region displays intensifying internal polarization, with mature cores such as Chengdu and Xi’an coexisting with all six remaining initial-stage cities. The northeastern region is the most severely affected, with 73.5% of cities in the decline stage and none in maturity. In the north–south direction, the development gap has widened continuously: the proportion of decline-stage cities in the north (37.4%) is nearly 2.8 times that in the south (13.5%), while the number of maturity-stage cities in the north is merely 45.5% of that in the south. The north–south gap has evolved from an early speed differential into a contemporary stage fault line and kinetic energy chasm. At the urban agglomeration level, with the exception of the Yangtze River Delta and Pearl River Delta—which are driven by a “maturity–growth” dual engine—the remaining agglomerations exhibit a pronounced “core–periphery” structure, with most surrounding cities in the Transition or decline stage, indicating that inter-city coordinated development capacity remains insufficient.

7.2. Policy Recommendations

First, the stage identification results provide a decision-making reference for differentiated governance. In 2023, the total number of cities in the Transition and decline stages reached 148, accounting for 52%, and decline-stage cities were highly concentrated in the north (49 cities, 37.4% of the northern total). This stage pattern suggests that incorporating urban development stage identification as a prerequisite for territorial spatial planning, fiscal transfer payments, and the layout of major projects could help enhance the stage-specificity of resource allocation. From a life cycle perspective, for regions with concentrated decline—such as Northeast China (73.5% of cities in the decline stage) and resource-based cities in North China—structural repair could revolve around special central fiscal support, industrial succession and substitution, and ecological compensation mechanisms, thereby creating a buffer space for potential stage recovery. For the central region, where 46.3% of cities are concentrated in the Transition stage, investments in innovation infrastructure, the reconstruction of vocational education systems, and urban renewal serve as potential levers for promoting stage succession and averting the loss of growth momentum. Cities in the growth stage currently occupy a transition window from scale accumulation to structural optimization; guiding them from extensive expansion to intensive enhancement is conducive to smooth evolution toward the maturity stage. The 16 maturity-stage cities could accumulate experience for high-level urban governance in areas such as institutional opening-up, integration into global innovation networks, and the radiating effects on surrounding regions.

Second, the phenomenon of inter-regional stage fault lines implies the necessity of differentiated regional policies. The eastern region possesses 62.5% of the country’s maturity-stage cities, with a decline rate of only 5.8%. This stage advantage provides a favorable foundation for further enhancing high-quality development capacity and releasing spillover effects to surrounding growth-stage cities. In the central region, 46.3% of cities stand at a critical juncture for stage transitions; shortening their dwell time in the transformation pathway may require relying on platforms such as national industrial transformation and upgrading demonstration zones and regional sci-tech innovation centers. Internal stage differentiation is prominent in the western region, which harbors both mature cores like Chengdu and Xi’an and all six of the country’s initial-stage cities, with decline-stage cities accounting for 27.9%. This characteristic suggests that policymakers should attend to the balance between “core strengthening” and “short-board mitigation,” avoiding a scenario in which core cities leapfrog in stages while peripheral stagnation worsens. In the northeast, no city has reached maturity, and 25 decline-stage cities account for 73.5% of the regional total; this systematically low stage positioning implies that initiating institutional and mechanism reforms and promoting central–local collaborative pilot programs for the regeneration of resource-based cities could represent a potential exploratory direction. Particularly noteworthy is that there are only five maturity-stage cities in the entire north (3.8%), with the number of decline-stage cities equaling that of growth-stage cities. This stage fault line suggests that, in the national layout of major productive forces (e.g., new energy bases, computing hubs, advanced manufacturing clusters), fully considering the stage status of northern cities and exploring the establishment of a “Northern Urban Revitalization Fund” might provide the foundational momentum to systematically improve the stage evolution environment for northern cities.

Third, the “mature center–lagging periphery” pattern characteristic of urban agglomerations suggests potential directions for collaborative governance. The Yangtze River Delta urban agglomeration already possesses five maturity-stage cities, but some cities in Anhui have entered the Transition or decline stage. This stage gradient suggests that cross-provincial institutional coordination and public service sharing may be important directions for facilitating stage transmission from the center to the periphery. The Pearl River Delta presents a stage structure of “Guangzhou–Shenzhen dual core + region-wide growth”; its “maturity–growth” collaborative pattern serves as a valuable reference for other national urban agglomerations seeking to understand the laws of stage coordination. The internal stage fault lines within the Beijing–Tianjin–Hebei agglomeration are conspicuous, with 6 of 11 cities in Hebei currently in the Transition or decline stage. How to create conditions for stage leapfrogging in surrounding cities through the effective matching of Beijing’s non-capital function dispersal with Hebei’s carrying capacity constitutes a critical stage coordination challenge. Although the Chengdu–Chongqing and Middle Reaches of the Yangtze River agglomerations have mature cores, over 60% of their peripheral cities remain in the Transition stage. From the perspective of stage evolution, shortening the dwell time of peripheral cities in the transformation pathway through mechanisms such as intercity transport connectivity, industrial chain collaborative parks, and basin-wide ecological co-governance could help prevent the further solidification of a stage differentiation pattern characterized by “core idling and peripheral marginalization”.