The “Scale Expansion Trap” in Cross-River Urbanization: Building Stock Vacancy and Carbon Lock-In for Nanchang, China

Abstract

Understanding spatial characteristics of urban building systems is critical for unraveling urban building stock growth patterns, addressing housing vacancy challenges, and advancing urban carbon neutrality. However, existing research on built environment stocks and housing vacancy spatial distribution remains limited, particularly in underdeveloped cross-river cities—where rapid urbanization often prioritizes scale expansion over demand matching, leading to unresolved issues of resource waste and environmental pressure. This study integrated material stocks analysis (MSA) and geographical information system (GIS) to uncover the spatial patterns of urban building material stocks and housing vacancy at a high spatial resolution for Nanchang, China—a typical underdeveloped cross-river city facing the “scale expansion trap” in its urbanization across the Ganjiang River. Results show that (1) Nanchang’s building stock exhibits a “butterfly-shaped” spatial pattern centered on the Ganjiang River, with simultaneous horizontal expansion (40-fold urban area growth since 1949) and vertical growth (super high-rises in new west-bank districts), reflecting aggressive cross-river scale expansion; (2) the total building material stock reached 1034 Mt (204 t/cap) in 2021, with over 85% accumulated post-2000—coinciding with large-scale cross-river development. Vacant buildings locked in 405 Mt of materials (39.17%), which is a direct consequence of the “scale expansion trap” where construction outpaced actual demand; (3) total embodied carbon emissions from building materials amounted to 264 Mt, with 104 Mt (39.39%) attributed to vacant stocks. This “vacant carbon lock-in” stems from mismatched urban construction and actual demand in the process of cross-river scale expansion; (4) spatially, high-value clusters of material stocks and carbon emissions overlapped at two cores (old town and Honggutan CBD), while housing vacancy rates were significantly higher in the urban periphery and Ganjiang’s west bank—the primary areas of cross-river scale expansion—than in the old town and east bank. These findings empirically demonstrate how the “scale expansion trap” in cross-river urbanization drives building stock vacancy and carbon lock-in. These findings also offer data-driven strategies for optimizing urban resource allocation, reducing housing vacancy, and promoting low-carbon transitions, especially for other underdeveloped cross-river cities globally.

1. Introduction

Urbanization is one of the most significant global megatrends of the last century [1,2]. It represents population concentration [3], the expansion of construction land [4], and the accumulation of materials in buildings [5] and infrastructure (built environment) [6,7], and defines the physical space for urban residents’ activities. The construction, maintenance, and demolition of urban built environment stocks cause major sustainability challenges for cities [6,8,9,10], including resource demand [11], energy use [12], greenhouse gas (GHG) emissions [13,14], and construction and demolition (C&D) waste generation [15]. Additionally, the particular phenomenon of overbuilding during a highly dynamic urbanization process in China has directly resulted in housing vacancies [16,17,18].

Therefore, to understand the complexity and sustainability of urban building system development [19], it is important to uncover the spatial patterns of urban built environment stock development at a high resolution [20]. Moreover, to alleviate the severe issue of urban housing vacancy [21], fulfill the climate ambitions of peaking before 2030 and neutrality before 2060 [22], reduce resource consumption [23], and lower carbon emissions [20,24], it is crucial to thoroughly benchmark and understand the built environment stocks across and within cities at various stages of development. However, our existing knowledge of the spatial patterns of built environment stocks across cities, especially within cities, remains limited [6]. This is largely due to a lack of high spatial resolution data, which are highly data-intensive and labor-intensive. As a typical form of urbanization in river-crossing cities, cross-river development often becomes a key strategy for underdeveloped cities to pursue spatial expansion and image upgrading. However, this model is prone to falling into the “scale expansion trap”. This study conceptualizes the “scale expansion trap” as an urbanization syndrome driven by land-finance fiscal models, where the aggressive construction of new districts systematically outpaces organic population and economic growth [25,26]. This state-sponsored overbuilding creates a spatial mismatch that locks material resources and embodied carbon into underutilized building stocks, reflecting a fundamental disconnect between planning incentives and market demand [25]. Unfortunately, existing research has rarely focused on the internal mechanism of the “scale expansion trap” driving building stock vacancy and carbon lock-in in underdeveloped cross-river cities, leaving a research gap that needs to be filled urgently.

The estimating approaches for built environment stocks can be generally categorized into top-down and bottom-up approaches [27]. The top-down approach builds on the mass-balance principle where the change in stock is the result of the difference between inflows and outflows of a material over time, typically over 1 year [28]. This provides a complete set of theoretical foundations and algorithm models, which can be used to effectively perform large-scale material flow analysis at the global and national levels [24]. Conversely, the bottom-up method permits fine-grained stock estimation by gathering the detailed physical measurements of buildings and infrastructure, along with associated material composition indicators [24,29,30]. However, this approach is labor-intensive and the scope of the bottom-up method is often restricted to the city-level or lower geographical regions. Because data can be directly derived from information on stock inventory, the results of bottom-up studies are typically considered more accurate than those obtained through a top-down approach. Several studies have attempted to characterize infrastructure development and its environmental effects at the district level in China [19,31,32,33]. For example, Wang et al. [34] established a 4D-GIS model of Longwu Village in Shenzhen to illustrate the spatio-temporal patterns and material metabolism evolution of buildings in the village, while Li et al. [35] quantitatively estimated the in-use stock of 662 residential buildings in the Huilongguan community of Beijing. Due to the large amount of data required to conduct these studies, the scope of bottom-up studies often cannot be extended to larger geographical scales [19] and less developed regions. However, in the era of big data, more data sources have provided new research directions for stock calculation [24], such as point of interest (POI) [19,20] and emerging big geodata. However, due to data availability limitations, the majority of studies investigating fine spatial resolution stock have focused on developed urban areas, such as Wakayama City [36], Manchester [36], Vienna [37], and Philadelphia [38]. Subsequently, several studies have quantified the weight and distribution of building material stocks across cities in China, such as Beijing [20], Shanghai [19] and Shenzhen [39]. However, our systematic knowledge of the quantity, structure, and material composition of buildings in entire cities and their impacts on resource utilization and spatial patterns remains limited [19], especially for underdeveloped cities [24], and there are even fewer studies on the spatial distribution of housing vacancy [40,41,42]. Paradoxically, the most pressing urban challenges have arisen in rapidly developing, yet still underdeveloped regions in recent decades.

As the capital city of Jiangxi Province in Southeast China, Nanchang is also one of the typical underdeveloped cross-river cities implementing river-crossing development strategies. Its urbanization process has been dominated by large-scale spatial expansion across the Ganjiang River since the late 1990s. However, this aggressive scale expansion has not been matched by synchronous population and industrial agglomeration, leading to a severe “scale expansion trap”. Nanchang’s housing vacancy rates (HVRs) were 22% and 33% in 2010 and 2020 respectively, ranking first among 28 surveyed cities [43], and a large amount of building materials and embodied carbon are locked in vacant buildings. Meanwhile, as an economic depression zone among neighboring provincial capitals, Nanchang’s experience is highly representative of underdeveloped cross-river cities facing similar challenges. Therefore, taking Nanchang as the case study, this research focuses on exploring the relationship between the “scale expansion trap”, building stock vacancy and carbon lock-in, which is of great significance for enriching the research on urban sustainable development in underdeveloped cross-river regions.

This study is therefore designed to answer the following core research question: How does the “scale expansion trap” spatially drive building stock vacancy and carbon lock-in in an underdeveloped cross-river city? To address this, we integrated big data mining techniques, material stocks analysis (MSA) and geographical information system (GIS) to pursue three specific objectives: (1) to develop a high-resolution spatial database of building stocks and vacancy in Nanchang, (2) to quantify the embodied carbon locked in both total and vacant building stocks, and (3) to analyze the spatial mismatch between construction scale and socio-economic demand as evidence of the trap’s mechanism.

Following the introduction in Section 1, this study is structured as follows. Section 2 details the research methodology, data sources, and analytical framework. Then, Section 3 presents the spatial results, discusses the role of the “scale expansion trap,” provides a comparative analysis, and addresses study limitations. Finally, Section 4 concludes with the key findings and policy implications.

2. Research Methods and Data Processing

2.1. Research Methods

The framework of this study is shown in Figure 1. Material stocks analysis (MSA), geographic information system (GIS), and big data mining were combined to construct a high-resolution database model of Nanchang’s building stock. Based on the research findings of Shi et al. [16] on housing vacancy rates, growth patterns were characterized and the housing vacancy distribution of the building stock was mapped using the high-resolution database. The building material stocks and their embodied carbon emissions were also quantified, with spatial mapping executed at fine-grained resolution.

2.2. Scope and Data Collection

The research scope of this study is defined as the urban built-up area of Nanchang City. According to data availability, the urban building stock in Nanchang in 2021 was selected as the research subject for analysis. Nine types of construction materials were considered, namely steel, timber, cement, brick, gravel, sand, asphalt, lime, and glass. The selection of these materials is based on their status as the fundamental constituents of the dominant brick–concrete and steel–concrete structures in China’s building stock, ensuring our assessment captures the material stock and embodied carbon [17,18,44,45]. In addition, ArcGIS 10.8 software was used for spatial analysis, data processing, and results visualization.

The geospatial database of Nanchang integrated big geodata (including footprint area, height, location, and morphological attributes) from the National Platform for Common GeoSpatial Information Services (https://www.tianditu.gov.cn/), combined with POI data from Amap (https://ditu.amap.com/). Building age and functional types were supplemented by data from major real-estate platforms including Anjuke and Lianjia. The housing vacancy rate (HVR) was derived from Shi et al. [16], who mapped housing utilization efficiency using census data at the block level. Spatial proximity analysis was performed in ArcGIS to unify these multi-source datasets. To validate accuracy, 1000 random sampling points were sampled, yielding an overall accuracy of 86.36%, exceeding the 85% threshold specified in the standards of the United States Geological Survey. This database systematically synthesizes building characteristics, vacancy metrics, and spatiotemporal attributes to support urban analysis.

2.3. Building Material Stock Estimation

The bottom-up method for estimating building material stocks was based on two key parameters (Equation (1)): (i) the physical size of the structure (m²) and (ii) the material intensity (MI) for each component (kg/m²). The MI varied according to different building types, structures and ages. The MI database for Nanchang was compiled from previous studies (Table S1 in the Supplementary Materials [46,47]), and buildings in the study area were categorized into two types based on the number of building floors and the footprint area, i.e., brick–concrete structures and steel–concrete structures [48,49].

$${MS}_{m,i,j}\left(t\right)=\sum _{m,i,j}\left({PS}_{i,j}\left(t\right)\times {MI}_{i,j}\left(t\right)\right)$$

(1)

where MS_m,i,j(t) is the stock of material m in type i (residential and non-residential) with structure j (brick–concrete and steel–concrete) in year t, PS_i,j(t) is the physical size of a building of type i with structure j in year t, which was calculated by multiplying the footprint area of the building by floor number to determine floor area of each building, and MI_i,j(t) denotes the intensity of building materials of type i with structure j in year t.

2.4. Embodied Carbon Emission Calculation

Building materials have a significant environmental impact throughout their life cycle, with 82–87% of carbon emissions originating from their production [50]. Therefore, we calculated and analyzed the carbon emissions in the materials production stage. The carbon emission factors for the construction materials (unit: kg CO₂-eq/t), which reflect the cradle-to-gate processes such as raw material extraction, manufacturing, and processing, were used to calculate the embodied carbon emissions in the building material stocks (Equation (2)):

$${EI}_m=\sum \left({EF}_m\times {MS}_m\right)$$

(2)

where EI_m is the total carbon emissions of material m, EF_m is the carbon emission factor of material m (Table S2 in the Supplementary Materials [15,20]), and MS_m is the material stocks of type m.

3. Results

3.1. Building Spatial Pattern

The spatial distribution of building stock in Nanchang City exhibited a strong temporal and spatial correlation with the urbanization process. As a typical underdeveloped cross-river city, Nanchang’s urbanization has been dominated by large-scale spatial expansion, which has directly shaped the spatial pattern of its building stock and laid the foundation for the subsequent “scale expansion trap”. In general, the urban area of Nanchang exhibited an east–west symmetrical pattern along the Ganjiang River (Figure 2a), with 46,017 buildings on the west bank and 95,809 buildings on the east bank. Notably, while the east bank has over twice as many buildings as the west bank, its total floor area does not reach twice the west bank’s. As urban expansion progressed, the urban building stock in Nanchang evolved into a distinct butterfly-shaped spatial distribution centered around the Ganjiang River, with building clusters along the eastern and western shorelines extending outward like wings.

Regarding building height, the average height declined progressively along the central axis of the Ganjiang River (Figure 2b). The scale of Nanchang’s building stock and the average height of its buildings grew substantially and steadily during the urbanization process. Completed in 1961, the Jiangxi Hotel, with a height of 30 m, was an iconic landmark of the city’s skyline at that time. In 1999, with the establishment of the Honggutan New Area, Nanchang began its cross-river development. By 2001, after Nanchang’s administrative center was relocated to the west bank of the Ganjiang River, Shajing Street experienced swift development, with the average building height reaching 19.42 m. Subsequently, urban construction significantly expanded the development space of Nanchang while also increasing the average height of the buildings. Additionally, the development of the Jiulong Lake area in 2011 led to an increase in building height, with the average reaching 23.25 m. Finally, the completion of Honggutan Central Business District (CBD) in 2014 marked a new milestone in urban development. The Honggutan CBD, which features the Nanchang Greenland Central 303 Twin Towers, covers an area of 12.27 km² and has an average building height of 24.40 m, making it the area with the second-highest average building height in Nanchang. Notably, the Honggutan CBD features several super high-rise buildings, including the 303-m-tall Nanchang Greenland Central 303 Twin Towers, along with other prominent skyscrapers. Since the founding of New China, the maximum building height in Nanchang has increased approximately 10-fold, from 30 to 303 m. However, with rapid urbanization and decreasing land resources, Nanchang has encountered challenges in optimizing its building system. Specifically, in the Honggutan area, dozens of skyscrapers have been built, indicating that Nanchang is reshaping its skyline to gain more development space. However, this aggressive vertical expansion, driven by the pursuit of urban scale and image, has exceeded the current carrying capacity of population and industries, becoming a key manifestation of the “scale expansion trap” and foreshadowing high vacancy rates in new urban areas.

Regarding building density, the distribution of buildings in Nanchang presents a clear spatial pattern of multi-core dispersion, as shown in Figure 2c. High-density areas are primarily located in the old town, the Yuxin Road North Area in the Xinjian District and the Chenghu East Road Area in Nanchang County. High-density buildings in the old town exhibit a patchy rather than point-like spatial distribution, particularly in locations such as Tengwang Pavilion, Baihuazhou, Dunzitang, and Ximazhuang around East Lake. According to the statistics, the urban population in Nanchang reached 225,000 in 1949, while the total urban area was only 8 km². Nanchang’s primary development focus remained in the area to the east of the Ganjiang River until the 1990s, accompanied by a significant increase in the residential population in the old town. In the 1990s, the old town, with a population of 400,000 over just 8 km², was one of the most densely populated urban areas in China, which significantly impeded urbanization. After decades of development, the urban area has expanded significantly to 365.51 km², while the population reached 2.84 million in 2021. Over the past 70 years, Nanchang’s urban area has expanded more than 40-fold from its 1949 size. This drastic horizontal expansion, coupled with uncoordinated population agglomeration (the urban population increased only 12.6 times during the same period), has led to scattered building distribution and inefficient use of construction land, a typical symptom of the “scale expansion trap”. Meanwhile, emerging commercial centers such as Honggutan CBD and Chaoyangzhou feature taller buildings but much lower building density compared to the old town. This difference underscores the varying socioeconomic conditions, planning strategies, and construction methods adopted in Nanchang across its development stages. In summary, Nanchang’s building stock is characterized by both horizontal expansion and vertical growth. Table S3 in the Supplementary Materials provides details on the relevant boundaries and data sources described in this section. The spatial distribution of all key urban development sites in Nanchang mentioned above is illustrated in Figure S1.

3.2. Building Stock Vacancy

A total of 141,826 buildings with a floor area of 482 million square meters (Mm²) were counted in the built-up area of Nanchang by this study in 2021. The spatial distribution of buildings classified by different construction time periods (building age cohorts) is shown in Figure 3a. Most old buildings are mainly distributed in the old urban area, while the buildings on the west bank of the Ganjiang River are much newer. This is consistent with the development history of Nanchang City.

Table 1. Building number, floor area and material stocks in different age cohorts.

Age Cohorts	–1959	1960–1979	1980–1989	1990–1999	2000–2021
Building numbers	29 (0.02%)	207 (0.15%)	1344 (0.95%)	21,901 (15.44%)	118,345 (83.44%)
Floor area (106 m2)	0.06 (0.01%)	0.50 (0.10%)	3.24 (0.67%)	65.06 (13.50%)	413.18 (85.72%)
Material stocks (106 t)	0.10 (0.01%)	0.92 (0.09%)	6.65 (0.64%)	139.65 (13.50%)	887.15 (85.76%)

illustrates the figures and proportions of building numbers, floor area, and material stocks by age cohorts. The distributions of building numbers, floor area, and building material stocks across various time periods exhibited similar trends. In terms of age, the floor area of buildings constructed or reconstructed after 2000 increased to 413 Mm², accounting for 85.73% of the total building floor area stock, overshadowing the 1990s value of 65 Mm² (13.49%), the 1980s value of 3 Mm² (0.66%), the 1960–1979 period’s value of 504,050 m² (0.10%), and the insignificant value of 55,576 m² (0.01%) before 1959. Similarly, almost 887 million tons (Mt) (85.76%) of building material stocks were found in buildings constructed or reconstructed after 2000. By contrast, only around 13.50% (140 Mt) of building material stocks were found in buildings built during the 1990s. This explosive construction scale, driven by cross-river development and new area expansion strategies, has far exceeded the actual housing and commercial demand derived from population and economic growth, leading to 405 Mt of materials being locked in vacant buildings. Moreover, the proportions for the 1980s, 1960–1979 period, and before 1959 were only 0.64%, 0.09%, and 0.01%, respectively, with the corresponding building materials consumed at 7 Mt, 0.92 Mt, and 0.10 Mt, respectively. These results indicated significant growth both in building material stocks and floor area after 2000, which was consistent with statistical data from the Nanchang Statistical Yearbook [52,53]. Compared to Beijing City, the newly built floor area in Nanchang increased sharply after 2003, while Beijing reached its peak in 2005 [15], further emphasizing the underdevelopment of Nanchang. To track the utilization of various building materials, the spatial distributions for nine key materials were quantified and mapped. This information was valuable for supporting predictions of future material demand, identifying the spatial distribution of waste generation, and developing vacant building databases.

Urban buildings in Nanchang exhibit significant spatial differences in housing vacancy rate distribution (Figure 3b). The urban periphery exhibits higher HVR than the old town, consistent with findings by Zheng et al. [21], i.e., housing vacancy rates exhibit significant variations between urban cores and suburban areas. A distinct north–south disparity exists, with southern districts characterized by more concentrated vacancy clusters and higher vacancy levels. A comparative analysis between building age and HVR patterns reveals a key paradox. The old town, despite its older building stock, maintains among the lowest HVR in the city; conversely, the Honggutan CBD, dominated by newer high-rises, shows higher HVR. The west bank, as the main area of cross-river expansion since the 1990s, has undergone large-scale and high-standard construction, but the lagging supporting facilities and industrial agglomeration have led to a serious mismatch between building supply and actual demand, making it a concentrated area of the “scale expansion trap”.

Table 2. Composition of building material stocks (unit: 10⁶ t).

Material	Steel	Timber	Cement	Brick	Sand	Gravel	Lime	Glass	Asphalt
Stock value (106 t)	26.31	14.39	148.97	180.26	324.06	320.63	17.92	0.96	0.96
Proportion of total stock (%)	2.54	1.39	14.40	17.43	31.33	30.99	1.73	0.09	0.09

illustrates the breakdown of building material stocks by material type in the urban area of Nanchang in 2021. The total material stock reached 1034 Mt, of which vacant building materials accounted for 405 Mt (39.17%); this indicates that nearly 40% of building materials are in a state of low or no efficiency. In terms of total building material stocks, concrete’s key components—sand (324 Mt, 31.33%), gravel (321 Mt, 30.99%), and cement (149 Mt, 14.40%)—together make up 76.72% of the total material stock. This proportion highlights their dominance in urbanization-driven construction and underscores significant potential for concrete recycling. These stocks demonstrate significant urban mining potential. For instance, steel and concrete from demolished reinforced concrete structures can be directly reused; crushed concrete can be processed and repurposed for new wall construction.

The spatial pattern of the total material stock (Figure 4a and Figure S2 in the Supplementary Materials) was closely correlated with the spatial characteristics of urban building stock, especially building height, density, and age. As illustrated in Figure 4a, the total building material stocks in Nanchang’s built-up areas exhibit a multi-centric and uneven distribution, with Honggutan CBD and the old town core as the primary high-value clusters. An analysis of the spatial distribution of high-value building material stocks reveals that buildings in these areas are generally tall and dense, and typically have a relatively new age profile. In contrast, low-value areas for building material stocks are dominated by low-rise buildings, which have lower material stock per unit. Figure 4b shows the building material stocks locked in vacant buildings, which are mainly related to housing vacancy rates. The high-value areas of vacant material stocks closely coincide with the hotspots of housing vacancy rates.

3.3. Carbon Footprint

The embodied carbon emissions of nine building material types were quantified (Figure 5a and Figure S3 in the Supplementary Materials), including the distribution of embodied carbon emissions from the production of materials locked in vacant buildings (Figure 5b). The results showed that the total embodied carbon emissions from building material stocks in Nanchang in 2021 reached 264 Mt, with 104 Mt (39.39%) attributed to vacant stocks, which is approximately six times the carbon emissions from the city’s annual electricity consumption (28.7 billion kW·h [54]). Among the vacant carbon, 85.6% comes from buildings constructed after 2000.

This substantial carbon footprint, dominated by the embodied emissions from materials, is closely related to Nanchang’s building structure characteristics and construction periods. From the perspective of building structure, the city’s buildings are mainly composed of brick–concrete and steel–concrete structures. Among them, steel–concrete structures (accounting for 52.28% of the total buildings in 2021) have high demand for cement, steel, sand, and gravel, while brick–concrete structures rely heavily on bricks and cement. From the construction period, more than 80% of the buildings were completed after 2000 (Table 1), and this stage coincided with the rapid development of Nanchang’s urbanization, with large-scale construction focusing on high-rise residential buildings and commercial complexes. These buildings not only require more steel for load-bearing structures but also consume a large amount of cement for concrete pouring, further amplifying the carbon emission contribution of these two materials.

As shown in Figure 6a, the spatial distribution of the total carbon footprint closely corresponds to the total stock of building materials, with higher values in areas with greater material stocks. Specifically, the high-carbon-emission clusters are mainly concentrated in two core areas: one is the old town (including Bayi Square and Tengwang Pavilion surrounding areas), where the dense distribution of brick–concrete buildings (built in the 1980s–1990s) has accumulated a large amount of cement and brick carbon emissions; the other is the Honggutan CBD, where a large number of steel–concrete super high-rise buildings (such as Nanchang Greenland Central 303 Twin Towers) have led to a surge in steel and cement carbon emissions.

Significant disparities in carbon emission distribution are observed between the eastern and western riverbanks, particularly in emissions from vacant buildings, with the western bank showing markedly higher emission intensity, strongly correlating with its higher vacancy rate (Figure 6b). The western bank (mainly Honggutan New Area and Jiulong Lake Area) has a vacant building carbon emission of 47 Mt, accounting for 45.19% of the city’s total vacant carbon emissions; while the eastern bank (old town and Nanchang County surrounding areas) has a vacant carbon emission of 57 Mt, accounting for 54.81% of the total. It is worth highlighting that the Honggutan CBD has a total carbon emission of 12 Mt, of which vacancy-related carbon emission is 9 Mt, accounting for 75% of the CBD’s total carbon emission.

However, significant disparities were found in their spatial distribution among the different materials due to variations in factors such as material intensity and stock distribution (Figure 6 and Figure S3 in the Supplementary Materials). For instance, Figure 6a shows that the Honggutan CBD and the old town had high steel-related carbon emissions, primarily due to substantial steel consumption. In contrast, the carbon footprint of bricks had a more uniform spatial distribution, attributable to their high material intensity in both brick–concrete and steel–concrete structures (Figure 6b).

4. Discussion

4.1. Spatial Patterns and Driving Mechanisms of the “Scale Expansion Trap”

The serious vacancy of building stock in Nanchang is a direct consequence of the “scale expansion trap” in cross-river urbanization—large-scale construction after 2000 has not been matched by effective demand, resulting in massive material lock-in. A spatial analysis of Nanchang’s housing vacancy rate pattern highlights that the old urban area, as the early core of urban development, is characterized by complete infrastructure, high population density, and well-equipped living facilities. Residents’ housing needs are well satisfied here, leading to a relatively low vacancy rate. After Nanchang initiated cross-river development, Honggutan District experienced rapid development following the relocation of the urban administrative center and the construction of a large number of commercial and office projects. However, excessive planning in some areas led to an oversupply of commercial and office spaces, which exceeded market demand, resulting in high vacancy rates for office buildings and shops, as shown in Figure 3b. Additionally, in newly built high-end residential areas, high housing prices turned many units into investment properties, further increasing the vacancy rate. The impact of infrastructure construction on vacancy distribution patterns indicates that in newly developed areas, residential buildings have low occupancy rates due to a lack of schools, hospitals, and supermarkets, despite their modern architectural design. Regarding the correlation between industrial development and vacancy rates, Honggutan CBD, dominated by the agglomeration of financial and commercial enterprises, is characterized by super high-rise office buildings. However, fierce market competition and economic factors have caused some enterprises to fail or delay occupancy, increasing office vacancy rates. In contrast, the old town, with its strong commercial atmosphere, comprises traditional commercial streets and small-scale businesses, featuring diversified housing uses and relatively low vacancy rates.

4.2. Quantifying “Developmental Vacancy” and Its Resource Consequences

The characteristics of building materials in Nanchang’s vacant housing represent a material projection of its “urbanization model”: the strategy of “scale expansion priority” during cross-river development has led to a substantial accumulation of new materials in vacant buildings, while lagging organic renewal in the old town has hampered the efficient utilization of traditional materials. Furthermore, this pattern gives rise to the “vacant carbon lock-in” phenomenon—an environmental consequence of the “scale expansion trap.” The large-scale construction of brick–concrete and steel–concrete buildings has generated massive embodied carbon emissions, yet high vacancy rates prevent these emissions from realizing their corresponding service value. This high proportion of vacant-related carbon emissions underscores the inefficiency of Nanchang’s urban construction model. Despite accelerated cross-river development and new area construction, lagging population and industrial agglomeration have left numerous buildings vacant, resulting in carbon lock-in. This situation creates a double loss: resources are wasted, and the environment bears the cost of emissions without receiving the benefits of usable services. Compounding this issue, 85.6% of the vacant carbon comes from buildings constructed after 2000, directly highlighting the severe environmental cost of blind scale expansion.

The spatial pattern of carbon emissions (Figure 5 and Figure 6) reflects the “dual-core” structure of Nanchang’s urban development. The old town, as the historical core, has accumulated carbon emissions through long-term construction, while the Honggutan New Area, as the new urban core, has emerged as a high-carbon zone due to large-scale new construction. The disparity between the eastern and western riverbanks is a spatial manifestation of the mismatch between development strategy and actual demand. The western bank has seen heavy investment in high-end residential and commercial buildings, but insufficient public services and slow industrial agglomeration have led to high vacancy rates and locked-in carbon emissions.

The Honggutan CBD serves as an extreme case that fully demonstrates this failure. Here, nearly three-quarters of the carbon emissions are in an idle state, representing a severe waste of carbon resources. This case exemplifies the flaw of a scale-oriented development model, where prioritizing scale and image over actual demand leads to the typical manifestation of the “scale expansion trap” in cross-river urbanization.

4.3. Policy Implications

For future urban construction, urban planners should rationally plan the proportion of residential, commercial, and office land in line with market demand and population growth trends to avoid oversupply and reduce vacancy.

To operationalize this planning principle and address the documented issues of material stock and carbon lock-in, a comprehensive management framework is essential. Urban managers should implement “material flow management” (e.g., material registration for vacant buildings and circular utilization indicators) to balance construction scale with efficiency. Furthermore, from the perspective of carbon emission reduction, Nanchang’s building material carbon footprint can be addressed through two key pathways: On the one hand, for existing buildings, the focus should be on renovating vacant buildings to release locked-in carbon emissions. For example, vacant commercial buildings in the Honggutan CBD can be transformed into rental housing or shared offices through functional conversion, thereby activating their idle carbon emissions. Similarly, old buildings in the old town can undergo low-carbon retrofits, such as replacing traditional cement mortar with low-carbon alternatives and adding thermal insulation layers to reduce operational energy consumption. On the other hand, for new buildings, the priority is to optimize building material structures and promote low-carbon materials. This includes, under the premise of ensuring structural safety, increasing the proportion of timber and other low-carbon materials, and promoting the use of recycled aggregate concrete to reduce emissions from cement and gravel.

To support these strategies, a robust monitoring mechanism is needed. The carbon emission management of vacant building materials requires a full-life-cycle tracking system. Specifically, a “vacant building carbon account” can be established using the GIS-MSA integrated method. This account would record the material type, quantity, and embodied carbon of each vacant building, enabling the formulation of differentiated reduction measures based on vacancy duration and building type.

4.4. Comparative Analysis and Limitations

In 2021, the per capita stock of building materials in Nanchang was approximately 204 tons per person. Compared to some cities in developing countries, such as Ezhou [55], Shanghai [19], Chiclayo [56], and Beijing [20], the per capita level of building environment stock in Nanchang was slightly higher. This relatively high per capita material stock, combined with low total stock and high vacancy rate, reflects the inefficiency of Nanchang’s urban construction model—excessive investment in building scale per capita without sufficient economic and population support, a key feature of the “scale expansion trap” in underdeveloped cross-river cities. However, in terms of the total building environment stock, the value of Nanchang was significantly lower than some more developed cities, including Wakayama city center [36], Vienna [37], Padua [57], and Beijing [20]. This disparity underscores notable differences in socio-economic development and population trends among different cities.

Admittedly, the database presented in this work also had uncertainties that may affect the accuracy of HVR-related analyses. First, the HVR data used in this study are completely cited from Shi et al. [16], which are based on block-level census data. This data granularity can only reflect the average HVR of a regional unit (e.g., a neighborhood) and cannot capture the differences in vacancy status between individual buildings (e.g., the HVR of high-end residential buildings vs. ordinary residential buildings in the same block). Second, the existing HVR calculation does not distinguish between “temporary vacancy” (e.g., houses under decoration, short-term unoccupied rental houses) and “long-term vacancy” (e.g., investment properties left unoccupied for more than one year). In Nanchang, where investment-driven housing construction is prominent, long-term vacancy is the main contributor to the high overall HVR, but the current data cannot quantify this proportion, which limits the formulation of targeted vacancy reduction policies. Third, the lack of dynamic HVR data (e.g., annual changes in HVR from 2010 to 2021) makes it impossible to analyze the coupling relationship between HVR evolution and urban development stages (e.g., whether the HVR decreases with the maturity of new areas), thus affecting the prediction of future vacancy trends.

In the future, to improve the accuracy and depth of HVR research, three aspects of work can be carried out. First, integrate multi-source big data to build a building-level HVR calculation model: use water and electricity consumption data (monthly consumption < 10 kW·h for vacant houses), property management records, and even high-resolution remote sensing images (to identify unlit rooms at night) to calculate the HVR of individual buildings, thereby realizing fine-grained identification of vacancy hotspots. Second, classify HVR by vacancy duration and housing type: through household surveys and follow-up visits, distinguish temporary vacancy from long-term vacancy, and clarify the HVR differences between residential, commercial, and office buildings, which can provide a more accurate basis for formulating differentiated policies (e.g., converting long-term vacant commercial buildings into rental housing, optimizing the supply structure of residential buildings). Third, establish a dynamic monitoring database of HVR: by updating HVR data annually, analyze the temporal and spatial evolution law of HVR and its correlation with factors such as population inflow, industrial structure adjustment, and public service improvement, so as to provide scientific support for optimizing urban construction plans and reducing vacant buildings.

In summary, the high HVR and its extreme spatial heterogeneity in Nanchang are typical manifestations of the “developmental vacancy” faced by less-developed cities during rapid urbanization. Resolving this problem requires not only adjusting the urban construction model from “scale expansion” to “quality improvement” but also strengthening the matching between housing supply, population agglomeration, and industrial development through precise urban planning and policy guidance. At the same time, improving the accuracy and granularity of HVR data is an important prerequisite for formulating effective vacancy reduction strategies and promoting the efficient use of building materials.

5. Concluding Remarks

This study integrated MSA and GIS techniques to construct a high-spatial-resolution database of Nanchang’s urban building stock, systematically uncovering the spatial patterns of building stocks, housing vacancy distribution, and embodied carbon emissions in this underdeveloped cross-river city. The findings not only fill the gap in fine-scale research on built environment stocks and housing vacancy in less-developed Chinese cities but also provide empirical insights for addressing “developmental vacancy” and carbon lock-in challenges during rapid urbanization.

First, Nanchang’s building stock presents a Ganjiang River-centered butterfly-shaped spatial pattern, with horizontal expansion (over 40-fold urban area growth since 1949) and vertical growth (ten-fold increase in maximum building height) occurring simultaneously. The old town maintains the highest building density (dominated by 1980s–1990s brick–concrete structures), while the Ganjiang River riparian areas, especially Honggutan CBD, has the tallest buildings (concentrated steel–concrete super high-rises). This pattern reflects the typical “cross-river development” trajectory of underdeveloped cities, where administrative center relocation and new area construction drive spatial restructuring, yet lag behind in population and industrial agglomeration.

Second, the building material stock in Nanchang reached 1034 Mt (204 t/cap), with over 85% accumulated post-2000. It indicates a phase of explosive construction during rapid urbanization. Sand (31.33%), gravel (30.99%), and brick (17.43%) dominate the material composition, while vacant buildings lock in 405 Mt of materials (39.17%). This implies nearly 40% of building materials are in low-efficiency or idle service, a direct reflection of the scale expansion trap. Spatially, building stock high-value clusters align with two cores, which are the old town (dense brick–concrete structures) and Honggutan CBD (steel–concrete high-rises), respectively. Notably, housing vacancy exhibits significant spatial heterogeneity: higher rates in the urban periphery than the old town, and on the Ganjiang River’s west bank (Honggutan, Jiulong Lake) than the east bank.

Third, Nanchang’s building material embodied carbon emissions totaled 264 Mt in 2021, with 104 Mt (39.39%) attributed to vacant stocks. This “vacant carbon lock-in” stems from mismatched urban construction and actual demand: large-scale steel–concrete building construction (52.28% of total buildings) drives high emissions from cement, steel, sand, and gravel, yet slow population/industrial agglomeration leaves buildings vacant. Spatially, high-carbon clusters overlap with material stock hotspots: the old town (cement–brick emissions from brick–concrete buildings) and Honggutan CBD (steel–cement emissions from super high-rises), with the west bank accounting for 45.19% of vacant carbon emissions.

Building on the empirical findings of high vacancy rates and carbon lock-in, the implications of this study extend beyond Nanchang. The identified “scale expansion trap” offers a cautionary framework for underdeveloped cross-river cities globally, particularly those relying on land-finance models to drive physical expansion. To avoid the inefficient lock-in of resources and carbon, urban planning must shift from a paradigm of scale expansion to that of quality integration. This entails (1) synchronizing land supply and new district construction with population and industrial growth trends, (2) implementing material flow management and circular economy principles to activate vacant building stocks, and (3) prioritizing low-carbon building materials and designs to mitigate embodied carbon.

For future research, integrating multi-source big data (e.g., water–electrical consumption, nighttime light remote sensing) to establish building-level dynamic vacancy monitoring will further refine vacancy classification (temporary vs. long-term) and improve policy targeting. Overall, this study demonstrates that high-spatial-resolution built environment stock analysis is a powerful tool for diagnosing urban sustainability challenges; the framework and findings can be extended to other underdeveloped cities to avoid the “scale expansion trap” and advance resource-efficient, low-carbon urban transitions.