The Carbon Emissions from Public Buildings in China: A Systematic Analysis of Evolution and Spillover Effects

Abstract

Public buildings, being the primary source of carbon emissions in China’s construction sector, present a pressing need for emission control. This imperative task not only ensures the sustainable progression of China’s building industry but also holds pivotal significance in the realm of global energy conservation and emission curtailment. This study analyzed the spatiotemporal evolution of carbon emissions from public buildings in China and assessed the spatial influence of related factors using a dataset covering 30 provincial units from 2006 to 2021. The analysis employed Theil’s index, Moran’s I index, standard deviation ellipse, and the spatial Durbin model. The study revealed an upward trajectory in carbon emissions from public buildings in China, although the growth rate was generally decreasing. Disparities in emission distribution among provincial units stem largely from intra-regional distinctions, notably prominent in the Low-Intensity High-Economy regions. Provincial carbon emissions from public buildings exhibited significant spatial correlation, manifesting as clusters of high–high and low–low patterns, indicative of mutual influence among adjacent areas. Additionally, the shift in carbon emission focal points from the northeast to the southwest underscored a more pronounced surge in the southwestern regions. Variables such as energy intensity, energy structure, per capita public building area, disposable income per capita, openness level, and environmental governance directly impact carbon emissions from public buildings. Among these, energy intensity, energy structure, disposable income per capita, and environmental governance also had spatial spillover effects. These findings provide a scientific reference and a foundation for policy-making, aiding local administrations in crafting strategies to mitigate carbon emissions from public buildings and foster sustainable progress.

1. Introduction

In recent years, rising temperatures and the trend of global warming have intensified rapidly, posing serious challenges to sustainable development, such as the frequent occurrence of extreme weather and rising sea levels, and addressing climate change has emerged as a critical global concern. As outlined in the sixth Synthesis Report from the United Nations Intergovernmental Panel on Climate Change (IPCC), the average surface temperature worldwide has surged by 1.1 °C between 2011 and 2020 in comparison to the period spanning 1850 to 1990. The report advocates for constraining global warming to below 1.5 °C relative to pre-industrial levels [1]. Since 2009, China has outpaced the United States to become the leading emitter of carbon dioxide globally [2]. To combat this, China introduced a “dual carbon” objective during the 2020 United Nations General Assembly, targeting carbon peaking by 2030 and achieving carbon neutrality by 2060 [3].

Greenhouse gases are the main cause of temperature rise, predominantly stemming from the combustion of fossil fuels and energy utilization, and buildings, alongside industry and transportation, rank among the key triad of energy-consuming sectors [4]. Buildings contribute to 39% of total carbon emissions in the United States [5], 25% in Latin America [6], and up to 60% in Hong Kong [7]; within the European Union, buildings account for 59% of total electricity consumption [8]. In 2021, China’s total building process carbon emissions reached 5.01 billion t CO₂, accounting for 47.1% of the national energy-related carbon emissions, with public buildings exhibiting higher energy intensities compared to other building categories. In Australia, commercial and office buildings alone consume approximately 60% of total energy, while in Singapore, commercial electricity consumption accounts for about 37% of the overall electricity consumption [7]. China’s public buildings consume 42% of the energy and contribute 41% of the carbon emissions in the operation phase with only 21% of the floor area. This underscores their pivotal role as a primary source of energy consumption and carbon emissions in the operational sphere [9].

Public buildings have the largest carbon emission potential among all building types [10], yet prevailing research predominantly concentrates on the construction sector holistically, sidelining this critical facet; at present, China is actively optimizing its industrial framework and developing the service sector, and the booming development of the service industry will also drive the development of public buildings. Therefore, a comprehensive analysis of carbon emissions from public buildings in China is needed to effectively assess the factors that impact public buildings in China, and to formulate carbon reduction strategies for public buildings, thereby reinforcing the adequacy of China’s carbon emission control strategies.

The organization of the subsequent chapters of this research is described below. In Section 2, an extensive review of carbon emissions is undertaken. Section 3 furnishes details on the methodology employed and the sources of data. Section 4 presents the specific findings of the study and provides an in-depth discussion based on them. Section 5 summarizes the findings of this study and explores their potential applications and implications for practical policymaking.

2. Literature Review

2.1. Spatial and Temporal Analysis of Carbon Emissions

Spatio-temporal analysis is a method of comprehensively considering the spatial correlation in spatial data as well as the randomness and complexity of the temporal dimension. Scholars across diverse fields have employed various research methodologies to scrutinize the spatio-temporal aspects of carbon emissions.

Within the realm of investigating the spatial and temporal characteristics of carbon emissions, a notable direction is to explore the correlation between their spatial agglomeration characteristics and inter-region. Researchers have revealed the spatial distribution patterns of carbon emissions in different industries through exploratory spatial data analysis such as the Moran Index. These studies show that there is not only significant spatial agglomeration of carbon emissions but also interdependence and influence of carbon emissions between different regions. For example, there is a high degree of agglomeration of carbon emissions in agriculture [11], and an escalating spatial correlation of emissions across industrial sectors [12].

The analysis of regional differences and heterogeneity is another important area of spatio-temporal research on carbon emissions. Scholars have explored regional disparities and heterogeneities in carbon emissions in terms of population, economy, space, and other dimensions [13] through methods such as the Gini coefficient, Thile index, and kernel density estimation. Studies have delineated China’s emission spatial distribution marked by an “east high, central and west low” trend [14], and that there are significant differences in carbon emissions within and among different regions [15].

Regarding dynamic evolution and trend analysis, researchers have utilized diverse methods such as the gravity model, cold and hot spot analysis, and the standard deviation ellipse technique to unveil the spatial and temporal evolution characteristics of carbon emissions and forecast future trends. Observations indicate a northwest shift in China’s carbon emissions center of gravity, surpassing the pace of urbanization and industrialization [16]. Furthermore, evolution trends in carbon emissions across various industries exhibit diversification, exemplified by the transition in carbon emissions distribution within the power sector from northeast–southwest to east–west [17]; the “H–H” concentration pattern of carbon emissions in the service industry, which is spreading from the east to the central region [18].

2.2. The Influencing Factors and Spillover Effects of Carbon Emissions

In the field of researching the influencing factors of carbon emissions, scholars have adopted a variety of models and methods to explore the key factors related to carbon emissions in depth, which also provides a certain foundation for the research of this paper. Currently, prevalent models employed to identify the factors influencing carbon emissions predominantly encompass the factor decomposition approach [19] and econometric modeling analysis. Factor decomposition methods mainly include Index Decomposition Analysis (IDA) and Structural Decomposition Analysis (SDA), with the Log Mean Divisor Index (LMDI) method standing out as the most favored within IDA [20]. Relevant scholars use the LMDI model to explore the differential impact of different drivers on carbon emissions [21,22]. Shi et al. used SDA to conduct an in-depth study on the influencing factors of the dynamics of carbon emissions in the construction industry and their contributions [23]. The econometric models mainly include OLS, STIRPAT, GMM, DID, etc. Hussain et al. used the STIRPAT model to scrutinize the carbon emissions of the countries along the “Belt and Road” route. Chen, Wang, Fan, and Guan et al. studied the carbon emissions of China’s provincial and county level, evaluating the primary drivers of CO₂ emissions [24,25,26,27]. Moreover, Kais and Dong et al. empirically analyzed the carbon emission influencing factors of 58 countries and the Yangtze River Economic Belt, respectively, by using the GMM model [28,29]. The advantage of the econometric model is that it can explore the mechanism and influence of the independent variables on carbon emissions. However, the carbon effect is independent between regions, which is inconsistent with the specific carbon emission, weakening the relevance of the conclusion [12].

Based on the abovementioned issues, some scholars have introduced spatial correlation effects and conducted empirical research using spatial econometric models, proving the existence of spatial spillover effects in carbon emissions. If this spillover effect is ignored in actual analysis, it will lead to inaccurate analysis results [30,31,32]. Spatial econometric models mainly include spatial Durbin model (SDM), spatial lag model (SAR), and spatial error model (SEM). SDM takes into account the spatial lag effect of the dependent and independent variables at the same time, and it can reveal the intrinsic laws and interrelationships of the spatial data in a more comprehensive way. In the analysis of factors affecting carbon emissions, it is possible to more comprehensively reveal the inherent laws and interrelationships of the data, which helps to prevent the impact of spatial spillover effects on endogeneity, bolster the robustness of the analysis results, and provide a more accurate basis for decision-making. Following a systematic review of existing literature, it has become apparent that further exploration is essential concerning the spatiotemporal effects of carbon emissions originating from public buildings. Existing studies on the factors influencing carbon emissions from public buildings have predominantly focused on heterogeneity [33,34], but have not fully accounted for spatial spillover effects. Given that the spatial measurement models are more reliable in analyzing the influencing factors, leveraging the spatial measurement models in analyzing these influencing factors remains a valuable pursuit.

The primary contributions of this study can be summarized in the following three key aspects. First, this study not only clarified the method of carbon emission calculation in the operation stage of public buildings but also deeply explored the spatial disparities and correlations of carbon emissions from public buildings in China, which helps policymakers take into account regional interdependencies and regional differences when formulating sustainable development policies. Second, the evolution characteristics of the spatial pattern of carbon emissions from public buildings in China’s interprovince were analyzed by applying the standard elliptical deviation, offering a valuable reference for decision-makers to identify key areas for reducing carbon emissions in public buildings. Finally, by incorporating spatial factors into the general econometric model and utilizing a spatial econometric model, this study explored the influence of related factors on carbon emissions and analyzed the spillover effects of each influential factor. This approach enables policymakers to thoroughly consider the impact of neighboring regions when formulating policies, thereby setting scientifically sound carbon emission reduction targets and contributing to broader sustainable development objectives.

3. Method and Data

3.1. Methodology

3.1.1. Carbon Emissions Measurement

Public buildings’ operational energy consumption encompasses heating, air-conditioning, lighting, elevators, comprehensive service equipment and facilities, etc. Within this research, the carbon emissions of public buildings were assessed using the IPCC technique [35], which relies on end-sector energy usage for measurement.

This paper focuses on carbon emissions during the operational phase of public buildings, with China’s Energy Statistics Yearbook serving as the source of data on carbon emissions from public buildings. The spectrum of industries encompassed within public buildings comprises transportation, storage, postal services, wholesale and retail trade, accommodation, and food services, among others. Energy consumption measurements span across electricity, gas (natural gas, gas, and liquefied petroleum gas), fuel oil (specifically diesel), and coal combustion. Notably, a proportional allocation is made for oil consumption, primarily attributed to automobile transportation within public buildings [36]. Adhering to the assessment methodology outlined by the IPCC, direct sources of carbon emissions from public buildings include raw coal, coke, gasoline, kerosene, diesel, LPG, and natural gas, and heat and electricity are acknowledged as indirect sources of carbon emissions. The specific calculation range is detailed in

Table 1. Calculation boundaries for carbon emissions from public buildings.

Renewable Energy		Transportation, Storage and Postal Services	Wholesale and Retail Trade and Accommodation and Restaurants	Else
direct source	raw coal	0%	100%	100%
	coke (processed coal used in the blast furnace)	0%	100%	100%
	diesel	0%	5%	5%
	gasoline	0%	100%	100%
	diesel fuel	0%	65%	65%
	liquefied petroleum gas	0%	100%	100%
	petroleum	0%	100%	100%
indirect source	electrical power	100%	100%	100%
	thermodynamic	100%	100%	100%

.

$$\mathrm{C}{\mathrm{O}}_2=C_{dir}+C_{ind}=\left({\sum }_{i=1}^7{C}_i\times S_i\times {EF}_i\right)\times {\displaystyle \frac{44}{12}}+\left(C_e\times {EF}_e+C_h\times {EF}_h\right)$$

(1)

$$\mathrm{C}{\mathrm{O}}_2=C_{dir}+C_{ind}=\left({\sum }_{i=1}^7{C}_i\times {CAL}_i\times {CC}_i\times {CO}_i\times {\displaystyle \frac{44}{12}}\right)+\left(C_e\times {EF}_e+C_h\times {EF}_h\right)$$

(2)

where $\mathrm{C}{\mathrm{O}}_2$ represents the aggregate public building carbon emissions (PBTCE), and $C_{dir},C_{ind}$ delineate direct and indirect carbon emissions, respectively. $C_i$ signifies the primary energy consumption, while $S_i$ and ${EF}_i$ stand for the standard coal conversion factor and carbon emission factor of energy, correspondingly. $C_e$ and $C_h$ represent electricity and heat consumption, whereas ${EF}_e$ and ${EF}_h$ denote the carbon emission factors associated with electricity and heat, respectively. Among them, the carbon emission factor for electricity is typically sourced from government or relevant departments’ standardized data repositories, ensuring ease of access and authoritative data origins. ${EF}_e$ uses the National Development and Reform Commission’s “Average Emission Factors of China’s Regional and Provincial Power Grids in 2010” and “Average $\mathrm{C}{\mathrm{O}}_2$ Emission Factors of Provincial Power Grids in 2012”; the “Letter on the Business Request to Provide the Self-Assessment Report on the Implementation of the Provincial People’s Government’s Responsibility for the Goal of Controlling Greenhouse Gas Emissions in 2018”; and the Ministry of Ecology’s Environmental Environment and Environmental Planning Institute’s “Study on $\mathrm{C}{\mathrm{O}}_2$ Emission Factors for Regional Power Grids in China 2023” in the average value. ${EF}_h$ refers to the calculation method provided in the “Research Report on Energy Consumption of Buildings in China (2018)” [37]; for the measurement, ${EF}_h=$ total carbon emissions from various heating energy sources/total thermal energy production.

${CAL}_i$ represents the low calorific value of energy type $i$. ${CC}_i$ denotes the default carbon content of energy type $i$. ${CO}_i$ reflects the carbon oxidation rate of energy type $i$.

Per capita public building carbon emissions (PBPCE) is the ratio of total public building carbon emissions to the population of the area, and is calculated by the following formula:

$$\mathrm{P}\mathrm{C}{\mathrm{O}}_2={\displaystyle \frac{\mathrm{C}{\mathrm{O}}_2}{\mathrm{P}}}$$

(3)

3.1.2. Theil Index

The Theil Index serves as a widely used metric for gauging inequality, assessing the extent of disparity in income or wealth distribution within a specific group or economic framework. A heightened Theil Index signifies amplified inequality [38]. By decomposing the Theil Index into intra-group and inter-group indices, the inter-regional and intra-regional differences in carbon emissions from public buildings and their contribution to the overall differences can be comprehensively revealed, and this analytical method can make up for the limitations brought by the hierarchical differences and different weights to the spatial analysis of carbon emissions, fostering a more holistic and precise evaluation of spatial heterogeneity [39].

In this study, we established a two-dimensional panel framework to divide China’s 30 provinces into regions based on GDP per capita and carbon intensity (carbon emissions from public buildings/added value of the tertiary industry), which can realize the pursuit of carbon emission reduction without affecting economic development [40]. Finally, the 30 provinces and municipalities in China were divided into four regions. For details, see

Table 2. Regional division.

Region	Provinces
High-Intensity High-Economy I	Tianjin, Inner Mongolia,
Low-Intensity High-Economy II	Beijing, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Hubei, Guangdong, Chongqing
Low-Intensity Low-Economy III	Anhui, Jiangxi, Henan, Hunan, Guangxi, Hainan, Sichuan, Yunnan, Gansu
High-Intensity Low-Economy IV	Hebei, Shanxi, Liaoning, Jilin, Heilongjiang, Guizhou, Shaanxi, Qinghai, Ningxia, Xinjiang

. The calculation formula for the Theil index is provided below:

$$T={\displaystyle \frac{1}{k}}{\sum }_{\mathrm{j}=1}^k\left({\displaystyle \frac{C_j}{\overline{C}}}\times \mathrm{l}\mathrm{n}{\displaystyle \frac{C_j}{\overline{C}}}\right)=T_{WR}+T_{BR}$$

(4)

$$T_i={\displaystyle \frac{1}{k_i}}{\sum }_{\mathrm{j}=1}^{k_i}\left({\displaystyle \frac{C_{i\mathrm{j}}}{\overline{C_i}}}\times \mathrm{l}\mathrm{n}{\displaystyle \frac{C_{i\mathrm{j}}}{\overline{C_i}}}\right)$$

(5)

Intra-regional variations

$$T_{WR}={\sum }_{i=1}^4\left({\displaystyle \frac{k_i}{k}}\times {\displaystyle \frac{\overline{C_i}}{\overline{C}}}\times T_i\right)$$

(6)

Interregional differences

$$T_{BR}={\sum }_{i=1}^4\left({\displaystyle \frac{k_i}{k}}\times {\displaystyle \frac{\overline{C_i}}{\overline{C}}}\times ln{\displaystyle \frac{\overline{C_i}}{\overline{C}}}\right)$$

(7)

The general Theil index for PBPCE is represented by $T$, $j$ denotes the province, $k$ signifies the number of provinces, $C_j$ depicts the public PBPCE of province $j$, and $\overline{C}$ denotes the national mean of PBPCE; $T_i$ denotes the overall variance Theil index of the synthesis of region $i$, and $k_i$ denotes the number of provinces and cities in region $i$. $C_{ij}$ denotes PBPCE of the $j$th provincial administrative unit in region $i$. ${\mathrm{C}}_i$ denotes the average value of PBPCE in region $i$. In order to delve deeper into the impact of intra-regional disparities and inter-regional disparities on carbon emissions from public buildings in China, the intra-regional contribution rate is defined as ${\displaystyle \raisebox{1ex}{$T_{WR}$}\!\left/ \!\raisebox{-1ex}{$T$}\right.}$ and the inter-regional contribution rate is ${\displaystyle \raisebox{1ex}{$T_{BR}$}\!\left/ \!\raisebox{-1ex}{$T$}\right.}$.

3.1.3. Moran’s Index

Moran’s I index stands as a pivotal metric for discerning spatial autocorrelation [41]. The index can be subdivided into global Moran’s I index and local Moran’s I index predicated on the study’s spatial scope. The global Moran’s I index is predominantly leveraged to scrutinize the spatial distribution patterns across an entire region, whereas the local Moran’s I index delves into uncovering the disparities within specific locales [42]. In order to make the research results more comprehensive, this paper selected the global Moran’s I index and local Moran’s I index to analyze the spatial correlation of carbon emissions from public buildings.

(1)

Global spatial autocorrelation index:

$$\mathrm{I}={\displaystyle \frac{{\sum }_{i=1}^n{\sum }_{j=1}^n{\omega }_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{S^2{\sum }_{i=1}^n{\sum }_{j=1}^n{\omega }_{ij}}}$$

(8)

$$S^2={\displaystyle \frac{1}{n}}{\sum }_{n=1}^n\left(x_i-\overline{x}\right)$$

(9)

where $x_i$ and $x_j$ are the carbon emissions stemming from public building construction within the $i$th and $j$th provincial administrative units, respectively. ${\omega }_{ij}$ denotes the spatial weighting matrix. We selected the spatial adjacency matrix (0–1 matrix), and Moran’s I index is constrained within the interval of [−1,1]. If I > 0, there is a positive spatial dependence, and the dependence of the spatial distribution increases with the value; if I < 0, the spatial dependence is negative, and if I = 0, there is no spatial correlation, showing a random distribution in space.

(2)

Local spatial autocorrelation index

By utilizing a local Moran’s I scatterplot, detailed dependencies between the observed region and its neighboring regions can be elucidated, complementing the global perspective provided by the global Moran’s I index [42]. The scatterplot representing the global Moran’s I index was segmented into four quadrants, each conveying distinct implications as delineated in

Table 3. Types of spatial clustering.

Region	Type of Agglomeration	Hidden Meaning
I	High–high	Spatially positively correlated; areas with high carbon emissions neighboring
II	Low–high	Spatially negative correlation; areas with low carbon emissions are adjacent to areas with high carbon emissions
III	Low–low	Spatially positively correlated; areas with low carbon emissions are adjacent to areas with low carbon emissions
IV	High–low	Spatially negative correlation; areas with high carbon emissions are adjacent to areas with low carbon emissions

.

3.1.4. Standard Deviation Ellipse

In spatial analysis, the standard deviation ellipse, initially introduced by Lefever in 1926 [43], serves as a spatial analysis technique employed to quantify the directional and distributional attributes of a dataset by utilizing parameters such as the ellipse’s center, standard deviation along the major and minor axes, the rotation angle, etc., and to intuitively reveal the spatial distribution feature and evolutionary trends of the public carbon emissions by identifying the changes of the center position and the direction of movement of the data [44,45,46].

3.1.5. Spatial Econometric Model

The spatial econometric model, an extension of traditional econometric models, integrates geospatial considerations to analyze spatial correlation and dependence statistically [47]. The SDM model serves as the fundamental framework of spatial econometric analysis, introduces spatial lag terms and spatial error terms to capture the mutual influence and dependence between geographically neighboring regions, solves the problem of spatial autocorrelation, and decomposes the coefficients into the direct effect and spillover effect through the partial differentiation method, which better explains the spatial spillover effect of the variable, improves the accuracy and robustness of the estimation results [48,49], and provides more accurate policy recommendations.

The basic steps are shown in Figure 1.

(1)

Initially, a spatial autocorrelation examination is conducted to validate the appropriateness of the spatial measurement model. On this basis, the traditional mixed panel model is first estimated using OLS. The spatial autocorrelation of the residuals is then tested using LM-lag, LM-error, and its robust forms (R-LMlag, R-LMerror);

(2)

Based on the results of the abovementioned inspection, determine the applicability of SAR and SEM. If the results of LM lag, LM error, R-LMlag, and R-LMerror statistics are all significant, SDM is suitable for the problem that needs to be studied; Otherwise, select SAR or SEM based on the corresponding results;

(3)

After initially determining the model, the Hausman test is employed to differentiate fixed effects from random effects;

(4)

Finally, verify whether the nested SDM model can be simplified into SAR and SEM models. Use the Wald test and LR test. If the result exceeds the significance level of 5%, SDM cannot be simplified as SLM or SEM; otherwise, it can be simplified as SAR or SEM.

Due to the individual differences between regions and the time factor [41], and considering that the regional public building carbon emission level is influenced not only by neighboring regions but also by the economic and technological factors of other areas, this study initially selected the nested spatial and temporal two-way fixed effect SDM model to analyze the impact of key factors on the carbon emission of China’s public buildings. Subsequent examinations were conducted to validate the model’s accuracy. The SDM model expression used in this paper is as follows:

$$Y_{it}=\rho W{Y}_{it}+\beta X_{it}+\theta W{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(10)

$$W_{ij}=\left\{\begin{array}{c}1,\mathrm{G}\mathrm{e}\mathrm{o}\mathrm{g}\mathrm{r}\mathrm{a}\mathrm{p}\mathrm{h}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{l} \mathrm{u}\mathrm{n}\mathrm{i}\mathrm{t} \mathrm{I} \mathrm{i}\mathrm{s} \mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{t}\mathrm{o} \mathrm{o}\mathrm{r} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{n}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d} \mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h} \mathrm{J} \\ 0, \mathrm{G}\mathrm{e}\mathrm{o}\mathrm{g}\mathrm{r}\mathrm{a}\mathrm{p}\mathrm{h}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{l} \mathrm{u}\mathrm{n}\mathrm{i}\mathrm{t} \mathrm{I} \mathrm{a}\mathrm{n}\mathrm{d} \mathrm{J} \mathrm{a}\mathrm{r}\mathrm{e} \mathrm{n}\mathrm{o}\mathrm{t} \mathrm{a}\mathrm{d}\mathrm{j}\mathrm{a}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{o}\mathrm{r} \mathrm{h}\mathrm{a}\mathrm{v}\mathrm{e} \mathrm{n}\mathrm{o} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{n}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\end{array}\right.$$

(11)

where $Y_{it}$ represents the carbon emission from public buildings in year $t$ within the $i$ th provincial administrative unit. The spatial autoregressive coefficient is denoted by $\rho$.$W$ denotes the spatial weight matrix, with the geographic neighbor matrix chosen for this study.$\beta$ symbolizes the vector of regression coefficients of explanatory variables. γ stands for the coefficient of the spatial lag term. ${\mu }_i$ and ${\lambda }_t$ represent the spatial and temporal fixed effects, respectively. ${\epsilon }_{it}$ denotes the random perturbation term.

3.2. Data Sources

Compared with cross-sectional data, panel data offer the ability to mitigate heteroskedasticity across various individuals or regions and rectify biases stemming from omitted variables [50]. Considering the availability and completeness of data, we utilized panel data encompassing 30 Chinese provinces (excluding Tibet Autonomous Region, Hong Kong Special Administrative Region, Macao Special Administrative Region, and Taiwan Province) over a 16-year period from 2006 to 2021. For some missing data, linear interpolation or annual average growth rate was used for calculation.

3.2.1. Data on Provincial Carbon Emissions

The energy data pertinent to computing the carbon emissions for each province were sourced from the China Energy Statistics Yearbook.

3.2.2. Data on Impact Factors

When selecting the determinants of carbon emissions from public buildings in China, our approach was rooted in the IPAT model [51], with the three levels of population, affluence, and technology as the basic entry points, expanding the dimensions of openness and the environment. Drawing insights from relevant literature, we identified key influencing factors: energy intensity (EI) [19,52], per capita public building area (PCPBA) [34], disposable income of all residents (DPI) [53], energy structure (ES) [54], the level of opening up to the outside world (FDI) [21], and the ability to govern the environment (EM) [41,55] as the explanatory variables. On this basis, the specific influencing factor variables were obtained by combining the background of the operation stage of public buildings and the availability and representativeness of data (see

Table 4. Specification of influencing factors.

Name	Specifications	Unit	Data Sources
Public floor space per capita	The ratio of total public floor space to total population	m3/person	China Statistical Yearbook China Urban Construction Statistical Yearbook
energy intensity	The ratio of total energy consumption consumed during the operational phase of a public building to the total public floor area	Tce/m2	China Energy Statistics Yearbook
energy structure	The ratio of electric energy consumption to total energy consumption	%	China Energy Statistics Yearbook
Disposable income of the population as a whole	Gross disposable income of the population as a whole over a given period of time	ten thousand dollars	China Statistical Yearbook
Egypt’s open-door policy towards the outside world	Amount of foreign direct investment as a ratio of GDP	%	China Statistical Yearbook
Environmental governance capacity	The ratio of investment in environmental pollution control to total GDP	%	China Environmental Statistics Yearbook

).

4. Results

4.1. Time Series Analysis of Carbon Emissions from Public Buildings

To enhance the depiction of the temporal trajectory of PBCE in China, we calculated the annual PBTCE and PBPCE spanning the period from 2006 to 2021, as shown in the following Figure 2 and Figure 3.

The annual PBTCE shows an upward trend in Figure 2. Over the span from 2006 to 2021, the annual PBTCE grew from 314.34 million tons to 1159.24 million tons, with an overall growth rate of 268.79% and an average annual increase of 9.20%. The most rapid growth rate, 22.89%, occurred between 2007 and 2008. This may be related to the economic growth, accelerated urbanization, and large-scale construction of public buildings at that time. In addition, since the commencement of reform and opening-up policies, China’s industrial structures have undergone optimization, and the “three-two-one” industrial pattern has been gradually consolidated [56], and the frequency of use and energy consumption of public buildings have been increasing, which further aggravates the growth of carbon emissions.

PBPCE also showed an upward trend over the 16-year period, growing from an initial 9.90 tons per capita to 28.78 tons per capita, marking a cumulative growth of 190.60% and an average annual growth rate of 7.43% (Figure 3). Again, the fastest annual growth rate of 16.07% was recorded in 2007–2008, after which there was a rapid decline. This may be attributed to the steady growth of China’s economy since the Five-Year Plan, accompanied by the rapid expansion of basic public construction, and the repercussions of the 2008 financial crisis, which impeded the tertiary industry’s progress [57], so the annual growth rate has declined since then.

In addition, there has been a notable deceleration in the growth trajectory of both annual PBTCE and PBPCE. This trend signifies the efficacy of China’s carbon emission reduction strategies amidst the consistent execution of the five-year plan [58]. Another surge in the growth rate in 2021 may be due to the gradual recovery of the economy after the new crown epidemic has been brought under some control.

4.2. Analysis of Spatial Differences in Carbon Emissions from Public Buildings

The average and maximum annual carbon emissions of 30 provinces in China from 2006 to 2019 were computed using the PBTCE and PBPCE calculation formulas (see

Table 5. Annual average PBTBE and annual average PBPCE in each province from 2006 to 2021.

Province	Average Annual PBTCE (Mt CO2)	Annual Max PBTCE (Mt CO2)	Average Annual PBPCE (Tons per Capita)	Annual Maximum PBPCE (Tons per Capita)
Beijing	49.53	62.90	2.42	2.87
Tianjin	17.83	25.53	1.32	1.86
Hebei	47.47	85.04	0.65	1.14
Shanxi	25.16	34.73	0.72	1.00
Inner Mongolia	34.29	49.04	1.41	2.00
Liaoning	27.85	39.41	0.65	0.93
Jilin	17.82	24.46	0.69	1.03
Heilongjiang	30.90	55.04	0.90	1.59
Shanghai	30.33	41.55	1.27	1.67
Jiangsu	41.61	78.02	0.50	0.92
Zhejiang	36.94	61.41	0.62	0.94
Anhui	18.99	40.54	0.31	0.66
Fujian	12.57	22.36	0.32	0.53
Jiangxi	11.17	24.79	0.25	0.55
Shandong	56.82	85.85	0.58	0.84
Henan	31.86	65.10	0.33	0.66
Hubei	21.10	29.91	0.36	0.51
Hunan	22.24	37.96	0.34	0.57
Guangdong	52.74	95.37	0.46	0.75
Guangxi	8.86	15.87	0.18	0.32
Hainan	4.39	8.66	0.46	0.85
Chongqing	9.24	14.97	0.30	0.47
Sichuan	12.24	18.38	0.15	0.22
Guizhou	34.12	50.68	0.92	1.33
Yunnan	6.53	10.86	0.14	0.23
Shaanxi	23.39	38.38	0.61	0.97
Gansu	8.58	17.21	0.34	0.69
Qinghai	3.20	4.86	0.56	0.82
Ningxia	4.57	7.26	0.67	1.00
Xinjiang	16.99	40.84	0.70	1.58

). Analysis of the table reveals that within the PBTCE calculation framework, the three regions of Shandong, Guangdong, and Beijing emerged as the top three in terms of average annual carbon emissions, and their combined contribution accounted for 22.12% of the total average annual carbon emissions of all provinces in China, while Qinghai, Hainan, and Ningxia ranked lowest in both average annual carbon emissions and maximum carbon emissions.

In PBPCE, the results showed a different trend. Beijing, Inner Mongolia, and Tianjin ranked among the top three, with the sum of the three accounting for 21.21% of the annual per capita carbon emissions of all provinces in China. These findings present a comprehensive outlook on the PBCE of China’s provinces under different assessment systems.

There are some differences between the results of PBTBE and PBPCE. By correlating carbon emissions with population size, we can provide a fairer comparison, a more refined assessment, better reveal the differences between different regions, and help formulate targeted policies and measures to reduce carbon emissions. Therefore, this paper selected “per capita carbon dioxide emissions from public buildings” (hereinafter referred to as “carbon emissions from public buildings, PBCE”) as the dependent variable for subsequent analysis and empirical testing.

To visually illustrate spatial disparities in carbon emissions from public buildings, we selected four feature points, 2006, 2011, 2016, and 2021. Employing the natural breakpoint technique within ArcGIS 10.8, the PBCE is categorized into four classes.

As showcased in Figure 4, a noticeable uptrend in regions exhibiting high emissions is evident from 2006 to 2021. In terms of geographic distribution, per capita CO₂ emissions were significantly higher in the eastern coastal and northern regions than in the south and west. Over time, the high-emission regions gradually spread to the central and western regions. As of 2021, the per capita carbon emissions from public buildings in Beijing, Tianjin, Shanghai, Heilongjiang, Inner Mongolia, and Xinjiang reached the highest level. Possible explanations for this phenomenon are that provinces such as Heilongjiang, Inner Mongolia, and Xinjiang are moderately developed provinces and municipalities, whose energy consumption relies mainly on fossil energy [59]; additionally, developed provinces such as Beijing, Tianjin, and Shanghai have a large proportion of tertiary industries and relatively small population bases in these provinces and municipalities, which together lead to high per capita carbon emissions from public buildings. The surge in PBCE in Guizhou around 2016 can be linked to the burgeoning big data industry [60]. In addition, provinces such as Sichuan and Guangxi all belonged to the low-emission region during the study period, aligning with findings by Li et al. [61], suggesting prolonged sluggish growth in public building-related industries within these provinces.

Subsequently, the Theil index was employed to delve into the spatial variances in PBCE across the major inter- and intra-regions (see Figure 5). As shown in the illustration, the overall Theil index of PBCE declined from 0.4064 to 0.148 during the period of 2006–2021, showing a decreasing trend in general, and the trends of intra-region and inter-region differences coincide with the trend of the overall Theil index. During 2011–2013, interregional inequality was greater than intraregional inequality, suggesting that the role of the economy began to increase during this period. In addition, intra-regional inequality was always larger than inter-regional inequality, mirroring the broader landscape of China’s construction industry [62]. The collective contribution rate of intra-regional variances stands at 54.31%, underscoring that the primary disparities in PBCE in China stem from intra-regional distinctions; this observation resonates with the conclusions drawn by Li and Luo et al. [62].

In addition, the overall Theil Index of the four regions exhibited a consistent downward trajectory, among which the Theil Index of region II (Low–High) was the highest and had the biggest decrease from 0.498 to 0.200, with an overall decrease rate of −59.84%, but it was still at the highest level as of 2021, which indicates that provinces and municipalities in region II achieved obvious results in the control and management of carbon emissions from public buildings. This may be due to the region’s policy promotion and practical efforts in green building, technological innovation, and industrial adjustment in recent years. Nevertheless, disparities in PBCE persist within region II and are still more prominent compared with other regions. The Theil Index of region I and region IV are always at a low level, indicating minimal discrepancies in per capita carbon emissions from public buildings among the provinces and municipalities in these areas. Region III belongs to the low-economic and low-intensity region, and its Theil index decreased from 0.092 to 0.086, with an overall decrease rate of only −6.52%, which may be attributed to the fact that these regions adhere to the traditional concept of economic development in eastern and central China, have a weak economic foundation, have low carbon intensity, and it is difficult for them to transform their industries and upgrade their technologies, so the speed of reducing the difference in PBCE is relatively slow.

4.3. Spatial Correlation Analysis of Carbon Emissions from Public Buildings

Leveraging the spatial adjacency matrix, an in-depth analysis was conducted to calculate and examine the spatial correlation of carbon discharges from public structures across China’s 30 provincial-level administrative regions over the period spanning from 2006 to 2021. The outcomes of statistical tests on the overall Moran’s I index value and Z value are detailed in

Table 6. Moran Index of carbon emissions from public buildings in China from 2006 to 2021.

Year	I	E(I)	Sd(I)	Z	p-Value
2006	0.280	−0.034	0.091	3.462	0.000
2007	0.303	−0.034	0.092	3.656	0.000
2008	0.248	−0.034	0.094	2.994	0.001
2009	0.229	−0.034	0.098	2.679	0.004
2010	0.244	−0.034	0.104	2.688	0.004
2011	0.206	−0.034	0.107	2.237	0.013
2012	0.242	−0.034	0.110	2.513	0.006
2013	0.230	−0.034	0.114	2.323	0.010
2014	0.230	−0.034	0.114	2.318	0.010
2015	0.304	−0.034	0.117	2.904	0.002
2016	0.334	−0.034	0.117	3.143	0.001
2017	0.324	−0.034	0.117	3.071	0.001
2018	0.366	−0.034	0.116	3.464	0.000
2019	0.381	−0.034	0.115	3.617	0.000
2020	0.366	−0.034	0.116	3.454	0.000
2021	0.401	−0.034	0.115	3.779	0.000

, showcasing that all the years from 2006 to 2021 passed the significance test of order 5%; these data indicate that the PBCEs across China’s 30 provinces are characterized by a high degree of autocorrelation.

To delve deeper into the spatial dynamics, this study employed the local autocorrelation method to scrutinize the local spatial clustering patterns within each province. By leveraging the local Moran’s I index, the spatial clustering characteristics were categorized into four correlation modes: H–H, L–H, L–L, and H–L agglomeration. As shown in Figure 6, the local Moran’s I indices across the 30 regions throughout the study duration predominantly reside in the first and third quadrants, indicating a prevalent trend of strong positive spatial correlation among most provincial-level administrations and their neighboring provinces. Regions positioned in the first quadrant revealed a scenario where high carbon emission regions adjoin other high carbon emission regions, including Beijing, Tianjin, Shanghai, and other regions where the per capita carbon emissions from public buildings are at a high level, and various factors of production are able to form a high-speed bidirectional flow between provinces, so the positive spatial spillover effect is significant. Conversely, regions in the third quadrant showcased a pattern where provinces neighboring low carbon emission regions also exhibited lower carbon emissions, such as Sichuan, Yunnan, Fujian, and other regions that may have superior ecological environments and lower urbanization levels, resulting in lower carbon emission levels. On the other hand, regions such as Liaoning and Gansu located in the second quadrant indicate that the PBCE are not at a high level, per-se, and are instead surrounded by the rest of the high-level provinces. By contrast, the carbon emissions of public buildings in quadrant IV provinces, such as Guizhou, have higher than those in neighboring provinces, thus occupying a key position in the polarization center. The polarization effect of these provinces is more prominent than the diffusion effect.

4.4. Trends in Spatial and Temporal Evolution of Carbon Emissions from Public Buildings

Employing the center of gravity-standard deviation ellipse model, this research delved into the spatial progression of PBCE in China. Through the utilization of ArcGIS software, relevant parameters were calculated (see

Table 7. Elliptic parameters of carbon emissions from interprovincial public buildings.

Year	Center of Gravity Coordinate (E, N)	Long Semi-Axis/km	Short Semi-Axis/km	Area/km2	Elliptical Corner/°	Direction of Travel
2006	(114.2606° E, 36.4868° N)	1018.03	957.37	3,061,737.33	43.2672
2011	(113.9597° E, 36.4029° N)	1052.68	898.14	2,970,076.70	23.1413	southwestern
2016	(113.6224° E, 36.5950° N)	1177.22	1002.80	3,708,521.18	37.2780	northwestern
2021	(113.1380° E, 36.1560° N)	1138.28	1082.36	3,870,347.83	53.9990	southwestern

), and four characteristic time points, namely, 2006, 2011, 2016, and 2021, were selected to draw the trajectory map of the center of gravity of the PBCE in China (see Figure 7) to visualize the spatial distribution attributes of the PBCE in China and its dynamic evolution trend.

The standard deviation ellipse of PBCE in China showed a northeast–southwest orientation between 2006 and 2021, with the elliptical area continuing to expand., indicating an increasing spatial agglomeration trend. The short semiaxis decreased and then increased from 2006 to 2011, suggesting a transition from a weakening to a strengthening influence of carbon emissions in the southeast and northwest regions; the long semiaxis continued to grow, indicating that the carbon emissions in the northeast–southwest direction were enhanced. The center of gravity shifted towards the southwest, showing a larger increment in the southwest than in the east. The ellipse azimuth rotated counterclockwise from 2006 to 2011, with the northern cities having an enhanced impact, while it rotated clockwise from 2011 to 2021, with the southern cities having an enhanced impact, and the final northeast–southwest oriented distribution was strengthened.

4.5. Analysis of Spatial Spillover Effects of Carbon Emission Influencing Factors in Public Buildings

In this study, carbon emissions per capita from public buildings served as the explanatory variables, and energy intensity, energy structure, public building area per capita, disposable income per capita of all residents, openness level, and environmental governance capacity were taken as the explanatory variables.

Through the results of the LM test, Wald test, and LR test (see

Table 8. Results of the spatial measurement model tests.

Test Methods	Statistic	p-Value
Moran’s I	21.449 ***	0.000
LM Spatial Lag	321.112 ***	0.000
Robust LM Spatial Lag	53.609 ***	0.000
LM Spatial Error	434.210 ***	0.000
Robust LM Spatial Error	166.707 ***	0.000
LR test for SAR	81.69 ***	0.000
LR test for SEM	75.49 ***	0.000
Wald test for SAR	38.07 ***	0.000
Wald test for SEM	54.70 ***	0.000
Hausman test	52.50 ***	0.000

Note: z-statistics in parentheses *** p < 0.01.

), it can be concluded that at a 1% significance level, the statistical p-values of the LM flag and LM error were significant, indicating rejection of non-spatial hypotheses. At a significance level of 1%, both Wald-SAR and LR-SAR statistics had significant p-values, thus rejecting the hypothesis of spatial autoregressive models. At a 5% significance level, the p-values of Wald-SEM and LR-SEM statistics were both significant, thus rejecting the hypothesis of the spatial error model. Based on the Hausman test, the statistic passed the significance test, and the spatial econometric model with fixed effects was selected in this paper. Thanks to the individual differences between regions and the time factor, and considering the economic significance of the explanatory variables [63], this paper selected the SDM model based on a time–space two-way fixed approach to conduct regression analysis and explore the impact of the primary influencing factors.

To guarantee the robustness of the results, maximum likelihood estimation of the parameters of the three modeling forms of SDM, SAR, and SEM was performed with per capita carbon emissions as the explanatory variable, and the results are shown in

Table 9. Spatial measurement model regression results.

Influencing Factors	SDM	SAR	SEM
EI	0.011 *** (22.53)	0.012 *** (22.82)	0.012 *** (24.12)
ES	−0.001 * (−1.70)	−0.001 (−0.83)	0.000 (0.11)
PPBA	0.011 *** (4.54)	0.010 *** (4.05)	0.012 *** (4.95)
DPI	0.060 *** (3.05)	0.051 *** (3.26)	0.047 *** (2.84)
FDI	−0.008 ** (−2.18)	−0.009 ** (−2.43)	−0.007 * (−1.92)
EM	0.038 *** (5.15)	0.031 *** (4.07)	0.043 *** (5.49)
Rho(lambda)	0.311 *** (6.08)	0.233 *** (5.28)	0.350 *** (6.54)
sigma2_e	0.006 *** (15.44)	0.007 *** (15.43)	0.007 *** (15.30)
Log-likelihood	530.4564	499.8168	505.3782

Note: z-statistics in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1.

. Consistency in the signs, magnitudes, and statistical significance of coefficients across the explanatory variables suggests a high level of result robustness.

The regression results of the SDM model indicate the elasticity of each explanatory variable, but cannot reflect the full picture of the spatial effects of the factors influencing the per capita carbon emissions of public buildings. Therefore, to gain a more profound insight into these spatial implications, the cumulative effect, direct impact, and spill-over consequences of each explanatory factor were meticulously calculated and deconstructed using the variance–covariance matrix extracted from the SDM estimation outcomes. The detailed results are outlined in

Table 10. Decomposition of effects of factors influencing carbon emissions in public buildings.

Influencing Factors	Direct Effects	Indirect Effects	LR_Total
EI	0.012 ***	0.000	0.012 ***
ES	−0.001 **	−0.006 ***	−0.008 ***
PPBA	0.010 ***	−0.018 ***	−0.008
DPI	0.066 ***	0.113 **	0.179 ***
FDI	−0.008 **	0.002	−0.006
EM	0.033 ***	−0.089 ***	−0.056 **
rho	0.311 ***
R-squared	0.628

Note: z-statistics in parentheses *** p < 0.01, ** p < 0.05.

.

Based on the findings presented in Table 10, the spatial spillover effects of the factors are outlined as follows:

(1)

Technical factors: The direct impact of energy intensity was significant, 0.012, and the indirect impact was not significant. A rise in local energy consumption per floor area unit exerted a notable adverse effect on energy conservation and emission reduction within the local public building sector, given that energy consumption stands as a primary source of carbon emissions [64]. Comparatively, EI had a negligible effect on neighboring areas; the effects of energy structure, both direct and indirect, were remarkably negative. This suggests that an increase in the proportion of power structures will inhibit the growth of carbon emissions from local and neighboring public buildings. China has been continuously optimizing its power structure in recent years to control fossil fuel consumption and promote the development of new energy sources such as wind and solar power; the proportion of total primary electric energy production increased from 8.5% to 20.6% between 2006 and 2021, and simultaneously, a system of inter-regional collaboration exists to drive structure optimization and promote low-carbon development. This cooperative framework generates substantial benefits, effectively curbing PBCE in neighboring regions [65];

(2)

Demographic factors: Public floor space per capita promotes PBCE locally and suppresses PBCE in neighboring areas. The degree of influence was 0.010, −0.018 respectively, indicating that for every 1 m² increase in public building area, PBCE from local increased by 0.01 t, while PBCE in neighboring areas decreased by 0.018 t. A large public building area per capita means a high population density in a limited space, requiring more floor space to meet the demands of life, leading to high energy consumption and carbon emissions. In areas with large public building areas per capita, there are usually abundant public building resources and facilities, and neighboring areas can learn from their experience in energy management strategies and environmental protection measures to mitigate carbon emissions;

(3)

Economic development factors: Both the direct and indirect effects of the economic development level exhibited significance, registering at 0.066 and 0.113, respectively, indicating that the disposable income per capita of all residents can significantly increase PBCE in local and neighboring areas. The reason for this phenomenon may be related to the change in consumption behavior of residents, where increased income stimulates their consumption of public buildings, such as shopping malls and entertainment venues. In addition, local economic prosperity may lead to the development of neighboring areas, influencing carbon dioxide emissions [66];

(4)

Openness to the outside world: The level of foreign investors can significantly inhibit local public building carbon emissions, with a direct impact coefficient of −0.008. However, no substantial effect was observed on the PBCE in neighboring regions. The introduction of foreign investment also introduces advanced technology and equipment to improve the local environment, which in turn improves the efficiency of PBCE and reduces carbon emissions [54]. The impact on neighboring areas is not large, probably because the neighboring cities have limited technological upgrading in the process of opening up and development. Simultaneously, it accepts the outflow of foreign enterprises into the city, which makes its carbon emissions rise. Therefore, the inhibitory effect of open development on the PBCE of neighboring cities is not obvious [67];

(5)

Environmental governance: Environmental governance capacity promotes local public building carbon emissions while suppressing carbon emissions in neighboring areas. The influence coefficients were 0.033 and −0.089, suggesting that China currently does not impose a high level of regulatory requirements for the process of reducing carbon emissions from public buildings [14]. The environmental governance policy may also have an effect, prompting neighboring regions to adopt more rigorous environmental protection measures.

5. Conclusions and Recommendations

5.1. Conclusions

In this research, 30 provinces in China were selected as the subjects of study from 2006 to 2021. The spatial variability, autocorrelation, spatio-temporal evolution of PBCE, and their influencing factors were analyzed using GIS technology, Theil’s index, Moran’s I index, standard deviation ellipse, and the SDM model. The study outcomes reveal the following key points: (1) PBTCE and PBPCE in China are growing gradually at a growth rate of 9.20% and 7.48%, but the growth rate is declining, indicating that China’s efforts to reduce the PBCE are bearing fruit. (2) PBCE are heterogeneous among provinces and regions, with per capita carbon dioxide emissions from public buildings in the eastern coastal and northern regions significantly higher than those in the southern and western regions. The average intra-group contribution was 54.31%, indicating an overall difference mainly stems from intra-regional differences but is on a narrowing trend. (3) During the study period, the Moran Index rose from 0.280 to 0.401, suggesting that PBCE in different provinces exhibits a high degree of autocorrelation, with this correlation becoming increasingly pronounced, mainly characterized by H–H and L–L agglomeration. (4) The geographical center of gravity of PBCE is concentrated near Hebei Province, progressively shifting from east to west. It indicates that the rise in PBCE is more prominent in the central and western regions compared to the east. Carbon emissions demonstrated a trend of aggregation towards the northeast–southwest direction and dispersion along the northwest–southeast direction. (5) Energy intensity, energy structure, public building area per capita, disposable income per capita of all residents, openness level, and environmental governance capacity may affect carbon emissions from public buildings, with varying direct and indirect effects. It is evident that strategies aimed at reducing carbon emissions from public buildings must be tailored and consider the broader macroeconomic environment’s influence.

5.2. Recommendations

This study offers an in-depth analysis of PBCE from 2006 to 2021, revealing the development trend of PBCE from multiple perspectives, such as time, space, and influencing factors. These insights play a pivotal role in steering endeavors toward energy conservation and emission reduction within the construction sector. The study yields the subsequent policy suggestions:

(1)

Recognizing the distinctive spatial attributes of PBCE, when formulating policies related to energy conservation and emission reduction, it is necessary to consider both uniform standards on a national scale and the special circumstances of each region to foster low-carbon progress in public building domains. Intra-regional disparity is the main factor contributing to the total difference in PBCE, so narrowing the gap between carbon emissions in urban areas should be regarded as an effective method for mitigation policy development. As there is a large difference between the south and the north, it is necessary for the government to prioritize measures to narrow the gap between carbon emissions in the southern and northern provinces and cities. In the cold northern regions, there is a high demand for heating in the winter, so more attention may need to be paid to the optimization of the heating system and the improvement of energy efficiency in energy-saving and emission-reduction strategies. In the southern region, due to the warmer climate, air conditioning is used frequently; therefore, more attention may need to be paid to the energy efficiency of air conditioning systems and the application of green building materials in energy conservation and emission reduction; carbon emissions show a strong positive correlation spatially, and it is necessary to establish a regional cooperation system to facilitate information exchange and cooperation in order to accelerate the emission reduction process. In addition, the center of gravity of PBCE is gradually moving from east to west, and more attention should be paid to this phenomenon and more support and assistance should be given to promote energy conservation and emission reduction in the field of their public buildings through the provision of financial and technological support;

(2)

In terms of driving factors, energy, technology, population, economy, environment, foreign trade, and other factors all affect carbon emissions from public buildings, necessitating comprehensive consideration in policy formulation. For example, to further promote the development of clean energy, accelerate the development and utilization of clean energy through technological innovation and policy support, and enhance energy utilization efficiency. Concurrently, optimizing the energy framework is important to diminish fossil energy reliance. Urban construction and public building project planning should meticulously weigh population distribution and economic developmental trends and rationalize public building scale and layout to avert haphazard expansion and excessive construction. The foreign investment environment should be optimized, attracting more foreign investment to enter the construction industry and fostering technological innovation and industrial upgrading. Environmental governance should be strengthened, environmental management capacity enhanced, and the construction industry steered toward green, low-carbon, and recyclable pathways. At the same time, awareness should be raised through robust energy conservation and emission reduction advocacy to cultivate a public ethos of environmental stewardship and foster broad societal engagement in sustainability efforts.

In general, the formulation of public building energy conservation and emission reduction policies should take full account of the actual situation of each place, and adopt targeted measures to promote the realization of low-carbon development in the field of public buildings nationwide. The collaborative implementation of these policies necessitates concerted efforts from various stakeholders, including governmental bodies, enterprises, the general populace, and other entities, all actively engaging in energy-saving and emission-reduction endeavors to collectively advance the realization of a green, low-carbon society.

This study may be subject to the following limitations. Owing to data constraints, the analysis in this paper is confined to examining carbon emissions from public buildings solely at the provincial level in China during the operational phase, omitting assessments at district and county levels and across the entire life cycle. In addition, the PBCE is affected by many complex factors, so this paper only examines the main factors that are representative of them. Once again, this paper only considers geographic matrices to analyze spatial relationships, and further research could consider using more spatial matrices to explore the spatial spillover of influencing factors.