Evaluation of Carbon Emission Efficiency in the Construction Industry Based on the Super-Efficient Slacks-Based Measure Model: A Case Study at the Provincial Level in China

Abstract

Rapid urbanization and an increasing carbon footprint have underscored the need for sustainable practices in the construction industry. With the aim of prioritizing global sustainable development, the measurement of carbon emission efficiency in the construction industry (CEECI) has emerged as a critical indicator. Nevertheless, a comprehensive exploration of carbon emission efficiency within the Chinese construction sector remains limited, despite the pressing demand to mitigate carbon emissions. To address this research gap, this study aims to provide valuable policy recommendations for effectively reducing carbon emissions. We conducted a thorough assessment of both the total carbon emissions and the carbon emission intensity in 30 provinces and cities across China from 2010 to 2020. Utilizing the slacks-based measure (SBM) model with non-desired outputs, we evaluated the static CEECI, including the spatial correlation analysis and the evaluation of the carbon reduction potential in the construction industry (CRPCI). Additionally, the dynamic CEECI was quantified using the Malmquist–Luenberger (ML) index model, followed by an index decomposition analysis. The findings reveal several noteworthy insights: (1) There exists a positive correlation between carbon emissions in the construction industry and the economic scale. Generally, less developed areas (e.g., central and western regions of China) exhibit higher levels of carbon emission intensity (CEICI), while more developed areas (e.g., eastern regions of China) demonstrate lower levels of CEICI. (2) The CEECI across various provinces and cities demonstrates a clear spatial positive autocorrelation, while the CRPCI exhibits a negative correlation with the CEECI, with larger CRPCI values observed predominantly in western China. (3) Technological progress (MLTC) emerges as a crucial factor influencing the CEECI in our dynamic analysis. These findings offer valuable insights for policymakers to develop focused strategies to effectively mitigate carbon emissions nationwide.

1. Introduction

The unprecedented growth of the global economy has resulted in a surge in energy demand and greenhouse gas emissions, leading to global climate change and unsustainable environmental conditions [1]. This situation has garnered widespread attention from the public [2,3]. As the fastest-developing country globally, China has also become the world’s largest carbon emitter and is thus facing mounting domestic and international pressure to mitigate emissions [4]. According to the BP Statistical Review of World Energy 2021 [5], China accounts for approximately 30% of global carbon dioxide emissions, and its carbon emissions will rise by more than 50% in the next 15 years if no appropriate control measures are implemented [1,6]. To mitigate carbon emissions, China pledged that its carbon emissions would peak by 2030 and that carbon neutrality would be achieved by 2060 [7]. The Chinese construction industry is a major contributor to carbon emissions, which is characterized by high energy consumption and low energy efficiency, with total energy consumption in building operations accounting for up to 20% of the national total [8]. Carbon emission efficiency in the construction industry (CEECI) is a crucial indicator to measure the relationship between economic growth and carbon emissions in the construction sector [9]. However, the absence of official annual carbon emissions data poses a challenge to the exploration of CEECI in China [1], and there is a dearth of research that focuses on quantifying the CRPCI in China. To effectively adopt targeted strategies to mitigate carbon emissions caused by the Chinese construction industry, it is imperative to measure the CEECI and quantify the CRPCI in various regions of China.

An accurate measurement of carbon emissions is crucial for analyzing the CEECI, and the calculation of carbon emissions for the construction industry can be divided into two parts, namely direct carbon emissions (DCEs) and indirect carbon emissions (ICEs) [10]. Direct carbon emissions are typically calculated using the carbon emission coefficient method, which involves multiplying the energy consumption by corresponding carbon emission coefficients to obtain carbon emission estimates [11]. However, the calculation of DCEs in previous studies mainly focused on a limited number of energy sources, which failed to achieve a comprehensive and systematic calculation of those emissions [9,12]. ICEs, on the other hand, are divided into type I indirect carbon emissions (${ICE}^1$) (i.e., the consumption of electricity and heat) and type II indirect carbon emissions (${ICE}^2$), where ${ICE}^2$ is the carbon emissions produced by other industries associated with the construction industry [13]. ${ICE}^1$ is commonly calculated using the IPCC carbon emission coefficient method [14], while ${ICE}^2$ is mostly measured using the input–output method [15] and the carbon emission coefficient method. While some researchers [16,17,18] have adopted the input–output method to calculate ${ICE}^2$, the calculation of annual provincial carbon emissions is constrained by poor data availability and accuracy, as China’s input–output tables are published at five-year intervals. To avoid the aforementioned limitations, this study selects more comprehensive energy sources to achieve an accurate estimation of DCEs and adopts the carbon emission coefficient method to calculate ${ICE}^2$. By calculating the carbon emissions generated by the primary building materials (i.e., steel, wood, cement, glass, and aluminum), the present research proves that this method is effective and reliable for calculating ${ICE}^2$ [19].

The research into assessing carbon emission efficiency is primarily focused on two aspects: static and dynamic analysis. Indicators adopted to explore the static performance of carbon emission efficiency can be measured using the single factor index and the total factor index [20]. However, the assessment of carbon emission efficiency solely based on the single factor method has been criticized for being one-sided as it fails to consider other production factors to assess the substantive level of carbon production. To address this limitation, Solow [21] introduced the concept of total factor productivity, incorporating input factors such as capital, labor, and energy into the research framework. The total factor carbon emission efficiency method has a more accurate and objective reflection on the relationship between carbon emissions and each factor, which has been widely employed by various researchers [22,23,24]. The two major methods used to measure the total factor carbon emission efficiency are the Stochastic Frontier Analysis (SFA) and the Data Envelopment Analysis (DEA). The SFA, a parametric method that only takes into account one output indicator, has been applied by some researchers [25,26] but may not align well with the actual situation. In contrast, the DEA with nonparametric models is more popular, as it avoids the influence of setting weights artificially on evaluation results [20,27,28,29,30,31,32,33,34]. Despite this, traditional DEA models are not applicable to efficiency studies involving non-desired outputs. Consequently, Tone [35] proposed the slacks-based measure (SBM) based on non-expected outputs, which has been widely employed to conduct analytical studies on the carbon emission efficiency of different regions [36,37,38]. Incorporating super-efficiency [39,40,41,42] into SBM models can effectively rank and analyze the decision units, making it more applicable for measuring carbon emission efficiency as the efficiency of the decision unit is no longer limited to the range of [0, 1]. Therefore, this paper employs a super-efficient SBM model that includes non-desired outputs to better analyze the CEECI. Compared to the vast number of studies on static analysis, limited research has focused on dynamic carbon emission efficiency. Researchers have mainly employed the Malmquist–Luenberger (ML) index model to evaluate carbon emission efficiency in various industries such as transportation [43,44,45], agriculture [46,47,48], and manufacturing [49,50]. Nonetheless, relevant studies on the construction industry are lacking. To address this knowledge gap, the CEECI in China is assessed from both static and dynamic perspectives, providing a comprehensive understanding of the current carbon emission situation [51,52], which can provide valuable insights for policy designers to develop targeted strategies in China. Hence, this paper aims to provide effective policy recommendations for carbon emission reductions by analyzing the CEECI in 30 provinces and cities in China.

The main contributions of this paper are as follows: First, it contributes to the existing literature by adopting a more comprehensive energy consumption indicator system to precisely calculate carbon emissions in 30 provinces and cities in China, enhancing the understanding of carbon emission footprints in the context of China. Second, it establishes a standard formula to effectively calculate the CEICI at the provincial level of China, which can be applied to future research assessing carbon emission intensities in other contexts globally. Third, this study introduces and quantifies the CRPCI in 30 provinces and cities in China, facilitating the rational allocation of provincial resources and providing valuable insights for the formulation of energy conservation and carbon emission reduction policies.

2. Research Methods and Models

2.1. Carbon Emissions Accounting Model

The concept of ICEs was introduced due to the construction industry’s characteristic of being highly interrelated with other industries, such as the manufacturing and transportation industries [19,53]. TCE is the sum of DCE and ICE, which is formulated as follows:

$$TCE=DCE+ICE$$

(1)

where TCE represents the total carbon emissions from the construction industry (10⁴ tons), DCE represents the direct carbon emissions from the construction industry (10⁴ tons), and ICE denotes the indirect carbon emissions from the construction industry (10⁴ tons).

2.1.1. DCE Accounting Model

At the inter-provincial construction industry level, in line with the principle of “whoever consumes it, takes it” (e.g., the principle of energy consumption), the carbon emissions directly consumed in the construction industry are regarded as DCEs. In order to enhance the comprehensiveness and accuracy of the measurement results, an extensive literature review has been conducted [9,12], resulting in the elicitation of 21 sources of energy consumption (see

Table 1. Sources of energy for direct carbon emission calculation.

Energy Type	NCVi (TJ/104 Tons)	Ci (Tons/TJ)	Oi
Raw coal	209.08	26.37	0.94
Washed refined coal	263.44	25.41	0.96
Other washed coal	83.63	25.41	0.96
Briquette	174.60	33.60	0.90
Coal gangue	83.63	27.30	0.98
Coke	284.35	29.50	0.93
Coke oven gas	292.71	13.60	0.99
Other gas	522.70	12.20	0.99
Crude oil	418.16	20.10	0.98
Gasoline	430.70	18.90	0.98
Kerosene	430.70	19.50	0.98
Diesel oil	426.52	20.20	0.98
Fuel oil	418.16	21.10	0.98
Naphtha	439.07	20.00	0.98
Lubricant	413.98	20.00	0.98
Petroleum asphalt	389.31	22.00	0.98
Petroleum coke	319.47	27.50	0.98
Liquefied petroleum gas	501.79	17.20	0.98
Refinery dry gas	460.55	18.20	0.98
Other petroleum products	409.80	20.00	0.98
Natural gas	292.54	15.30	0.99

Note: The average low calorific value of energy is obtained from the General Rules for Calculation of the Comprehensive Energy Consumption (GB/T2589-2020) in China [55]. The carbon content per unit calorific and carbon oxidation rate of energy are from the Guidelines for the Preparation of Provincial Greenhouse Gas Inventories [56].

). The consumption of energy in the construction industry is derived from the China Energy Statistical Yearbook (2010–2020) [54]. The DCE calculation formula is as follows:

$$DCE={\sum }_{i=1}^{21}{E}_i\times {NCV}_i\times C_i\times O_i\times \frac{44}{12}.$$

(2)

where E_i represents the consumption of energy i in the construction industry (10⁴ tons), NCV_i represents the average low calorific value of energy i (TJ/ 10⁴ tons), C_i is the carbon content per unit calorific value of energy i (tons/TJ), and O_i is the carbon oxidation rate of energy i.

Table 1 presents the average low calorific value, carbon content per unit calorific value, and carbon oxidation rate of the 21 energy sources.

2.1.2. ICE Accounting Model

ICEs can be divided into ${ICE}^1$ and ${ICE}^2$, which are generated from the electricity and heat energy consumptions within the construction industry and emissions from other related industries, respectively. The accounting formula is as follows:

$$ICE={ICE}^1{+ICE}^2$$

(3)

where ${ICE}^1$ represents type I indirect carbon emissions from the construction industry (10⁴ tons) and ${ICE}^2$ denotes type II indirect carbon emissions from the construction industry (10⁴ tons).

(1) ${ICE}^1$ accounting model

The electricity consumption and heat consumption in the construction industry are derived from the China Energy Statistical Yearbook (2010–2020) [54]. The specific accounting formula is as follows:

$${ICE}^1=EP\times {\alpha }_{EP}+HP\times {\alpha }_{HP}$$

(4)

where EP represents electricity consumption in the construction industry (10⁸ kW·h), α_EP is electricity carbon emission factor (kgCO₂/kWh), HP denotes heat consumption in the construction industry (10¹⁰ kJ), and α_HP stands for the heat carbon emission factor (tCO₂/GJ), with 0.11 tCO₂/GJ as the accounting basis.

Table 2. Average carbon emission factors of regional power grids in China.

Name of the Grid	Coverage of Provinces	Carbon Emission Factor (kgCO2/kWh)
North China region	Beijing, Tianjin, Hebei, Shanxi, Shandong, and Inner Mongolia	0.8843
Central China region	Henan, Hubei, Hunan, Jiangxi, Sichuan, and Chongqing	0.5257
East China region	Shanghai, Jiangsu, Zhejiang, Anhui, and Fujian	0.7035
Northeast China region	Liaoning, Jilin, and Heilongjiang	0.7769
Northwest China region	Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang	0.6671
Southern China region	Guangdong, Guangxi, Yunnan, Guizhou, and Hainan	0.5271

Note: The electricity carbon emission factors were obtained from the 2012 China Regional Grid Average CO₂ Emission Factor issued by the Department of Climate Change of the National Development and Reform Commission [57]. The heat carbon emission factors were obtained from the Greenhouse Gas Emission Accounting Methodology and Reporting Guide for Enterprises in Other Industries [11].

presents the carbon emission factors of regional power grids in China.

(2) ${ICE}^2$ accounting model

The carbon emission coefficient method adopted from the IPCC [11] was employed in this study to calculate ${ICE}^2$, which includes the carbon consumption of five main construction materials (i.e., steel, wood, cement, glass, and aluminum) [19]. The consumption of each construction material is derived from the China Statistical Yearbook on Construction (2010–2020) [58]. The formula is as follows:

$${ICE}^2={\sum }_{j=1}^5{M}_j*f_j*\left(1-r_j\right).$$

(5)

where M_j represents the consumption of the jth construction material (10⁴ tons), f_j represents the carbon emission factor of the jth construction material (kg/kg, kg/m³, or kg/weight box), and r_j stands for the recycling factor of the jth construction material.

Table 3. Carbon emission factors and recovery factors of major building materials.

Construction Materials	Carbon Emission Factor	Recovery Factor
Steel	1.789 kg/kg	0.8
Wood	10.45 kg/m3	-
Cement	0.822 kg/kg	-
Glass	35.78 kg/weight box	-
Aluminum	2.6 kg/kg	0.85

Note: The carbon emission factor of construction materials and the recovery factor of construction materials were derived from references [59,60,61,62].

demonstrates the carbon emission factors and recovery factors of major building materials.

DCEs and ${ICE}^1$ fall under the category of carbon emissions during the operation stage of a building. On the other hand, ${ICE}^2$ pertains to the carbon emissions associated with the materialization stage of the building (e.g., material production and building construction). As a result, the TCEs can be further categorized into two different stages, as represented by the following formulas:

$$OCE=DCE+{ICE}^1$$

(6)

$$MCE={ICE}^2$$

(7)

where OCE represents the carbon emission during the building operation stage (10⁴ tons) and MCE denotes the carbon emission during the building materialization stage (10⁴ tons).

2.1.3. The CEICI Accounting Model

The concept of carbon emission intensity was proposed by President Hu Jintao [63] with the aim of effectively assessing the level of China’s economic development, technological progress, and energy use efficiency. This indicator is generally calculated by dividing carbon emissions by GDP (gross domestic product). The smaller the carbon emission intensity, the lower the carbon emissions per unit of economic output, which indicates better coordination between regional or industrial economic developments and low-carbon environmental protection. Therefore, the CEICI can be defined as the ratio of carbon dioxide emissions from the construction industry to the total construction output, which reflects the low-carbon development level of the construction industry in a specific region. The specific calculation formula is as follows:

$$CEICI=\frac{TCE}{P}$$

(8)

where CEICI is the carbon emission intensity of the construction industry (tons/10⁴ RMB) and P represents the total construction industry output (10⁸ RMB).

2.2. The Static CEECI Measurement Model

2.2.1. Factor Selection and Data Source

Referring to relevant research on measuring static carbon emissions [9,12,64] and considering the characteristics of the Chinese construction industry, energy, labor, capital, and mechanical equipment are selected as input factors, while the total construction output and carbon emissions are taken as the desired and non-desired outputs, respectively. The specific indicators of each element and data sources are shown in

Table 4. Input–output indicators and data sources for the CEECI.

Category	Elements	Specific Indicators	Data Source
Input elements	Energy	Total energy consumption in the construction industry (104 tons)	China Energy Statistics Yearbook [54]
	Labor	Year-end number of employees in the construction industry (104 people)	China Statistical Yearbook [65]
	Capital	Construction industry enterprise assets (108 RMB)	China Statistical Yearbook on Construction [58]
	Mechanical equipment	Total power of own construction mechanical equipment at the end of the year (104 kilowatts)	China Statistical Yearbook on Construction [58]
Expected output	Total industry output	Total construction industry output (108 RMB)	China Statistical Yearbook on Construction [58]
Non-desired output	Carbon emissions	Carbon emissions from the construction industry (104 tons)	Calculated by Equations (1)–(5)

.

2.2.2. Super-Efficient SBM Models including Non-Desired Outputs

Within the total factor productivity framework, the primary methods for measuring static efficiency include the SFA and DEA. The SFA considers the influence of stochastic factors during the production process while examining cross-period panel data. This method requires the input indicators to be independent of each other and only permits one output indicator, making it challenging to achieve in actual production settings in the construction industry. In comparison, the DEA does not require the specific form of the constructed model and is not affected by the scale of operations, making it suitable for problems involving multiple inputs and multiple outputs. However, the DEA model is not applicable for evaluating efficiency in scenarios involving non-expected outputs, as it is constrained to the setting of output maximization. In addition, most traditional DEA models employ radial and angular measures, which fail to capture all the slack variables, resulting in flaws in efficiency evaluations. In order to address the issues of non-expected output and slack variables concurrently, the SBM model proposed by Tone [35] is employed in this study. To effectively account for non-desired outputs, carbon emissions are utilized as an indicator, and the super-efficient SBM model containing non-desired outputs is selected to measure the CEECI. Matlab software was mainly used to construct the model, and the specific equations [66] are as follows:

$$\rho =min\frac{1+\frac{1}{m}\sum _{i=1}^m\frac{S_i^x}{x_{i0}}}{1-\frac{1}{S_1+S_2}(\sum _{k=1}^{S_1}\frac{S_k^y}{y_{k0}}+\sum _{l=1}^{S_2}\frac{S_l^z}{z_{l0}})}.$$

(9)

$$s.t.\left\{\begin{array}{c}{x}_{i0}\ge \sum _{j=1,\ne 0}^n{\lambda }_j{x}_j-S_i^x,\forall i;\\ y_{k0}\ge \sum _{j=1,\ne 0}^n{\lambda }_j{y}_j+S_k^y,\forall k;\\ z_{l0}\ge \sum _{j=1,\ne 0}^n{\lambda }_j{z}_j-S_l^z,\forall l;\\ 1-\frac{1}{S_1+S_2}\left(\sum _{k=1}^{S_1}\frac{S_k^y}{y_{k0}}+\sum _{l=1}^{S_2}\frac{S_l^z}{z_{l0}}\right)>0;\\ S_i^x\ge 0,S_k^y\ge 0,S_l^z\ge 0,{\lambda }_j\ge 0,\forall i,j,k,l;\end{array}\right..$$

(10)

where ρ is the efficiency of the decision unit; n is the decision unit; x, y, and z are the input, desired output, and undesired output, respectively; m, s₁, and s₂ represent the number of variables of the input, desired output, and undesired output, respectively; and S^x, S^y, and S^z refer to the slack variables of the input, desired output, and undesired output, respectively.

2.3. The Dynamic CEECI Measurement Model

2.3.1. The ML Index Model

The common method for measuring dynamic efficiency is the Malmquist index model, which was pioneered by the Swedish economist Sten Malmquist [67] in 1953. The Malmquist index has significant advantages in dealing with efficiency changes and their decomposition. Fare et al. [68] further construct the Malmquist productivity index model based on the DEA measurement model to study both the static and dynamic aspects of carbon emission efficiency to provide a comprehensive measurement perspective. Yet, it was found that the Malmquist index is unable to assess dynamic efficiency when non-desired outputs such as carbon emissions are taken into consideration. In this light, Chung [69] et al. addressed this deficiency by incorporating non-desired outputs within the ML framework of efficiency measurements. This approach aims to solve practical problems related to dynamic efficiency measurements. Building upon the super-efficient SBM model that considers non-desired outputs, the CEECI is further analyzed dynamically through the construction of the ML index model. Adopting Cheng’s [66] method using the non-desired output SBM, the ML index is calculated using the following formula:

$${ML}_c^{t+1}\left(x^t,y^t,z^t,x^{t+1},y^{t+1},z^{t+1}\right)={\left[\frac{E_c^t\left(x^{t+1},y^{t+1},z^{t+1}\right)}{E_c^t\left(x^t,y^t,z^t\right)}\bullet \frac{E_c^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)}{E_c^{t+1}\left(x^t,y^t,z^t\right)}\right]}^{\frac{1}{2}}.$$

(11)

where the subscript c represents constant returns to scale (CRS); t stands for the time duration; x, y, and z are the inputs, desired outputs, and non-desired outputs, respectively; and $E_c$ is the efficiency value in the case of CRS.

2.3.2. ML Exponential Decomposition

The ML index, as a new branch of the Malmquist index, offers the advantage of assessing the control of pollutant emissions while measuring dynamic productivity and technological progress, thus effectively addressing the issue of non-desired outputs such as carbon emissions in the research process [70]. Chung and Fare [71] derived the ML index (i.e., the M-index with undesired outputs) based on the Malmquist index by using the output-oriented directional distance function. Under the assumption of CRS, the ML index can be decomposed into technical efficiency (MLEC) and technological progress (MLTC) with the following equations:

$${ML}_c^{t+1}={MLEC}_c*{MLTC}_c.$$

(12)

$${MLEC}_c=\frac{E_c^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)}{E_c^t\left(x^t,y^t,z^t\right)}$$

(13)

$${MLTC}_c={\left[\frac{E_c^t\left(x^{t+1},y^{t+1},z^{t+1}\right)}{E_c^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)}\bullet \frac{E_c^t\left(x^t,y^t,z^t\right)}{E_c^{t+1}\left(x^t,y^t,z^t\right)}\right]}^{\frac{1}{2}}$$

(14)

where the subscript c is CRS; t is the time duration; and x, y, and z represent the inputs, desired outputs, and non-desired outputs, respectively.

Further, the MLEC can be decomposed into pure technical efficiency (MLPEC) and scale efficiency (MLSEC) under the variable returns to scale (VRS) assumption, as follows:

$${ML}_c^{t+1}={MLEC}_c*{MLTC}_c=MLPEC*MLSEC*{MLTC}_c.$$

(15)

$$MLPEC=\frac{E_v^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)}{E_v^t\left(x^t,y^t,z^t\right)}$$

(16)

$$MLSEC=\frac{E_c^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)/E_v^{t+1}\left(x^{t+1},y^{t+1},z^{t+1}\right)}{E_c^t\left(x^t,y^t,z^t\right)/E_v^t\left(x^t,y^t,z^t\right)}$$

(17)

where the subscript c represents CRS; the subscript v represents VRS; and t stands for the time duration; x, y, and z represent the input, desired output, and non-desired output, respectively.

3. Empirical Results

3.1. Carbon Emissions from the Construction Industry

3.1.1. Results of the TCEs

Based on the equations of (1)–(5) and the obtained data in relation to carbon emissions, the carbon emissions of the construction industry in 30 provinces and cities from 2010 to 2020 were calculated. The annual amount of each province’s or city’s carbon emissions in the construction industry from 2010 to 2020 is presented in

Table A1. Carbon emissions of the construction industry from 2010 to 2020 (Unit: 10⁴ tons).

Provinces	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	Average
Beijing	4141.10	4247.77	3435.60	3855.11	3991.27	3858.62	3655.28	3876.19	4616.35	4977.63	4636.69	4117.42
Tianjin	2180.10	2962.44	2921.30	3827.69	5124.88	3377.18	3066.20	2683.86	3917.30	2345.50	2887.00	3208.49
Hebei	11,931.56	50,978.83	50,098.14	18,619.91	8787.23	9145.52	6541.50	7026.63	5645.86	6563.21	5100.96	164,03.58
Shanghai	2683.57	2862.43	2711.69	2771.60	2850.15	2352.96	2424.72	2659.57	2582.21	5715.11	3413.71	3002.52
Jiangsu	17,446.34	29,870.23	32,186.46	24,667.14	26,506.67	24,510.53	24,102.46	24,053.74	25,803.79	24,940.07	26,773.66	25,532.83
Zhejiang	22,551.46	27,728.49	29,545.27	33,489.57	34,282.08	34,412.04	34,716.80	37,087.15	31,773.29	25,682.84	24,995.17	30,569.47
Fujian	6784.00	6664.42	8351.58	11,196.03	14,788.44	15,257.24	17,283.81	21,591.42	25,139.66	27,867.68	26,267.89	16,472.02
Shandong	9632.56	9778.67	22,368.62	10,686.56	11,149.21	10,562.08	9843.76	9792.52	10,052.95	11,333.09	9600.75	11,345.53
Guangdong	7009.04	10,127.94	10,306.67	9755.95	10,390.67	8726.73	8005.24	10,334.96	10,671.74	12,435.93	12,798.55	10,051.22
Hainan	268.49	430.32	415.05	515.16	405.61	266.53	302.59	331.99	454.03	465.23	393.19	386.20
Shanxi	4494.84	3095.55	3138.88	3297.47	3857.80	3518.09	3427.76	3630.51	3882.76	3787.12	3537.35	3606.19
Anhui	4519.68	4863.35	4905.01	6093.21	6607.09	5599.87	6943.01	8886.85	10,135.06	18,269.11	12,359.16	8107.40
Jiangxi	2285.25	3530.82	3801.32	4651.75	2297.18	6068.00	6466.47	8027.91	9170.38	10,437.02	9898.34	6057.68
Henan	7087.71	7281.73	8859.62	8525.09	11,723.72	8563.09	10,555.01	11,561.57	16,238.15	12,341.31	12,957.67	10,517.70
Hubei	4824.10	9355.86	16,766.38	15,564.60	18,350.34	14,754.54	19,784.64	17,360.74	17,852.62	14,421.98	14,508.45	14,867.66
Hunan	6142.03	6000.92	6912.98	7275.36	7619.36	7962.16	8714.58	9489.04	11,208.90	12,309.58	12,425.32	8732.75
Inner Mongolia	2063.54	2376.58	2290.73	2114.21	2085.44	2207.67	2553.85	3316.38	2515.35	1794.59	1811.30	2284.51
Guangxi	1907.17	2156.22	2940.48	3169.30	3533.47	3114.67	3957.46	4445.54	3766.46	4456.99	3960.43	3400.75
Chongqing	5007.53	5428.01	4810.03	5721.15	5850.80	5749.34	6325.33	6374.27	6328.41	6573.56	6385.13	5868.51
Sichuan	10,775.14	13,150.94	19,530.44	21,545.78	23,840.78	11,331.86	12,908.92	16,079.06	20,112.77	20,516.96	21,457.48	17,386.38
Guizhou	1040.72	1318.54	1624.78	2784.48	3403.56	4407.27	7880.68	3703.11	4292.21	4622.59	3836.98	3537.72
Yunnan	2347.06	2078.75	2576.87	5915.49	6472.38	3208.55	3807.53	4192.47	4956.26	5176.64	5747.87	4225.44
Shaanxi	3788.10	6280.58	4513.80	4821.53	5515.13	5627.84	6204.58	6354.95	7734.02	8005.90	8752.45	6145.35
Gansu	975.35	2122.99	1682.12	2090.38	2279.04	1988.81	2132.31	1862.40	1938.57	1647.23	2190.77	1900.91
Qinghai	573.40	330.01	346.55	366.79	396.53	391.06	416.94	507.48	519.98	463.93	503.29	437.81
Ningxia	506.34	646.52	633.60	821.17	953.18	702.07	628.94	626.39	693.35	909.82	558.23	698.15
Xinjiang	1144.42	1671.97	1927.40	2078.28	2602.63	1955.34	1928.41	1909.97	1693.89	1875.38	1910.06	1881.61
Liaoning	6602.81	10,681.55	9277.68	16,766.37	16,541.03	6068.53	5681.10	3467.76	3314.42	3084.19	3153.76	7694.47
Jilin	1741.40	2005.27	4815.41	7033.20	7466.20	2607.52	1857.23	1547.13	2251.32	2182.85	1243.56	3159.19
Heilongjiang	1432.66	2111.97	1653.23	1704.99	1736.29	1274.77	1270.41	1186.11	1199.92	2057.54	1428.20	1550.55

in Appendix A, and the total construction industry output of 30 provinces and cities in China is presented in

Table A2. Total construction industry output of 30 provinces and cities in China from 2010 to 2020 (Unit: 108 RMB).

Provinces	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	Average
Beijing	5196.02	6046.22	6588.30	7407.09	8209.80	8436.73	8841.19	9736.71	10,939.76	11,999.36	12,905.87	8755.18
Tianjin	2424.49	2986.45	3258.57	3670.53	4123.49	4488.90	4891.81	4262.35	3791.10	4096.50	4388.79	3853.00
Hebei	3231.46	3972.66	4865.09	5203.92	5625.75	5252.57	5517.69	5655.96	5740.25	5847.97	5948.09	5169.22
Shanghai	4300.19	4586.28	4843.44	5102.84	5499.94	5652.47	6046.19	6426.42	7072.21	7812.47	8277.04	5965.41
Jiangsu	12,405.92	15,122.85	18,423.55	21,712.16	24,592.93	24,785.81	25,791.76	27,956.71	30,846.66	33,099.18	35,251.64	24,544.47
Zhejiang	12,007.89	14,907.42	17,332.74	20,066.42	22,668.19	23,980.59	24,989.37	27,235.83	28,756.20	20,390.20	20,938.61	21,206.68
Fujian	2935.94	3692.62	4424.54	5459.41	6689.21	7605.81	8531.45	9993.65	11,548.82	13,164.42	14,118.01	8014.90
Shandong	5496.59	6482.90	7281.33	8332.70	9313.46	9381.72	10,087.43	11,477.75	12,898.29	14,269.29	14,947.30	9997.16
Guangdong	4715.46	5774.01	6514.43	7729.24	8356.50	8865.68	9652.31	11,372.05	13,714.37	16,633.41	18,429.84	10,159.75
Hainan	199.48	255.47	283.11	285.31	276.33	278.63	307.76	322.76	339.22	365.98	391.37	300.49
Shanxi	2143.46	2324.91	2668.17	2983.81	3103.49	2931.26	3318.47	3566.57	4071.46	4653.28	5113.64	3352.59
Anhui	2864.96	3597.26	4230.44	4970.34	5482.93	5695.94	6047.29	6829.67	7888.45	8503.26	9365.12	5952.33
Jiangxi	1690.02	2095.47	2789.57	3459.53	4122.63	4602.49	5179.03	6166.81	6993.40	7944.80	8649.16	4881.17
Henan	4400.61	5279.36	6009.08	7082.37	7911.89	8047.65	8807.99	10,086.58	11,360.52	12,701.68	13,122.56	8619.12
Hubei	4345.20	5586.45	7043.42	8343.40	10,059.59	10,592.86	11,862.40	13,390.73	15,133.87	16,979.67	16,136.11	10,861.24
Hunan	3161.73	3915.02	4407.92	5255.98	6020.97	6630.82	7304.22	8423.00	9581.44	10,800.62	11,864.03	7033.25
Inner Mongolia	1125.58	1394.68	1441.00	1540.48	1402.93	1123.47	1220.81	1122.19	1040.12	1086.06	1134.44	1239.25
Guangxi	1222.31	1553.07	1867.06	2271.39	2608.91	2953.42	3449.19	4210.07	4671.72	5407.31	5853.24	3278.88
Chongqing	2534.36	3328.83	3975.67	4731.88	5552.21	6256.94	7035.81	7605.66	7819.42	8222.96	8974.97	6003.52
Sichuan	4163.07	5256.65	6240.33	7239.49	8066.66	8768.24	9959.68	11,400.34	12,983.75	14,668.15	15,612.70	9487.19
Guizhou	622.96	824.72	1039.22	1365.00	1640.24	1947.74	2362.95	2932.96	3329.98	3714.89	4080.24	2169.17
Yunnan	1510.96	1868.40	2383.66	2888.82	3054.67	3268.93	3867.22	4726.36	5458.52	6122.09	6724.82	3806.77
Shaanxi	3063.61	3216.63	3529.39	3993.81	4557.71	4752.61	5329.23	6227.47	7120.15	7883.89	8501.13	5288.69
Gansu	751.99	925.84	1364.63	1708.27	1814.52	1849.02	1947.24	1825.42	1796.43	1916.35	2049.28	1631.73
Qinghai	279.61	319.42	325.76	396.39	432.91	409.51	410.62	406.93	435.14	460.72	512.24	399.02
Ningxia	342.69	427.92	466.95	564.66	625.16	524.53	511.25	549.21	565.04	601.41	641.81	529.15
Xinjiang	963.72	1320.37	1622.31	2071.52	2306.28	2255.74	2258.24	2418.70	2110.05	2278.17	2693.12	2027.11
Liaoning	4690.31	6217.52	7547.39	8743.37	7851.12	5413.76	3926.71	3688.33	3528.41	3554.45	3815.32	5361.52
Jilin	1348.78	1626.65	1990.43	2200.15	2521.00	2216.31	2283.56	2218.37	2183.63	1863.10	2005.78	2041.62
Heilongjiang	1769.70	2029.16	2373.96	2450.57	2150.75	1680.39	1716.61	1560.07	1194.28	1181.35	1206.37	1755.75

in Appendix A.

Table A1 illustrates that there is a substantial variation in carbon emissions across China. Zhejiang, Jiangsu, Sichuan, Fujian, and Hebei are the top five provinces with average carbon emissions above 150 million tons. Among them, Zhejiang (305.695 million tons) and Jiangsu (255.328 million tons) exhibit the highest carbon emissions, accounting for 13.11% and 10.95% of the national total, respectively. In contrast, Hainan, Qinghai, and Ningxia are the bottom three provinces with average carbon emissions below 10 million tons. Notably, Hainan demonstrates the lowest average carbon emissions of only 3.862 million tons (accounting for 1.26% of Zhejiang), significantly below other regions in China. By examining Table A2, it is found that the TCEs in most provinces is positively correlated with the total output from the construction industry, which is consistent with Du et al. [72], who emphasized a positive correlation between carbon emissions and economic growth in most provinces in China.

3.1.2. Results of the CEICI

Based on Formula (8) and relevant data, ArcGIS was used to depict the spatial distribution of the average CEICI from 2010 to 2020, as shown in Figure 1.

The results reveal pronounced regional differences in the CEICI of China. In particular, the CEICI is found to be higher in western and central regions, which are regarded as less developed areas with a relatively low per capita GDP (data from the China Statistical Yearbook [65]), whereas the CEICI is found to be lower in well-developed eastern regions such as Beijing and Shanghai. As an exception, Fujian manifests a high level of CEICI, which can be attributed to the mounting carbon emissions arising from the expanding construction enterprises coupled with a subdued construction output in this area. Additionally, it is found that Hebei, Fujian, Sichuan, Guizhou, and Inner Mongolia exhibit a notably heightened CEICI, which aligns with the findings of related studies [73]. For example, Zhang et al. [7] noted an increased carbon emission intensity in western China.

3.2. The Static CEECI

3.2.1. Results of the CEECI

According to Formulas (9)–(10), Matlab was used to calculate the CEECI for each province in China with the super-efficient SBM model. Then, ArcGIS was further used to visually present the results of the CEECI, allowing for a detailed analysis of its spatial distribution characteristics. Due to technical limitations, it is not feasible to plot the CEECI bar charts for all years on a single map. Therefore, we have selected specific cross-sectional years, namely 2010, 2013, 2017, and 2020, for our analysis [39]. Figure 2 illustrates the average spatial distribution of the CEECI from 2010 to 2020, along with the bar chart depicting the CEECI for the years of 2010, 2013, 2017, and 2020.

Figure 2 and the annual statistics (see

Table A3. CEECI by provinces from 2010 to 2020.

Provinces	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	Average
Beijing	1.300	1.370	1.396	1.428	1.405	1.475	1.460	1.452	1.465	1.479	1.553	1.435
Tianjin	1.043	1.101	1.054	1.065	1.065	1.002	1.018	0.569	0.384	0.452	0.423	0.834
Hebei	0.446	0.451	0.415	0.505	0.676	0.540	0.593	0.540	0.538	0.623	0.599	0.539
Shanghai	1.125	1.114	1.033	1.036	1.038	1.126	1.138	1.120	1.180	1.267	1.165	1.122
Jiangsu	0.825	0.679	0.628	0.789	0.805	1.013	1.036	1.041	1.060	1.089	1.134	0.918
Zhejiang	1.064	1.097	1.089	1.111	1.117	1.086	1.086	1.058	1.105	0.739	0.751	1.028
Fujian	0.587	0.695	0.610	0.593	0.552	0.567	0.561	0.531	0.497	1.032	1.043	0.661
Shandong	0.408	0.455	0.359	0.447	0.524	0.552	0.556	0.571	0.601	0.650	0.589	0.519
Guangdong	0.467	0.520	0.414	0.492	0.523	0.552	0.554	0.525	0.488	0.656	0.548	0.522
Hainan	1.111	1.233	1.225	1.125	0.682	1.017	1.006	0.696	0.521	0.534	0.509	0.878
Shanxi	0.487	0.542	0.459	0.490	0.514	0.506	0.524	0.510	0.506	0.622	0.549	0.519
Anhui	0.522	0.605	0.497	0.501	0.545	0.685	0.630	0.568	0.535	0.512	0.585	0.562
Jiangxi	1.005	0.900	1.032	0.825	1.171	0.774	0.781	0.710	0.661	1.018	1.021	0.900
Henan	0.585	0.661	0.506	0.603	0.619	0.605	0.633	0.612	0.521	0.662	0.580	0.599
Hubei	0.533	0.585	0.464	0.538	0.620	0.647	0.707	0.675	1.048	1.064	1.048	0.721
Hunan	0.622	1.027	0.669	1.015	1.027	0.642	0.655	0.623	0.581	0.632	0.575	0.733
Inner Mongolia	0.495	0.500	0.421	0.428	0.380	0.334	0.332	0.294	0.243	0.290	0.278	0.363
Guangxi	0.719	1.012	1.020	1.121	1.141	1.099	1.152	1.180	1.040	1.110	1.108	1.064
Chongqing	0.690	0.690	0.590	0.605	0.759	0.848	0.830	0.754	0.750	0.783	0.762	0.733
Sichuan	0.494	0.545	0.468	0.493	0.502	0.595	0.562	0.506	0.482	0.583	0.490	0.520
Guizhou	0.438	0.534	0.458	0.431	0.458	0.422	0.354	0.403	0.354	0.367	0.336	0.414
Yunnan	0.490	0.576	0.423	0.442	0.437	0.530	0.513	0.517	0.488	0.571	0.456	0.495
Shaanxi	0.691	0.707	0.593	0.559	0.550	0.620	0.639	0.653	0.596	0.656	0.584	0.622
Gansu	0.447	0.470	0.442	0.477	0.486	0.569	0.546	0.454	0.392	0.383	0.328	0.454
Qinghai	0.527	0.575	0.396	0.460	0.463	0.468	0.390	0.338	0.316	0.367	0.347	0.422
Ningxia	0.557	0.615	0.535	0.533	0.540	0.484	0.464	0.420	0.424	0.449	0.471	0.499
Xinjiang	0.760	0.813	0.667	0.541	0.568	0.549	0.744	0.550	0.389	0.414	0.428	0.584
Liaoning	0.641	0.804	0.643	0.482	0.555	0.511	0.400	0.421	0.423	0.487	0.450	0.529
Jilin	1.067	1.013	0.433	0.540	1.019	0.593	0.673	0.681	0.523	0.451	0.493	0.681
Heilongjiang	1.277	1.122	1.187	1.147	1.110	1.007	1.011	0.702	0.436	0.406	0.394	0.891

in Appendix A) reveal significant variations in the CEECI in China, with high levels mainly distributed in eastern and coastal areas and low levels mainly distributed in western areas. The average CEECI in Beijing (1.435), Shanghai (1.122), Zhejiang (1.028), and Guangxi (1.064) from 2010 to 2020 exceeded 1.0, indicating a higher level of carbon efficiency in comparison to other provinces in China. In contrast, provinces such as Inner Mongolia, Gansu, Qinghai, and Guizhou in the western region demonstrate low average annual efficiency levels of less than 0.5. The average annual CEECI in the western region of Inner Mongolia is 0.363, ranking the least among 30 provinces and cities in China.

3.2.2. Spatial Correlation Analysis

Moran’s I index is an important measure of spatial autocorrelation [74,75], which is divided into a global Moran’s I index and a local index. Specifically, the global Moran’s I index is used to detect the spatial pattern of the whole region, while the local index is used to reveal the heterogeneity of the local region [76]. To further analyze the spatial correlation in each province, GeoDa software was used to calculate the global Moran’s I of the CEECI from 2010 to 2020 based on the geographic distance spatial weight matrix, which is used to describe the average correlation degree observed in the whole region with the surrounding areas [77].

When −1 < Moran’s I < 0, the spatial correlation is negative. When 0 < Moran’s I < 1, the space is positively correlated. The calculation results are shown in

Table 5. Global Moran’s I for CEECI from 2010 to 2020.

Year	Moran’s I	p-Value	Z-Value	E[I]	Mean	sd
2010	0.2529	0.023	2.3583	−0.0357	−0.0344	0.1218
2011	0.1625	0.056	1.6087	−0.0357	−0.0344	0.1224
2012	0.1130	0.108	1.2471	−0.0357	−0.0359	0.1194
2013	0.1937	0.036	1.9185	−0.0357	−0.0364	0.1199
2014	0.2220	0.028	2.1057	−0.0357	−0.0379	0.1235
2015	0.2410	0.016	2.3652	−0.0357	−0.0391	0.1185
2016	0.2213	0.026	2.1764	−0.0357	−0.0407	0.1204
2017	0.1554	0.056	1.7155	−0.0357	−0.0427	0.1155
2018	0.1983	0.035	2.0591	−0.0357	−0.0438	0.1176
2019	0.2049	0.022	2.1295	−0.0357	−0.0452	0.1174
2020	0.1772	0.040	1.9105	−0.0357	−0.0453	0.1165

.

The results of the global Moran’s I reveal that every year, except for 2012, passed the significance test. The global Moran’s I of the CEECI is consistently greater than 0, indicating an obvious positive spatial autocorrelation. To further examine the correlation between the internal spatial units and the adjacent units, a local spatial correlation test was conducted. Due to space constraints, Figure 3 displays the Moran local scatter plots for the years 2010, 2013, 2017, and 2020.

Figure 3 illustrates that most provinces and cities are distributed in the first quadrant or the third quadrant, which indicates that there is a significant positive spatial correlation between the clustering of high CEECI areas and the clustering of low CEECI areas. The provinces and cities in the first quadrant are mostly located in the eastern region while those in the third quadrant are mostly located in the western region, indicating a mutual driving effect of the CEECI among the 30 provinces and cities. In general, the local distribution characteristics of the CEECI in each province exhibited relatively minor changes during 2010–2020. The number of provinces and cities in the first and third quadrants of the index remained relatively stable, and the overall spatial characteristics did not exhibit substantial changes between 2010 and 2020.

3.2.3. Results of the CRPCI

In accordance with Formulas (9) and (10), the relaxation variable S^z of carbon emissions was obtained, and the ratio of the relaxation variable (S^z) to the TCEs is the CRPCI [78,79,80]. The CRPCI represents the potential to reduce greenhouse gas emissions and achieve sustainable development goals through the implementation of emission reduction measures and technological innovations [81]. As the building industry is a significant contributor to carbon emissions, reducing its carbon footprint can lead to environmental improvements such as air quality enhancements, resource conservation, and economic benefits [82]. Therefore, quantifying the CEECI and evaluating the CRPCI can disclose the current challenges related to high carbon emissions and identify opportunities for improvement [83]. This assessment aids in the development and implementation of appropriate emission reduction measures, policies, and technological innovations, promoting the transition of the building industry to a low-carbon trajectory, reducing greenhouse gas emissions, and enhancing resource utilization efficiency. A larger CRPCI means that the region has excessive carbon emissions with more space for its reduction. The CRPCI of China was measured from 2010 to 2020, and the results are shown in

Table 6. The carbon emission reduction potential in the construction industry of 30 provinces and cities in China (2010–2020) (Unit: %).

Provinces	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	Average
Inner Mongolia	60.11	38.24	62.47	56.84	63.45	80.77	83.02	87.25	86.08	58.88	75.65	68.43
Guizhou	51.54	36.43	57.88	69.04	55.91	66.07	66.69	71.09	79.35	58.06	71.33	62.13
Ningxia	50.48	38.72	52.73	55.96	53.08	57.48	68.59	64.57	67.80	55.35	48.99	55.80
Sichuan	70.25	56.39	78.90	78.10	54.83	28.59	39.31	28.47	37.51	49.01	62.04	53.04
Hebei	78.07	88.68	92.00	79.61	35.97	35.51	27.35	40.05	34.99	27.64	43.29	53.02
Liaoning	42.49	22.49	39.35	69.29	59.27	44.95	70.04	51.40	56.59	16.84	44.90	47.06
Qinghai	60.52	15.73	36.56	28.97	27.82	41.78	56.66	70.32	74.27	39.00	64.68	46.94
Gansu	37.58	54.69	44.91	45.96	56.22	26.12	24.70	42.35	60.44	35.61	66.40	45.00
Shanxi	61.39	31.42	41.22	44.80	44.86	32.80	57.51	58.07	55.39	9.94	40.45	43.44
Yunnan	47.88	23.62	38.06	68.53	65.52	36.10	39.11	35.01	39.91	25.24	56.79	43.25
Guangdong	56.82	56.53	54.32	49.22	48.90	34.65	26.77	34.41	39.99	2.83	36.74	40.11
Anhui	48.68	23.83	41.69	45.90	54.15	13.99	25.90	30.41	26.47	65.11	59.00	39.56
Shaanxi	21.61	40.37	47.69	46.13	39.97	26.00	34.16	31.02	41.90	33.21	56.60	38.06
Xinjiang	28.75	14.62	40.84	45.06	44.15	38.79	19.96	33.75	58.94	25.23	49.03	36.28
Shandong	53.80	30.13	77.00	45.77	38.01	35.02	24.99	21.00	17.08	8.92	18.68	33.67
Guangxi	30.98	35.34	33.91	34.27	35.63	38.30	34.16	35.63	28.49	30.09	25.40	32.93
Hubei	27.08	41.75	71.47	64.32	51.32	28.00	43.76	32.68	0.00	0.00	0.00	32.76
Fujian	40.55	11.38	34.39	39.52	38.57	36.35	39.43	53.45	55.79	0.00	0.00	31.77
Jilin	0.00	0.00	72.10	79.76	0.00	30.02	3.96	0.63	44.34	38.53	34.85	27.65
Henan	38.54	12.45	52.27	31.33	58.91	0.73	0.00	14.33	28.42	19.01	38.56	26.78
Chongqing	43.80	26.47	47.75	47.77	22.26	6.91	7.17	0.00	13.61	14.24	25.45	23.22
Heilongjiang	0.00	0.00	0.00	0.00	0.00	0.00	0.00	21.75	62.86	66.62	66.65	19.81
Jiangxi	14.05	16.09	14.39	13.74	15.24	18.38	12.76	23.39	29.44	27.68	21.19	18.76
Hunan	44.25	0.00	13.81	11.78	0.00	11.76	10.88	9.16	15.89	31.59	38.95	17.10
Hainan	0.00	0.00	0.00	0.00	28.63	0.00	0.00	5.88	47.25	41.71	53.14	16.05
Tianjin	0.00	0.00	0.00	0.00	0.00	0.00	0.00	38.35	70.60	12.10	48.59	15.42
Jiangsu	4.51	33.27	41.86	1.36	19.25	0.00	0.00	0.00	0.00	0.00	0.00	9.11
Zhejiang	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Beijing	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Shanghai	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00

.

This result shows the wide variation in the CRPCI values in different regions of China. It is worth noting that Inner Mongolia (68.43), Guizhou (62.13), Ningxia (55.80), Sichuan (53.04), and Hebei (53.02) have higher CRPCI values, with an average annual rate of more than 50%. On the other hand, the CRPCI value of Zhejiang, Beijing, and Shanghai is 0, which is significantly lower than other provinces and cities in China. In addition, the average annual CRPCI in Jiangsu is relatively small at less than 10%.

3.3. The Dynamic CEECI

By considering the static characteristics, the dynamic CEECI of 30 provinces and cities in China was further analyzed based on the ML index (see

Table 7. ML index of the construction industry in 30 provinces and cities in China (2010–2020).

Provinces	2010–2011	2011–2012	2012–2013	2013–2014	2014–2015	2015–2016	2016–2017	2017–2018	2018–2019	2019–2020	Average
Beijing	1.142	1.131	1.019	0.988	1.009	0.993	0.996	1.004	0.993	1.188	1.047
Tianjin	1.081	0.838	0.984	0.899	0.555	1.228	0.597	0.728	1.062	1.117	0.909
Hebei	0.982	0.985	1.132	1.120	0.648	0.887	0.617	1.047	1.082	1.139	0.964
Shanghai	0.925	0.972	0.887	0.998	1.073	0.987	0.972	1.174	0.982	1.026	1.000
Jiangsu	0.690	1.036	1.126	0.983	1.008	0.939	1.027	0.971	1.010	0.938	0.973
Zhejiang	1.030	1.011	0.997	1.003	0.974	0.988	0.984	1.041	0.646	1.077	0.975
Fujian	1.148	0.936	0.930	0.904	0.989	0.955	0.984	0.933	0.981	0.997	0.976
Shandong	1.001	0.913	1.167	0.993	0.917	0.972	1.025	1.102	0.960	1.275	1.033
Guangdong	1.010	0.895	1.122	0.935	0.930	0.994	0.923	1.002	1.152	1.036	1.000
Hainan	1.073	0.992	0.769	0.589	0.961	0.832	0.678	0.822	0.945	1.128	0.879
Shanxi	1.030	0.973	1.001	0.932	0.776	1.034	0.950	1.072	1.061	1.094	0.992
Anhui	0.991	0.970	0.928	0.938	1.079	0.866	0.922	0.961	0.866	1.259	0.978
Jiangxi	0.859	1.061	0.878	1.399	0.623	0.935	0.958	0.918	1.210	1.005	0.985
Henan	0.930	0.907	1.033	0.895	0.841	0.954	1.014	0.864	1.168	0.984	0.959
Hubei	0.984	0.908	1.082	0.987	0.759	1.010	0.804	1.600	1.109	0.990	1.023
Hunan	1.666	0.655	1.004	1.025	0.573	0.955	0.985	0.953	1.000	0.962	0.978
Inner Mongolia	0.985	0.876	0.962	0.819	0.793	0.997	0.878	0.908	1.165	1.076	0.946
Guangxi	1.028	1.138	1.123	0.995	0.962	1.030	1.061	0.882	0.996	1.026	1.024
Chongqing	0.907	1.067	0.979	1.095	1.178	0.876	0.956	0.988	0.971	1.204	1.022
Sichuan	1.018	0.976	1.012	0.927	1.100	0.919	0.908	0.985	1.119	0.951	0.991
Guizhou	1.088	1.009	0.892	0.931	0.871	0.813	1.121	0.943	0.988	1.001	0.966
Yunnan	1.051	0.877	0.970	0.862	1.076	0.945	0.997	0.994	1.063	0.919	0.975
Shaanxi	0.995	0.768	0.879	0.849	0.778	0.994	1.027	0.941	1.010	0.993	0.923
Gansu	0.930	1.100	1.004	0.881	0.999	0.922	0.829	0.915	0.884	0.959	0.942
Qinghai	0.984	0.821	1.079	0.874	0.864	0.834	0.852	1.000	1.069	1.093	0.947
Ningxia	1.039	1.004	0.917	0.891	0.639	0.957	0.876	1.114	1.001	1.217	0.966
Xinjiang	1.014	0.894	0.952	1.005	0.844	0.817	0.775	0.533	0.935	1.247	0.902
Liaoning	1.270	0.958	1.161	0.509	0.616	0.778	1.032	1.087	1.007	1.113	0.953
Jilin	0.980	0.453	1.173	1.265	0.552	1.104	0.992	0.832	0.789	1.681	0.982
Heilongjiang	0.878	1.087	0.924	0.956	0.725	0.997	0.690	0.642	0.882	1.064	0.884

) obtained from Formula (11). The ML index of the construction industry in 30 provinces and cities in China is decomposed by Formulas (12)–(17).

Table 8. ML index decomposition of the construction industry in 30 provinces and cities in China for major years.

Provinces	2010–2011			2013–2014			2016–2017			2019–2020
	MLTC	MLPEC	MLSEC	MLTC	MLPEC	MLSEC	MLTC	MLPEC	MLSEC	MLTC	MLPEC	MLSEC
Beijing	1.084	1.049	1.005	1.004	0.980	1.004	1.002	0.992	1.002	1.130	1.049	1.001
Tianjin	1.019	1.052	1.003	0.900	0.999	1.001	0.981	0.574	0.973	1.193	0.922	1.016
Hebei	0.970	0.932	1.086	0.837	1.010	1.325	0.989	0.894	1.017	1.185	0.957	1.004
Shanghai	0.935	0.990	1.000	0.996	0.999	1.003	0.988	0.984	1.000	1.116	0.916	1.004
Jiangsu	0.838	0.920	0.894	0.963	0.984	1.037	1.022	1.017	0.988	0.901	0.984	1.058
Zhejiang	0.999	1.043	0.989	0.997	1.018	0.988	1.010	0.982	0.992	1.059	1.031	0.986
Fujian	0.970	1.193	0.992	0.972	0.942	0.988	1.039	0.972	0.974	0.986	0.991	1.019
Shandong	0.898	1.058	1.054	0.848	1.031	1.136	0.998	1.054	0.974	1.407	0.965	0.939
Guangdong	0.907	1.088	1.024	0.880	0.961	1.105	0.974	0.947	1.000	1.240	1.287	0.649
Hainan	0.967	0.670	1.657	0.971	1.063	0.571	0.978	1.145	0.604	1.184	1.141	0.835
Shanxi	0.926	1.086	1.025	0.888	1.049	1.000	0.976	0.982	0.991	1.240	0.883	1.000
Anhui	0.854	1.089	1.065	0.863	0.955	1.139	1.022	0.912	0.989	1.102	1.226	0.932
Jiangxi	0.959	0.991	0.904	0.986	1.175	1.207	1.055	0.910	0.998	1.002	1.006	0.997
Henan	0.823	1.391	0.812	0.873	0.866	1.185	1.049	1.013	0.954	1.123	1.000	0.876
Hubei	0.896	0.988	1.110	0.857	1.011	1.140	0.834	0.957	0.997	1.006	0.996	0.989
Hunan	1.009	1.650	1.001	1.017	1.011	1.001	1.035	0.981	0.970	1.057	1.163	0.783
Inner Mongolia	0.976	1.020	0.989	0.921	0.913	0.973	0.994	0.894	0.987	1.123	0.951	1.007
Guangxi	0.730	1.043	1.349	0.978	1.060	0.960	1.036	1.025	0.999	1.028	0.942	1.059
Chongqing	0.907	0.993	1.007	0.873	1.186	1.058	1.052	0.945	0.962	1.238	1.022	0.952
Sichuan	0.922	1.032	1.071	0.910	0.977	1.042	1.008	0.931	0.968	1.133	0.901	0.932
Guizhou	0.892	1.159	1.052	0.875	1.072	0.992	0.985	1.103	1.032	1.096	0.922	0.991
Yunnan	0.894	1.179	0.997	0.871	0.988	1.001	0.988	0.996	1.013	1.152	0.803	0.993
Shaanxi	0.997	1.284	0.797	0.863	0.978	1.006	1.006	1.007	1.015	1.115	0.894	0.997
Gansu	0.883	0.994	1.060	0.864	1.017	1.002	0.995	0.853	0.976	1.121	0.845	1.013
Qinghai	0.902	1.165	0.936	0.870	0.829	1.213	0.982	0.885	0.981	1.155	0.985	0.961
Ningxia	0.941	0.621	1.778	0.880	1.091	0.929	0.968	0.934	0.969	1.161	1.120	0.936
Xinjiang	1.156	0.817	1.307	0.883	1.032	1.017	0.846	0.761	0.972	1.204	1.013	1.022
Liaoning	1.012	0.998	1.258	0.873	1.328	0.868	0.980	1.067	0.987	1.206	0.920	1.004
Jilin	1.034	0.973	0.976	0.670	1.873	1.007	0.980	1.022	0.990	1.536	1.128	0.970
Heilongjiang	0.999	0.859	1.024	0.988	0.954	1.014	0.994	0.978	0.711	1.095	1.007	0.965

presents the results for the years of 2010–2011, 2013–2014, 2016–2017, and 2019–2020.

Overall, the ML index of most provinces and cities is greater than or close to 1, and the ML value of the eastern region is relatively lower than that of the central and western regions. According to the ML index decomposition model, the CEECI in the eastern region is affected by MLTC and MLPEC, while the increase of MLTC, MLPEC, and MLSEC in the western region is conducive to the increase of the CEECI, especially MLTC.

4. Discussion

4.1. Discussion and Analysis of TCEs

Due to limited studies attempting to achieve a systematic calculation of carbon emissions in various provinces in China, this study selected more comprehensive energy sources to achieve an accurate estimation of DCEs and employed the carbon emission coefficient method to measure ICE². In order to further explore the results of the proportion of the OCEs and MCEs in the TCEs of Chinese provinces and cities, the average carbon emission and its composition histogram, as shown in Figure 4, were further analyzed.

By comparing the collective share of the carbon emissions during the building operation stage (OCEs) and the share of the carbon emissions during the building materialization stage (MCEs) within the TCEs, it is found that the proportion of the MCEs is significantly higher than that of the OCEs in all provinces. The MCEs involve emissions generated from the construction stage, along with upstream industries (production and transportation of materials) and downstream industries (e.g., demolition). This is consistent with existing research findings [13,84]. For example, Du et al. [73] found that the MCEs accounted for about 90% of the TCEs. In addition, the prevalence of lower temperatures, extensive centralized heating areas, and the large proportion of heating energy within building operations contribute to a relatively elevated proportion of the OCEs in some specific northern provinces (e.g., Inner Mongolia, Tianjin, and Qinghai).

4.2. Discussion and Analysis of the Static CEECI

From the results of the CEECI and its spatial distribution map in Figure 2, it can be seen that the level of the CEECI in the eastern region is significantly higher than that in the central and western regions. This may be attributed to the heightened focus and strategic emphasis on the developed cities (Beijing, Shanghai, and Zhejiang) at the national level, along with a set of carbon emission reduction activities and a higher level of energy mix measures. Beijing, Shanghai, and Zhejiang are committed to promoting the development of clean energy, especially wind and solar energy, and increasing the proportion of renewable energy in their energy structures. In addition, the government has implemented incentive measures for green building projects to further promote carbon reduction activities [85]. The higher CEECI level in Guangxi Province is mainly attributed to its focus on green and energy-efficient building developments, resulting in significantly lower carbon emissions compared to other provinces and cities with similar building scales. For instance, Guangxi Province has issued the “Guangxi Green Building Action Plan” and established an energy-saving supervision system [86]. In addition, the relatively poor carbon emission efficiency of areas such as Inner Mongolia may be due to its unreasonable energy structure and the lack of technical personnel in the construction industry in the western areas. Mengna et al. [9] also found that the lower CEECI in Inner Mongolia was mainly attributed to conventional coal-dominated energy structures. This relationship underscores the positive correlation between regional economic developments and CEECI levels.

In general, the eastern region has a relatively open economy with a mature construction industry and a balanced allocation of production factors. Notably, a significant proportion of energy consumption in the eastern region comprises low carbon emission coefficient sources, such as natural gas and hydropower, contributing to the high level of CEECI in this area. On the contrary, the low level of CEECI in the western region can be attributed to its relatively less-advanced economic developments, which results in a less attractive environment for high-quality talents in the construction industry, leading to lower quality construction personnel and weaker technological advancements. Yang [27] also highlighted that although the CEECI is higher in the eastern region than in the western region, most provinces in China are experiencing a fluctuating upward trend. Du et al. [39] explained that the reasons for the lower CEECI levels in marginal cities (e.g., Xinjiang and Jilin) are the economic downturn, a continuous contraction of the construction market, and the lagging optimization of the energy structure, along with poor management practices in those areas.

4.3. Discussion and Analysis of the CRPCI

In order to better analyze and show the distribution of the CRPCI in different regions of China, the annual spatial distribution of the CRPCI was drawn, as shown in Figure 5.

It can be seen from Figure 5 that the central and western regions demonstrate a higher CRPCI compared to the eastern region. Specific provinces and cities in eastern China, including Shanghai, Beijing, and Zhejiang, demonstrate zero potentials, indicating that they have already achieved an advanced level of CRPCI. In contrast, the central and western regions, such as Inner Mongolia, Guizhou, Ningxia, Sichuan, and Hebei, exhibit an average CRPCI exceeding 50%, highlighting a substantial scope for improvement in energy conservation and carbon emission reduction in these provinces. This finding is consistent with existing studies suggesting that there is great potential for emissions reduction in heating districts such as Inner Mongolia and Hebei [87]. Moreover, an examination of the CEECI reveals an inverse correlation with the CRPCI; that is, provinces and cities with a higher CRPCI tend to have a lower CEECI, whereas those with a lower potential exhibit a higher CEECI [88].

4.4. Discussion and Analysis of the Dynamic CEECI

Using the ML index model to study the dynamic CEECI, the ML index in the majority of the 30 provinces and cities is greater than or close to 1, suggesting a steady growth of the CEECI with a slowed-down growth rate in recent years. Some studies have also indicated that areas with a higher CEECI exhibit a relatively lower ML index and slower CEECI growth [89]. The eastern provinces generally have lower ML values with a slow growth rate, which may be due to the fact that they have achieved an excellent CEECI level [12]. Additionally, Beijing and Shanghai are exhibiting high values of ML while maintaining a high level of CEECI. This is mainly attributed to the advanced economic development of Beijing and Shanghai, which has driven local governments to allocate substantial funds for the adoption of low-energy and renewable building materials and to establish prominent standards and regulations for the green development of the construction sector [90,91]. For example, Beijing has issued a standard specification on the management of green building incentive funds [86], while Shanghai has mandated the adoption of assembly-type constructions for new civil and industrial buildings [92].

It can be concluded from Table 8 that the enhancement of the CEECI in all provinces and cities is mainly due to the impact of MLTC, which is consistent with existing studies [93,94]. From a regional perspective, the CEECI in the eastern region is affected by MLTC and MLPEC, and the decision-making unit is constantly moving to the production frontier and approaching the optimal scale. The construction industry in the eastern region has reached a certain scale, the economic development level is high, and the MLSEC has not significantly affected the improvement of the CEECI. The improvement of MLTC, MLPEC, and the MLSEC in the central and western regions is all conducive to promoting the improvement of the CEECI, indicating that these regions should pay particular attention to MLTC, accelerate the growth of low-carbon technologies, and improve the ML index. However, the ML index of the central and western regions as a whole shows a fluctuating upward trend, indicating that, with the continuous development of the construction industry in the central and western regions, it is approaching the optimal scale.

Despite the fact that this paper has conducted a relatively complete and systematic analysis of the CEECI of China, there are still some limitations in its calculation of MCEs and the scope of the analysis of the CEECI. First, the consumption data for building materials in China is primarily recorded in the China Statistical Yearbook on Construction [59], which only includes information on the consumption of steel, wood, cement, glass, and aluminum. Consequently, the calculated MCEs only take into account the carbon emissions generated by these five main building materials, potentially introducing a slight deviation into the calculation. Hence, it is recommended for researchers to further explore the impact of carbon emissions from building materials beyond these five main categories to determine the contribution of these additional materials to the total carbon emissions in the construction industry. This can provide a more comprehensive understanding of the carbon emissions profile of the sector. Second, conducting an analysis at the provincial level offers a macroscopic perspective but may overlook the subtle differences and variations within each province’s construction industry. Future research is recommended to focus on the municipal level within a province to investigate the variations of the CEECI among different cities. This can help evaluate the strengths and weaknesses of the emission reduction policies and measures implemented in these cities to complement the findings of the present study.

5. Conclusions and Policy Recommendations

This paper establishes a comprehensive carbon emission accounting model for the accurate measurement and analysis of carbon emissions in the construction industry in China. By employing the super-efficiency SBM model and the ML index model containing non-expected outputs, both the dynamic and static analyses of the CEECI in 30 provinces and cities were conducted. A spatial analysis of the CEECI and CRPCI was performed, which can provide valuable implications for policy designers to effectively mitigate carbon emissions. The main conclusions and policy recommendations are as follows:

(1) The TCEs of the Chinese construction industry shows a fluctuating upward trend, and the ${ICE}^2$ generated by the construction industry is the main component of the TCEs, indicating that the construction industry has a strong industry driving role. Consequently, the government should establish differentiated carbon emission reduction responsibility mechanisms for different industries and establish low-carbon management and implementation rules. For instance, the establishment of low-carbon awareness among different industries (e.g., steel and concrete) and the implementation of reward and penalty systems based on their levels of emissions can effectively mitigate the carbon emissions generated in the materialization stage.

(2) The CEECI varies significantly among provinces in China, and the CEECI in the eastern region is significantly higher than that in the central and western regions. The government should strengthen regional cooperation and mutually beneficial partnerships in order to overcome regional constraints and reduce disparities in the CEECI. Efforts should be made to facilitate the transfer of technology from the eastern regions to the central and western regions, with a particular focus on attracting skilled professionals in the construction industry. The positive role of high-efficiency areas in driving the development of low-efficiency areas should be fully utilized to achieve regionally coordinated developments. Additionally, since Beijing and Shanghai have the best CEECIs, their green building policies and regulations are advocated for adoption by other provinces and cities in China to facilitate green building developments nationwide.

(3) The CRPCI of China in different provinces varies significantly, and the CRPCI is found to be negatively correlated with the CEECI. The government should tailor economic and technological support based on the characteristics of each region, comprehensively taking local conditions into account. Provinces with a high CRPCI but a low CEECI should fully utilize the advantages of low-carbon policies and accelerate the development of clean technologies (e.g., renewable materials). Provinces with a high CRPCI should focus on optimizing industrial structures and resource allocations to further improve their CEECI. Taking inspiration from the clean energy initiatives promoted by Beijing, Shanghai, and Zhejiang, such as wind and solar power, efforts should be made to reduce reliance on traditional coal-dominated energy structures and to increase the proportion of renewable energy in the overall energy mix.

(4) From the dynamic decomposition results of the CEECI, MLTC is the main factor affecting the CEECI in each province and has a significant positive effect on the improvement of the CEECI. The government should increase scientific and technological investments and make overall plans. Technical issues such as heating and cooling efficiency and energy electrification should be prioritized. In addition, the construction industry in China’s provinces should strengthen technological transformations, such as replacing technical equipment that is not suitable for technical or economic use with new technical equipment.