The Spatial Relationship between CO₂ Emissions and Economic Growth in the Construction Industry: Based on the Tapio Decoupling Model and STIRPAT Model

Abstract

Studying the relationship between carbon emissions in the construction industry and economic growth in the construction industry is of great significance for the sustainable development of the industry and the country. This paper aims to analyze the decoupling status of various provinces, the spatial relationship between carbon emissions and economic growth, and provide targeted suggestions for promoting the green and sustainable development of the construction industry. The relevant panel data on CO₂ emissions from the construction industry in 30 provinces and municipalities during China’s “Thirteenth Five-Year Plan” period was collected and organized in this paper. A Tapio decoupling model and a spatial econometric model which is combined with the STIRPAT model are used in the paper. The research indicates that as a whole, China’s construction industry is in a state of weak decoupling, and actions to reduce carbon dioxide emissions, bear fruit, and energy efficiency have been improved; the economic growth still promotes CO₂ emissions and the relationship between the two is still in the rising phase of the CKC (Carbondioxide Kuznets Curve) curve; economic growth has significant spatial spillover effects, and economic growth in other regions will inhibit carbon dioxide emissions in the region, but its effect is lower than the promotion effect of direct effects. At this stage, the employee factor and technology development level also promote carbon dioxide emissions, and the spatial spillover effect is small. At the same time, the degree of influence of each influencing factor varies in different decoupling provinces.

1. Introduction

China’s economic development is inseparable from the rapid development of the construction industry. The proportion of the added value of the construction industry to GDP has grown significantly, from 6.64% in 2010 to 7.17% in 2019. The construction industry, as a pillar industry of the national economy, has made increasingly prominent contributions. It is well-known that economic growth is usually accompanied by the consumption of various resources. In particular, during the development process of the traditional construction industry, there are problems such as extensive construction methods, high pollution emissions, and high energy consumption. The carbon emission problem caused by high energy consumption is particularly serious, and the ratio of CO₂ emissions in the construction industry to the total emissions of the whole society is quite large [1]. The “China Building Energy Efficiency Development Report (2016): Building Energy Efficiency Operation Management” mentions that construction and operation energy consumption accounts for up to 36% of the total primary energy consumption of the whole society [2]. The “China Building Energy Consumption Research Report (2020)” indicated that the proportion of total carbon emissions in the whole process of construction increased to 51.3% of the total carbon emissions in China in 2018. Therefore, if China wants to achieve their “dual carbon” goals, including carbon peaking and carbon neutrality, it must first realize the low-carbon transformation of the construction industry, strengthen energy conservation, and promote emission reduction.

The State Council issued the “Thirteenth Five-Year Plan” Energy Conservation and Emission Reduction Comprehensive Work Plan. In this plan, the “dual control” target of total quantity and intensity during the “Thirteenth Five-Year Plan “period was set. China has shifted the goal of high-speed economic development to high-quality development, and the construction industry has also simultaneously explored a high-quality development path to seek low-carbon transformation. However, to promote economic growth, the consumption of resources is inevitable. At the same time, as the emission reduction work continues, it will become more and more difficult, leaving less and less space for emission reduction work. Based on this, this paper chooses to study the relationship between the CO₂ emissions and the economic growth of the construction industry during the “Thirteenth Five-Year Plan” period, make the goal of low-carbon development of the construction industry to be realized as soon as possible, and achieve the goal of reducing carbon emissions while taking into account the economic development, organically combining “low carbon” and “development” [3].

2. Literature Review

In the 1970s, Ehrlich et al. proposed the IPAT equation to assess environmental pressure and believed that the factors that caused the increasing pressure on resources and environment included population growth, material economic level improvement, and scientific and technological progress [4]. In research on the influencing factors of carbon emissions, generally, various scholars agree that the economic development is the main factor leading to the increase of carbon emissions [5]. The Environment Kuznets Curve (EKC) proposed by Grossman believes that the linear correlation curve between environmental quality and income is inverted U-shaped [6]. With the in-depth study of EKC by scholars, another relationship curve has gradually been derived, that is, the Carbon-dioxide Kuznets Curve (CKC) [7,8]. In addition, the decoupling theory proposed by Samouilidis and Mitropoulos in 1984 [9] was able to judge whether carbon emissions and economic growth develop synchronously, and is widely used to study the short-term dynamic correlation between the two [10]. In 2005, Tapio proposed the Tapio decoupling index system, which is based on the elastic method and decoupling theory, to research the greenhouse gas emissions, transportation volume, and industry output value increase of the European transportation industry, and subdivided the decoupling into eight different states [11]. This index system has been widely used in subsequent studies [12,13].

Various scholars have conducted practical research on the relationship, carbon emissions and economic growth, at the national level. For example, Klaus Hubacek [14] evaluated China’s population, economic level, technological level, and related carbon emission driving forces based on the IPAT equation. Its research results show that economic growth, one of the key influencing factors, determines the extent of carbon emissions. Although technological progress can also reduce carbon emissions, its effect is smaller than that of economic growth, and the impact of population growth is not significant. Saboori researched the linear relationship between per capita CO₂ emissions and per capita GDP in Malaysia and found an inverted U-shaped curve [15]. Kang et al. [16] studied the carbon dioxide Kuznets curve of economic growth and CO₂ emissions in China and found that it was an inverted-N shape. The research has also been refined to the industry level, and the construction industry has become a key area of low-carbon research due to its large carbon emissions and rapid economic scale expansion. Huang L et al. [17] studied the CO₂ emission levels caused by construction activities in 40 countries around the world based on the world environmental input–output table. Conrado et al. [18] calculated the cost-effectiveness of CO₂ emission reduction in the Brazilian construction industry. Onat calculated carbon emissions, which include the direct and the indirect ones, of buildings in the United States by using the life cycle method, and these buildings include residential and commercial buildings [19]. Feng Bo et al. [20] researched the influencing factors of carbon emissions in the construction industry and the decoupling state from 2004 to 2011 by using the Tapio decoupling model. With the rise of spatial metrology, research on the spatial effect of carbon dioxide emissions has been carried out by more and more scholars. Bai [21] investigated the relationship between the interregional flow of R&D factors, the spillover effect of spatial knowledge and China’s economic growth. The spatiotemporal characteristics of carbon emissions from the construction industry were analyzed by Fan et al. [22], and they used spatial autocorrelation and kernel density function methods. Zhang et al. (2018) [23] studied the spatial spillover effects of population urbanization on CO₂ emissions.

According to the literature review, In the study of the construction industry, few scholars have focused on the study of the relationship between carbon dioxide and economic growth through spatial effects. Therefore, the innovation of this paper is that it establishes a theoretical model, performs spatial regression analysis and decomposes spatial effects. It studies the impact of economic growth on carbon dioxide emissions and the spatial spillover effect, during the Thirteenth Five-Year Plan period, in China’s construction industry. The regions of different decoupling states are divided, and the spatial relationship between carbon emissions and economic growth in different decoupling provinces is innovatively analyzed, which will be conducive to the formulation of targeted strategies.

3. Theoretical Models and Data Sources

3.1. Theoretical Model

3.1.1. Construction of Decoupling Model

The Tapio decoupling model for carbon emissions is set as follows [24]:

$$e_{\left(C,GP\right)}=\frac{C_t-C_0}{C_0}/\frac{G{P}_t-G{P}_0}{G{P}_0}=\frac{\mathsf{\Delta }{C}_t/C_0}{\mathsf{\Delta }G{P}_t/G{P}_0}$$

(1)

In the formula, $e_{\left(C,GP\right)}$ is the decoupling elasticity value of the construction industry, C is the carbon dioxide emissions (10,000 tons), GP is the total output value (100 million yuan), the subscript t represents the current period, and the subscript 0 represents the base period.

Table 1. Eight different decoupling states of Tapio decoupling model.

Title 1	Decoupling State	Value of Decoupling Elasticity	ΔC	ΔGP
Negative decoupling state	Expansion negative decoupling state	e > 1.2	>0	>0
	Weak negative decoupling state	0 < e < 0.8	<0	<0
	Strong negative decoupling state	e < 0	>0	<0
Decoupling state	Recession decoupling state	e > 1.2	<0	<0
	Weak decoupling state	0 < e < 0.8	>0	>0
	Strong decoupling state	e < 0	<0	>0
Connection state	Growing connection state	0.8 < e < 1.2	>0	>0
	Decaying connection state	0.8 < e < 1.2	<0	<0

lists the judging criteria for 8 decoupling states.

3.1.2. Construction of STIRPAT Model

The STIRPAT equation is an improved IPAT equation, and the IPAT equation is expressed as:

$$\mathrm{I}=\mathrm{P}\times \mathrm{A}\times \mathrm{T}$$

(2)

From Equation (2), it can be seen that the premise of the IPAT equation is that population factors, wealth factors and technological development level factors have a fixed coefficient of influence on environmental pressure, which is obviously not in line with objective reality and has obvious limitations. Therefore, Dietz et al. proposed an improved IPAT equation, that is, the Stochastic Impacts by Regression on Population, Affluence and Technology (STIRPAT):

$$I_{it}=a{P}_{it}^b{A}_{it}^c{T}_{it}^d{e}_{it}$$

(3)

Among them, a represents the constant term of the model; b, c, and d, the superscripts, represent the influence coefficients of each variable on the environmental pressure, respectively, and e is the error interference term. To eliminate heteroskedasticity, take the logarithm of both sides of the STIRPAT equation to get:

$$Ln{I}_{it}=a_0 + bLn{P}_{it} + cLn{A}_{it} + dLn{T}_{it} + e_{it}$$

(4)

Based on the availability of data, the environmental impact factor index (I), the population factor (P) and the wealth factor (A) are measured by the following relevant data for the construction industry: the carbon dioxide emissions (C), the number of employees (LB), and the gross output value (GP). Technological development level (T) is measured by technical equipment rate (TER) and industrial structure (IS). The available empirical model is:

$$Ln{C}_{it}=a_0 + bLnL{B}_{it} + cLnG{P}_{it} + d_{1}LnTE{R}_{it }+ d_{2}LnI{S}_{it} + e_{it}$$

(5)

3.2. Data Sources

Except for carbon dioxide emissions, the rest of the variable data are compiled from the data published directly on the official website, some from the “China Statistical Yearbook”, and the other from the “China Construction Industry Statistical Yearbook” over the years. The industrial concentration of the construction industry, the proportion of the total output value of general contractors with special and first-class qualifications in each region to the total output value of all contractors, is used to measure the industrial structure. The methodology for measuring CO₂ emissions is described below.

3.2.1. Measurement of Carbon Dioxide Emissions

CO₂ emissions from buildings include the sum of direct and indirect CO₂ emissions (Equation (6)). Direct emissions include carbon dioxide emissions from the utilization of fossil energy sources and the use of electricity in the production process, while indirect emissions refer to CO₂ emissions from the consumption of building materials.

$$C^j=C_{dir}^j+C_{ind}^j$$

(6)

In the formula: $C^j$ represents the total carbon dioxide emissions in the jth province; $C_{dir}^j$ is the direct carbon emissions in the jth province; and $C_{ind}^j$ is the indirect carbon emissions in the jth province.

3.2.2. Measurement of Direct Carbon Emissions

This paper adopts the IPCC carbon accounting method, that is, the default emission factor method, as the method for calculating direct carbon emissions. Ten types of energy sources, including raw coal, coke, gasoline, kerosene, diesel, fuel oil, liquefied petroleum gas, natural gas, thermal natural gas and electricity are used as direct sources of carbon emissions.

$$\begin{array}{c}{C}_{dir}^j={{\displaystyle \sum }}_{i=1}^9{C}_i^j+C_p^j\\ ={{\displaystyle \sum }}_{i=1}^9{E}_i^j{\delta }_i^j{r}_i^j+E_p^j{r}_P^j\end{array}$$

(7)

In the formula: $C_i^j$ and $C_p^j$ respectively represent the CO₂ emissions from the combustion of 9 main fossil energy sources and the carbon emissions from electricity. Meanwhile, $E^j$ represents the energy consumption (data source: China Energy Statistical Yearbook). ${\delta }_i^j$, another symbol, is the average low calorific value calculated by the conversion coefficient of standard coal (data source: “General Principles of Comprehensive Energy Consumption Calculation GBT2589-2020”, the corresponding value of natural gas is calculated based on the average value of the data range). $r_i^j$ is the default CO₂ emission factor for each energy released by IPCC (data source: 2006 IPCC Guidelines for National Greenhouse Gas Inventories). $r_P^j$ is the CO₂ emission coefficient of electricity (data source: “2019 Annual Emission Reduction Project China Regional Power Grid Baseline Emission Factor”).

3.2.3. Measurement of Indirect Carbon Emissions

The main building materials officially announced by the state are cement, steel, glass, wood and aluminum. Therefore, the carbon dioxide formed from the consumption of these five materials is usually regarded as indirect carbon emissions.

$$C_{ind}^j=\frac{44}{12}{{\displaystyle \sum }}_{i=1}^5{E}_i^j{\epsilon }_i^j(1-{\alpha }_i^j)$$

(8)

In the formula: 44/12 is the conversion factor for CO₂. $E_i^j$ is the consumption of main materials (data source: China Construction Industry Statistical Yearbook, in order to maintain the unity of data units, the average density of wood is 0.54 t/m³, and the glass is converted according to 50 kg per weight box). ${\epsilon }_i^j$ represents the carbon emission coefficient [25]. Meanwhile, ${\alpha }_i^j$ is the recyclability coefficient of the material [26].

4. Empirical Analysis

4.1. Analysis of Carbon Dioxide Decoupling Effect

Substitute the carbon dioxide emissions and the total output value in the construction industry in China from 2016 to 2020 into Equation (1), and obtain the decoupling state of the carbon emissions in each region in

Table 2. Decoupling state of construction industry carbon emissions in China’s provinces from 2016 to 2020.

	Decoupling State	Area
Negative decoupling state	Expansion negative decoupling state	Shanghai, Anhui, Hainan, Xinjiang
	Weak negative decoupling state	Tianjin
	Strong negative decoupling state	Heilongjiang, Shandong
Decoupling state	Recession decoupling state	Inner Mongolia, Liaoning, Jilin, Zhejiang
	Weak decoupling state	Nationwide, Beijing, Shanxi, Jiangsu, Jiangxi, Henan, Chongqing, Yunnan, Shaanxi, Qinghai
	Strong decoupling state	Hebei, Hubei, Guangxi, Guizhou, Ningxia
Connection state	growing connection state	Fujian, Hunan, Guangdong, Sichuan, Gansu

. It can be seen that China’s construction industry as a whole shows a weak decoupling state, and at the same time, the majority of provinces and municipalities shows a weak decoupling state among the 30 provinces and municipalities, indicating that the construction industry emission reduction work in the whole country and 9 provinces including Beijing and Shanxi has achieved initial results, and the carbon emission situation is relatively optimistic. According to the analysis, it was found that from 2016 to 2020, the total output value of the construction industry has increased steadily except for a few provinces, which is due to the fact that the construction industry in each province has formed a relatively stable development scale, so the decoupling state of carbon emissions mainly depends on the growth of carbon emissions. In the carbon emissions of various provinces, the direct carbon emissions of raw coal and other energy sources have changed less in recent years, and the growth has tended to be stable, so the indirect carbon emissions formed by construction materials [27], which account for an average of more than 90%, are the most direct factors affecting the total carbon emissions and the decoupling state. For example, among the several provinces in a weak decoupling state, Beijing, Jiangsu and other provinces with large construction industries have higher material consumption, carbon dioxide emissions are still rising, the carbon emissions growth rate is lower than the growth rate of the total output value, and energy efficiency improved. Smaller provinces such as Qinghai are more conducive to precise control due to their low consumption of materials.

There are relatively few provinces in a negative decoupling state, and economic growth in these areas usually comes at the cost of accelerated carbon dioxide emissions, large consumption of building materials, large indirect carbon emissions, and extremely low energy efficiency, and green production technologies in these areas need to be further strengthened to improve energy efficiency. Among the five provinces in the state of growth connection, the indirect carbon emissions of the construction industry in these provinces have reached more than 95% on average, and some provinces have reached 99%, which means that the province actually consumes a large amount of building materials on the basis of the increased output value, which brings higher carbon emissions, and the economic growth and carbon emission change rate are basically the same, and future emission reduction work faces great challenges.

4.2. Spatial Autoregressive Correlation Analysis

4.2.1. Model Construction and Analysis

Before constructing a spatial autoregressive model, the Moran’s index was firstly analyzed on the panel data of CO₂ emissions combined with the spatial weight matrix. In

Table 3. Moran’s I test results.

Year	I	E (I)	sd (I)	z	p-Value
2016	0.224	−0.034	0.112	2.304	0.011 **
2017	0.268	−0.034	0.113	2.684	0.004 ***
2018	0.275	−0.034	0.120	2.584	0.005 ***
2019	0.396	−0.034	0.122	3.546	0.000 ***
2020	0.305	−0.034	0.121	2.802	0.003 ***

**, *** represent the significance levels at 5%, and 10%, respectively.

, the test results show that the Moran’s I values are all positive, and they all passed the 5% significance test. It indicates that a significant spatial correlation of carbon dioxide emissions exists in China’s construction industry. The above test results meet the premise of spatial measurement, so the following analysis could be carried out.

We added spatial weights to Equation (5), and constructed a spatial autoregressive model (SAR, Equation (9)). Next, we performed the Hausman test, the result is 11.86, and the original “random effects should be used” hypothesis was rejected with a significance level of 5%. Therefore, this paper adopts an individual fixed-effects spatial autoregressive model.

In order to avoid the deviation of regression results caused by multicollinearity, and to avoid the influence of other variables on the final result, this paper adopts the method of successively increasing explanatory variables for regression. At the same time, a spatial autocorrelation model (SAC, Equation (10)) was established to ensure the robustness of the model application.

Table 4. Regression results of spatial econometric model.

Variable	SAR (1)	SAR (2)	SAR (3)	SAR (4)
LnGP	0.5585 ***	0.4489 ***	0.4575 ***	0.4294 ***
LnLB		0.3702 ***	0.4350 ***	0.4771 ***
LnTER			0.0917 *	0.09712 *
LnIS				0.1806
rho	−0.1862 *	−0.2369 **	−0.2349 **	−0.2385 **
lambda
R_sq	0.8720	0.9283	0.9205	0.9200

*, **, *** represent the significance levels at 1%, 5%, and 10%, respectively.

shows the regression results.

$$Ln{C}_{it}={\alpha }_i + {\gamma }_t + \rho w_i^{\prime }Ln{C}_t + bLnL{B}_{it} + cLnG{P}_{it} + d_{1}LnTE{R}_{it} + d_{2}LnI{S}_{it} + {\mu }_{it}$$

(9)

$$Ln{C}_{it}={\alpha }_i + {\gamma }_t + \rho w_i^{\prime }Ln{C}_t + bLnL{B}_{it} + cLnG{P}_{it} + d_{1}LnTE{R}_{it} + d_{2}LnI{S}_{it} + \lambda m_i^{\prime }{v}_t + {\mu }_{it}$$

(10)

In the above equation, $w_i^{\prime }$ is the spatial weight matrix, with 0–1 indicating whether each province is adjacent or not; if the adjacent is represented as 1, the reverse is represented as 0. $\rho$ is the spatial lag term coefficient of the explanatory variable. $b, c, d_1, d_2$ are expressed as coefficients of elasticity for each explanatory variable. ${\alpha }_i$ and ${\gamma }_t$ represent the constant term results under individual effects and point-in-time effects. ${\mu }_{it}$ is a perturbation term under the spatial autoregressive model. $\lambda m_i^{\prime }{v}_t$ + ${\mu }_{it}$ represents the perturbation term under the spatial autocorrelation model.

The results of each SAR model show that the coefficients of the spatial autoregressive models are all significant, which means that the spatial dependence between the growth of the gross output value of construction industry and carbon dioxide emissions in each province and city is significant. The effect of the growth of the gross output value on the carbon dioxide emissions of the construction industry is 0.5585, 0.4489, 0.4575, and 0.4294, and all of them passed the 1% significance test. It shows that the gross output value of the construction industry in these provinces and urban areas still has a significant role in promoting CO₂ emissions. Then, if we combine this with the CKC curve theory, we can find that although China has begun to vigorously promote the green and low-carbon transformation of the construction industry, the relationship between the economic growth and CO₂ emissions is still in the rising stage of the inverted U-shaped CKC curve. The reason for this is that the current industrial upgrading strategy has not helped the construction industry to reduce the input of resource demand, and it has not yet reached the level that does not affect the overall resource consumption at all. The construction industry is a traditional industry with huge demand for resource elements. Although under the requirement of rapid economic development to high-quality development, construction enterprises are also striving to promote high-quality development, they still need to invest in elements to achieve economic growth. Therefore, with the continuous consumption of various energy sources, carbon dioxide emissions increase with the continuous rise of the gross output value.

In the SAR (2) model, the SAR (3) model and the SAR (4) model, the population factor has a significant role in promoting carbon dioxide emissions. It shows that the growth of employees at this stage is also one of the factors that promote CO₂ emissions. It shows that the energy consumption level of workers in the work activities at the construction site and other work processes remains high, which means that the staff’s awareness of carbon reduction and the ability to use and transform green new technology in work activities are still insufficient [28]. In the level of technological development, each variable has a promotion effect. The result indicates that the carbon reduction capacity of the current construction industry technology is insufficient. Therefore, in order to promote high-quality economic development and achieve the goal of reducing carbon dioxide emissions from improving the level of technological development, it still takes a certain amount of time, and green production technology needs to be further developed. Among them, the effective adjustment of the industrial structure does not have a significant promoting effect on CO₂ emissions. From the results, we can see that optimizing the industrial structure can effectively reduce emissions in the construction industry, but it still needs to be optimized.

4.2.2. SAR Model Spatial Effect Decomposition

In Table 4, the RHO coefficients are negative, which are −0.1862, −0.2369, −0.2349 and −0.2385, respectively, indicating that the economic growth in adjacent region has a spillover effect of inhibiting carbon dioxide emissions in the region. This paper further decomposes the spatial effect of the model SAR (4) to further study the spatial effect distribution of each variable on the CO₂ emissions.

Table 5. Decomposition results of spatial effects.

Variable	Direct Effect	Indirect Effect	Total Effect
LnGP	0.4294 ***	−0.0857 *	0.3437 ***
LnLB	0.4895 ***	−0.0975 *	0.3920 ***
LnTER	0.1039 **	−0.0201	0.0838 **
LnIS	0.1892	−0.0374	0.1519

*, **, *** represent the significance levels at 1%, 5%, and 10%, respectively.

shows the specific results.

The direct effect of economic growth was 0.4294, the indirect effect was −0.0857, and the total effect was 0.3437. The three effects passed the significance test. Among them, the direct effect shows a significant positive direction. It means that the economic growth of the regional construction industry has significantly increased the carbon dioxide emissions in the region. Compared to direct effects, the indirect effect is negative, and its significance level is slightly lower, which means that the economic growth of neighboring regions will inhibit the carbon dioxide emissions of the region through spatial spillover effects. The reason is that when the economic growth in the adjacent area is significant, the enterprises will shift from the local area to the adjacent area to continue production based on development considerations. Therefore, the number of high-energy-consuming enterprises in the area will decrease, and the local carbon dioxide emissions will decrease accordingly. However, the spillover effect is not strong and is significantly lower than the direct effect.

The spatial effect decomposition results of construction workers show a similar trend. The negative indirect effect can also be explained by the good economic development of construction enterprises in adjacent areas, which makes the adjacent areas more attractive to employees and can accommodate more employees. Therefore, the number of people flowing into adjacent areas increases, and the number of employees in local enterprises decreases, thereby reducing the carbon emissions of local enterprises, but the inhibitory effect is small. At the technological development level, the indirect effects of each variable are not significant, demonstrating that there is no significant spatial spillover effect in the technological development, and carbon dioxide emissions are mainly for the local technological level to directly promote.

4.2.3. Regional Analysis

In order to better analyze the impact of relevant factors on carbon emissions in the construction industry, this paper combines Table 2 to further analyze the data samples from negative decoupling, decoupling, and growth connection regions. From the results (

Table 6. Analysis results of different decoupling state areas.

Variable	R1	R2	R3
LnGP	0.8024 **	0.0773	0.9732 ***
LnLB	−0.7131 *	0.6770 ***	0.3869
LnTER	−0.0630	0.1245 **	0.1336 **
LnIS	−0.1600	0.3164	−0.6991 *

*, **, *** represent the significance levels at 1%, 5%, and 10%, respectively.

), in the provinces in the state of negative decoupling (R1), the growth of carbon emissions in the construction industry is mainly due to the promotion role brought by the economic growth of the construction industry, while the human factor has a significant inhibitory effect on it, and the carbon reduction effect brought by the technical level and industrial structure is not obvious. For the decoupled provinces (R2), we can find that there is no significant promotion relationship between economic growth and carbon emissions, which is consistent with the decoupling status of these provinces, whose carbon emissions mainly come from the insufficient transformation of labor and technology level to “green production”. Among the Connection state provinces (R3), the catalytic effect of economic growth is significant and much higher than that of other factors, indicating that carbon emissions in these provinces are also increasing significantly as the economy grows, which is also consistent with their growth status. Among other related factors, the lack of green technology is another contributing factor, and the adjustment of its industrial structure has already contributed to its carbon reduction work.

4.2.4. Robustness Test

In order to ensure the small error in model selection and the validity of the SAR model estimation results, this paper builds the spatial autocorrelation model (SAC) again to test the robustness of the above estimation results. Judging from the estimation results of the SAC model, the regression results are shown in

Table 7. Robustness test.

Variable	R0	R1	R2	R3
LnGP	0.2124 **	0.6401 **	0.0798	1.0209 ***
LnLB	0.3817 ***	−0.6928 **	0.6495 **	0.3189
LnTER	0.1232 **	0.0987	0.1402 *	0.1215 **
LnIS	0.2901	0.1211	0.2296	−0.5026 *

*, **, *** represent the significance levels at 1%, 5%, and 10%, respectively.

, and the significance of the regression results of each variable and the direction of its effect on CO₂ emissions are consistent with those of the SAR models in the whole country (R0) and in the decoupling state area, which shows that the model selection, estimation results and research conclusions in this paper are robust.

5. Conclusions

First, the national construction industry as a whole is in a state of weak decoupling of carbon emissions. From a regional perspective, a state of decoupling is shown in 60% of the 30 provinces and municipalities. Most of these regions are in a weak decoupling state, and some of these provinces are already in an ideal strong decoupling state. In the construction industry, emission reduction and energy conservation work in most provinces has achieved results. Another 23% and 17% of provinces, respectively, are in a negative decoupling state and a growth connection state. These provinces need to take measures to further promote energy conservation and do their best to reduce emissions. The analysis found that indirect carbon emissions caused by the consumption of building materials are the main factors affecting the decoupling state.

Second, through the establishment of a spatial measurement model, it is found that the growth of the total output value has a significant promoting effect on CO₂ emissions. Meanwhile, a spatial spillover effect existed between the two. Therefore, this paper believes that, in the construction industry, the relationship between the economic growth and CO₂ emissions has not yet reached the peak of the CKC curve, and the economic growth will still promote regional CO₂ emissions. The inhibitory effect of the spatial spillover effect of economic growth is small, and the carbon dioxide emissions in this region are still mainly directly affected by the economic growth of local enterprises. At this stage, population factors and technical production level factors still play a critical part in promoting carbon emissions, and the application of green processes, technologies and materials is still not comprehensive enough. After further analysis, it was found that the influencing factors of carbon emissions under different decoupling states have different degrees of influence on it. In the provinces with negative decoupling status, economic growth showed a promoting effect and staff showed a restraining effect. In the decoupled provinces, the staff and technical levels showed a promoting role. In the provinces with growth connection, economic growth and technological level played a promoting role, and the industrial structure presented a restraining effect.

6. Policy Recommendations

In view of the above empirical conclusions, in order to transform the high-speed economic growth of the construction industry into high-quality growth, it is necessary to further explore the path of green and low-carbon development under the premise of ensuring economic development, and it is necessary to carry out comprehensive carbon reduction work in the construction industry from the use of building materials, technical production level, personnel and employment, and so forth so as to promote the decoupling of carbon emissions in the construction industry as soon as possible in the future. The following recommendations are therefore made.

First of all, construction enterprises comprehensively strengthen the carbon reduction effectiveness from each process in the whole process management of construction projects. Set phased carbon reduction targets, and run them through the entire building life cycle from design, material selection and manufacturing, to construction to operation to renovation and demolition, and ultimately achieve high-quality growth in a green economy. Specific measures can start from the following aspects: the use of whole-process integrated design in the design stage of construction projects, and overall control of carbon emissions at all stages of construction projects. In the selection and manufacture of materials, try to choose green materials with high recyclability rate, and strengthen the development and utilization of new green materials. Strengthen the skills training of construction personnel during the construction process, reduce embodied carbon emissions by changing personnel behavior, and improve technical production level, reduce resource consumption, and improve energy utilization. In the next life cycle of the construction project, we will always pay attention to energy consumption, including electricity and energy, and strive to achieve zero net carbon emissions throughout the process.

Secondly, while formulating policies to promote the carbon reduction of the construction industry in their regions, local governments should also pay attention to the economic development and carbon reduction strategies of the construction industry in other regions, especially adjacent regions, and jointly promote the sustainable and green development of the construction industry through synergies between regions and the use of regional advantages. At the same time, each region should formulate targeted carbon reduction strategies based on the current decoupling status of the region. For example, provinces in negative decoupling should continue to strengthen the application of green technologies, strengthen the research and development of carbon-neutral technologies such as zero-carbon technologies, carbon reduction technologies, and carbon-negative technologies, and further optimize the industrial structure; Provinces in decoupling status should focus on training staff on carbon reduction behaviors to promote further decoupling; Provinces in the state of growth connection are too closely related to carbon emissions and economic growth, and it is necessary to strengthen all-round carbon reduction work on green materials, technology, personnel, and so forth, and reduce carbon emissions with high energy consumption such as building materials and electricity, formulate macro carbon reduction strategies, and market access criteria for carbon emissions, strengthen the effectiveness of external factors that promote enterprise carbon control and carbon reduction, and accelerate the digital transformation of construction enterprises.