Introduction

Globally, the construction industry, serving as a foundation for numerous economies, substantially contributes to energy consumption. It is estimated that the construction sector consumes 40% of energy and releases 1/3 of carbon dioxide emissions associated with energy worldwide1. Accordingly, the construction industry is critical for reducing carbon footprint and fostering sustainability.

The construction sector in China, a key component of the national economy, pushes forward societal advancement significantly. Figure 1a demonstrates that the construction industry’s value added as a percentage of GDP (CVAG) climbed from 5.68% in 2007 to 7.00% in 2021, before stabilizing at 6.9 and 7.2%. This indicates the robust economic growth of China’s construction sector, a significant economic driver in the country. Nonetheless, numerous energy resources are consumed behind the industry’s flourishing growth2. As shown in Fig. 1b, the construction industry’s energy consumption increased by approximately 2.2 times between 2007 and 2021, from 42.03 Mtce to 96.08 Mtce, which has made the greenhouse effect even worse. Thus, the construction industry’s green low-carbon advancement is crucial for attaining the “dual carbon” (carbon peaking and carbon neutrality) targets proposed by China.

Unlike other countries, China’s unique political system enables government governance to play a pivotal force in environmental protection. Specifically, government environmental governance (EG) primarily involves the establishment and enforcement of policies, regulations, economic incentives, and other measures to guide and compel industries toward sustainable practices3,4. Through these measures, the government ensures compliance with environmental regulations among enterprises, which is particularly critical in the industry’s emissions reduction efforts5,6. It is one of the most significant ways to combat climate change, guarantee energy security, and achieve emission abatement7. Currently, EG is measured using quantitative standards that emphasize the amount of resources allocated by the government for environmental governance4. For instance, some scholars assessed it using government investments related to environmental pollution8,9. A greater level of investment indicates a higher intensity of environmental governance9. Notably, measuring EG through these quantitative indicators underscores the direct actions undertaken by governments in environmental management, including formulating and enforcing regulations, making financial investments, and so on. This approach emphasizes the processes involved in environmental governance rather than the outcomes of these efforts10.

Fig. 1
figure 1

The CVAG and the industry’s energy consumption from 2007 to 2021.

Full size image

Current research primarily centers on the upstream and downstream sectors of the construction industry’s supply chain, including manufacturing11, metal12, and transportation13 sectors. However, there is relatively little research specifically addressing the construction industry itself. Among the few studies that do exist, several limitations are evident. Firstly, most measured EG through indicators related to costs and investments in environmental pollution control8,9. They primarily explore the effects from a financial support perspective, emphasizing the importance of government allocation of economic resources to address environmental issues and promote sustainable societal development. While substantial financial investments indicate the government’s commitment to environmental governance, they do not necessarily guarantee effective outcomes or improvements in environmental conditions. Without the backing of appropriate policies and their correct implementation, these funds may fail to yield the expected environmental benefits. As a result, incorporating policy orientations into measurement is crucial, as this more comprehensively reflects the alignment between governmental policies and actions in environmental governance, so as to facilitate a thorough assessment of EG. Additionally, these studies have not fully considered the role produced by the unique institution of the Chinese government8,14. Apparently, there is space for further deepening in this research area.

Consequently, the contribution is reflected in three aspects: (1) previous studies have explored the influence of EG, primarily targeting the upstream and downstream sectors of the construction industry’s supply chain. There is a lack of research on the construction industry, positioned midstream in the supply chain. This study empirically analyzes how EG affects the carbon intensity of the construction sector (CCEI), filling a void in the relevant research. (2) In this research domain, most scholars predominantly use financial investments to gauge EG. This paper, however, posits that financial investments constitute the material foundation of environmental governance, whereas policy orientations offer guidance on its direction. These elements jointly influence the government’s actions in environmental governance. Consequently, this paper measures the EG from the perspectives of financial support and policy orientations. It seeks to thoroughly represent the government’s financial contributions as well as its policy directions and intentions in environmental governance, thereby enabling a more comprehensive evaluation of its actual effectiveness and sustained influence. (3) Given the critical role of government institutions in optimizing industrial structure under China’s socialist system, this paper considers China’s distinctive fiscal decentralization system. It incorporates two typical features of government behavior in this context—vertical decentralization (VD) and horizontal competition (HC)—into the analytical framework. This analysis explores the threshold effects of EG on the CCEI from an institutional perspective, thereby enriching the theoretical research on fiscal decentralization.

Theoretical analysis and hypotheses

Environmental governance by the government serves as a crucial part for addressing environmental issues; however, its impact on emission reduction can be complex. As a result, this section will first review the relationship between EG and emission reduction. After that, it will review how fiscal decentralization affects the EG’s influence in the Chinese institutional context.

Government environmental governance and carbon emissions

A number of scholars are also committed to researching how EG influences carbon emissions to obtain policy implications. Some scholars argue that EG can significantly contribute to mitigate carbon emissions15,16. This reason could be that the implementation of stringent regulations and policies can promote the adoption of clean technologies and enhance environmental awareness, thereby effectively reducing emissions15. Other scholars contend that the relationship between EG and emissions abatement can be characterized by an intricate U-shaped curve, where both promoting and inhibiting effects are present17,18,19. They believe that the implementation of environmental policies by the government may lead to an increase in pollution emissions by businesses in the short term, thereby exacerbating environmental issues. This phenomenon is referred to as the green paradox18,20. The causes may include ineffective policy effects and the delays in policy implementation, particularly when policies fail to adapt promptly to changes in the market19,20.

In the area of the construction industry, research is relatively limited, and there is scholarly disagreement concerning the impact of EG. Some scholars argue that EG can effectively mitigate emissions in this sector by encouraging construction companies to adopt cleaner technologies and sustainable practices8,14. Conversely, others contend that these measures may inadvertently lead to short-term increases in emissions, as firms might boost production or continue using existing, less efficient methods before transitioning to greener alternatives9.

Based on the analysis above, it is evident that most scholars believe the impact of EG on carbon emissions is not linear. Therefore, specifically within the context of the construction industry, this paper proposes that the impact of EG on the CCEI shows non-linearity (H1).

Moreover, some scholars have also explored the spatial spillover effects resulting from EG on carbon emissions. Several studies argue that EG’s influence can exhibit negative spatial spillover effects, meaning it can lead to a reduction in emissions in adjacent areas21,22. This phenomenon may stem from the current tendency of various regions to adopt a “race to the top” strategy, where governments implement stricter environmental policies compared to neighboring provinces21,23. By doing so, they aim to cut regional carbon dioxide emissions and enhance energy efficiency. However, other scholars believe that it may exert positive spatial spillover effects11,24,25. For example, Liu, Xin, and Li (2022)11 concluded that an increase in the local EG significantly increased carbon emissions in adjacent manufacturing sectors. This may be due to the “free-riding” effect, where improvements in one area lead neighboring governments to relax regulations, allowing adjacent areas to benefit without making similar efforts to mitigate emissions11,25.

In the construction field, there is a debate among the limited existing studies8,9. Some scholars suggest that strengthening local EG can effectively reduce carbon emissions in adjacent construction sectors8. This may be because enhancing local EG can raise environmental standards and promote the adoption of green technologies, which can diffuse and subsequently lower emissions in neighboring areas. On the other hand, some scholars argue that there is no spatial spillover effect, which may be attributed to the relatively low level of environmental regulation in the study areas9. In such environments, even if EGI is strengthened, its effects may not effectively extend to neighboring regions, resulting in no significant reduction in carbon emissions.

According to the analyses, although there is no consensus supporting the universal existence of such effects, the diffusion mechanism does exist; specifically, local environmental standards and the application of green technologies may diffuse through economic and social networks. Consequently, we posit that the impact of EG on CCEI exhibits spatial spillover effects (H2).

Government environmental governance and fiscal decentralization

In 1994, China implemented the tax-sharing system, granting local governments some authority to decide fiscal revenues and expenditures, thus creating a unique fiscal decentralization system26,27. Simultaneously, the competition for scarce resources for local economic development has gradually formed fiercely competitive mechanism among local governments28. In this context, it has been observed that the dual typical phenomena of fiscal decentralization—vertical decentralization (VD) and horizontal competition (HC)—have substantially driven rapid economic development. Concurrently, these processes have inevitably resulted in the degradation of ecological environments. Therefore, it is imperative to investigate the connection between EG and emission mitigation from the institutional aspect.

Regarding VD, there are two main viewpoints. On the one hand, several studies have proven that implementing EG may not be beneficial for abating emissions under China’s fiscal decentralization system. In fact, it could have the opposite effect, known as the “green paradox”29,30. The main reason can be that VD in China is characterized by political centralization and economic decentralization31. This strengthens the self-governance of local authorities in economic development, but weakens the central government’s power to monitor and intervene in local affairs32. When there is a high degree of VD, local governments are likely to decrease financial support and relax standards in environmental governance to draw in more investment, thereby winning the “political tournament,” even if it harms the environment30. Conversely, when the degree of VD is low, local governments will be subject to strict supervision and intervention by the central government. Consequently, environmental policies and measures will be strictly enforced by local authorities, fulfilling their responsibility for environmental protection with due diligence32. On the other hand, other studies have found that in areas with higher levels of VD, local governments have a stronger propensity to allocate more financial resources toward combating pollution and advancing environmentally friendly technologies, thereby reducing the environmental footprint33,34.

Concerning HC, existing studies have three viewpoints. First, some scholars believe HC appears as the “race to the bottom”. In other words, authorities may relax environmental protection standards to compete for mobility resources, thus aggravating emission pollution30. Furthermore, the externalities of the natural environment allow neighboring governments to “free-riding” on the benefits of local government environmental management, fueling the “bottom‑to‑bottom competition”11. Second, HC can appear to be the “race to the top”23. This is mainly based on Tiebout’s “voting by foot” theory35. Namely, local governments strongly desiring the high-quality environment seek to attract more resources that value it by strengthening EG11. Third, it has been argued that the impact of HC on EG is complex and subject to great uncertainty36.

Consequently, the impact of fiscal decentralization on EG is likely not a simple linear relationship, but rather exhibits complex nonlinear characteristics as it varies with VD and HC. Currently, there is a lack of exploration in the construction industry in this area. Considering that local governments may adopt varying emission reduction measures under VD and that HC may influence environmental standards in the construction sector, we propose that the influence of EG on the CCEI exhibits non-linear characteristics as vertical decentralization and horizontal competition increase (H3). Figure 2 shows the flowchart of this study.

Fig. 2
figure 2

The flowchart of this study.

Full size image

Methods

Modelling

Benchmark regression model

We employ the two-way fixed effects panel regression model for the study and analysis. This serves to mitigate the effects of unobservable factors in the time and individual dimensions. The model equations are given by Eq. (1).

$$CCE{I_{it}}={\beta _0}+{\beta _1}E{G_{it}}+{\beta _2}{X_{it}}+{u_i}+{v_t}+{\varepsilon _{it}}$$
(1)

where \(CCE{I_{it}}\) and \(E{G_{it}}\) individually show the carbon intensity of the construction sector and the government environmental governance in region i in year t. \({X_{it}}\)is the control variable. \({u_i}\)and \({v_t}\)individually refers to the fixed effect of region and time. \({\varepsilon _{it}}\)is the error term.

Spatial econometric model

The global Moran’s I index is employed to assess the spatial autocorrelation of CCEI, determining the presence of spatial dependency. The formula employed is as follows:

$$I={{\sum\nolimits_{{i=1}}^{n} {\sum\nolimits_{{j=1}}^{n} {{w_{ij}}\left( {{X_i} - \bar {X}} \right)\left( {{X_j} - \bar {X}} \right)} } } \mathord{\left/ {\vphantom {{\sum\nolimits_{{i=1}}^{n} {\sum\nolimits_{{j=1}}^{n} {{w_{ij}}\left( {{X_i} - \bar {X}} \right)\left( {{X_j} - \bar {X}} \right)} } } {{s^2}\sum\nolimits_{{i=1}}^{n} {\sum\nolimits_{{j=1}}^{n} {{w_{ij}}} } }}} \right. \kern-0pt} {{s^2}\sum\nolimits_{{i=1}}^{n} {\sum\nolimits_{{j=1}}^{n} {{w_{ij}}} } }}$$
(2)

where X, \(\bar {X}\), and \({s^2}\)individually represent CCEI’s value, mean and standard deviation. \({w_{ij}}\)is the element of spatial weight matrix. The value of I ranges from [-1, 1]. Specifically, if 0<I≤1 (or -1≤I<0), it indicates that the CCEI exhibits positive (or negative) spatial autocorrelation.

On this basis, it is necessary to select an appropriate model for analysis. A spatial autoregressive (SAR) model is chosen for spatial econometric testing. To verify the validity of this model, the LM test and the Hausman test are utilized for validation. The results can be found in the Section “Analysis of spatial effects,” which confirms the model’s appropriateness. The formula for this model is as follows:

$$CCE{I_{it}}={\alpha _0}+rho \cdot W \cdot CCE{I_{it}}+{\alpha _1}E{G_{it}}+{\alpha _2}EG_{{it}}^{2}+{\alpha _3}{X_{it}}+{\mu _i}+{\nu _t}+{\varepsilon _{it}}$$
(3)

where \(rho\)is the coefficient of the spatial lag term, andW denotes the spatial weight matrix. Referencing Parent and Lesage (2008)37, this paper constructs an economic-geographical matrix based on provincial geographic and economic distances, in order to better reflect the effect of geo-economic factors on the spatial distribution characteristics of CCEI38. The expression of the matrix is: \({W_{eg}}={W_e} \times {W_g}\). In the expression, \({W_e}\) refers to the matrix of economic distances, with its elements showing the reciprocal of the absolute difference in the per capita value-added of the construction industry between two provinces. \({W_g}\)is the matrix of geographical distances, with its elements indicating the reciprocal of the squared latitude and longitude distance between two provinces.

Additionally, referencing Shao et al. (2022)39, this study constructs a nested spatial weight matrix that integrates geographical and economic distances for robustness testing. The expression of the matrix is: \({W_{egq}}=\lambda {W_e}+(1 - \lambda ){W_g}\). \(\lambda\) denotes the weight, with a value of 0.5. Simultaneously, a spatial geographical matrix \({W_g}\) and a spatial adjacency matrix \({W_a}\) are also used to verify the results’ reliability.

Dynamic threshold model

The threshold regression model is employed to further explore threshold effects of the Chinese institutional context. However, emitting CO2 is a dynamic process, which means current emissions may be influenced by past emission activities. Consequently, we adopt the dynamic threshold model proposed by Seo and Shin (2016)40, which adds the lagged terms of explained variables to explanatory variables. In this way, estimation biases caused by endogeneity problems in previous static threshold models could be avoided41. Based on these, the one-period lagged CCEI is included in the model shown in Eq. (4).

$$CECI_{{it}} = x^{'} _{{it}} \alpha + k\left( {1,x^{'} _{{it}} } \right)1\left\{ {q_{{it}} > \gamma } \right\} + u_{i} + \varepsilon _{{it}}$$
(4)

where \(x^{'} _{{it}}\) denotes all explanatory variables that contain the one-period lagged CCEI. \(\alpha\) shows the coefficient. k is the slope parameter. \(1\left\{ \cdot \right\}\) represents the indicator function, yielding an output of 1 when the condition within the parentheses is True, and 0 when it is False. \({q_{it}}\) refers to the threshold variable, which in this study denotes government environmental governance. \(\gamma\) is the threshold parameter. Specific information on this model can be found in detail in Seo and Shin (2016)40.

Variable description

The explained variable is carbon intensity of the construction industry (CCEI). Referring to Li, Sun, et al. (2017)42, we obtain the CCEI by dividing the industry’s carbon emissions by its total output value. Notably, we refer to IPCC (2006)43 and Sun, Hao, and Long (2023)9 to measure the carbon emission. Specifically, the eight primary energy sources consumed by this sector are used as direct emission sources, and heat and electricity are utilized indirect emission sources. Please refer to Sun, Hao, and Long (2023)9 for specific calculation details.

The core explanatory variable is government environmental governance (EG). As previously mentioned, existing research generally uses indicators like environmental protection costs and investments to represent it. Although these indicators reflect the financial support from authorities, they fail to reveal the government’s policy orientation. Without policy guidance and implementation, mere financial investment may not achieve the expected environmental protection outcomes due to mismanagement or improper resource allocation. Consequently, the EG is comprehensively measured from both financial support and policy orientation perspectives in this paper. Specifically, in terms of financial support, following Lin and Chen (2020)12, EG is expressed by the proportion of industrial pollution control investment to industrial value added. Regarding policy orientation, referencing Chen and Chen (2018)5, the frequency of 15 keywords related to environmental protection, emission reduction, energy consumption, and other topics in the annual work reports of local authorities is used as a measurement. A detailed list of these keywords is provided in the Appendix Table A1. A high frequency of these keywords indicates an increased urgency of local governments towards environmental governance and a corresponding rise in policy orientation44,45. Finally, using these data, the entropy weighting method is applied to objectively measure the EG. Table 1 illustrates the composition the indicator.

Table 1 The composition of government environmental governance (EG).
Full size table

Vertical decentralization (VD) and horizontal competition (HC) are threshold variables. As for VD, the indicator is measured using fdr/(fdr + fdc) as suggested by Chen and Chang (2020)46. In the expression, fdr and fdc denote provincial fiscal expenditure per capita and central fiscal expenditure per capita, respectively. As for the “yardstick competition” among local governments, relevant studies believe that HC mainly manifests as economic competition47,48. In the competition process, foreign direct investment (FDI) is not solely a driver of rapid local economic growth but also an essential way for local governments to compete for resources48. Hence, we use FDI/GDP to measure this indicator. The greater the value of the indicator, the more intense the competition.

The control variables include urbanization (UR)49,50, industrial structure (IS)51, technological innovation (TI)9, foreign trade (FT)52, and industrial agglomeration (IA)53. Specific descriptions are detailed in Table 2.

Table 2 Descriptive statistics of variables.
Full size table

Data sources

Due to data availability, we selected data from 2007 to 2021 for 30 provinces in China, excluding Tibet, Taiwan, Hong Kong, and Macau, resulting in a total sample size of 450. The data derives from China Statistical Yearbook, China Real Estate Statistics Yearbook, China Statistical Yearbook on Environment, China Statistical Yearbook on Construction, China Energy Statistical Yearbook, China Statistical Yearbook on Science and Technology, and EPS China Data.

Results and discussion

Analysis of benchmark estimation results

Estimation results of Eq. (1) are displayed in Table 3. From column (3) to column (7), the effects of control variables on CCEI are progressively controlled.

Table 3 Benchmark estimation results.
Full size table

Obviously, the coefficients of EG and EG2 consistently show significantly positive and negative, respectively. This indicates the presence of an inverted U-shaped relationship, thereby confirming H1. In other words, the initial phase will see an increase in CCEI due to the enhancement of EG. When the inflection point is crossed, EG will have an abatement effect. Notably, the results align with the discoveries made by Huang and Tian (2023)54, Yin et al. (2022)55, and Xu and Xu (2023)56, reinforcing the validity of our conclusions.

The reasons for this relationship may be attributed to the “green paradox” and “reverse emissions reduction” effects54,55. Specifically, the initial announcement of stringent environmental policies and governance measures is likely to result in energy suppliers adopting a pessimistic outlook for the future. Under these circumstances, the supply of fossil energy to the construction market will increase before policy implementation. This short-term behavior not only reflects a response to policy uncertainty but may also be related to economic decisions made by enterprises to maximize short-term profits. This can lead to higher carbon emissions in the short term, illustrating the “green paradox” effect18,19,55.

Subsequently, as environmental policies and governance measures continue to be implemented, along with strengthened regulatory efforts, construction enterprises will gradually increase their investments in green production technologies and renewable energy, driven by policy incentives. At this stage, companies will reduce their reliance on fossil fuels and achieve emission reductions through technological innovation and optimized resource allocation, a process referred to as “reverse emissions reduction54,55.” From economic perspective, as green technologies mature and economies of scale are realized, the long-term operational costs for businesses will decrease, leading to enhanced profitability. This will underscore the importance for governments to consider economic incentives for enterprises when designing environmental policies to promote sustainable development and improve overall economic efficiency.

To further analyze the relationship, we plot the curve (see Fig. 3) for analysis. In this graph, the EG values span from a low of 0.047 to a high of 0.862. The turning point occurs when the EG is about 0.515. The curve indicates that before the inflection point, CCEI increases with the increase in EG. After the inflection point, CCEI decreases as EG further increases, showing a negative correlation. This further confirms the previously mentioned inverted U-shaped relationship.

Fig. 3
figure 3

Inverted U-shaped curve.

Full size image

On this basis, multiple tests are performed to determine whether the result is robust. First, referencing Lind and Mehlum (2010)57, the UTEST is employed to verify the relationship between EG and CCEI. The results, presented in Appendix Table A2, align with the baseline regression analysis. In the table, the extreme point is 0.515 with a corresponding P-value of 0.067, rejecting the null hypothesis of a non-U-shaped relationship. Moreover, the curve’s slope individually exhibits positive and negative values within the examined range. Thus, this relationship is verified.

Apart from that, robustness checks are conducted from other perspectives. Appendix Table A3 presents the test results, which confirm the reliability of the core conclusion. First, the core explanatory variable is substituted. Following Zhao, Tao, et al. (2023)58, the EG is measured using the proportion of investment in industrial pollution control to GDP, as shown in column (1) of Table A3. Second, given that it takes time for environmental policies and measures to affect the construction industry in a region or country, we lag the EG by one and two periods for the regressions, respectively. The findings for the one-period lag term (see column (2)) and the two-period lag term (see column (3)) both confirm the significant inverted U-shaped effect. Third, the sample range is adjusted. The results after excluding the municipalities are demonstrated in column (4). Ultimately, to test the endogeneity of the model, we introduce CCEI’s one-period lagged term as an instrumental variable in Eq. (1) and then estimate the effect using the systematic GMM7,21,30. The finding (see column (5)) indicates that there is no autocorrelation in the residual term of the estimation equation, and that all instrumental variables are strictly exogenous in the absence of over-identification, suggesting the reasonableness of this model setting.

Analysis of spatial effects

The analysis begins with an examination of the spatial autocorrelation of CCEI, as illustrated in Fig. 4. From 2007 to 2021, the values of Moran’s I indexes are consistently positive and statistically significant at the 1% level, which means the distribution of CCEI is not random but exhibits strong spatial dependence at both geographical and economic levels.

Fig. 4
figure 4

The Moran’s I index of CCEI from 2007 to 2021.

Full size image

Furthermore, to determine the appropriate spatial econometric model, tests are conducted. The results, presented in Table 4, indicate that the LM test results are significant at the 1% level. In the robust LM test, only the R-LM-lag test result is significant. This finding suggests a strong spatial lag effect among the variables, with no significant spatial error effect, indicating that the spatial autoregressive (SAR) model is suitable. Additionally, the Hausman test result for this model is 35.72, which is significant. Consequently, the SAR model with fixed effects is employed for the analysis.

Table 4 The test results.
Full size table

The analysis results are displayed in Table 5, where columns (1) and (2) represent the regression coefficients without the inclusion of control variables in the equation, while columns (3) and (4) do. As shown in the first row of Table 5, rho is significantly positive at the 1% level, indicating a significant positive correlation in CCEI across provinces. The impact of EG

Table 5 Estimation results of spatial effects.
Full size table

on the CCEI is further analyzed through the direct and indirect effects, confirming H2.

To be more specific, the direct effects reveal the inverted U-shaped connection between EG and CCEI in the local area. In terms of indirect effects, EG’s influence exhibits the spatial spillover effect. That is to say, local EG not only affects local CCEI but also exerts a similar inverted U-shaped impact on neighboring regions. This may be because, in the short term, spatial disparities in EG across regions likely cause construction companies that cause more serious environmental pollution to move to neighboring regions with lower governance intensity. This behavior can be understood economically, as firms often seek to minimize costs associated with compliance and operational expenses. Consequently, the pollution refuge is formed23. However, over the long term, emission reduction technologies developed by local construction companies will gradually mature as environmental policies and regulations continue to be strengthened. Given the interconnection pattern between construction industries in most adjacent regions9, the widespread dissemination and utilization of green technologies will cut carbon emissions in the surrounding areas54. Furthermore, a comparison of the direct and indirect effects reveals that the absolute value of the direct effect coefficient exceeds that of the indirect effect coefficient. This suggests the impact of local EG on CCEI within provinces is greater than that of EG in neighboring regions, demonstrating a localization phenomenon.

To ensure result reliability, multiple spatial weight matrices are employed for robustness testing. Following the substitution of We.g. with Wegq, Wg, and Wa, the SAR model with fixed effects is utilized for analysis, with relevant results detailed in Appendix Table A4. Under different weight matrices, the spatial autocorrelation coefficient remains significantly positive. Furthermore, the inverted U-shaped connection is verified through direct, indirect, and total effects, strengthening the robustness of findings.

Analysis of threshold effects

Considering the potential impact of China’s unique fiscal decentralization system, this paper explores the dynamic threshold effects of this distinctive system, characterized by vertical decentralization (VD) and horizontal competition (HC). Table 6 presents the results, which provide strong evidence supporting H3.

Table 6 Estimation results of threshold effects.
Full size table

Clearly, as the level of VD increases, the effect of EG on CCEI shifts from promotion to suppression. Specifically, when the threshold is below 0.896, ER significantly contributes to the rise in CCEI. When the threshold is above 0.896, however, the effect shifts to be significantly dampening. The finding suggests that when local governments have more financial autonomy, they will invest more in enhancing environmental management capacity and increasing environmental regulation, particularly for the highly polluting and energy-intensive industries like the construction industry, thereby advancing sustainability. Economically, this shift reflects a strategic decision by governments to allocate resources more effectively, aiming to achieve long-term sustainability rather than short-term growth. In fact, since the implementation of the 12th Five-Year Plan, the promotion of environmental sustainability in the construction sector has become the guiding principle for the industry’s advancement. Subsequently, local governments have increasingly prioritized environmental concerns within the construction industry. As a result, when governments have greater financial independence, they tend to allocate more resources to implement intensive green and low-carbon initiatives.

As for the horizontal competition, environmental governance significantly reduces the CCEI when the threshold is below 0.287. After the threshold is crossed, the role of EG transforms from significant inhibition to significant promotion, showing the “bottom‑to‑bottom competition” among governments. This result further indicates that heightened competition among municipal governments leads to uncompensated negative externalities of environmental contamination, exacerbating environmental issues. One possible explanation is that local governments have a more pronounced need for economic growth. When competition is fierce, local governments may adopt relatively lenient environmental policies to attract economic drivers such as skilled workers and international investment to achieve economic advance. This reflects a classic trade-off in public policy, where the immediate economic benefits can overshadow environmental concerns. Nonetheless, when government competition is low, local governments may have a greater tendency to devote more energy to environmental oversight47,48.

Moreover, Fig. 5 shows the likelihood ratio (LR) curves for the two threshold variables, reflecting the reliability and validity of the threshold values.

Fig. 5
figure 5

Likelihood ratio curves of threshold variables.

Full size image

To verify the reliability of findings, several tests are conducted. First, the dependent variable lagged by two periods is added to the explanatory variables for regression analysis, with the results displayed in columns (1) and (4) of Appendix Table A5. Second, the core explanatory variable is replaced. Consistent with the robustness check of the baseline regression, the proportion of industrial pollution control investment to GDP is used as a new measure of EG, with relevant results demonstrated in columns (2) and (5). Third, by excluding data from municipalities directly, the influence of sample selection is assessed, with the results presented in columns (3) and (6). Under these different test conditions, although the coefficient sizes and threshold values fluctuate, the direction of the impact of EG on the CCEI remains consistent, indicating the robustness of the findings.

Conclusions and implications

Based on provincial panel data from 2007 to 2021 in China, this paper first measures CCEI and then empirically analyzes EG’s impact on the CCEI using the panel regression model, spatial econometric model, and dynamic threshold model. The findings demonstrate that: (1) the influence of EG on CCEI follows a significant inverted U-shaped trajectory, which remains robust through a series of empirical tests. The turning point occurs when the EG value is about 0.515. With the improvement of environmental governance policies and measures, as well as the strengthening of regulatory efforts, EG can significantly contribute to the low-carbon transitions within the construction sector. (2) The influence of EG exhibits significant spatial spillover effects. Specifically, an increase in the level of local EG has a significant inverted U-shaped influence on the CCEI in both local and adjacent regions. Moreover, local EG has a stronger impact on emission reductions than the EG of neighboring regions. These underscore the spatial effects of localized ER’s impact and advocates for the adoption of integrated and coordinated policy measures to mitigate emissions within the construction sector across all regions. (3) The impact of EG on the CCEI has significant dynamic threshold effects under the regime of fiscal decentralization. More precisely, the impact of EG transitions from facilitative to inhibitory when vertical decentralization surpasses 0.896. Conversely, the impact of ER shifts from inhibitory to promotive when horizontal competition exceeds 0.287. Thus, flexible implementation and promotion of environmental policies are essential for effectively controlling CCEI amidst shifting dynamics of fiscal decentralization.

The findings have important significance for local governments in implementing scientifically and reasonably policies and measures. At the same time, they can also provide insights and experiences for other countries, especially developing ones, which are undergoing low-carbon transitions in the construction sector.

First, the inverted U-shaped relationship indicates the complex impact of EG on CCEI. Initially, stringent policies may lead to a temporary increase in carbon emissions, prompting policymakers to carefully consider the timing and implementation methods to avoid situations where companies reduce their environmental governance efforts in pursuit of short-term gains due to policy uncertainty. In the long run, promoting green investment and innovation is an effective way to reduce carbon emissions. Policymakers should provide financial support and incentives to encourage construction companies to invest in green technologies and sustainable practices, thereby lowering long-term emissions and fostering the development of the green economy. Additionally, the coordination of environmental governance measures and policies between provinces should be emphasized, with a focus on identifying their spatial impact effects. Enhanced communication among provincial departments can accelerate the development of an inter-provincial emission reduction mechanism in the construction industry, leading to more effective governance and emission control.

Second, it is crucial to reasonably allocate the powers of central and local governments. According to the empirical results, it is necessary to grant local governments with greater financial independence in order to enable them to effectively implement government initiatives aimed at lowering environmental footprints. Also, oversight of the capital flow of local authorities should be strengthened to ensure the efficient execution of investment funds allocated for environmental management. Apart from that, local governments should actively cooperate with local construction enterprises on low-carbon and energy-saving technologies to enhance their green innovation capabilities.

Third, the criteria for assessing the performance of local governments should be optimized. The findings indicate that the competition among local authorities in China is characterized as “bottom-to-bottom competition,” which can lead to significant ecological and environmental damage. Given that local governments in China are heavily influenced by performance evaluation, the central government should continue to abandon the “GDP-only” evaluation criteria, enhance the significance of environmental management in evaluating local officials, and reasonably guide each local government to shift to “top-to-top competition”.

There are still some limitations in this paper. First, the limitation is reflected in the indicators. There are various indicators and methods for measuring EG, and different indicators and measurement approaches may affect the CCEI differently. Therefore, based on different perspectives, subsequent studies could break down different types of EG to research the mechanism of its effect on the CCEI. Additionally, future studies need to explore more dimensions of control variables to reflect the influence of EG more comprehensively and accurately.