First, LGDG impacts carbon emissions through divestment of financing platform functions, debt replacement and the standardization of budget management. Local governments have always been at the centre of the economic and social system, and the financing platforms representing them have financing advantages in the traditional credit market, resulting in the continuous expansion of government debt. Demand-side competition theory suggests that expanding lending by finance platforms will crowd out the credit resources of private firms and reduce their access to finance (Asteriou et al., 2021). From the perspective of price competition theory, an increase in the demand for government debt pushes up the demand for funds in the financial market, leading to a rise in the interest rate on government borrowing (Boly et al., 2022). The capital asset pricing model shows that financial institutions will set interest rates on deposits and loans using the interest rate on government debt as an essential reference for their decisions. As a result, increasing the government’s financing costs will increase corporate lending rates (Zhou et al., 2023). Carbon emission reduction requires long-term financial support, and companies are less motivated to reduce emissions when faced with “difficult and expensive financing.” This precisely confirms the law revealed by the Environmental Kuznets Curve (EKC): Before the inflection point of the EKC, debt-financed economic growth was accompanied by high carbon emissions (Grossman and Krueger, 1995). After LGDG, on the one hand, local governments” financing platforms functions were stripped, reducing their indirect intervention in the credit market and thus increasing corporate credit lines. On the other hand, the replacement bonds issued by local governments to repay the financing platform’s stock debt have lower interest rates and longer terms, which reduces the demand of local governments for bank credit funds and lowers the actual market interest rate level, thereby reducing corporate financing costs. Zhang and Song (2021) point out that alleviating financing difficulties has increased corporate environmental protection investment, such as purchasing energy-saving and emission-reduction equipment, and increasing investment in green technologies. This has lowered the tipping point of the EKC and is conducive to developing a low-carbon economy (Riti et al., 2017). Additionally, the new local debt governance plan strictly regulates the use and repayment of debt funds. It also includes debt funds in budget management, improves the efficiency of budgetary fund allocation, and ensures that funds are allocated to environmental protection governance and supervision. Local governments often promote enterprises” green and low-carbon transformation through non-market tools, such as increased environmental supervision and law enforcement, and market-based tools, such as financial subsidies and tax incentives, thereby reducing carbon emissions. On this basis, the following Hypothesis 1 is formulated.

Hypothesis 1

LGDG reduces carbon emissions.

In China’s evaluation and promotion system for officials, which prioritizes economic growth, local governments have established a competitive model for attracting investment, characterized by low industrial land prices and increased infrastructure investment. The fiscal decentralization system of “centralizing financial power and decentralizing administrative power” has resulted in a continual increase in local government fiscal deficits. Local governments have increasingly turned to borrowing through financing platforms to alleviate fiscal pressure and attract investment, resulting in a continuous rise in debt. In 2014, the central government issued a new Budget Law and Opinions to regulate government debt financing behavior (Asteriou et al., 2021). Reducing “misallocation of land resources” and “economic infrastructure investment” is the main policy tool for LGDG.

Reducing “land resource mismatch.” On the one hand, local governments obtain bank loans based on credit or by pledging land assets to financing platforms. Loose borrowing relaxes the budgetary constraints of local governments, giving them more fiscal space for investment promotion. Industrial capital with a substantial tax base has muscular mobility and drives the development of upstream and downstream industries. Thus, it has become the main target of governments competing to attract capita (Mao et al., 2022). The fierce competition for investment promotion has led to expanding industrial land scale, distortion of industrial land prices and even “competition for the bottom line of quality (Bai et al., 2024).” Therefore, many high-carbon industrial enterprises have gathered in industrial parks, resulting in severe overcapacity and the formation of industrial clusters with high energy consumption and carbon emissions. On the other hand, debt expansion has intensified the debt servicing burden, prompting local governments to restrict commercial land use and sell commercial land at elevated prices to obtain land premiums as a guarantee for debt repayment. High commercial land prices raise the production cost in the service sector, leading to a decline in the profitability for high-value-added service industries, further hindering industrial restructuring. The delay in upgrading the industrial structure has resulted in increased energy consumption, reduced overall sector energy efficiency, and, increased urban carbon emissions. Additionally, the long-term low prices of industrial land suggest distortion from administrative intervention in the factor market. This distortion creates rent-seeking opportunities that enable enterprises to acquire production factors at lower costs and gain excess profits, thereby diminishing their incentive to innovate technologically. At the same time, the limited supply of commercial land has driven up housing prices. Profit-seeking enterprises tend to invest their capital in real estate for arbitrage, which crowds out capital investment in technological innovation and hinders regional carbon emission reductions. LGDG has stripped local financing platforms of their functions. It prohibits local governments from providing guarantees by injecting assets, such as land, into financing platforms, effectively reducing local government intervention in the land market. This reduces the problems of overcapacity resulting from the low price of industrial land and the decline in service industry profits caused by the high cost of commercial land. Low-efficiency, high-carbon-intensive enterprises are relocating due to rising industrial land costs, while the service industry is expanding due to increased profit margins, facilitating industrial restructuring and upgrading. At the same time, LGDG has decreased local governments’ reliance on land-related fiscal revenues. Enhanced efficiency in land resource allocation reduces opportunities for rent-seeking arbitrage, boosts corporate investment in technological innovation, and thus encourages carbon emissions reductions in the region. Therefore, the second hypothesis is proposed.

Hypothesis 2

LGDG reduces carbon emissions by lowering land resource mismatches.

Reducing “economic infrastructure investment.” The World Bank (1994) categorizes infrastructure investment into economic and social infrastructure investments. Local governments allocate substantial debt funds to the production sector to construct infrastructure, achieving economic growth. The allocation of financial funds has two effects on carbon emissions. First, in the case of financial constraints, local governments” preference for “production-heavy” investment squeezes out environmental protection inputs, reducing financial funds available for pollution control and environmental protection supervision, which results in decreased pollution control and emission reduction by enterprises. Second, local officials are often keen to invest debt funds in infrastructure development to highlight political performance and promotion (Zhao et al., 2023). In general, activities related to the transportation and construction sectors are the primary sources of carbon emissions. The impact of infrastructure development on carbon emissions is mainly reflected in the construction and operation periods. During the construction period, the increased demand for transportation infrastructure will drive demand for products in upstream industries such as steel and cement, producing many high-carbon-density building materials and high energy consumption, thus increasing carbon emissions (Santos, 2017). During the operational period, transport infrastructure facilitates the development of the logistics sector and the improvement of population agglomeration patterns. However, the frequent utilization of transport vehicles has increased pollution emissions. As Xie et al. (2017) point out, transport infrastructure promotes regional economic growth but damages regional ecosystems. The construction and operation of buildings also affect carbon emissions. Fossil fuel consumption for maintenance and construction is a direct source of carbon emissions, while electricity and heat are indirect sources of carbon emissions. Moreover, the increase in construction has driven the demand for transportation infrastructure, thereby increasing carbon emissions. The LGDG programme clarifies that the use and repayment of debt financing should be subject to a strict repayment plan and a stable source of debt-servicing funds and should be incorporated into budget management. At the same time, information on the budget and final accounts has been made public, increasing financial transparency. These provisions have prompted local governments to scientifically plan the use and investment of debt funds, optimizing the structure of financial expenditures. It reduces duplicative construction and “performance projects” and reduces energy consumption. Moreover, LGDG can also increase funding for environmental protection governance and supervision, including expanding the area of community green space, increasing forest resources, and subsidizing energy conservation and emission reduction by enterprises. This is conducive to zero carbon emissions (Zhang et al., 2022). Thus, a third hypothesis is proposed.

Hypothesis 3

LGDG reduces carbon emissions by lowering economic infrastructure investment.

The independent variable is carbon emissions per capita (Per_carbon). Following Cong et al. (2014), all direct emissions within the urban area, energy-related indirect emissions outside the metropolitan, and other indirect emissions from the spillover of activities within the city are calculated separately. The total carbon emissions of each prefecture-level city are obtained by summing them and then divided by the total population to get the per capita carbon emissions. In the robustness test, this paper regresses the ratio of all direct emissions in the urban jurisdiction to the total local population (Per_carbon1) and the ratio of energy-related indirect emissions outside the urban jurisdiction to the total local population (Per_carbon2), respectively, as proxies for the independent variable (Per_carbon).

The financing platform’s interest-free debt is non-financial and does not involve debt risk. Accordingly, the interaction term between the balance of interest-bearing debt of financing platforms and the time of reform implementation (Debt × post) is employed as the independent variable, and its coefficient elucidates the impact of the reform. Following the research by Hu et al. (2022), we use the average value of outstanding interest-bearing debt issued by financing platforms during the 3 years prior to the implementation of LGDG (2012–2014) as the value for Debt. The higher the value, the more substantial the dependence of each region on financing platforms, the greater the impact of policy. post is a dummy variable for implementing the reform, and the assignment is consistent with Equation 1.

First, this paper collects the amount and area of each land market transaction in the country and selects matching prefecture-level city data from each land data set. The average price and average area of commercial land and the average price and average area of industrial land in each prefecture-level city are compiled based on industry classification information. The logarithmic ratio of the average price of industrial to commercial land (Land1), the logarithmic ratio of the average area of industrial to commercial land granted (Land2), and the ratio of commercial land to the total area (Land3) are used to measure the mismatch of land.

Second, the World Bank (1994) states that economic infrastructure consists mainly of public utilities and works, and other transport sector facilities, while social infrastructure includes education, environmental protection and medical care. As local government debt funds are primarily invested in economic infrastructure construction (Huang et al., 2020), according to Fourie (2006), the logarithmic of the ratio of fixed asset investment in municipal public facilities to regional GDP and the ratio of highway mileage to end-of-year population are employed as the proxy variables for public utilities and works (Facilities) and transport infrastructure (Transport), respectively. Social infrastructure is measured regarding library collections per capita (Social) and environmental protection inputs (Environmental) with logarithmic treatment.

This paper controls for the following city characteristic variables: 1) Economic development (Per_gdp) is measured by taking the natural logarithm of GDP per capita. The level of economic development reflects the allocation of input factors such as labour, capital, and energy, which determine a city’s carbon emissions (Chai et al., 2023). 2)The proportion of the permanent urban population in the total permanent population of urban and rural areas measures the urbanization rate (Urban). The agglomeration effect of urbanization improves production efficiency and reduces carbon emission intensity. 3) Upgrading industrial structures can reduce a city’s dependence on traditional energy sources, influencing carbon emissions. This is measured by the proportion of the secondary industry (S_gdp) and the tertiary industry (T_gdp) in each city’s GDP. 4) Foreign capital dependency (FDI) is measured by the proportion of foreign capital actually utilized in each municipality to GDP for the year. 5) The fiscal deficit (Fiscal deficit) is the discrepancy between fiscal expenditures and revenues within the general budget. The larger the fiscal deficit in a region, the larger the debt is, and the more difficult it is to implement LGDG. 6) Regional economic development pressure (Pressure) is quantified by the natural logarithm of the age of the municipal party secretary (Mao et al., 2022). Pressure on regional economic development is the main reason for debt expansion. The smaller the pressure of economic development and the lower the productive expenditure, the more minor the crowding out of funds needed for carbon governance is conducive to carbon emission reduction. 7) Fiscal decentralization (Fiscal_dec) is determined by the share of municipal revenue in the sum of municipal, provincial and central revenue. The higher the autonomy of local finance, the stronger its ability to intervene in environmental protection and carbon reduction strategies, thereby affecting regional carbon emissions (Saveyn and Proost, 2008).

Table 1 demonstrates that the variation in the coefficient of carbon emissions (Per_carbon) in each municipality is about 79%, indicating significant differences in carbon emissions in different regions. The extreme difference of Debt is 1000.1900, indicating significant differences in the interest-bearing debt balance of local financing platforms in different cities, and the effect of reform would also differ. The minimum of Land1 is 0.0346, and the maximum is 21.2313. This indicates differences in the ratio of industrial to commercial land unit prices in different regions. Furthermore, it can be inferred that the lower the ratio, the stronger the motivation of local officials to be promoted. The mean value of Land2 is 1.4057, which is lower than the median of 1.4110. This suggests that, in most areas, the area of industrial land granted is greater than that of commercial land. The mean value of Land3 is 0.1197, indicating that the average ratio of commercial land area to the region’s total area is only 11.97 percent, distorting land resource allocation. Facilities and Transport have significant extreme differences, suggesting substantial variations in economic infrastructure across different regions. The mean and median of Social are close to each other, suggesting less variation of environmental investment in different regions. The standard deviation of the control variable, the fiscal deficit, is significant, indicating that the fiscal gap and the scale of debt borrowing exhibit considerable variation across different regions. The results for the remaining variables were as expected. The above results suggest the existence of notable regional heterogeneity.

Refer to Hu et al. (2022) to verify that the changing trend between the control and treatment groups was the same before the reform. The model is designed as shown in Equation 2. p e r _ c a r b o n i , t = ρ 0 + ∑ t = 2010 2019 ρ t P o s t t × D e b t i + ∑ γ j C o n t r o l i , t + ∑ C i t y + ∑ Y e a r + ε i , t (2)

The first period of the sample, 2009, is used as the base period, and the parallel trends are judged based on the significance of the net effect coefficients ρ t . The 95% confidence intervals for the pre-reform period 2010–2014, shown in Figure 2, all contain zero, so the coefficient on the interaction term is insignificant. There is no notable discrepancy in carbon emissions between the treatment and control groups compared to the baseline period, and the parallel trend test remains valid. Therefore, the premise for testing the effectiveness of LGDG using the IDID model is valid.

The paper conducts the following placebo test to eliminate the influence of unobserved factors during policy implementation: 1) Local financing platforms” balance of interest-bearing debt was randomly allocated across prefecture-level cities. This simulated reform variable interacted with the previous debt balance to re-perform the IDID regression. The process above was repeated 500 times to obtain the density distribution of the regression coefficients. As illustrated in Figure 3, the estimated coefficient of Debt × post is observed to be distributed around 0 under random assignment, tending to a normal distribution. In contrast, the coefficient of the baseline regression, −1.1851, clearly belongs to an extreme value, proving that carbon emission reductions are influenced by the LGDG rather than by the interference of other unobservable variables. 2) Suppose the reform is brought forward by 2 years. In Table 3 (1), the interaction term Debt × Time1 between the simulated variable of the reform time 2 years earlier and the mean value of the interest-bearing debt size of the financing platforms in each municipality does not significantly affect carbon emissions. In contrast, the shock caused by actual reform Debt × post is notable, suggesting that the decrease in carbon emissions is not due to the timing of the policy implementation or unobservable factors.

This study conducts substitution tests on the core explanatory variable and the dependent variable to eliminate endogeneity issues caused by variable selection bias. Referring to Cong et al. (2014), this paper uses direct per capita emissions within the urban jurisdiction (Per_carbon1) and indirect energy-related per capita emissions outside the urban jurisdiction (Per_carbon2) as the proxy variables for the explanatory variable per capita carbon emissions (Per_carbon) respectively. The results in Table 5 (2) and (3) demonstrate that the coefficient estimate on LGDG is significantly negative at the 1% level, consistent with the sign of the forecast in Table 5 (3).

The local financing platform debt is implicit debt outside the statutory debt limit, which may underestimate the debt size on local government statistics books. Therefore, referencing Brixi and Schick (2002), explicit and implicit in the local government debt matrix were used to carve out the debt boundaries. Data on explicit debt with legal repayment obligations is published only at the provincial level. Drawing on the research of Mao and Huang (2018), we allocated the explicit debt balance of provincial-level local governments to prefecture-level cities based on each prefecture-level city’s GDP proportion to the provincial GDP, thereby obtaining the explicit debt of prefecture-level cities. The results in Table 5 (6) show that the coefficient of LGDG is significantly negative, which is the same conclusion as in Table 2.

To reduce the impact of period selection on the benchmark regression results, this paper regresses the mean value of interest-bearing debt balances of financing platforms in the four and 5 years before the policy’s implementation as a proxy variable for Debt, respectively. Table 5 (4) and (5) show that the magnitude and significance of the LGDG are consistent with the results in Table 2.

To exclude the endogeneity issue stemming from sample selection bias, Debt × post is employed as the dependent variable, and per capita GDP, urbanization rate, industrial structure upgrading, regional economic development pressure, dependence on foreign investment and fiscal decentralization are utilized as covariates. The Logit model is used to ascertain the propensity scores of the observations to test for endogeneity problems. Based on scores, kernel density maps were created by matching 1:1 nearest neighbours of the experimental and control groups. Figure 4 shows a larger overlap area between the two groups, indicating that the data characteristics of the two groups are close and the matching effect is improved. After re-regression, the results in Table 6 (1) indicate that the coefficient for LGDG is −1.1028, essentially the same size and significance level as the baseline regression results.

China launched low-carbon pilots in 2013, 2014 and 2016. According to Luo et al. (2024), to eliminate the interference of low-carbon city pilot policies, cities belonging to the pilot region are assigned a value of 1 and vice versa as 0. The existence of duplicate pilot cities is defined as the earliest implementation of a low-carbon pilot. If a region is approved as a low-carbon pilot, all cities are low-carbon city pilots. Table 6 (2) shows that the low-carbon pilot policy has no significant effect on carbon emissions, preventing the policy from interfering with the empirical results.

The findings presented above indicate that LGDG can potentially influence carbon emissions. However, from a logical perspective, the implementation of unreasonable carbon reduction policies may also result in a reduction of tax sources, an increase in fiscal pressure, and the emergence of local debt risks. We perform instrumental variable estimation to eliminate the endogeneity resulting from reverse causality. This paper refers to Demirci et al. (2019), which uses healthcare expenditures (Medical) in prefecture-level cities as an instrumental variable for balancing interest-bearing debt of financing platforms. According to econometric principles, instrumental variables need to meet the correlation and exogenous assumptions. On one hand, the debt incurred by financing platforms was primarily used for investment expenditures, such as infrastructure and land development. However, LGDG has curtailed such spending. Under the strict constraints of the fiscal expenditure structure, the proportion of healthcare expenditure has increased, indicating a correlation between the outstanding interest-bearing debt of financing platforms and healthcare expenditure. On the other hand, healthcare expenditures are primarily influenced by demographic characteristics and do not directly involve industrial activities, such as energy consumption. Therefore, the instrumental variable has no direct relationship with carbon emissions and satisfies the exogenous assumption. The regression results in Table 6 (3) show that the interaction term Medical × post between the instrumental variable and the policy shock has a significant positive correlation with LGDG. Column (4) shows that the coefficient on LGDG is significantly negative at the 1% level and much more significant in absolute value than the estimated coefficient in Table 4 (3). This indicates that the endogeneity problem underestimates the role of LGDG in carbon abatement. The F-value for the first-stage regression of the instrumental variable is considerably greater than 10, indicating no weak instrumental variable problem.

In conclusion, the sign and significance of the coefficients on Debt × post are generally consistent with Table 4 (3), regardless of whether one conducts a parallel trends test, a placebo test, a substitution of explanatory variables and a PSM-DID test, a policy interference exclusion test, or instrumental variable estimation. Thus, LGDG significantly reduces carbon emissions, which is a robust result and further evidence of the first hypothesis.

The theoretical analyses in Section 2 demonstrate that LGDG is more effective in regions with higher debt dependence. LGDG constrains the government’s land transfer behaviour. It reduces the mismatch of land resources, leading to a reduction in the area, an increase in the price of industrial land, an increase in the area and a decrease in the price of commercial land (Bai et al., 2024). Therefore, LGDG helps reduce excess capacity and improve energy efficiency, reducing carbon emissions. Table 7 (1) and (2) show that the coefficient of Debt × post is significantly harmful. The coefficient of Debt × post in column (3) is significantly positive, indicating that LGDG reduces the distortion in industrial land prices in areas with high treatment intensity, reduces industrial land area, increases commercial land area, and reduces land resource mismatch. The second hypothesis is supported. The reason is that LGDG has reduced the local government’s dependence on land finance, reduced competition to attract investment by land, improved the quality of investment attraction and environmental protection policy standards, and reduced energy consumption and carbon emissions from industrial enterprises within the jurisdiction. At the same time, LGDG has increased the scale of commercial land use and reduced costs. Sufficient funding has promoted technological innovation in the service industry, improved labour productivity and energy efficiency, and reduced carbon emissions from the production end. Furthermore, LGDG reduces administrative intervention and optimizes the allocation of land resources, thereby decreasing the crowding out of green technology investments by rent-seeking arbitrage capital, which contributes to reducing carbon emissions.

The theoretical analysis demonstrated that LGDG has prompted more apparent improvements in areas heavily dependent on local government debt. The reduction in financial funds for economic infrastructure investment and the growth in financial investment in social infrastructure, such as environmental pollution control, have a more pronounced impact on optimizing the fiscal expenditure structure, thereby reducing carbon emissions (Xie et al., 2017). Table 8 (1) and (2) show that the coefficient for Debt × post is significantly negative, at least at the 5% level, indicating that LGDG is conducive to reducing investment in economic infrastructure. Columns (3) to (4) show that LGDG significantly increases the construction of social infrastructure, thus confirming the third hypothesis. LGDG reduces local governments” preference for productive expenditure. It lowers energy consumption and carbon emissions related to economic infrastructure construction, including various industrial parks, water and power supply facilities, and road transport. Investment in environmental protection, supervision, and other social public goods has increased, including tax breaks and emission reduction subsidies, thereby promoting the low-carbon development in the region and achieving carbon emission reductions.

Under an evaluation and promotion system of officials centered on GDP growth, economic development pressures are a crucial factor affecting the relationship between LGDG and carbon emissions. Regions with low economic development pressures demonstrate strong fiscal self-sufficiency and depend less on economic infrastructure investments and industrial land subsidies, resulting in weaker constraints on short-term growth. In light of the dual-carbon target requirement, these regions tend to utilize the funds released from LGDG to address pollution and enhance the environment, thereby attracting more production factors. At the same time, regions with lower pressure are less reliant on debt financing, less impacted by LGDG, and can effectively enhance environmental governance, resulting in more substantial carbon emission reductions. Accordingly, this section employs the age of the municipal party committee secretary as a proxy variable for economic development pressure to divide the groups. Older municipal secretaries with fewer promotion incentives tend to adopt conservative policies as the group with low economic development pressures, and younger municipal secretaries as the group with high pressures (Mao et al., 2022). The results in Table 9 (1) and (2) indicate that the coefficient of Debt × post is significantly negative in the smaller group, whereas it is not significant in the larger group. The findings indicate that LGDG exerts a more pronounced influence on carbon emission reductions when the pressure is lower.

The degree of marketization influences the green allocation efficiency of debt funds, thereby impacting the effectiveness of LGDG. A higher degree of marketization means mature product and factor markets, a sound legal and regulatory system, and high resource allocation efficiency. LGDG regulates local governments” debt financing behavior at the legal level, fosters a favourable development environment for economic entities, and facilitates cities” transition to green and low-carbon practices. When the degree of marketization is low, the inefficient allocation of market resources hinders the effectiveness of LGDG. Therefore, the regulatory effect of LGDG is more evident in areas with a high degree of marketization. Based on this, this study uses the marketization index method proposed by Fan et al. (2011) to ascertain each region’s degree of marketization. The pre-policy sample mean was employed as the basis for grouping, with those greater than or equal to the mean being the higher marketization group and those less than the mean being the lower group. Table 9 (3) and (4) demonstrate that the coefficient of Debt × post in the lower group is not statistically significant, while it is significantly negative in the higher group. This suggests that a higher degree of marketization results in a more significant effect of LGDG on carbon emission reduction.

Environmental regulations are direct measures for regulating carbon emissions and influence the green investment preference of debt. Stronger environmental regulations will raise the financing costs of high-carbon projects, prompting local governments to enhance environmental protection investments and green low-carbon subsidies to meet carbon emission reduction targets (Bai et al., 2024). Therefore, environmental regulation increases investments in debt funds for environmental protection. LGDG has limited debt expansion, with regions subject to stricter regulations significantly affected. Drawing on Zhou et al. (2024), this study uses the comprehensive index of various pollutant emissions in a city to represent the strength of environmental regulation. The median value of the index is used as the criterion to divide the sample into two groups: those with stronger ecological regulation and those with weaker environmental regulation. Table 9 (5) and (6) demonstrate that the coefficient of Debt × post is significantly negative in the high-intensity group, indicating that the carbon abatement effect of LGDG is more pronounced in regions with stronger environmental regulations.

The impact of LGDG on carbon emissions will also exhibit heterogeneity due to the varying resource endowments and industrial structures of cities across different geographical locations. The central and western regions face significant pressures for economic development, and local governments often rely on non-market methods, such as debt financing, to stimulate economic growth. This manifests in suppressing industrial land prices to attract high-tax-base industrial enterprises while increasing investment in economic infrastructure to promote investment attraction. These projects typically have high carbon emission characteristics. The eastern region is economically advanced, with the service sector accounting for a relatively high proportion of the regional economy. Debt funds primarily invest in low-carbon emission sectors, such as high-tech and modern service industries. Therefore, when local governments implement measures to control debt expansion, the central and western regions will be significantly affected. Moreover, under pressure from economic development, central and western regions may ease environmental regulations, and LGDG could be enforced by limiting financing for high-carbon projects, thus compelling enterprises to reduce carbon emissions. Based on this, this paper categorizes the sample cities into eastern, central, and western regions. Table 10 shows that the coefficient of LGDG is significantly negative, at least 5%, in the central and western regions. Meanwhile, it is not significant in the eastern areas. The results indicate that LGDG significantly impacts reducing carbon emissions in the central and western cities, but not in eastern cities.