In past studies, research on government bonds has consistently been a focal point for scholars. Since China allowed local governments to directly issue bonds and implemented debt substitution policies in 2015, academia has continuously engaged in in-depth research and investigation on this topic. Many scholars have primarily focused their studies on local government bonds in China. For instance, Xie and Chen (2020) conducted research on the structure of bond issuance and found that stronger local governments with higher leverage tend to opt for public issuance. Fu and Wang (2020) analyzed the theoretical mechanisms of local government debt reform practices from the perspective of budget constraints. They argued that during the period of economic new normal and stagnant real estate prices, general constraints rather than strict ones might be the optimal choice for preventing local government bankruptcy costs. Cheng and Ye (2020) discussed the latest developments in government debt substitution and suggested enhancing financial risk prevention and control. Qiu et al. (2022) found that local government bonds used for substitution could reduce the issuance scale of urban construction bonds through a “substitution effect.” On the other hand, debt substitution could lower the financing cost of urban construction bonds through an “implicit guarantee effect,” thereby weakening the “substitution effect.”

Regarding the impact of debt substitution on commercial banks, Jia (2017) conducted a study on the influence and effects of local government debt substitution on commercial banks, suggesting that while debt substitution could have short-term impacts on banks, it could fundamentally change local governments’ financing behavior and financing methods, thereby influencing banks’ long-term development models. Kong and Xie (2020) empirically examined the impact of local government bonds on the rate of return on commercial bank wealth management before and after debt substitution, discovering that local government bonds lowered and raised the rate of return on commercial bank wealth management respectively before and after debt substitution. Liang et al. (2020) found that debt substitution significantly altered the asset structure of commercial banks and improved their liquidity, with the change in liquidity being influenced by the absolute scale of debt substitution and the relative proportion of debt issuance. Wu and Xu (2017), using Jingdezhen City as an example, studied the positive effects of local government debt substitution on commercial banks and analyzed the adverse effects of local government debt substitution on the transformation and development of commercial banks. However, some scholars have expressed concerns regarding the risks associated with debt substitution. Liang and Hao (2019), using a five-sector DSGE model, found that while extending the debt term structure through debt substitution could alleviate debt accumulation and mitigate macroeconomic risks to some extent, it could also reduce the effectiveness of fiscal policy.

Scholars from developed countries have continually conducted in-depth research and exploration on banks’ holdings of government bonds, offering diverse perspectives that provide valuable insights. Fratianni and Marchionne (2017) found that banks worldwide adjusted their asset composition by replacing loans with securities to reduce overall regulatory credit risk in response to subprime and sovereign debt crises. Gennaioli et al. (2018) discovered that during normal times, there was a significant volume of bond holdings, and during sovereign defaults, there existed a strong negative correlation between banks’ holdings of government bonds and their loans. Crosignani (2021) found that if sovereign risk is sufficiently high, low-capital banks will reduce loans to the productive sector and further increase their holdings of domestic government bonds. Affinito et al. (2019) found that the high liquidity and high yield of government bonds, along with their convenience in terms of capital cost, make them suitable to meet banks’ needs during periods of declining bank liquidity, profitability, and loan quality. However, some scholars hold dissenting views on banks’ holdings of government bonds. Gros (2017) argues that the notion of allowing banks to purchase a substantial amount of their own sovereign bonds to stabilize the market in crises is flawed. Banks are intermediaries for private savings, and their financing costs exceed those of sovereign banks. Thus, purchasing bonds in large quantities during crises might exacerbate financial market pressure, rather than alleviate the crisis. Instead, governments should facilitate direct holdings of government debt by households (and other genuine monetary investors).

The financial ecological environment is a concept with distinct Chinese characteristics. Since Zhou Xiaochuan (2004) first proposed the financial ecological environment, Chinese scholars have basically formed a consensus: the financial ecological environment refers to the general term for the external basic environment of economy, politics, culture, society, law, and other factor on which various economic entities rely for survival and development. It is a multi-dimensional and complex system formed by the interaction and mutual influence of various factors relying on the fund chain and credit chain. In measuring the development quality of the financial ecological environment, most studies have chosen indicators such as economic foundation, financial development, credit and legal environment, local government public services, and social security to construct an evaluation system for the financial ecological environment (Deng and Chen, 2014). These indicators are considered key factors in maintaining and optimizing the financial ecological environment at the regional level. However, there is a lack of research that specifically analyzes the role of local government debt as a variable in the construction and optimization of the financial ecological environment. Existing studies mostly indirectly analyze the impact of local government debt on economic and financial development or analyze it from the perspective of local government behavior.

Firstly, it is difficult to draw a consistent conclusion on the impact of local government debt on financial ecological environment. Local government debt stimulates economic growth by supplementing local financial resources and promoting infrastructure development (Martins, 2021). However, the expansion of debt leads to long-term changes in the fund structure and interest rates, resulting in crowding-out effects on commercial bank loans and private capital investment. This, in turn, has a negative impact on capital accumulation and economic growth (Demirci et al., 2019; Cheng et al., 2022; Johnson and Yushkov, 2022).

Local government debt has different effects on regional economic growth due to variations in its scale, structure, and types. Furthermore, the impact of local government debt on regional financial stability follows two paths: first, it operates through the transmission channel of local government debt—shadow banking—financial risk securitization—systemic financial risk (Yarba and Güner, 2020; Hassan et al., 2022). Second, the spiral growth of local government debt and dependence on land finance drives the expansion of the real estate asset price bubble, potentially triggering systemic financial risk (Gelpern and Gerding, 2016; Chang et al., 2022; Tian, 2022). Lastly, according to Xingzhi and Wang (2009), the fundamental cause of the differences in the financial ecological environment among Chinese regions lies in the behavioral intervention of the central and local governments. Government behavior plays a crucial and dominant role in the evolution of the financial ecological environment but also leads to its deterioration. This is mainly manifested in the strengthening of administrative tendencies in financial resource allocation, a reduction in the efficiency of market-based allocation of financial resources, the speed of innovation in the financial industry, and risk resistance capabilities, as well as an increase in moral hazard issues (Zhang and Dai, 2007).

According to Table 1, significant progress has been achieved in related studies. There are still limitations and further areas of exploration. First, there is a lack of direct research on the impact of local government debt on the regional financial ecological environment. It is necessary to further explore the mechanisms and effects of the role of local government debt in the financial ecological environment from a systemic perspective. Second, it has been indicated in some literature that the debt structure and its efficiency of use also require attention. Third, previous research has confirmed the spatial spillover effects of local government debt, but the spatial impact of local government debt growth on the financial ecological environment needs to be examined. Based on these considerations, this study will examine the impact of local government debt on the financial ecological environment from the dual perspectives of debt scale and structure, and derive relevant insights. This will contribute to a better understanding of the role of local government debt in financing the construction of the financial ecological environment and provide valuable references for future policy-making.

The local government debt data includes estimates of both explicit and implicit debt. The explicit debt is represented by the balance of local government bonds issued by cities since 2009. However, for the years 2009–2014, where city-level data was not available on the Ministry of Finance website and only provincial-level data was provided, we followed the approach of Mao Jie and Huang Chunyuan (2018) to allocate the provincial bond balance to individual cities based on their respective GDP contribution to the provincial GDP. As for implicit debt, bank loans generated by local financing platforms constitute the main component. Therefore, in estimating the scale of implicit debt, we selected the interest-bearing debt of financing platforms as a proxy indicator for implicit debt balance. This approach avoids underestimating the size of local government debt by solely considering local government financing vehicle bonds, thus maximizing coverage of local government implicit debt.

For the financial ecological environment data, a regional financial ecological environment evaluation system was constructed, incorporating four primary indicators: economic foundation, financial development, government governance, and social credit foundation. It also included 13 secondary indicators such as economic development level, economic vitality, and economic openness. In selecting specific tertiary indicators, we adhered to principles of scientific validity, objectivity, comprehensiveness, comparability, and measurability based on indicator data. For some indicators lacking data sources, such as the closure rate of financial cases, their exclusion would not fundamentally affect the evaluation of the financial ecological environment and could be considered as a technical adjustment. Data for this study was collected from various official sources including the “China Statistical Yearbook,” “China Financial Yearbook,” “China City Statistical Yearbook,” as well as the websites of local statistical bureaus and the Ministry of Finance of China.

The entropy method is an objective weighting method that can determine the weight of indicators based on the size of the information provided by the observed values of various indicators. It avoids the bias caused by human subjective factors and ensures that the results can be better explained. Therefore, the entropy method is used to assign weights and scores to the financial ecological environment evaluation system constructed for 227 prefecture-level cities in China. Based on normalized and standardized data, the calculation formulas for the weight W j of the j -th indicator and the financial ecological environment evaluation score F i of the i -th region using the entropy method are respectively: W j = g j ∑ 1 m g j , F i = ∑ W j P i j (1)

In formula (1), P i j represents the weight of the j th indicator for the i th prefecture-level city, and g j represents the coefficient of variation of the j th indicator.

Taking the panel data of 227 prefecture-level cities in China as an example, this study conducts an in-depth investigation into the relationship between local government debt and the financial ecological environment. It analyzes the scale and structural effects of local government borrowing on the financial ecological environment and further explores the regional heterogeneity of the impact of local government debt expansion on the financial ecological environment. Accordingly, based on Hypothesis 1 and Hypothesis 2, the following benchmark regression models are constructed to examine the relationship between the size and structure of local government debt and the financial ecological environment: f e e i t = α 0 + α 1 d e b t g d i t + ∑ α i c o n t r o l i t + μ i + v t + ε i t (2) f i t = α 0 + α 1 h − t i t + ∑ α i c o n t r o l i t + μ i + v t + ε i t (3)

Where i represents the prefecture-level city and t represents the year, μ and v represent unobservable regional and time effect random variables used to control fixed effects that cannot be directly observed at the regional and time levels. ε represents a random error term. The explained variable fee represents the quality of the regional financial ecological environment. The core explanatory variable debtgdp represents the local government debt level of the region and is replaced by the debt rate debtfin in the robustness test. h t is the proportion of local government implicit debt to total debt scale, used to represent local government debt structure. The higher the proportion of implicit debt, the higher its structural risk may be. Other control variables include economic growth quality, population size, openness to the outside world, credit structure, industrial structure, fiscal dependence, etc., which are measured by per capita real GDP growth rate (pggdp), total resident population (popula), import and export amount as a percentage of GDP (open), financial institution loan-to-deposit ratio (credit), tertiary industry output value ratio (indust), and fiscal revenue ratio (revenue).

The expansion of local government debt not only affects the development of the financial ecological environment in the region but may also have an impact on surrounding areas. Therefore, according to h y p o t h e s i s 3 , a spatial econometric model is set up and a widely used spatial Durbin model is used to test the spatial spillover effect of local government debt on the financial ecological environment: f e e i t = ρ ∑ j = 1 n W i j f e e j t + β ∑ j = 1 n W i j X j t + γ X i t + ∑ δ i c o n t r o l i t + α i + μ t + ε i t (4) ε i t = λ ∑ j = 1 n W i j ε i t + v i t (5)

Where X includes debtgdp and h t , α i , μ t and ε i t represent individual effects, time effects and random error terms respectively. The coefficients of each variable are α , β , γ , δ , κ , λ and W i j is the spatial weight matrix, using two types of weight matrices: geographic distance and economic distance. The geographic distance weight matrix is the reciprocal of the geographic distance between cities, and the economic distance weight matrix is the product of the reciprocal of the average GDP of 227 prefecture-level cities in China from 2000 to 2020 and the geographic distance weight matrix, which is then standardized. The geographic matrix takes into account physical world mapping and geographical factors. The economic matrix considers influences from economic, social, cultural, and other aspects. In subsequent tests, matrix forms are changed to ensure the robustness of the results.

Table 2 presents the Pearson correlation coefficients for the main variables. Notably, there is a significant positive correlation at the 1% level between local government debt (debtgdp) and the financial ecological environment index (fee). This suggests an initial indication that local government debt is conducive to promoting the financial ecological environment. Furthermore, local government debt (debtgdp) shows a significant positive correlation at the 1% level with economic foundation (eco), financial development (fin), government governance (gov), and social credit basis (soc). This preliminary evidence suggests that local government debt enhances the financial ecological environment by fostering economic foundation, financial development, government governance, and social credit basis. Regarding control variables, most of them exhibit significant correlations with local government debt at the 1% level, indicating the reasonability of the chosen control variables. Moreover, the correlation coefficients among variables are all below 0.5, implying the absence of severe multicollinearity issues among variables.

The test for multicollinearity shows that the variance inflation factor ( V I F ) of the related variables is all less than 5, indicating that there is no multicollinearity. Tables 3, 4 report the test results of the impact of local government debt size and structure on the financial ecological environment using a two-way fixed effects model after dealing with heteroscedasticity. Columns (1) to (5) are the regression results of taking the financial ecological environment index ( f e e ) and four dimensions of economic foundation ( e c o ), financial development ( f i n ), government governance ( g o v ), and social credit foundation ( s o c ) as dependent variables. Table 3 shows that the impact of local government debt on the regional financial ecological environment is significantly negative at the 1 % level, initially confirming that the process of local government borrowing financing has inhibited the construction and optimization of the financial ecological environment, and hypothesis 1 has been verified. This indicates that during the sample observation period, the expansion of local government debt has fully demonstrated its inhibitory effect on the overall development of the financial ecological environment, and various risks have increased. Specifically, the expansion of debt size has a promoting effect on the construction of economic foundation and financial development foundation in the region, and has an inhibitory effect on the social environment and government governance efficiency in the region. Since the effect of the former is significantly smaller than that of the latter, local government debt expansion overall inhibits regional financial ecological environment development.

Table 4 reveals that, in terms of debt structure, a higher proportion of implicit debt is detrimental to the optimization of the financial ecological environment, which is statistically significant at the 1% level and aligns with the inference of Hypothesis 2. This could be attributed to the increased repayment and interest pressure faced by local governments as the proportion of implicit debt rises, impeding their investments in projects supporting innovation, talent attraction, and industrial structure upgrading. Consequently, local government credit and governance capacity are compromised, regional financial risks are heightened, and the quality of the local financial ecological environment is diminished. Specifically, the debt structure variables exhibit significant negative correlations with the social environment and government governance, highlighting the threatening nature of local government implicit debt to the sustainable and healthy development of the financial ecological environment. Furthermore, it exerts a notable inhibitory effect on the local social environment and government efficiency. The main reason behind this lies in the close association between local government implicit debt, borrowed through financing platforms and shadow banking, and local financial institutions. The inherent problems of implicit debt, such as low investment returns and uncertain repayment funds, may diminish the asset quality of financial institutions, increase their liquidity risks, and raise their non-performing loan ratios, thereby hindering local financial development and the optimization of the financial ecological environment. On the other hand, local government implicit debt can facilitate economic foundations and financial development. This is mainly attributed to the fact that local government implicit debt, primarily obtained through urban investment bonds, enables land financing. Subsequently, local governments utilize land finance for real estate development, thereby stimulating the local economy and economic growth. However, since the inhibitory effect of the former is significantly greater than the latter, the expansion of local government implicit debt overall suppresses the favorable development of the regional financial ecological environment.

Taking into account the potential variations in the impact of changes in local government debt on the financial ecological environment, we further employ the threshold regression model proposed by Hansen to examine whether the effect of local government debt scale exhibits a threshold effect. Based on the results of both single and double threshold tests, the F-statistic and its corresponding p-value suggest that the null hypothesis cannot be rejected. This implies that there is no threshold effect of local government debt on the financial ecological environment during the sample period. This relationship is consistently maintained in a linear manner.

The spatial autocorrelation test reveals significant results for the global Moran’s I index of the financial ecological environment across 227 prefecture-level cities nationwide, indicating the presence of spatial correlation in the regional financial ecological environment. Subsequently, through LM, Hausman, and LR tests, we determine the adoption of the spatial Durbin model (SDM) with time and space fixed effects. For comparison purposes, Table 5 reports the regression results of both the spatial autoregressive (SAR) and spatial error (SEM) models. Columns (1) to (3) demonstrate that the coefficients of local government debt obtained from different regression models are negative and significant at the 1% level. The spatial lag term also passes the significance test, indicating positive spatial spillover effects of debt expansion on the development of financial ecological environments in other regions. Columns (4) to (6) reveal a negative relationship between debt structure indicators and the financial ecological environment. In all model regressions, the coefficients of the proportion of implicit debt are significantly negative, while the spatial lag term coefficients are positive. This suggests that as the proportion of implicit debt increases, local government debt expansion becomes detrimental to the optimization of the local financial ecological environment. However, it can promote the development of financial ecological environments in neighboring regions. A higher proportion of implicit debt implies greater debt burdens and financial risks, which hinders credit construction and impedes the development of financial ecological environments. Nevertheless, the expansion of implicit debt in the local region may indirectly drive financial resources in neighboring areas, thereby stimulating the development of financial ecological environments in those regions as well.

To further understand the spatial impact of local government debt on the financial ecological environment, Table 6 presents the decomposition results of the direct effect, indirect effect, and total effect of the SDM model. The results indicate a significant spatial spillover effect of local government debt on the regional financial ecological environment. During the observation period, the expansion of local government debt scale significantly inhibits the development of the local financial ecological environment while promoting the construction of financial ecological environments in neighboring regions. Moreover, an increase in the proportion of implicit debt also hinders the continuous optimization of the financial ecological environment in the local region while stimulating the development of financial ecological environments in adjacent areas.

To address the endogeneity issue resulting from the bidirectional causal relationship between local government debt and the financial ecological environment, this study employs instrumental variable (IV) approach to conduct endogeneity tests. Lagged values of debt scale and debt structure are introduced as instruments, and the two-stage least squares (IV-2SLS) method is applied for regression analysis. Firstly, there is a strong correlation between the lagged one-period local government debt and the current-period local government debt, as they exhibit a time-series relationship. Secondly, the lagged one-period local government debt reflects debt incurred in previous periods and is unrelated to the current-period error term (the impact of financial environment). The lagged one-period variables for debt scale and debt structure meet the requirements of correlation and independence ¹ . The decision-making regarding local government debt is typically influenced by factors such as policies, economic cycles, and fiscal planning. Unobservable factors that lag by one period result from past economic and policy events. Therefore, they are relatively independent over time and should not exhibit serial correlation. The results are presented in Table 7. The first stage results show that both the local government debt ratio and the proportion of implicit debt are significantly positively correlated with their lagged values. In the second stage, after controlling for endogeneity issues, the local government debt ratio is found to be negatively correlated with the financial ecological environment, and the proportion of implicit debt is also negatively correlated with the financial ecological environment. These results pass the significance test at the 1% level, and are consistent with the baseline regression model, further confirming the robustness of the regression results.

To improve the reliability of the regression analysis results, this study replaces the debt ratio indicator with the debt-to-GDP ratio and retests Eq. 2. According to Table 8. The results indicate that the estimated coefficients of the debt-to-GDP ratio variable, after replacing the core variable, exhibit similar signs and levels of significance compared to the baseline regression results. Additionally, the significance of some control variables is enhanced. These findings confirm the robustness of the baseline regression results.

To enhance the robustness of the spatial effects analysis on the impact of local government debt on the financial ecological environment, further regression verification was conducted using an economic distance weight matrix. The regression results are presented in Table 9. After replacing the spatial weight matrix with the economic distance matrix, the estimated coefficients of the main core explanatory variables obtained from the regression remain consistent with the previous results, thus confirming the reliability of the earlier findings.