Does Local Government Debt Affect Corporate Innovation Quality? Evidence from China

Abstract

This study investigates the impact of local government debt levels on the behavior of individual firms, which is crucial for understanding the systemic risks associated with local government debt and fostering economic vitality. Using data from publicly listed companies on the Shanghai and Shenzhen stock exchanges between 2013 and 2022, this study empirically examines the effect of local government debt on corporate innovation quality. The findings demonstrate that local government debt expansion has a significant negative impact on corporate innovation quality. The negative impact remains robust across endogeneity tests and multiple robustness checks. Channel analysis indicates that as local government debt increases, innovation subsidies and procurement funding led toward firms’ decline, while both tax and non-tax revenue demands indicated firm increases. This resource reallocation contributes to the observed decline in corporate innovation quality. Further heterogeneity analysis reveals that regions with lower levels of government intervention and fiscal pressure exhibit a smaller negative effect of local government debt on innovation quality. Finally, examining the economic outcomes reveals that the decline in innovation quality, resulting from current local debt expansion, significantly reduces total factor productivity and firm value in the subsequent year, posing challenges for sustainable corporate development.

1. Introduction

Since the 2008 global financial crisis, China has implemented policies to stimulate domestic demand, including a CNY 4 trillion investment plan (USD 547.95 billion), to counter economic slowdown. Local governments have primarily funded these initiatives through bank loans and bond issuance via financing platforms, which has led to rapid local government debt growth [1,2]. According to China’s National Bureau of Statistics, the new local government debt totaled approximately USD 639.73 billion in 2023, bringing the national balance to nearly USD 4.79 trillion. While debt expansion has achieved certain goals, such as alleviating fiscal pressures and stimulating local economic growth, it also poses risks [3]. For instance, under high debt pressures, local governments tend to reduce innovation investments and cut fiscal subsidies to firms to ease their financial burden, which can undermine sustainable corporate innovation. Additionally, when public funds are insufficient, governments are more likely to increase taxes or impose additional fines to cover expenditures and service debt obligations. However, these actions heighten firms’ tax burdens and non-tax expenses, weakening their innovation environment and, ultimately, their innovation quality [4,5].

High-quality innovation is a primary driver of long-term economic growth [6]. As the key providers of products and services, firms not only hold high-quality innovations such as patents but also serve as primary actors in applying patented technologies for product development and capacity expansion. Therefore, firms play a critical role in innovation activities. Corporate innovation performance should not be measured merely by patent counts; rather, innovation quality is crucial. Research has extensively examined firms’ innovation behaviors, capabilities, influencing factors, and economic outcomes [7,8,9], yet innovation quality has received limited attention. In a government-led economy like China, the government plays a particularly vital role in resource allocation and economic guidance, thereby substantially influencing firms. Local government debt, which is essentially a special type of resource, can create an “innovation crowding-out effect” when it is excessively high, resulting in various adverse impacts on corporate innovation.

From the new institutional economics perspective, the negative impact of local government debt on the innovation ecosystem can be viewed as a form of institutional crowding-out. North’s theory of institutional change posits that an imbalanced institutional environment can significantly impact the incentive mechanisms of micro-entities [10]. Rapid local government debt expansion can disrupt the institutional equilibrium for regional innovation, thereby affecting firms’ innovation behaviors. As local government debt increases, associated debt risks gradually increase. This diminishes its positive effects on the macroeconomy, supporting the real economy and driving economic development, while amplifying negative impacts [11,12,13].

Meanwhile, from a resource supply perspective, high debt pressure drives local governments to significantly reduce fiscal subsidies for corporate R&D and procurement scales, weakening the external support for corporate innovation. Additionally, to bridge fiscal deficits, local governments often increase tax intensity and non-tax charges (e.g., administrative donations and fines), transferring fiscal pressures to firms [14]. This institutional resource crowding-out not only restricts firms’ marginal financial resources for R&D but also distorts their innovation incentive structures at a deeper level. Consequently, firms may shift from pursuing long-term technological innovation to focusing on short-term survival strategies, thereby undermining the strategic sustainability of innovation inputs and ultimately degrading the quality of the corporate innovation ecosystem.

Extant research on measuring corporate innovation quality primarily focuses on patent counts and types. Common indicators include patent citation frequency and the number of invention patents [15,16,17]. However, these metrics have limitations. Patent citation data are often incomplete. Meanwhile, invention patent counts only reflect the quantity of innovation, lacking deeper insights into quality. Therefore, more robust measures are needed to accurately evaluate corporate innovation quality. Here, the patent knowledge breadth method offers a new perspective for assessing innovation quality. Aghion et al. [18] and Akcigit et al. [19] pioneered the use of knowledge breadth to measure patent quality, thus evaluating firms’ innovation quality. This approach effectively reflects the complexity of knowledge embodied in patents, addressing the limitations of patent quantity-based assessments. Therefore, this study uses the knowledge breadth method to estimate patent quality among Chinese firms, providing a novel perspective on innovation issues within the Chinese context.

To further investigate the consequences and underlying mechanisms of local government debt expansion on innovation behavior in microeconomic entities, this study draws on crowding-out effects and dynamic capability theory and empirically examines the impact of local government debt on corporate innovation quality. We use data from Chinese firms listed on the Shanghai and Shenzhen stock exchanges from 2013 to 2022. Using Python-based retrieval and web-scraping techniques, data on local government debt and corporate patent information were collected from public platforms such as China’s Local Government Bond Information Disclosure Platform, the China Government Procurement Network, and the China National Intellectual Property Administration. The city-level data were then matched with firm-level data to create a comprehensive “city–firm–year” database.

The results reveal that local government debt expansion significantly negatively affects corporate innovation quality. Crucially, this effect holds across endogeneity tests and multiple robustness checks. Channel analysis suggests that local government debt influences firms’ innovation behavior through the revenue enhancement and expenditure reduction channels. From the expenditure reduction perspective, local governments reduce innovation subsidies and procurement for firms. From the revenue enhancement perspective, local governments impose higher tax and non-tax burdens on firms. Heterogeneity analysis shows that lower government intervention and reduced fiscal pressure can mitigate the negative impact of local government debt on innovation quality. Additionally, economic outcome analyses indicate that the innovation quality decline due to current local government debt expansion significantly deteriorates firms’ total factor productivity (TFP) and firm value in the following year, posing challenges for sustainable corporate development.

This study makes three primary contributions. First, it expands the literature on the microeconomic effects of local government debt. Studies have primarily focused on macro-level impacts, such as the effects on regional economic growth, financial market resource allocation, and shadow banking development. Some studies have examined the negative externalities of local government debt on firms’ investment and financing behavior. Meanwhile, this study focuses on the impact of local government debt on corporate innovation quality, providing empirical evidence on the crowding-out effect of local government debt at the micro level. Second, from the corporate innovation quality perspective, this study measures innovation quality based on the knowledge complexity embedded in corporate patents. Extant research primarily uses patent counts and types to measure corporate innovation quality, which has limitations and may not adequately capture innovation quality. Employing the patent knowledge breadth method helps us provide a new perspective for examining innovation issues within the Chinese context. Finally, this study thoroughly examines the channels through which local government debt affects corporate innovation quality from both revenue enhancement and expenditure reduction perspectives, enriching the theoretical framework for understanding how local government debt impacts corporate innovation quality.

2. Literature Review and Hypothesis Development

2.1. Literature Review

2.1.1. The Economic Effects of Government Debt

As a crucial fiscal instrument, local government debt drives regional economic development by funding infrastructure projects and enhancing public services, thereby strengthening fiscal autonomy and providing the foundational support for corporate innovation. A moderate expansion of local government debt can positively influence economic growth and employment, fostering a stable business environment and market order [20,21]. However, excessive debt accumulation may introduce various adverse macroeconomic effects, including heightened fiscal risks, resource misallocation, regional inflation, and policies locked into short-term returns [22]. Such conditions not only threaten financial system stability but may also suppress public services and innovation investments, adding to the macroeconomic performance uncertainties [23]. Therefore, balancing economic growth with debt management has become a key focus for scholars and policymakers alike.

On a microeconomic level, while local government debt can create a favorable environment for firms, it can also impose burdens when debt levels become excessive. Moderate debt expansion supports improved business operations through policies such as infrastructure development, tax incentives, and credit support, cultivating external conditions conducive to innovation [24]. However, high debt levels often result in a “crowding-out effect”. Specifically, to meet debt obligations, governments may increase tax burdens or reduce support for innovative firms. Consequently, firms must operate under heavy tax burdens and constrained financing, thereby stifling their willingness to invest in innovation and transformation [25,26]. Furthermore, excessive debt may prompt local governments to favor low-value-added industries that yield quick returns, exacerbating structural imbalances and negatively impacting the development of high-tech industries and innovation ecosystems. Hence, local government debt levels not only affect macroeconomic stability but also have profound implications for corporate innovation vitality and long-term development potential [27]. The prudent control and optimized utilization of government debt are essential for promoting sustained corporate innovation and high-quality regional economic development.

2.1.2. Knowledge Management and Corporate Innovation

Knowledge management (KM) refers to the systematic process of acquiring, creating, storing, sharing, and utilizing knowledge within an organization. It involves the structured management of organizational knowledge assets to create value and meet both tactical and strategic objectives [28]. At its core, KM focuses on fostering the creation, acquisition, dissemination, and application of knowledge to enhance organizational competitiveness and drive innovation.

Since the 1960s, the classification of knowledge in knowledge management (KM) has evolved significantly, reflecting advancements in theory and practice [29]. Initially, knowledge was broadly categorized into tacit knowledge (intuitive and experiential, difficult to articulate) and explicit knowledge (codified and easily transferable). By the 1990s, additional classifications emerged, including declarative knowledge (“know-what”), procedural knowledge (“know-how”), conditional knowledge (“know-when”), and relational knowledge (“know-with”), emphasizing diverse applications of knowledge. The 2000s saw the inclusion of embedded knowledge (ingrained in systems, culture, or processes), contextual knowledge (dependent on specific environments), and cultural knowledge (reflecting shared organizational values). In recent years, the focus has shifted toward dynamic knowledge (adaptable and evolving), big data-driven knowledge (derived from analytics), strategic knowledge (aligned with long-term goals), and emerging forms like machine-generated knowledge and collective knowledge (facilitated by AI and collaboration). This progression illustrates a growing emphasis on integrating technology, adaptability, and collaboration into KM frameworks [30,31].

Corporate innovation can be defined as the systematic development and application of new ideas, technologies, products, services, or processes to enhance organizational performance and create competitive advantages [32]. Technical innovations focus on advancements in tools, processes, and technologies, driving efficiency and productivity, while social innovations aim to address societal challenges by reshaping behaviors, norms, or institutional structures. According to the OECD and Oslo manual, the four types of innovation—product, process, marketing, and organizational innovations—provide a comprehensive framework for analyzing innovation dynamics [33]. Among these, product innovation is closely linked to technological advancement and serves as a key indicator of a firm’s patent quality. Therefore, this study focuses on product innovation as the primary dimension for assessing corporate innovation quality.

Explicit knowledge, which can be codified and easily shared, spreads rapidly across firms and nations, contrasting with tacit knowledge—rooted in experience and intuition—that is indispensable for fostering genuine innovation. The interplay between education, tacit knowledge, innovation, and technology is critical; education not only builds foundational knowledge but also cultivates the cognitive and practical skills necessary for technological advancements and creative problem-solving. Expanding the analytical perspective to integrate these dimensions will provide deeper conceptual clarity and a holistic understanding of innovation in practice.

Tacit knowledge refers to the experiential, intuitive, and context-specific understanding that is challenging to articulate or codify but essential for problem-solving and creativity (Polanyi, 1966) [34]. It plays a critical role in the creation and interpretation of complex patents, as it enables inventors to integrate diverse knowledge domains, navigate uncertainties, and develop innovative solutions that go beyond explicit, codified information. Tacit knowledge underpins innovation quality by fostering interdisciplinary approaches and contextually rich solutions, which are essential for generating patents that embody high levels of knowledge complexity and practical relevance [34,35]. This interplay between tacit knowledge and explicit knowledge ensures that innovation is not only technically robust but also adaptable and impactful.

2.1.3. Measuring Corporate Innovation Quality and Its Determinants

In recent years, the concept of innovation quality has garnered increasing attention in corporate innovation research, with scholars exploring measurement methods from various perspectives. Haner [36] initially proposed evaluating corporate innovation quality based on three dimensions: product, management, and process improvements. Building on this framework, some scholars have defined innovation quality as the performance of final products or services, managerial innovation effectiveness, and operational process improvements. Krammer and Sorin [37] suggested using the number of invention patents, design patents, and utility models as quantitative indicators of innovation quality. Hsu et al. [38] employed patent citation frequency to assess patent quality. Thus, the literature commonly uses patent quantity and quality as primary indicators of innovation quality, although such measures have limitations. However, using patent citation frequency to assess patent quality has notable limitations. Citation practices can be biased by industry, region, or strategic considerations. Additionally, citations often reflect broader applicability rather than technological significance, while legal and administrative influences, such as examiners’ references, may skew results. This metric also emphasizes explicit knowledge, overlooking the tacit knowledge critical to innovation, and may underestimate the quality of patents. Large corporations like Toyota are adopting innovative approaches to assess patent quality, moving beyond traditional metrics like citation frequency. Toyota also emphasizes techniques incorporating metrics such as claim breadth, originality, and technical complexity. These strategies enable a more comprehensive and forward-looking assessment of patent quality.

Research on the determinants of corporate innovation quality has primarily focused on internal and external factors. Studies have widely examined the influence of firm size, R&D investment, and ownership structure on innovation quality [39]. Additionally, from a strategic perspective, some scholars have analyzed the effects of corporate social responsibility (CSR) strategies and institutional investment on innovation quality. For instance, CSR strategies can enhance innovation quality by reducing agency problems and strengthening corporate absorptive capacity [40]. At the macro level, government innovation incentives, the financial environment, and legal frameworks also play critical roles in shaping corporate innovation quality [38]. Some studies suggest that the impact of policies, such as tax incentives and government subsidies, on innovation quality varies with policy heterogeneity [24]. Furthermore, strengthened patent laws and intellectual property protections may influence innovation quality in complex ways [39]. However, few studies have explored the impact of local government debt—a macroeconomic shock—on corporate innovation quality.

In summary, the methods for measuring corporate innovation quality are largely based on patent quantity, market performance, and technological depth, with few studies adopting knowledge breadth as a measurement approach. Research generally considers factors affecting innovation quality across multiple dimensions, including firms’ internal R&D strategies, ownership structures, CSR strategies, and the external policy environment. However, the negative effects of government debt expansion and its impact on corporate innovation quality have received comparatively little attention.

2.2. Hypothesis Development

In pursuit of economic stimulus, China’s extensive local government debt initiatives have, in the short term, achieved outcomes such as balancing local fiscal budgets, stabilizing economic growth, and optimizing resource allocation. However, as local governments have not phased out debt according to central government plans, repayment pressures have risen significantly, thereby escalating regional financial risks. Generally, the scale of local government debt reflects the intensity of its repayment burden. Thus, increased debt pressures undoubtedly alter local governments’ fiscal arrangements. These changes in fiscal management not only influence government actions but also spill over, affecting firms at the microeconomic level within the region [41]. Specifically, local government debt levels influence corporate innovation activities through two primary channels: expenditure reduction and revenue enhancement.

From the expenditure reduction perspective, local governments may curtail innovation subsidies and reduce procurement levels for firms. Innovation subsidies, a key component of fiscal expenditure, represent direct or indirect funding transfers to firms without repayment obligations [42]. Providing subsidies effectively means the government forgoes potential returns on these funds. Consequently, when local government debt reaches high levels, subsidy funding is often reduced to minimize default risk [3,43]. Furthermore, government procurement, characterized by both administrative and market functions, is a crucial tool for protecting and supporting local firms, particularly in alleviating financing constraints. Since procurement is largely a flexible component of fiscal expenditure, high debt pressures tend to reduce governments’ willingness to engage in such expenditures.

From the revenue enhancement perspective, local governments may increase tax and non-tax revenues extracted from firms to fulfill their fiscal obligations [44]. Taxation is a primary source of government revenue. Under high debt conditions, local governments have an incentive to raise tax rates to fill fiscal gaps. As a direct reflection of fiscal policy, tax enforcement affects firms’ actual tax burdens. When local debt pressures are substantial, governments often increase tax collection efforts to boost revenue and ease repayment pressures. Additionally, non-tax revenues offer an alternative avenue for alleviating debt pressure. Non-tax revenues mainly include fines and corporate donations. Studies indicate that when local debt pressures are high, governments may increase fines (such as environmental or audit-related penalties) and solicit donations to enhance fiscal revenue. Notably, non-tax revenue sources have shorter time horizons, while local governments retain discretion over both itemized charges and rates. Hence, high debt levels create incentives for governments to increase non-tax revenues as a means of relieving fiscal stress.

These dynamics suggest that when local governments face debt pressures, they may resort to increasing both tax and non-tax revenues while simultaneously reducing innovation subsidies and procurement expenditures. Such actions have adverse implications for corporate innovation quality. On the one hand, firms facing unstable subsidies and reduced procurement are likely to shift resources toward low-risk projects, thereby compromising innovation quality. This not only hinders the development of a sustainable innovation ecosystem but also weakens regionally driven, innovation-based development models. On the other hand, reductions in fiscal support and increases in fines directly affect firms’ cash flows, worsening their financial conditions, intensifying financing constraints, elevating production costs, and reducing profit margins. Collectively, these pressures significantly suppress firms’ incentives to pursue innovation.

Accordingly, this study proposes the following hypothesis:

Hypothesis 1:

Local government debt expansion suppresses corporate innovation quality.

Hypothesis 2:

Local government debt might influence firm innovation through the revenue enhancement channel and the expenditure reduction channel.

3. Research Design

3.1. Data Sources

This study used annual data from prefecture-level cities and A-share listed companies in China from 2013 to 2022. Following extant studies, data on local government debt were primarily sourced from the China Local Government Bond Information Disclosure Platform and Wind database. Patent-related data were collected from the Chinese Research Data Services (CNRDS) and patent files from the China National Intellectual Property Administration. Additional firm-level data were obtained from the China Stock Market & Accounting Research (CSMAR) financial and economic research database, while regional-level data were sourced from the CEIC database and China City Statistical Yearbook. Company data were matched with city-level data based on each firm’s registered location.

The raw sample was processed according to the following principles: (1) excluding listed firms in the financial industry due to their distinct regulatory environments, unique capital structures, and differing financial reporting standards; (2) excluding samples of ST and *ST listed firms; (3) excluding samples with evident anomalies in variables; (4) excluding samples from companies with significant missing data; and (5) performing winsorization at the 1% and 99% levels for all continuous variables to mitigate the influence of outliers. After the above screening process, the final sample in this study consists of 3884 listed companies from 239 prefecture-level cities across 31 provinces in China.

3.2. Variable Measurement

3.2.1. Innovation Quality

Following Aghion et al. [18] and Akcigit et al. [19], this study measured patent quality—and consequently, firm innovation quality—by examining the breadth and complexity of knowledge embodied in patents. Specifically, we utilized the International Patent Classification (IPC) codes from patent documents provided by the National Intellectual Property Administration of China. The structure of IPC codes for invention patents follows a hierarchical format of Section–Class–Subclass–Main Group–Subgroup, as exemplified by “A03B03/00”. The first letter of the IPC code indicates one of eight main sections (A–H), the second and third digits represent the class, and the fourth letter specifies the subclass. The “/” symbol serves as a delimiter between the main group (positioned before the slash) and subgroup (positioned after the slash).

Since subgroup-level distinctions generally provide minimal informational variance, we assigned weights based solely on main group classification codes to derive a patent’s knowledge breadth. As invention patents tend to contain the highest levels of knowledge and innovativeness, we restricted our measurement of knowledge breadth exclusively to firms’ invention patents.

However, as solely relying on classification code information does not fully capture intra-patent differentiation, we adopted a weighting method inspired by the Herfindahl–Hirschman index (HHI), commonly employed in industry concentration analyses. This approach, which is based on main group classification, allowed us to more accurately reflect the knowledge diversity within patents. The measurement methodology is as follows:

$$InnoHH{I}_i=1-{\displaystyle \sum }{\lambda }^2$$

(1)

Here, λ represents the share of each main group within the IPC classification codes. Consequently, a higher InnoHHI value indicates greater differentiation among the IPC classification main groups, signifying that the knowledge utilized in the firm’s innovation process is more complex. This, in turn, reflects a higher innovation quality. Then, we aggregated knowledge breadth information to the firm level using a median-based approach across three dimensions—firm, year, and patent type. This process ultimately yielded firm innovation quality (innoqua).

3.2.2. Local Government Debt

Following Huang et al. [45], we measured the scale of local government debt using the ratio of the interest-bearing debt balance of local government financing platform companies to the gross domestic product (GDP). This interest-bearing debt balance includes the sum of short-term loans, accounts payable, notes payable, short-term bonds payable, long-term loans, bonds payable, and non-current liabilities due within one year, as reported on the financing platforms’ balance sheets. Specifically, we aggregated the interest-bearing debt balances of all local government financing platforms by prefecture-level city and year. This total was then standardized by dividing by the GDP of the respective city, yielding an annual prefecture-level measure of local government debt.

3.2.3. Control Variables

Following prior studies [32,37], this study included four control variables at the prefecture-level city scale: per capita GDP (pergdp), GDP growth rate (dgdp), total import and export volume (exim), and the scale of fixed asset investment (fixin). At the firm level, we controlled for company size (size), leverage (lev), return on assets (roa), cash flow ratio (cashflow), fixed asset ratio (fixed), revenue growth rate (growth), firm age (firmage), the proportion of independent directors (indep), the shareholding proportion of the top five shareholders (top5), and the institutional ownership ratio (inst). The specific calculation methods for these indicators are summarized in

Table 1. Variable definitions.

Variables	Definition
innoqua	Calculated using the knowledge breadth approach.
lodebt	The ratio of interest-bearing debt balance of local government financing platforms to GDP.
pergdp	The natural logarithm of local GDP per person of the cites.
dgdp	GDP growth rate of the cites.
exim	The natural logarithm of total imports and exports of the cites.
fixin	The natural logarithm of fixed asset investment of the cites.
size	The natural logarithm of total assets.
lev	The ratio of total liability divided by total asset.
roa	The ratio of net income divided by total assets.
cashflow	The ratio of the firm’s operating cash flow over its book assets
fixed	The proportion of fixed assets to total assets.
growth	(business income − business income last year)/business income last year.
firmage	The natural logarithm of one plus the firm’s established period.
indep	The ratio of independent directors on the board.
top5	The percentage of shareholdings of the largest five shareholders.
inst	The ratio of the number of shares held by institutions to the firm’s total shares.

.

3.3. Empirical Strategy

Following Huang et al. [45], we constructed the following model to examine the relationship between local government debt and firm innovation quality:

$$innoqu{a}_{it}={\alpha }_0+{\alpha }_{1}lodeb{t}_{ct}+\beta Z_{ct}+\delta X_{it}+{\gamma }_j+{\phi }_t+{\epsilon }_{it}$$

(2)

Here, the subscript i denotes firms, c represents regions, and t indicates years. The dependent variable, innoqua, measures firm innovation quality, while the independent variable, lodebt, represents the ratio of local government debt to GDP in the city where the firm operates. Control variables at the prefecture-level city and firm levels are denoted by Z and X, respectively. Industry fixed effects, ${\gamma }_j$, account for heterogeneity across industries, while year fixed effects, ${\phi }_t$, control for unobserved time-varying factors. As previously discussed, if an expansion in local government debt hinders firm innovation quality, we would expect the coefficient ${\alpha }_1$ to be significantly negative.

3.4. Summary Statistics

Table 2. Summary statistics.

VarName	Obs	Mean	SD	Median	Min	Max
innoqua	19,458	0.250	0.265	0.257	0.000	0.914
lodebt	17,363	0.324	0.417	0.209	0.011	0.629
pergdp	17,410	11.560	0.476	11.666	9.219	12.579
dgdp	17,416	6.585	3.590	7.000	−20.630	109.000
exim	19,458	11.993	8.088	16.242	0.000	19.853
fixin	19,458	5.247	7.979	0.000	0.000	18.966
size	19,458	22.131	1.253	21.922	19.570	26.452
lev	19,458	0.390	0.192	0.380	0.046	0.927
roa	19,457	0.048	0.066	0.046	−0.382	0.255
cashflow	19,455	0.051	0.065	0.049	0	0.266
fixed	19,458	0.201	0.141	0.174	0.002	0.719
growth	19,453	0.169	0.347	0.118	−0.653	3.894
firmage	19,458	2.898	0.307	2.944	1.792	3.611
indep	19,423	0.377	0.053	0.364	0.286	0.600
top5	19,426	0.544	0.149	0.545	0.188	0.892
inst	19,430	0.350	0.242	0.343	0.000	0.884

presents the descriptive statistics. The mean and median values for firm innovation quality are 0.250 and 0.257, respectively, with a standard deviation of 0.265. The maximum value is 0.914 and the minimum is 0, indicating that overall innovation quality among Chinese firms remains relatively low. The mean of local government debt is 0.324, with a maximum value of 0.629, suggesting that the debt-to-GDP ratio of local governments is generally high. The characteristics of the remaining variables are consistent with findings from the existing literature.

4. Empirical Results

4.1. Baseline Regression

Table 3. Estimation results of the effect of local government debt on innovation quality.

	(1)	(2)	(3)	(4)
	Innoqua	Innoqua	Innoqua	Innoqua
lodebt	−0.0009 **	−0.0012 ***	−0.0013 ***	−0.0013 ***
	(0.0004)	(0.0004)	(0.0004)	(0.0004)
pergdp		−0.0135 **	−0.0093	−0.0086
		(0.0056)	(0.0057)	(0.0057)
dgdp		−0.0004	−0.0003	−0.0002
		(0.0007)	(0.0007)	(0.0007)
exim		−0.0080 ***	−0.0081 ***	−0.0079 ***
		(0.0017)	(0.0016)	(0.0017)
fixin		0.0330 ***	0.0337 ***	0.0331 ***
		(0.0039)	(0.0039)	(0.0039)
size			0.0118 ***	0.0126 ***
			(0.0020)	(0.0021)
lev			0.0232 *	0.0215
			(0.0132)	(0.0133)
roa			0.0131	0.0267
			(0.0357)	(0.0363)
cashflow			−0.0503	−0.0482
			(0.0339)	(0.0341)
fixed			0.0612 ***	0.0626 ***
			(0.0170)	(0.0170)
growth			−0.0212 ***	−0.0212 ***
			(0.0058)	(0.0058)
firmage			0.0112 *	0.0098
			(0.0066)	(0.0067)
indep				−0.0222 ***
				(0.0046)
top5				−0.0329 **
				(0.0138)
inst				−0.0058
				(0.0092)
cons	0.8889 ***	1.3529 ***	0.9977 ***	1.0004 ***
	(0.0021)	(0.0535)	(0.0698)	(0.0729)
Indus FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
N	17,360	15,356	15,350	15,304
Adj. R2	0.163	0.174	0.180	0.181

Notes: *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively. The standard errors reported in the parentheses are clustered at the city level.

presents the regression results assessing the impact of local government debt on firm innovation quality. Column (1) reports the estimate with only the core explanatory variable. Column (2) introduces control variables at the prefecture-level, while Column (3) further incorporates firm-level financial controls. Column (4) includes additional governance-related variables at the firm level. Across all specifications, the coefficient for lodebt remains significantly negative at the 1% level, underscoring a robust and consistent negative relationship between local government debt and firm innovation quality. From an economic perspective, the results indicate that when the independent variable increases by one standard deviation (0.417), the dependent variable changes by 0.22% relative to its mean (calculated as 0.417 × 0.0013/0.250). While this figure may seem modest, it is economically significant. According to the estimations by the Social, Science, Culture, and Statistics Division of China’s National Bureau of Statistics in their “Research on the China Innovation Index”, China’s innovation performance improved by 0.4% in 2023 compared to the previous year. From this comparative perspective, our findings are both scientifically valid and credible.

Thus, local government debt increases are associated with reductions in firm innovation quality. Hence, Hypothesis 1 is supported. Overall, this finding is consistent with the crowding-out theory that high levels of public debt can distort local economic conditions and undermine private sector incentives. This finding is not exclusive to China, as similar correlations between high public debt and its crowding-out effects on private sector innovation have been observed in some Western countries. However, differences in political and economic systems, such as the multiparty structures in Western democracies, may moderate these effects by influencing fiscal policies and resource allocation. The socio-political system of a country significantly impacts the relationship between local government debt and innovation, as centralized systems like China’s often direct debt toward targeted innovation goals, whereas decentralized or pluralistic systems may lead to more diverse outcomes due to varying regional or political priorities.

4.2. Endogeneity Issues

Although local government debt expansion is nearly exogenous for individual firms, capturing all influencing factors is difficult. This can lead to endogeneity issues such as omitted variable bias. To address these concerns, this study follows Demirci et al. [46] and employs per capita public expenditure on healthcare and family planning (medexp) as an instrumental variable for local government debt. Higher per capita spending in these areas is likely to increase a region’s debt burden; however, this variable does not directly impact firm innovation quality.

Column (1) of

Table 4. 2SLS regression results.

	(1)	(2)	(3)	(4)
	First-Stage	Second-Stage	First-Stage	Second-Stage
medexp	0.4593 ***
	0.0557
Bartik			0.3537 ***
			0.0621
lodebt		−0.0147 **		−0.0551 ***
		(0.0072)		(0.0135)
Controls	Yes	Yes	Yes	Yes
Indus FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
F-value	68.024		32.394
Observations	13,411	13,411	13,505	13,505

Notes: ** and *** represent significance at the 5% and 1% levels, respectively. The standard errors reported in the parentheses are clustered at the city level.

reports the first-stage regression results from the two-stage least squares (2SLS) estimation. The coefficient of medexp is significantly positive at the 1% level, indicating that cities with higher per capita healthcare and family planning expenditures tend to have greater local government debt levels. The Cragg–Donald Wald F statistic of 68.024 rules out concerns regarding weak instrumentation. Column (2) presents the second-stage regression results, where the coefficient for lodebt remains significantly negative, consistent with the baseline regression results.

Additionally, following Nakamura and Steinsson [47], we construct a Bartik-style instrument by interacting initial local government debt levels with the national government debt trend over time. We then employ 2SLS estimation to assess this relationship. Columns (3) and (4) of Table 4 present the results. The Cragg–Donald Wald F statistic of 32.394 further confirms the validity of the instrument, ruling out concerns about instrument weakness. In the second-stage regression, the coefficient for lodebt remains significantly negative. These results indicate that, even after addressing potential endogeneity issues, the findings remain robust.

4.3. Robustness Analyses

1.

Alternative explanatory variable;

Given that local government debt is a stock concept, this robustness check replaces it with the change in local government debt (ΔLGD) as an alternative explanatory variable. Here, ΔLGD is defined as the difference between the current and previous years’ interest-bearing debt balance of local government financing platforms in each prefecture, divided by the current year’s GDP. Column (1) of Panel A in

Table 5. Robustness analyses results.

Panel A
	(1)	(2)	(3)	(4)
	Innoqua	Innoqua2	Innoqua	Innoqua
ΔLGD	−0.0007 **
	(0.0004)
lodebt		−0.0002 ***	−0.0024 ***	−0.0013 ***
		(0.0000)	(0.0008)	(0.0004)
L.lodebt			0.0009
			(0.0033)
teche				−0.0036
				(0.0025)
edun				−0.0074 ***
				(0.0024)
_cons	0.0806	0.9812 ***	1.0866 ***	0.9303 ***
	(0.0577)	(0.0731)	(0.0811)	(0.0821)
Controls	Yes	Yes	Yes	Yes
Indus FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
N	15,304	15,298	12,882	15,304
Adj. R2	0.145	0.082	0.084	0.081
Panel B
	(1)	(2)	(3)	(4)
	innoqua	innoqua2	innoqua	innoqua
lodebt	−0.0006 *	−0.0008 *	−0.0030 ***	−0.0074 ***
	(0.0003)	(0.0004)	(0.0007)	(0.0010)
_cons	1.2938 ***	1.1129 ***	0.9520 ***	0.7101 ***
	(0.0844)	(0.0840)	(0.0871)	(0.0931)
Controls	Yes	Yes	Yes	Yes
Indus FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
i.province#i.year	Yes	No	No	No
i.province#C.year	No	Yes	No	No
N	15,302	15,303	10,039	8752
Adj. R2	0.168	0.122	0.116	0.118

Notes: *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively. The standard errors reported in the parentheses are clustered at the city level.

reports the corresponding results, which indicate that the coefficient of ΔLGD is significantly negative at the 5% level, consistent with the baseline results.

2.

Alternative dependent variable;

Next, we recalculate knowledge breadth (innoqua2) using the mean method based on IPC classification codes. Column (2) of Panel A in Table 5 presents the results, showing that the coefficient of local government debt remains consistent in sign and significance, indicating that the results are robust to this alternative measurement approach.

3.

Alternative dependent variable;

Considering that the impact of local government debt expansion on firm innovation may involve a lag, we also re-estimate the model with the explanatory variable lagged by one period. Column (3) of Panel A in Table 5 shows that the conclusions remain unchanged, supporting the robustness of the original findings.

4.

Additional control variables;

Human capital and R&D investments at the prefecture level are major components of local government expenditure and may affect firm innovation. To control for these factors, we include the number of enrolled university students and government R&D expenditures as additional controls in the baseline model. These variables are sourced from the China Statistical Yearbook. Column (4) of Panel A in Table 5 shows no significant change in the coefficient for local government debt.

5.

Higher-level fixed effects;

In the baseline model, industry and time fixed effects are controlled. However, firm innovation may also be influenced by province-level policy shocks. To account for this, we include province-by-time fixed effects and province-specific time trends. Columns (1) and (2) of Panel B in Table 5 show results consistent with the baseline.

6.

Subsample regressions.

To address potential biases arising from the COVID-19 pandemic’s impact on local government debt and firm operations, we exclude post-2020 observations from the sample. Additionally, we remove observations for firms located in provincial capitals to control for administrative hierarchy effects. Columns (3) and (4) of Panel B in Table 5 again show findings consistent with previous findings.

Overall, these robustness checks demonstrate that the initial results are stable and reliable, reinforcing the conclusion that local government debt expansion negatively impacts firm innovation quality.

4.4. Channel Analysis

We examine two main channels through which local government debt can affect firm innovation within its jurisdiction: revenue enhancement and expenditure reduction.

4.4.1. Expenditure Reduction Channel

Information on government innovation subsidies provided to listed companies is disclosed in the “Non-operating Income” section under “Government Subsidies” in the financial statement notes of annual reports. Due to variations in reporting formats, this study utilizes Python for keyword-based text search to identify specific items classified as innovation subsidies. The identified items are then aggregated to compute the annual total of innovation subsidies for each firm, which is subsequently log-transformed for the analysis.

Data on government procurement contracts are obtained from the “China Government Procurement Network”, where details of procurement contracts are publicly available. Supplier names are matched to listed companies and their subsidiaries through both exact and fuzzy matching, with the manual verification of the fuzzy matches. This process allows us to compile the total annual procurement value for each firm, which is also log-transformed.

We then perform regressions using local government innovation subsidies and procurement spending as dependent variables, with local government debt as the independent variable.

Table 6. Expenditure reduction channel analysis results.

	(1)	(2)
	Innovation Subsidies	Procurement Spending
lodebt	−0.0485 ***	−0.0328 **
	(0.0048)	(0.0133)
Controls	Yes	Yes
Indus FE	Yes	Yes
Year FE	Yes	Yes
N	15,304	15,301
Adj. R2	0.280	0.057

Notes: ** and *** represent significance at the 5% and 1% levels, respectively. The standard errors reported in the parentheses are clustered at the city level.

shows that the coefficients for lodebt are significantly negative. Thus, local government debt expansion is associated with reductions in both innovation subsidies and procurement expenditures directed toward firms.

4.4.2. Revenue Enhancement Channel

Next, we examine whether the expansion of local government debt translates fiscal pressure onto firms by increasing their tax burden and non-tax expenditures. The corporate tax burden is measured as the total taxes and fees paid by the firm, minus any tax refunds received, divided by operating revenue. Non-tax expenditures represent the extent to which local governments increase fiscal revenue through non-tax channels. These are calculated as the sum of a firm’s forfeiture payments and public donations, divided by operating revenue.

We conduct regressions using firms’ tax burden and non-tax expenditures as dependent variables, with local government debt as the independent variable.

Table 7. Revenue enhancement channel analysis results.

	(1)	(2)
	Tax Burden	Non-Tax Expenditures
lodebt	0.0006 ***	0.0035 **
	(0.0001)	(0.0017)
Controls	Yes	Yes
Indus FE	Yes	Yes
Year FE	Yes	Yes
N	15,304	15,191
Adj. R2	0.415	0.119

Notes: ** and *** represent significance at the 5% and 1% levels, respectively. The standard errors reported in the parentheses are clustered at the city level.

shows that the coefficients for lodebt are significantly positive, indicating that local government debt expansion is associated with an increase in both corporate tax burdens and non-tax expenditures. Thus, local government debt expansion amplifies the fiscal pressure on firms, potentially through higher taxes and additional non-tax financial demands.

4.5. Heterogeneity Analysis

4.5.1. Local Government Intervention

The degree of local government intervention refers to the extent to which local governments exercise economic control over institutions and activities within their jurisdiction. The negative impact of local government debt on firm innovation quality may vary according to the level of government intervention. According to the political tournament theory, upper-level governments, particularly in China, often establish promotion criteria focused on economic growth, motivating lower-level governments to set corresponding economic targets. When a local government’s debt level is particularly high, the pressure to meet these targets intensifies, increasing the likelihood that it will raise taxes or shift fiscal burdens onto firms, thereby negatively affecting innovation quality. Consequently, we hypothesize that the effect of local fiscal pressure on firm innovation quality will be more pronounced in regions where government intervention in the economy is stronger.

To test this hypothesis, we construct a variable, govint, measuring government intervention at the prefecture level. This is the ratio of general budget expenditures to gross regional product (GRP), where higher govint values indicate a greater degree of government intervention. Based on the annual median, we classify regions into two groups: high (where govint exceeds the annual median) and low intervention (where govint is below the annual median). We then conduct a regression analysis for each group.

Columns (1) and (2) in

Table 8. Heterogeneity analysis results.

	(1)	(2)	(3)	(4)
	Innoqua	Innoqua	Innoqua	Innoqua
lodebt	−0.0024 ***	−0.0013	−0.0024 ***	0.0021
	(0.0007)	(0.0009)	(0.0005)	(0.0013)
Controls	Yes	Yes	Yes	Yes
Indus FE	Yes	Yes	Yes	Yes
N	10,960	4342	8766	6534
Adj. R2	0.084	0.124	0.089	0.126
Observations	13,411	13,411	13,505	13,505

Notes: *** represent significance at the 1% levels, respectively. The standard errors reported in the parentheses are clustered at the city level.

present the results for the high- and low-intervention groups, respectively. The findings reveal that in regions with higher government intervention, the coefficient of local government debt on firm innovation quality is significantly negative. In contrast, in regions with lower levels of government intervention, the negative impact of local government debt on firm innovation quality is not statistically significant.

4.5.2. Fiscal Pressure

The impact of local government debt on firm innovation quality may also be influenced by the degree of fiscal pressure faced by local governments. On the one hand, according to the crowding-out effect theory, when governments finance deficits through bond issuance, it raises financing costs for the private sector. This can “crowd out” corporate investment in innovation. This increase in financing costs hampers innovation activities, ultimately reducing overall innovation quality. On the other hand, in regions with greater fiscal pressure, governments may adopt more conservative budget allocations, resulting in reduced support for corporate R&D subsidies and innovation incentives, thereby directly limiting firms’ innovation investment. In contrast, in regions with lower fiscal pressure, the negative impact of local government debt on firm innovation quality may be substantially mitigated.

To test this hypothesis, we construct a fiscal pressure variable (fispre) at the city level. This variable is measured as the difference between budgetary expenditures and revenues, divided by budgetary revenues, with higher fispre values indicating greater fiscal pressure. Based on the annual median, we classify cities into two groups: high (where fispre exceeds the annual median) and low fiscal pressure (where fispre is below the annual median). Regression analyses are then conducted separately for each group.

The results shown in Columns (3) and (4) of Table 8 for the high and low fiscal pressure groups, respectively, support the following hypothesis: when local governments face high fiscal pressure, the coefficient for local government debt on firm innovation quality is significantly negative. In contrast, in regions with lower fiscal pressure, the negative effect of local government debt on firm innovation quality is not statistically significant.

4.6. Further Analysis

According to the dynamic capability theory, innovation quality decline indicates a deficiency in a firm’s ability to adapt to external changes and achieve technological breakthroughs, which undermines the firm’s agility in dynamic competition. A decrease in innovation quality can have substantial negative implications for a firm’s sustainable development, particularly manifesting as declines in core metrics such as TFP and firm value. Prior research has highlighted the critical role of innovation quality in technological advancement. Specifically, innovation quality declines can slow technology diffusion, thereby decreasing economic efficiency. Furthermore, lower innovation quality can erode a firm’s brand value and market influence, ultimately diminishing its overall market value. To examine this hypothesis, we draw on Kim et al. [48] and employ a two-stage model to analyze the effect of declining firm innovation quality, caused by local government debt expansion, on changes in TFP and firm value in the following year.

In the first stage, we transform the baseline regression model into a difference model. This allows us to estimate the effect of changes in local government debt on changes in firm innovation quality. Then, we obtain the fitted value for the change in firm innovation quality ${\widehat{\Delta innoqua}}_{it}$.

$$\Delta innoqu{a}_{it}={\alpha }_0+{\alpha }_1\Delta lodeb{t}_{ct}+\beta \Delta Z_{ct}+\delta \Delta X_{it}+{\gamma }_j+{\phi }_t+{\epsilon }_{it}$$

(3)

In the second stage, we regress the changes in TFP $\Delta tf{p}_{it}$ or firm value $\Delta Tobinq$ on the fitted values of firm innovation quality ${\widehat{\Delta innoqua}}_{it}$ obtained from the first stage. The results are reported in

Table 9. Influence of declining firm innovation quality caused by local government debt.

	(1)	(2)
	$\mathbf{\Delta }\mathit{t}\mathit{f}{\mathit{p}}_{\mathit{i}\mathit{t}}$	$\mathbf{\Delta }\mathit{T}\mathit{o}\mathit{b}\mathit{i}\mathit{n}\mathit{q}$
${\widehat{\Delta innoqua}}_{it}$	−0.7679 ***	−3.1249 ***
	(0.1264)	(0.6989)
Controls	Yes	Yes
Indus FE	Yes	Yes
Year FE	Yes	Yes
N	8532	8555
Adj. R2	0.587	0.220

Notes: *** represent significance at the 1% levels, respectively. The standard errors reported in the parentheses are clustered at the city level.

.

$$\Delta tf{p}_{it}/\Delta Tobinq={\alpha }_0+{\alpha }_1{\widehat{\Delta innoqua}}_{it}+\beta Z_{ct}+\delta X_{it}+{\gamma }_j+{\phi }_t+{\epsilon }_{it}$$

(4)

Column (1) reports the regression of TFP changes $\Delta tf{p}_{it}$ on the fitted values ${\widehat{\Delta innoqua}}_{it}$, while Column (2) reports the regression of changes in firm value $\Delta Tobinq$ on the fitted values ${\widehat{\Delta innoqua}}_{it}$. The estimated coefficients for both regressions are significantly negative at the 1% level. Thus, the decline in firm innovation quality resulting from local government debt expansion significantly worsens both TFP and firm value in the following year, thereby posing a threat to the firm’s sustainable development.

5. Conclusions

Research suggests that excessive government debt expansion may adversely affect microeconomic entities [37,38]. However, few studies have examined the impact of local government debt on corporate innovation quality. Building upon crowding-out effects and dynamic capability theory, this study investigates the influence of local government debt expansion on corporate innovation quality. We find a significant negative impact, with this negative effect remaining robust across endogenous checks and several robustness tests.

Channel analysis further indicates that local government debt levels affect local firms’ innovation behavior through two primary channels: revenue enhancement and expenditure reduction. From an expenditure reduction perspective, local governments tend to reduce innovation subsidies and procurement amounts for firms. From a revenue enhancement perspective, local governments rely more heavily on firms’ tax revenue and non-tax revenue. Heterogeneity analysis reveals that lower government intervention and fiscal pressures can mitigate the adverse effects of local government debt on corporate innovation quality. Furthermore, economic outcome analysis indicates that the innovation quality decline caused by local government debt expansion significantly impairs corporate TFP and firm value in the following year, thereby adversely impacting sustainable development.

Additionally, from a comparative perspective, China’s centralized fiscal system amplifies the direct impact of local government debt on corporate innovation, whereas in Western countries with decentralized structures, such as the United States and Germany, this relationship is moderated by market-driven mechanisms and diverse regional priorities. Our findings highlight systemic risks associated with local government debt in the context of China, as excessive debt expansion not only hampers corporate innovation but also exacerbates financial vulnerabilities and threatens overall economic stability. By impairing innovation—a key driver of long-term growth—local government debt poses risks to both microeconomic entities and broader macroeconomic systems.

Based on these findings, we argue that local governments must adopt measures to alleviate debt pressure and mitigate the negative effects on corporate innovation. First, local governments should improve debt management by setting reasonable debt limits to prevent excessive expansion. Second, optimizing fiscal budget allocation can ensure that funds are directed toward technological innovation and corporate R&D, thereby increasing support for innovative firms. Third, local governments should reduce dependence on corporate tax and non-tax revenue, thereby alleviating firms’ tax burdens and strengthening their capacity for innovation. Finally, enhanced transparency in local government debt information is essential to ensure that the use of debt funds is open, transparent, and subject to public scrutiny. This transparency can strengthen investor and corporate confidence in local fiscal conditions, facilitating a stable environment conducive to corporate innovation.

This study has three primary limitations due to data constraints. First, the analysis focuses on publicly listed companies. Future studies can extend the scope to include non-listed firms. Given that non-listed firms typically lack access to equity markets for financing, they are likely to experience a more pronounced crowding-out effect from local government debt. Second, the current model considers only the interaction between government and corporate behavior. Future research can incorporate bank behavior into the analytical framework to examine how financial system changes, driven by the pressure of local government expansion, may influence corporate innovation activities. Third, the relationship between local government debt and innovation in China provides important insights for other countries, particularly those with federal structures like the USA, Germany, and Switzerland, where fiscal decentralization may produce varying effects on regional innovation. While China’s centralized policies often link local government debt to targeted innovation goals, federal systems might exhibit less direct influence due to diverse regional priorities. Additionally, trends such as Toyota’s open patent policy in the EV industry demonstrate how knowledge sharing can drive innovation independent of financial constraints, suggesting that collaboration and intellectual property strategies could complement or even mitigate the role of local government debt in fostering innovation.