Does Local Governments’ Innovation Competition Drive High-Quality Manufacturing Development? Empirical Evidence from China

Abstract

This study aims to reveal the influence mechanism of innovation competition on the high-quality development of the manufacturing industry in Chinese local governments. Additionally, the study provides a theoretical basis for understanding how governments’ investment in science and technology breaks through key technological bottlenecks, enhances the innovation ability of enterprises, and promotes the high-quality development of the manufacturing industry. Based on balanced panel data of 269 prefecture-level and above cities in China from 2008 to 2021, the entropy value method is used to construct a comprehensive evaluation index of manufacturing development quality, and a two-way fixed-effect panel model is employed for the empirical analysis. The findings reveal that (1) for every 1% increase in local government investment in science and technology, the manufacturing high-quality development index will increase by 0.261%, indicating that local governments’ innovation competition significantly promotes the quality of manufacturing development; (2) enterprise innovation capacity plays a mediating role between government competition and manufacturing quality improvement; (3) the combined mechanism of innovation drive and promotion tournament results in a significant spatial strategic interaction of local governments’ innovation competition and a positive spillover effect on neighboring regions. Therefore, this study suggests that local governments implement different science and technology innovation investment strategies to optimize the allocation of innovation resources according to the regional manufacturing technology level.

1. Introduction

The history of industrial development shows that improvement in manufacturing quality is the core driver of modern economic growth. As of 2024, China’s industrial added value exceeded CYN 40.5 trillion, and the manufacturing industry has ranked first worldwide for 15 consecutive years [1]. This remarkable achievement is attributed to the traditional dividends of resources, labor, and participation in the international cycle, not to mention the strong impetus to the development of the manufacturing industry by local government competition [2]. However, China’s manufacturing sector currently faces three prominent challenges: the “low-end lock-in” of the global value chain; constraints of key technologies; and escalating technological embargoes of developed countries.

In the political economy with Chinese characteristics, local government competition is an important institutional variable for explaining economic growth. During the GDP appraisal-oriented period, local governments compete for economic growth [3], leading to problems such as insufficient investment in innovation. As China enters a new stage of development, the performance appraisal system of officials has included indicators that reflect the quality of development, such as innovation indicators, and the competition among local governments has gradually shifted to “competition for innovation” [4]. Against this background, this study focuses on three core questions: (1) Can local governments’ competition for innovation promote the high-quality development of the manufacturing industry? (2) Is there a strategic interaction between local governments and S&T expenditure? (3) What are the spatial effects of such competition on manufacturing development? Answering these questions has important policy implications for the Chinese government in realizing the strategic goal of “manufacturing power.”

The marginal contribution of this study lies in the following: First, theoretically, it expands the application of the theory of “promotion tournament” in the field of innovation competition by integrating the study of the innovation competition of local governments with the high-quality development of the manufacturing industry in an analytical framework. Additionally, it attempts to clarify the logical mechanism between the innovation competition of local governments and the high-quality development of the manufacturing industry, theoretically supporting the innovation competition of local governments and promoting the high-quality development of the manufacturing industry. Second, in terms of the construction of the indicator system, carbon emissions are incorporated into the manufacturing evaluation indicator system, strengthening the consideration of the green development dimension, enriching the research on the indicator system of high-quality development of manufacturing industry at the city level, and highlighting the realistic connotation of green development in the manufacturing industry. Third, from analytical and geographical perspectives, it examines the impact of the spatial strategic interaction effect of innovation competition among local governments on the high-quality development of the manufacturing industry and provides an empirical basis for the coordination of regional innovation policy in the manufacturing industry.

2. Literature Review

2.1. Historical Evolution of China’s Local Government Competition Theory

The development of local government competition theory exhibits distinct characteristics of “East-West dialogue” (as shown in

Table 1. Historical evolution of China’s local government competition theory.

Theory Name	Theoretical Basis	Theoretical Contributions	Shortcomings
1. First-generation fiscal federalism	1. Tiebout’s model;	1. Emphasizes the information advantage and competition mechanism of local governments in the provision of public goods;	1. Lacks explanation of local governments’ internal competition.
	2. Oates’ decentralization theory;	2. Providing a classic framework for modern fiscal decentralization practice; 3. It is the most important theoretical foundation for research on local government competition.
	3. Musgrave’s decentralization framework.
2. Second-generation fiscal federalism	1. Public choice theory;	1. Explains why some developing countries have been able to develop effective local government competition.	1. Failing to study what motivates the competitive behavior of local governments;
	2. Contract theory;		2. Failing to explain a series of local government competitions after the tax-sharing system.
	3. Incomplete contract theory.
3. Competition for growth	1. China has a fiscal federalism system;	1. Explaining the miracle of China’s economic growth and making up for the neoclassical theory;	1. It is difficult to explain the fervor of local officials for GDP growth rates and China’s sustained high growth rate for more than 30 years.
	2. Political tournaments;	2. Reveals that China’s dual incentives of “politics + finance” have shaped its unique growth model.
	3. Corporatization of local governments.
4. Competition for innovation	1. Innovation-driven fiscal federalism;	1. Explains how this provides a path for developing countries to “government-supported innovation in the manufacturing sector”.	1. Long-term subsidies lead to policy dependence of enterprises;
	2. Extension of the political tournament theory;	2. How local governments are the builders of the innovation ecosystem;	2. Local governments follow the trend of duplicated construction;
	3. Regional innovation system theory.	3. How local governments’ innovation incentives promote the development of the manufacturing sector.	3. Local protectionism leads to the fragmentation of technology routes.

). Early Western studies, grounded in Tiebout’s [5] “voting with feet” theory and Oates’ [6] decentralization theorem, established the analytical framework of “traditional fiscal federalism”. This framework was later introduced to Chinese academia by Weingast and Qian [7]. However, the “traditional federalism” theory failed to adequately explain the behavioral patterns of local governments in China’s decentralization reforms since the 1990s. Qian and Weingast subsequently proposed “market-preserving federalism”. To address the lack of a political micro-foundation for local governments’ institutional supply in the “second-generation federalism” theory, scholars such as Zhang and Zhou, building upon the “promotion tournament” theory and based on profound observations and analyses of China’s unique development model since the reform and opening up, proposed the economic theory of “competition for growth”. This theory attempts to explain the motivation behind local government competition during China’s economic growth process. As the negative effects of local government competition in China—including duplicate construction, environmental pollution, and lagging innovation—continued to accumulate, scholars such as Fu and Zhang put forward the concept of “competition for innovation”. This shift marked the beginning of academic attention toward the crucial role of local government innovation in promoting high-quality economic development.

2.2. Influencing Factors of High-Quality Development in the Manufacturing Industry

The high-quality development of the manufacturing industry is the result of a combination of factors. Technological innovation is the main driver for enterprises to improve product quality [8], with studies demonstrating its crucial role in enhancing production efficiency and promoting value chain upgrading [9]. Other relevant studies have emphasized that skill-biased technological progress and labor force adaptability significantly influence manufacturing upgrades [10].

Meanwhile, the impact of the policy environment is crucial. Babool and Reed [11] found that as environmental regulatory policies improved in the member countries of the Organization for Economic Co-operation and Development (OECD), environmental requirements increased production costs and reduced profits for most manufacturing enterprises. Zheng et al. [12] demonstrated that high-tech enterprise certification policies had limited positive effects on regional innovation performance, whereas corporate R&D investment exerted significant positive impacts. Based on the analyses of the pros and cons of various innovation incentive policy tools, Bloom et al. [13] revealed that R&D tax credits and direct public funding support yielded the fastest results in the short term, but policies that enhance human capital supply exhibited more substantial long-term effects.

The digital transformation of the manufacturing industry has received widespread attention in recent years. Dou and Gao [14] argued that with the emergence of a new generation of information technology, the digital economy has gradually become an inescapable reality for manufacturing companies involved in green innovation activities. Xie and Li [15] realized that digital infrastructure can significantly facilitate the governance of business model innovations in small and medium-sized manufacturing firms through linear regression analyses of 237 cases in Guangdong Province. Wang [16] showed that AI significantly increased the total factor productivity (TFP) of Chinese manufacturing firms, confirming the underlying mechanism through three important channels: technological innovation, human capital optimization, and market matching enhancement. However, the deeper application of digital technologies is constrained by the institutional environment, especially in developing countries [17].

2.3. The Impact of Local Governments’ Innovation Competition on the High-Quality Development of the Manufacturing Industry in China

Early studies predominantly focused on the inhibitory effects of GDP-oriented local government competition on manufacturing innovation. Song and Zhao [18] found that, under Chinese-style fiscal decentralization, local governments’ GDP growth-driven decision-making and behaviors exhibit suppressive effects on both technological innovation and its transformation efficiency. Similarly, Yang et al. [19] revealed that fiscal decentralization causes misalignment in local governments’ investment preferences, diminishing support and guidance for enterprise innovation, and ultimately constraining the enhancement of corporate innovation capabilities. However, Zhao et al. [20] investigated the impact of fiscal expenditure competition on regional innovation, confirming that competition in local fiscal science and technology expenditure significantly contributes to elevating regional innovation levels.

With the implementation of China’s innovation-driven strategy, the paradigm of local government competition has undergone a systemic transformation, characterized in this phase by the diversification of policy instruments, the expansion of competition dimensions, and the reconfiguration of spatial patterns. A study by Liu [21] revealed that Chinese government funding significantly enhances innovation efficiency, recommending direct subsidies as the priority policy tool complemented by tax incentives and financial policies to amplify policy effectiveness. Li and Wang [22] highlighted that high-tech zones, serving as hubs for China’s regional integrated economic development, not only mirror industrialized nations in fostering collaborative innovation clusters but also fulfill unique missions in attracting foreign investment and catalyzing “growth pole” effects for local industries. However, the negative effects of local governments’ innovation competition should not be overlooked. Xu et al. [23] further identified that intergovernmental competition not only significantly constrains low-carbon economic transition but also diminishes the catalytic effect of green technological innovation, thereby compromising the quality of green development in the manufacturing industry. Through combing the existing literature, it was found that with the dynamic evolution of China’s economic development strategy, local government competition has shifted from “innovation inhibition” to “innovation drive”, but there is little research on the impact of local governments’ innovation competition on the high-quality development of the manufacturing industry within the same analytical framework. Additionally, few studies have investigated the combination of local governments’ innovation competition and the high-quality development of the manufacturing industry using the same analytical framework to study the influence of the former on the latter. Based on this, to systematically explain the mechanism of the influence of local governments’ innovation competition on high-quality development in the manufacturing industry, in this study, we adopted the panel data of 269 prefectural-level and above cities in China from 2008 to 2021 to measure the quality level of manufacturing development in each city and constructed an empirical model to investigate the effect of local government’ innovation competition on high-quality development in the manufacturing industry.

3. Theoretical Analysis and Research Assumptions

With central government decentralization and concessions since the reform and opening up, local governments have become subjects of interest. Under the incentives provided through promotion tournaments centered on GDP, local governments in China have formed an expenditure structure that emphasizes growth and enhancement of people’s livelihoods, leading to recurrent infrastructure development, the distortion of fiscal expenditure structure [24], and environmental pollution. Nevertheless, this has had a crowding-out effect on investment in scientific and technological innovations [25], with biased financial expenditures restraining basic research and technological innovation.

The implementation of an innovation-driven development strategy has enabled the incorporation of technological innovation into the performance appraisal system [26], and local governments are competing for technological innovation, a complex, systematic project that requires the coordinated support of factors such as “human resources and materials”. Local governments’ financial investment in science and technology directly alleviates the funding gap of the jurisdiction’s innovation system [27]. Through the leverage effect of funds, it raises the R&D subsidies of the jurisdiction’s scientific research colleges and universities and manufacturing enterprises, driving enterprises to increase R&D investment, supporting technological innovation, and enhancing TFP [28] and the value chain upgrading of the manufacturing industry (as shown in Figure 1).

Meanwhile, for basic R&D activities and technological innovation with a long R&D cycle and high uncertainty, local governments effectively dissipate and reduce R&D risks through the establishment of an innovation fault-tolerance mechanism and a risk compensation fund to enhance the enthusiasm for technological innovation in scientific research colleges and universities and enterprises, overcome technological constraints restricting the development of the local manufacturing industry, and promote the high-quality development of the manufacturing industry. Howell [29] showed that the government often subsidizes new firms to promote innovation. Early government subsidies roughly double the probability of a firm receiving subsequent venture capital and have a significant positive impact on patent applications and revenue. However, programs for introducing high-level talent funded by local governments send strong signals to the market that they value talent and innovation [30], especially for scientists and engineers who have made outstanding contributions in key technology areas. This encourages them to join the wave of technological innovations in the manufacturing industry by providing generous opportunities and flexible research management strategies, thereby enhancing R&D capabilities in key technology areas.

Therefore, this study proposes the following hypothesis:

H1: 

Chinese local governments’ innovation competition has a promoting effect on the high-quality development of the manufacturing industry.

As the core leaders of regional innovation ecosystems, local governments in China have formed policy combinations through multidimensional governance tools and institutional innovations under innovation competition to systematically cultivate regional innovation capacity and drive the manufacturing value chain toward the two ends of the “smile curve”. Specifically, the enhancement of regional innovation capacity promotes the spatial agglomeration of innovation factors and reshapes the industrial ecosystem, and the synergistic network effect of innovation factors accelerates knowledge spillover and technology diffusion. Knowledge spillover is considered the main driver of the increase in the innovation rate of technology clusters [31], which improves the production process or technological flow of the manufacturing industry and increases the efficiency and quality of the manufacturing industry. According to Audretsch and Feldman [32], increased government investment in R&D can generate interfirm knowledge flows, and this spillover effect is exponentially amplified with the support of information technology. Meanwhile, enhancing regional innovation capacity also promotes the self-organized evolution of the innovation ecosystem, which in turn enhances the TFP of the manufacturing industry.

Consequently, the following hypothesis is proposed:

H2: 

Local governments’ innovation competition in China promotes an improvement in manufacturing quality by enhancing regional innovation capacity.

Technological innovation has characteristics similar to those of public goods, and the development of digital information technology has made the spatial spillover characteristics of technological innovation more significant. Local government officials, considered individuals who take into account rational economic strategies before making decisions on innovation investments, and based on their own financial income status and resource allocation, decide whether to expand innovation investments to overcome the constraints on the high-quality development of the manufacturing industry, technology, and talent short board. In fact, local governments’ financial investment decisions in science and technology may not be based purely on their own conditions, but with reference to, and imitation of, neighboring local governments that have already made such decisions. This implies that, under the influence of competition in innovation, when a local government increases its support for innovation of enterprises in its jurisdiction, technological innovations, product upgrades, iterations, and new business models of the jurisdictional enterprises are rapidly imitated, with a higher innovation level, by enterprises in its geographically adjacent regions. Such innovation by local enterprises in the jurisdiction is quickly imitated by enterprises in geographically neighboring jurisdictions, which promotes the application of valuable technologies and experiences at a lower cost, changes the production mode of the local manufacturing industry, and improves production efficiency. In addition, due to the relative performance “yardstick” competition in the field of science and technology innovation, local governments are sufficiently incentivized to engage in spatial strategic interactions in the competition for investment in innovation, increase investment in innovation, formulate talent introduction programs, and stimulate “talent-technology-systems”. The talent-technology-system coupling effect improves the efficiency of factor allocation, reduces the transaction cost of innovation, breaks through the traditional factors of production with diminishing return constraints, and contributes to the high-quality development of the manufacturing industry.

Therefore, the following hypothesis is proposed:

H3: 

There is a spatial strategic interaction in the innovation-oriented competitive behavior of Chinese local governments.

4. Research Design

4.1. Model Setting

To examine the direct impact of government innovation competition on the quality of manufacturing development, this study constructs the following econometric model:

$${HQDM}_{it}={\beta }_0+{\beta }_1{IC}_{it}+\delta X+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(1)

where ${HQDM}_{it}$ is the quality level of manufacturing development in city i in year t; ${IC}_{it}$ is the level of government innovation competition in city i in year t; X denotes a series of control variables reflecting regional characteristics; ${\mu }_i$ denotes the region’s fixed effect that does not vary with time; ${\gamma }_t$ denotes the time’s fixed effect that does not vary with individuals; and ${\epsilon }_{it}$ is the random disturbance term.

4.2. Variable Selection and Explanation

• Explained variable: the quality index of manufacturing development (HQDM): Based on the connotation and requirements of the quality of manufacturing development, this study draws on the practice of Pan et al. [33] by considering the availability of manufacturing data at the city level and the three major aspects of measuring the quality of manufacturing development in the city: industrial efficiency, technological innovation, and green development. They covered 12 tertiary indices, including the enterprise profit rate, the proportion of employment in the manufacturing industry, labor productivity, and the intensity of research and development investment, to construct an evaluation index system of manufacturing development quality in prefecture-level cities (as shown in

  Table 2. Evaluation index system of manufacturing development quality.

  Primary Indicators	Secondary Indicators	Tertiary Indicators	Indicator Attributes
  High-Quality Development of Manufacturing (HQDM)	Industrial efficiency	Total profit of industrial enterprises above designated size/business income of industrial enterprises above designated size	Positive
  		Industrial output value/industrial employees	Positive
  		Industrial output value	Positive
  		Percentage of industrial employees	Positive
  	Technological Innovation	Internal expenditure on R&D of industrial enterprises above scale/operating income of industrial enterprises above scale	Positive
  		Number of patent applications	Positive
  		Artificial intelligence enterprises	Positive
  		Penetration rate of industrial robots	Positive
  	Green development	Industrial sulfur dioxide emissions/total industrial output value	Negative
  		Industrial carbon dioxide emissions/total industrial output value	Negative
  		Industrial wastewater emissions/total industrial output value	Negative
  		Comprehensive utilization rate of industrial solid waste	Positive

  ). In addition, they adopted the entropy value method of objective assignment to determine the weight of each three-level index and measure the comprehensive score while considering the comprehensive score as the index for measuring the quality of urban manufacturing development.
• Core explanatory variable: local government’ innovation competition (IC): Most studies have used the GDP growth rate ranking or the construction of a composite index as a proxy variable for the degree of government competition; however, these indicators focus on measuring the degree of local governments’ efforts to “compete for growth” and are less concerned with local governments’ competition for innovation. To assess the impact of local governments’ innovation competition on enterprises’ innovation activities, this study employs the proportion of local governments’ science and technology expenditures to their fiscal expenditures in a benchmark regression to measure the degree of government innovation competition.
• Mediating variable: innovation ability (IA): Based on previous theoretical analyses, this study selects innovation ability, expressed by the number of effective invention patents authorized in the region, as the mediating variable.
• Control variables: To comprehensively analyze the role of local governments’ innovation competition in improving the quality of manufacturing development at the city level, drawing on existing research, this study considers the impact of economic development, population, consumption, education, and infrastructure on the quality level of manufacturing development and sets the control variables as follows: (1) the level of economic development (GDP), using the natural logarithmic value of GDP as the measure; (2) population density (POP), measured by the logarithm of the ratio of the total population to the area of the administrative region at the end of the year; (3) consumption level (CONSMP), measured by the ratio of total retail sales of consumer goods to GDP; (4) education investment level (EDU), measured by the proportion of education expenditure to the local financial expenditure in the general budget; and (5) infrastructure level (INFRA), expressed as the natural logarithmic value of total road mileage per 10,000 people.

4.3. Sample Selection and Data Sources

The study’s sample period was 2008–2021, considering the availability of relevant data at the city level, excluding cities with partially missing data, and identifying 269 sample cities, ultimately resulting in a total of 3766 observations with 14 years of balanced panel data. The data on AI enterprises were obtained from network data, and the data on other variables were derived from the China City Statistical Yearbook and the statistical annual reports of some prefecture-level cities. Missing data were filled in using the linear interpolation method. According to the statistical description of the relevant variables (as shown in

Table 3. Descriptive statistics of variables.

Variable	Number of Observations	Mean Value	Standard Deviation	Minimum Value	Maximum Value
HQDM	3766	0.036	0.038	0.004	0.496
HQDM_C	3766	0.040	0.047	0.005	0.577
IC	3766	0.016	0.017	0.001	0.207
GDP	3766	7.282	0.987	4.328	10.674
POP	3766	5.760	0.914	1.609	7.882
CONSMP	3766	0.376	0.106	0.026	0.996
EDU	3766	0.179	0.041	0.036	0.377
INFRA	3766	3.500	0.532	0.140	5.202

), the mean value of the quality index of manufacturing development (HQDM) was 0.036, standard deviation was 0.038, the minimum value was 0.004, and the maximum value was 0.496, suggesting that the quality of manufacturing development among the different sample cities varied considerably over the sample period.

5. Analysis of Empirical Results

5.1. Analysis of Benchmark Regression Results

In this study, the ordinary least squares method is used to estimate Model (1). The estimation results of the impact of local governments’ innovation competition on the improvement in the quality of manufacturing development are reported in

Table 4. Benchmark regression results.

	(1)	(2)	(3)	(4)	(5)	(6)
IC	1.458 ***	0.739 ***	0.685 ***	0.329 ***	0.422 ***	0.261 ***
	(0.029)	(0.029)	(0.159)	(0.102)	(0.124)	(0.091)
GDP		0.019 ***		0.017 ***		0.009 **
		(0.001)		(0.002)		(0.004)
POP		−0.005 ***		0.076		0.07
		(0.001)		(0.046)		(0.044)
CONSMP		−0.028 ***		−0.011		−0.009
		(0.004)		(0.007)		(0.009)
EDU		−0.049 ***		0.061 **		0.091 ***
		(0.01)		(0.024)		(0.027)
INFRA		−0.015 ***		−0.032 **		−0.044 ***
		(0.001)		(0.013)		(0.014)
_cons	0.012 ***	−0.017 **	0.025 ***	−0.424 *	0.029 ***	−0.293
	(0.001)	(0.008)	(0.003)	(0.231)	(0.002)	(0.225)
Individual fixed effect	No	No	Yes	Yes	Yes	Yes
Year fixed effects	No	No	No	No	Yes	Yes
N	3766	3766	3766	3766	3766	3766
R2	0.402	0.602	0.799	0.842	0.845	0.865
F-value	2529.323	948.873	18.688	26.993	11.551	8.739

Note: The regression coefficients are outside parentheses, and the robust standard errors of clusters aggregated at the province level are inside parentheses; *, **, and *** indicate statistical significance at the 0.1, 0.05, and 0.01, respectively. The same notes apply to the tables below.

, Columns (1)–(6), where individual and year fixed effects are not controlled for in Columns (1) and (2). Both individual fixed effects and year fixed effects are controlled for in Columns (5) and (6). Columns (1), (3), and (5) present the regression results without adding control variables. The estimated coefficients of the local governments’ innovation competition variable are significantly positive, indicating that local governments’ innovation competition significantly promotes the quality of manufacturing development. On this basis, Column (6) includes control variables and controls for individual and year fixed effects, in which the sign of the regression coefficient of the local governments’ innovation competition variable is still positive and passes the 1% level of significance, showing that a 1% increase in IC leads to a 0.26% improvement in HQDM. This indicates that local governments’ innovation competition has a significant positive impact on the improvement in the quality of manufacturing development during the examination period. Moreover, local governments’ innovation competition behavior can promote the manufacturing industry to achieve technological innovation, improve production efficiency, and promote the quality of manufacturing development. In Column (6) with control variables, the regression coefficients of the economic development level and the education input level are significant, indicating that the growth of GDP and education input improves the quality of manufacturing development. The regression coefficients of population density, consumption level, and other variables are not significantly positive or significantly negative, indicating that these factors do not have a significant positive effect on improving the quality of development in the manufacturing industry.

5.2. Robustness Testing

First, we substitute the measurement method. The TOPSIS entropy weight method is utilized to remeasure the quality level of manufacturing development (HQDM_C), and the newly derived composite index replaces the original value for regression. Second, we substitute the core explanatory variables. As there may be a potential bidirectional causal relationship between local governments’ innovation competition and the quality of manufacturing development, and the empowering effect of local governments’ innovation competition on the quality of manufacturing development has a time lag, the original variables are regressed again with the variable of local governments’ innovation competition that has a time lag of one period instead of the original variables. Third, we substitute the standard error clustering level. In the benchmark regression, standard errors are clustered at the city level according to general practice. However, this clustering ignores the strong economic linkages (e.g., talent introduction policies, science and technology expenditure inputs, etc.) among the innovation behaviors of municipal governments in the same province; therefore, this study re-clusters the standard errors at the provincial level and reports the estimation results as a robustness test in Column (3) of

Table 5. Robustness test.

	(1)	(2)	(3)	(4)	(5)
	Replace the Metric:HQDM_C	Replacement of Core Explanatory Variables:L.IC	Clustering to Provinces	Winsorization	Excluding Municipalities
IC	0.287 ***	0.245 **	0.261 ***	0.308 ***	0.313 ***
	−0.106	(0.108)	−0.084	−0.08	−0.091
_cons	−0.278	−0.300	−0.293	0.019	−0.299
	−0.257	(0.243)	−0.232	−0.09	−0.225
Control variable	Yes	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes
N	3766	3497	3766	3766	3710
R2	0.851	0.871	0.865	0.919	0.865
F	6.625	8.780	5.523	11.113	9.261

Note: The regression coefficients are outside parentheses, and the robust standard errors of clusters aggregated at the province level are inside parentheses; **, and *** indicate statistical significance at the 0.05, and 0.01, respectively. The same notes apply to the tables below.

. Fourth, the sample size is changed. The baseline model is re-estimated after a 1% bilateral shrinkage of all variables, and the four municipalities are excluded from the re-estimation of the baseline model, while considering the special characteristics of municipalities-economy, population, and location. Columns (1) to (5) in Table 5 show that the regression coefficients of the variable of local governments’ innovation competition have only changed in magnitude compared to the previous one, and the positivity and negativity does not change or pass the significance level, proving the conclusion that the promoting effect of innovation competition among local governments on improving the development quality of the manufacturing industry is robust.

5.3. Endogeneity Issues

Although the benchmark regression results pass a series of robustness tests, their accuracy may be challenged by endogeneity. First, we consider the problem of omitted variables. Despite the inclusion of firm and year fixed effects, third-party factors that simultaneously affect local governments’ innovation competition and the quality of manufacturing development may still be omitted. Second, there exists a reverse causation problem. Although the aforementioned robustness test adopts a one-period lag treatment for the innovation competition variable, it is not yet possible to completely overcome this problem. Therefore, this study addresses the omitted variable problem by introducing cross-fixed effects and adopting the instrumental variable approach to mitigate the potential endogeneity problem.

• Cross-fixed effects: In addition to the aforementioned control for firm and year fixed effects, province-year fixed effects are further introduced to control for possible omitted macro policy shocks. The results in Columns (1) and (2) of

  Table 6. Endogeneity test results.

  	(1)	(2)	(3)	(4)
  	Province-Year Fixed Effects		Instrumental Variable: One Period Lag
  IC	0.473 ***	0.389 ***	0.937 ***	0.598 ***
  	(0.146)	(0.109)	(0.26)	(0.227)
  Control variable	No	Yes	No	Yes
  Individual fixed effect	Yes	Yes	Yes	Yes
  Year fixed effects	Yes	Yes	Yes	Yes
  N	3696	3696	3497	3497
  R2	0.865	0.885	−0.032	0.142
  F value	10.476	5.038	12.975	9.243
  Kleibergen-Paap rk LM statistic			29.445	31.215
  p value			0.000	0.000
  Cragg-Donald Wald F statistic			902.535	632.051

  Note: The regression coefficients are outside parentheses, and the robust standard errors of clusters aggregated at the province level are inside parentheses; *** indicate statistical significance at the 0.01, respectively. The same notes apply to the tables below.

  show that the coefficient of innovation competition is positively significant at the 1% level, supporting previous core findings.
• Instrumental variables approach: On the one hand, science and technology expenditures, as an important aspect of local governments to promote the quality of development, can positively enhance the quality of manufacturing development. On the other hand, local innovation competition with a one-period lag is used as an instrumental variable of local governments’ innovation competition in the current period to perform a regression using the 2SLS. Columns (3) and (4) of Table 6 present the results. From the regression results, compared with the results of the benchmark regression in Column (6) in Table 4, the results in Columns (3) and (4) indicate that the coefficients of local governments’ innovation competition are both improved, and the sign of the coefficients does not change; thus, they are both still significantly positive at the 1% level. In addition, the p-value of the Kleibergen-Paap rk LM statistic in Columns (3) and (4) is 0.000, indicating the rejection of the original hypothesis of “under-identification of instrumental variables”; that is, there is no under-identification; Cragg-Donald Wald F-statistics are significant at 902.535 and 632.051, respectively. The coefficient of local governments’ innovation competition, still significantly positive at the 1% level, increases, but the sign of the coefficient does not change. The Cragg-Donald Wald F-statistics are 902.535 and 632.051, respectively, which are significantly larger than the critical value of the Stock-Yogo weak identification test; that is, there is no weak instrumental variable. This shows that local government innovation competition contributes significantly to the quality of manufacturing development after considering the endogeneity problem.

5.4. Mechanism Analysis

The previous analysis shows that local governments’ innovation competition promotes the quality of manufacturing development by improving innovation capacity. Therefore, the intrinsic influence mechanism of the driving role of innovation capacity is verified by constructing a mediation effect model.

$${IA}_{it}={\alpha }_0+{\alpha }_1{IC}_{it}+\delta X+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(2)

$${HQDM}_{it}={\theta }_0+{\theta }_1{IC}_{it}+{\theta }_2{IA}_{it}+\delta X+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(3)

Among them, the innovation ability ${IA}_{it}$ is the mediating variable, which is mainly measured by the number of patent applications and the innovation index. The other variables are consistent with those in Model (1). Considering the mediation effect test of MacKinnon (2002) [34], the specific steps are as follows: first, test whether the regression coefficient ${\beta }_1$ of the local government innovation competition (IC) index in Model (1) passes the significance test, and if it is significant, then continue the test; otherwise, it indicates that the local governments’ innovation competition does not have a significant effect on the quality of manufacturing development. Second, when ${\beta }_1$, ${\alpha }_1$, and ${\theta }_2$ in Models (2) and (3) are all significant, then there is an indirect effect. At this point, when ${\theta }_1$ is not significant, it indicates that there is a full mediation effect; when ${\theta }_1$ is significant, it indicates that there is a partial mediation effect, and the coefficient ${\alpha }_1\times {\theta }_2$ represents the mediation effect, reflecting the extent to which innovation capability, as a mediating variable, affects the improvement in the quality of manufacturing development.

Table 7. Results of the mediation effect test.

	HQDM	IA: Patents for Inventions	Score	IA: Innovation Index	HQDM
IC	0.261 ***	1.875 **	0.225 ***	7.432 ***	0.198 **
	(0.091)	(0.911)	(0.084)	(1.838)	(0.086)
IA: Patents for inventions			0.019 ***
			(0.006)
IA: Innovation index					0.009 ***
					(0.002)
_cons	−0.293	1.291	−0.318	0.172	−0.294
	(0.225)	(1.737)	(0.207)	(3.425)	(0.224)
control variable	Yes	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes
N	3766	3766	3766	3766	3766
R2	0.865	0.956	0.871	0.955	0.869
F	8.739	1.3	8.368	13.428	8.475

Note: The regression coefficients are outside parentheses, and the robust standard errors of clusters aggregated at the province level are inside parentheses; ** and *** indicate statistical significance at the 0.05, and 0.01, respectively. The same notes apply to the tables below.

reports the regression results of the mediation effect model. The results show that innovation capacity is a mediating variable when assessing the impact of local governments’ innovation competition on the quality of manufacturing development. Regardless of the measure, the regression coefficient of local governments’ innovation competition on the quality of manufacturing development is significantly positive at least at the 5% level, indicating that innovation capacity plays a mediating role in the impact of local governments’ innovation competition on the quality of manufacturing development.

5.5. Heterogeneity Test

Given China’s unique characteristics as a large country economy, regional differences may exist in the extent of the effects of local governments’ innovation competition, highlighting that regional heterogeneity should be analyzed in the impact of local governments’ innovation competition on the quality of manufacturing development. To examine the city-level and regional heterogeneity of the impact of local governments’ innovation competition on the quality of manufacturing development under different factor allocation structures and factor input levels, this study divides the 269 sample cities into provincial and non-provincial capitals according to the administrative level; then, it divides the sample cities into coastal and inland cities according to their location and then according to the National Resource-Based Cities Sustainable Development Plan (2013–2020). In addition, the sample cities are divided into resource-based and non-resource-based cities according to the National Sustainable Development Plan for Resource-based Cities (2013–2020).

Columns (1)–(4) in

Table 8. Heterogeneity test results.

	(1)	(2)	(3)	(4)	(5)	(6)
	Provincial Capital Cities	Non-Provincial Capital Cities	Coastal Cities	Non-Coastal Cities	Resource-Based Cities	Non-Resource-Based Cities
IC	0.348	0.178 **	0.279	0.241 ***	0.076 **	0.272 **
	(0.482)	(0.069)	(0.298)	(0.091)	(0.036)	(0.129)
_cons	0.111	−0.32	−0.123	−0.033	−0.058	−0.26
	(0.629)	(0.272)	(0.433)	(0.092)	(0.064)	(0.306)
control variable	Yes	Yes	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
N	364	3402	700	3066	1442	2324
R2	0.866	0.882	0.876	0.863	0.812	0.869
F	3.782	6.398	6.283	8.251	4.906	6.476

Note: The regression coefficients are outside parentheses, and the robust standard errors of clusters aggregated at the province level are inside parentheses; ** and *** indicate statistical significance at the 0.05 and 0.01, respectively. The same notes apply to the tables below.

present the estimation results of the tests for differences in city administrative levels and city location. The results show that the impact of governments’ innovation competition on the quality of manufacturing development is more significant in non-capital cities than in provincial capitals, probably because the seat of China’s provincial government, that is, the capital city, is the province’s political, economic, and cultural center and is one of the most important “growth poles” of the province’s economy. In manufacturing development, owing to the special position of provincial capital cities in the province and their inherent advantages in manufacturing, they are prioritized in terms of support from the central and provincial governments in a series of policies on industry, finance, science, and technology. In recent years, in particular, under the impetus of the strategy of “strengthening provincial capitals”, by increasing infrastructure construction and promoting technological innovation, manufacturing development has reached a certain level, even in its late stage. The development of the manufacturing industry in provincial capitals is not as affected by innovation investment in the late stage as the development of the manufacturing industry in non-capital cities.

The regression coefficient of innovation competition in non-coastal cities is significantly positive at the 1% level, indicating that the positive impact of local governments’ innovation competition on the quality of manufacturing development is stronger in non-coastal cities than in coastal cities. The reason may be that since the reform and opening up, driven by China’s long-standing “export-oriented” strategy, coastal local governments have seized the opportunities for development and perfected infrastructure development, which has relied on the manufacturing industry earlier than inland cities, and the industrial efficiency, technological innovation, and green development have reached a certain level, which makes the influence of the local government’s investment in science and technology relatively insignificant. As non-coastal cities are constrained by geographic location, transportation costs, and infrastructure, the level of development of the manufacturing industry lags behind, and the accumulation of technology is relatively poor.

Investments that promote technological innovation and local government innovation can alleviate the financial pressure of innovation and R&D, reduce the risk of innovation, and promote the development of the manufacturing industry, an effect that is relatively obvious. The influence of local governments’ innovation competition on the quality of manufacturing development is significant and insignificant in non-resource cities and resource cities, respectively. The possible reason is that resource cities often consider the development and utilization of local resources as the pillar industry, and by extending the industrial chain related to the resources, they absorb employment and promote the development of the local economy. Such a locally adapted industrial development strategy reduces the local government’s investment in science and technology innovation in manufacturing. Therefore, the effect of local governments’ investment in innovation in resource-based cities is relatively insignificant.

6. Further Discussion

To further examine the spatial spillover effect of local governments’ innovation competition on the improvement in the quality of manufacturing development, a spatial panel model is selected for regression. Commonly used spatial panel models include the spatial error model (SEM), the spatial lag model (SAR), and the spatial Durbin model (SDM). In the empirical selection process, not every model is suitable for adoption, and the applicability of the model needs to be tested. First, the Wald test or likelihood ratio (LR) test is used to determine whether the SDM model can be degraded to the SAR and SEM models; second, the Hausman test is used to select fixed versus random effects; and finally, the LR test is used to determine whether a single fixed effect versus a two-way fixed effect should be selected. In summary, the final test results indicate that time- and individual double-fixed SDM should be selected for regression.

Before the estimation, we first examine whether there is spatial autocorrelation in the data. Then, we adopt a spatial proximity matrix and apply the global Moran method to measure the Moran index value of the explanatory variables, i.e., local governments’ innovation competition and manufacturing development quality.

Table 9. Spatial correlation test.

	Year	2008	2009	2010	2011	2012	2008	2009
HQDM	Moran’s value	14.661	14.779	15.522	14.588	14.728	14.103	13.896
	p value	0.000	0.000	0.000	0.000	0.000	0.000	0.000
IC	Moran’s value	14.661	14.779	15.522	14.588	14.728	14.103	13.896
	p value	0.000	0.000	0.000	0.000	0.000	0.000	0.000
	Year	2015	2016	2017	2018	2019	2020	2021
HQDM	Moran’s value	13.667	13.787	12.478	11.981	10.618	10.602	9.078
	p value	0.000	0.000	0.000	0.000	0.000	0.000	0.000
IC	Moran’s value	13.667	13.787	12.478	11.981	10.618	10.602	9.078
	p value	0.000	0.000	0.000	0.000	0.000	0.000	0.000

shows that the Moran index values of the relevant variables during the sample period exhibit significant spatial autocorrelation, indicating that the variables exhibit obvious spatial distribution characteristics.

From the spatial effect results (as shown in

Table 10. Spatial effect test results.

	SDM		SAR
	Geographic Distance Matrix	Economic Distance Matrix	Geographic Distance Matrix	Economic Distance Matrix
$\rho$	0.692 ***	0.129 ***	0.771 ***	0.132 ***
	(0.048)	(0.013)	(0.032)	(0.012)
$\mathrm{IC}$	0.268 ***	0.294 ***	0.272 ***	0.309 ***
	(0.029)	(0.031)	(0.029)	(0.031)
$\mathrm{W}\times \mathrm{IC}$	0.884 ***	0.039
	(0.187)	(0.029)
Control variable	Yes	Yes	Yes	Yes
Individual fixed effect	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
N	3766	3766	3766	3766
R2	0.02	0.279	0.207	0.25
Log-likelihood	10,706.631	10,504.134	10,641.087	10,472.775
Comparison of Modeling Results
LR spatial lag term	131.09	63.22
	(0.000)	(0.000)
Wald spatial lag term	132.77	53.83
	(0.000)	(0.000)
LR spatial lag term	67.76	81.78
	(0.000)	(0.000)
Wald spatial lag term	63.43	82.37
	(0.000)	(0.000)

Note: The regression coefficients are outside parentheses, and the robust standard errors of clusters aggregated at the province level are inside parentheses; *** indicate statistical significance at the 0.01, respectively. The same notes apply to the tables below.

), it can be deduced that local governments’ innovation competition has a significant positive impact on the improvement in the quality of local manufacturing development. The coefficient of the spatial lag term of the dependent variable ranges from 0.129 to 0.771, which is also statistically significant, indicating that the quality of manufacturing development in neighboring cities has a significant positive impact on the quality of manufacturing development in the city. The significant positive externality of the quality of manufacturing development indicates that a city with a higher quality of manufacturing development leads to an improvement in the quality of manufacturing development in neighboring cities. In the SDM model, the coefficient of the core explanatory variable, with spatial lag term W×IC, is significantly positive when the geographic distance weight matrix is used, indicating that the level of governmental innovation in geographically neighboring cities also enhances the level of manufacturing development quality in the city to a large extent; this result indirectly verifies the influence of spatial geographic correlation on the quality of manufacturing development. The Wald and LR tests using the SDM estimation results consistently show that the original hypothesis of the SDM model, which can be simplified to an SAR model or an SEM model, can be rejected, thus focusing on the SDM estimation results. Evidently, innovation-oriented local governments generate numerous preferential policies in terms of enterprise innovation, and innovation-type competition tools in a certain region tend to promote the enhancement of the innovation capacity of local enterprises, while innovation itself carries a spillover effect. Thus, local government competition from an innovation perspective has a positive spatial spillover effect on the quality of the development of the manufacturing industry in other regions.

7. Conclusions and Policy Recommendations

7.1. Conclusions

By systematically analyzing the influencing mechanism of innovation competition among local governments on the high-quality development of the manufacturing industry in China, this study draws the following important conclusions: First, innovation competition among local governments significantly promotes technological innovation and productivity improvement in the manufacturing industry through an increase in science and technology financial expenditure and the optimization of innovation policy supply. The empirical results show that for every 1% increase in local government science and technology investment, the index of high-quality development of the manufacturing industry increases by 0.261%, verifying the validity of the theory of “promotion tournament” in the stage of innovation-driven development. Second, enterprise innovation capacity partially mediates the relationship between government competition and manufacturing upgrade, and the robustness of this mechanism is further confirmed by the instrumental variable method (using historical science and technology resources as the instrumental variable). More importantly, the spatial econometric analysis shows a significant positive spatial spillover effect of local governments’ innovation competition, and every 1% increase in science and technology investment in neighboring regions can raise the local manufacturing quality index by 0.884%, revealing the synergistic development characteristics of regional innovation systems.

7.2. Policy Recommendations

In the context of China’s implementation of innovation-driven strategies, in order to promote high-quality development in the manufacturing industry, regarding the innovation competition among local governments that is caused by the inefficiency of innovation investment, regional innovation capacity needs to be improved; factors such as industrial isomorphism, duplicated construction, etc., should be prevented; and theories of government intervention, synergistic development, and regional innovation, among others, should be used to build a systematic policy framework that synergistically guides local governments’ innovation competition and facilitates the development of the manufacturing industry.

First, resource allocation should be optimized, and core technology research should be the primary focus. Local governments should change the “egalitarian” mode of scientific research investment, set up industry-specific innovation funds, optimize resource allocation, play a leveraging role of financial funds, guide the investment of social capital, make up for the shortage in R&D funding, and promote core technology research. At the same time, local governments should establish risk tolerance strategies, reduce R&D risks, ease the pressure on R&D activities, incentivize the willingness of enterprises to invest in R&D, improve the quality of R&D, and promote technological innovation.

Second, regional innovation ecosystems should be built, enhancing regional innovation capacity. Local governments should establish the main position of manufacturing enterprises, joint colleges and universities, scientific research institutions, and other scientific research institutions, forming industry-university-research innovation alliance. Through this, they build a regional innovation ecosystem; promote the depth of docking between colleges and universities, scientific research institutions, and enterprises; rationalize the relationship between “basic research-technology research-result transformation”; and improve the regional innovation capacity.

Third, according to local conditions, different development strategies should be implemented. Non-provincial capitals and non-coastal cities should avoid the manufacturing industry isomorphism observed in provincial capitals and coastal cities and instead follow the principle of gradient transfer, choose to undertake potential industries, avoid industrial reconstruction, and consolidate the foundation of industrial development. At the same time, based on local resource allocation and according to local conditions, governments should formulate development planning, increase R&D investment, support technological innovation, and promote the development of the manufacturing industry.

Fourth, a regional synergistic development strategy should be established to promote a win-win cooperation. Based on the spatial strategic interaction observed in local governments’ innovation competition, a regional synergistic mechanism for innovation should be developed that breaks regional local protectionism and changes local governments’ innovation competition from a “zero-sum game” to a “win-win cooperation” and from ‘investment’ to “cooperation”. This results in a shift from “investment” to “quality”, avoids duplicate construction, realizes resource sharing, and guides the rational and orderly development of the manufacturing industry.