Innovation-Driven Policy and Low-Carbon Technology Innovation: Research Driven by the Impetus of National Innovative City Pilot Policy in China

Abstract

Since 2008, China has established innovative pilot cities in batches, with green and low-carbon principles and objectives as the core of the NICP policy. Therefore, it is of great significance to accurately evaluate the driving effect of the NICP policy on low-carbon technology innovation, to expand the coverage of pilot cities in a prudent and orderly manner. The research focuses on the economic and environmental potential of the national innovative city pilot (NICP) policy. However, the relationship between the NICP policy and low-carbon technology innovation remains to be examined. This article employs a sample of 274 prefecture-level cities in China spanning the years 2003 to 2020 for research purposes, and uses a series of methods such as time-varying DID and intermediary effect models to examine the impelling impact and intricate workings of the NICP policy on low-carbon technology innovation. The study found that: (i) The NICP policy possesses the potential to impel innovation in low-carbon technology, and the impact of the policy exhibits a fluctuating yet upward trajectory over time. (ii) The NICP policy promotes low-carbon technology innovation through financial technology investment, population aggregation, and digital construction. (iii) The innovation effect of the NICP policy is significantly influenced by resource endowment and the disclosure of environmental information. The impact of the NICP policy on innovation in low-carbon technology is more pronounced in resource-based cities than non-resource-based cities, and it is particularly noteworthy in well-established resource-based cities with abundant resource endowments. The impetus generated by the NICP policy towards the innovation of low-carbon technology is notably more substantial for cities that exhibit elevated levels of environmental information disclosure. Local governments should implement active environmental information disclosure at the city level. This paper not only enriches the relevant research on low-carbon technology innovation but also provides empirical evidence for promoting the NICP policy nationwide. Additionally, it serves as a policy reference for creating innovative characteristic cities under the “dual carbon” goal.

1. Introduction

The 14th Five-Year Plan explicitly highlights the urgency to expedite the advancement of eco-friendly and low-emission progress by reinforcing legal and policy assurances for environmentally sustainable development, fostering green finance, and implementing an innovation-driven development strategy with a market-oriented and green low-carbon technology focus. This statement serves to explicate the pivotal role of innovation in China’s comprehensive modernization endeavor and exalts innovation to unparalleled levels of eminence. According to the innovative-driven pilot practice, China had already established its first innovative pilot city in Shenzhen as early as 2008. Subsequently, in 2010, the policy was expanded to 41 cities (including county-level cities and districts), followed by 6 cities in 2011, 3 cities in 2012, 10 cities in 2013, 17 cities in 2018, and 25 cities in 2022, all of which were set up in batches as innovative pilot cities. As of 2022, the scope of innovative pilot cities has expanded to 103 cities. From the perspective of innovation-driven development, according to the BP Statistical Review of World Energy, 2021, China has been the world’s largest primary energy consumer for 12 consecutive years. In 2021, China’s primary energy consumption accounted for 26.1% of the world’s total, reaching a historic high. Furthermore, according to data from the International Energy Agency (IEA), China’s carbon dioxide emissions exceeded 11.9 billion tons in 2021, accounting for 33% of the global total. In the context of high-quality economic development, exploring energy conservation and emission reduction requires a focus on technology innovation to seek ways to improve low-carbon technology innovation. Therefore, in September 2020, China pledged to attain the “dual carbon” objective of reaching the peak of carbon emissions and accomplishing carbon neutrality by the year 2060, and took the lead in proposing ten major action plans for carbon peaking including the “Green Low-Carbon Technology Innovation Action”. Therefore, whether the NICP policy can drive low-carbon technology innovation and how to effectively drive low-carbon technology innovation through which mechanism is not only the core theme of building an innovative country but also the key to achieving the “dual carbon” goal.

Numerous scholars have undertaken extensive research on the impelling factors of technology innovation and the potential for the development from diverse standpoints [1,2,3,4]. In contrast to other technological advancements, the innovation of green and low-carbon technology exhibits greater levels of unpredictability and heterogeneity. Numerous erudite individuals have delved deeper into the impelling elements of green and low-carbon technology innovation, scrutinizing them through the lens of institutional theory, encompassing environmental regulations and low-carbon city pilot projects [5,6,7], natural resource theory [8,9], the stakeholder theory of government, enterprises, and public consumers [10], and market theory [11]. The research on the NICP policy mainly involves the implication of cities oriented towards innovation and the economic potential and environmental potential of the policy. Regarding the implication of cities oriented towards innovation, some scholars believe it is a creative city, emphasizing innovative ideas and culture [12], while others believe it is an innovative city, emphasizing innovative elements such as knowledge, technology, and talent [13]. Since the 1990s, foreign countries have been implementing the “innovative city” strategy, which primarily relies on urban characteristics, government support, innovative service systems, and urban culture to create innovative cities. These cities include Boston in the United States [14], Sydney in Australia [15], London in the United Kingdom [16], and 14 innovative cities in South Korea [17]. Regarding innovative service systems, South Korea’s innovative cities, for example, are based on the close integration of scientific research and industry, and they gather talent and resources based on market-oriented education, allowing scientific research achievements to be directly connected to enterprises. Additionally, Daejeon has made efforts to create an environment and launched the “100,000 Affordable Housing Construction Plan” to solve housing problems.

Regarding the economic potential and environmental potential of the NICP policy since China established Shenzhen as the first national innovation-oriented city in 2008, relevant literature has been developed around three aspects: Firstly, findings on the economic effect of the NICP policy indicate that it has a noteworthy and affirmative impact on the enhancement of Foreign Direct Investment (FDI) quality [18], industrial structure upgrading [19], green efficiency of the logistics industry [20], and green total factor productivity [21]. Secondly, findings on the environmental effect of the NICP policy show that it can significantly improve energy productivity [22], green total factor energy efficiency [23], carbon emission efficiency [24], carbon emission intensity [25], and carbon lock-in [26,27]. Thirdly, findings on the innovation effect of the NICP policy show that it is conducive to promoting the flow of knowledge between industry, academia, and research institutions [28,29], city brand [30], green innovation efficiency [31], and green technology progress [32]. In contrast to the government’s direct intervention in innovation through resource allocation, several studies suggest that the government should act as a coordinator or facilitator to promote innovation, and its policy formulation should focus on creating a favorable environment for innovation [33,34].

In summary, the literature on the pilot policy for innovative cities is based on quasi-natural experimental methods to assess its economic and environmental potential, but there is limited consideration of the literature related to green and low-carbon technology innovation. Firstly, literature related to green technology innovation is associated with the NICP policy. Green technology includes not only low-carbon technology innovations such as alternative energy production, transportation, energy conservation, carbon capture, and storage but also non-low-carbon technology innovations such as waste disposal, pollution control, and forestry technology. However, there is a scarcity of literature exploring the driving effects and intrinsic driving mechanisms of the NICP policy on low-carbon technology innovation. If these driving effects and mechanisms are not well understood, the marginal effects may be overestimated or underestimated, which could affect the government’s decision on whether to implement the NICP policy nationwide. Secondly, there is a severe lack of literature that accurately measures the level of low-carbon technology innovation by identifying low-carbon technology patents through the International Patent Classification (IPC) of the Green Patent List. Accurately measuring the level of low-carbon technology innovation is a prerequisite for evaluating the effectiveness of the NICP policy. Thirdly, there are many driving factors for low-carbon technology innovation, and there is limited literature discussing the necessary conditions for the NICP policy to have a driving effect. The lack of discussion on the necessary conditions for innovation may also lead to an underestimation of the marginal effects of the NICP policy, which is not conducive to the construction of the NICP policy system.

Therefore, the main contributions of this paper are as follows: Firstly, in terms of research content, the quasi-natural experimental method is used to evaluate the driving effect of pilot policies for innovative cities on low-carbon technology innovation, and the driving mechanism of pilot policies for innovative cities is explored in depth based on multiple factors such as government scientific research investment, human resource aggregation, and digitalization level, expanding the research on the effect and driving mechanism of pilot policies for innovative cities. Secondly, in terms of research methods, the International Patent Classification (IPC) of the Green Patent List is identified to obtain information on patent classifications related to climate mitigation, and low-carbon technology invention patents that match the characteristics of low-carbon technology innovation are selected based on the incoPat database, which helps to separate low-carbon technology innovation from non-low-carbon green technology innovation. Compared with other classification standards, the IPC code defines the technology of innovative activities more clearly and standardly, which is helpful for accurately evaluating the driving effect of urban low-carbon technology innovation behavior. Thirdly, from the perspective of urban resource endowment, environmental information disclosure level, and other multiple heterogeneities, the driving effect of the NICP policy on low-carbon technology innovation is studied, providing an empirical reference for designing pilot policies for innovative cities that are adapted to urban characteristics and low-carbon technology innovation types.

2. Analysis of Mechanisms and Research Hypotheses

2.1. The Driving Effects of the NICP Policy on Low-Carbon Technology Innovation

The evaluation criteria, principal objectives, and operational strategies of the NICP policy are deeply linked with the advancement of a sustainable, low-emission economy. To be precise, the primary criterion of an inventive urban evaluation encompasses the notion of eco-friendly and sustainable practices with a minimal carbon footprint. The “Guidelines for Building Innovative Cities” (https://www.safea.gov.cn/xxgk/xinxifenlei/fdzdgknr/fgzc/gfxwj/gfxwj2016/201612/t20161213_129574.html, accessed on 14 April 2023) proposes an indicator system for building innovative cities, which encompasses eco-friendly and sustainable benchmarks, such as the all-encompassing energy utilization per 10,000 CNY of GDP and the intensity of carbon dioxide emissions per unit of GDP. These indicators require pilot cities to make low-carbon technology innovation an important dimension of innovation. Secondly, the main task of the NICP policy is to achieve sustainable social development. The “Guiding Opinions on Further Promoting the Policy Work of Innovative Cities” (https://www.most.gov.cn/xxgk/xinxifenlei/fdzdgknr/fgzc/gfxwj/gfxwj2010before/201004/t20100419_76811.html, accessed on 14 April 2023) highlight that the primary objectives of innovative cities encompass “expediting the metamorphosis of economic development approaches and fostering harmonious and sustainable economic and social advancement”. Low-carbon technology innovation can reduce carbon emissions while improving economic benefits. Thirdly, the innovative driving mode proposed by the NICP policy is conducive to low-carbon technology innovation. Traditional carbon-based technology has led to path dependence in the economic and technological system, making it easy to fall into the innovation dilemma of carbon lock-in. A paradigm shift in technology is needed to unlock the traditional technology and economic system. The NICP policy requires the innovation driving force of cities to shift from traditional resources and capital to innovative factors such as technology and talent, which fosters the transition from conventional carbon-based technology paradigms to low-carbon technology paradigms. Drawing upon these, the ensuing hypothesis is posited:

H₁. 

The NICP policy has the potential to foster low-carbon technology innovation.

2.2. The Driving Mechanism for the NICP Policy on Low-Carbon Technology Innovation

Firstly, the NICP policy drives low-carbon technology innovation through innovative resource input. As green low-carbon technology innovation involves risks such as high capital investment, long investment cycles, and asymmetric information, this technology innovation possesses “dual externalities” of technology and the environment [35]. The motivation for enterprise low-carbon innovation investment is limited, requiring innovation resource support from the government and society. According to the theory of government intervention, government support helps to correct externalities in innovation [36]. In this regard, the “Guidelines for Building Innovative Cities” point out that the NICP policy focuses on the aggregation of innovative factors such as science and technology and talents. Therefore, this article further distinguishes innovation resources into science and technology investment and talent aggregation to identify the mechanism by which the NICP policy drives enterprise low-carbon technology innovation through science and technology investment and talent aggregation.

Secondly, the NICP policy drives low-carbon technology innovation through digital development. The “Guidelines for Building Innovative Cities” propose that the NICP policy should promote urban innovation governance through digital construction such as big data, and the Internet of Things. Digital construction can not only optimize the allocation of innovation resources through paths such as reducing transaction costs, improving management efficiency, and accelerating industrial integration but also provide a good public service platform and innovation carrier for innovation, reducing the phenomenon of information asymmetry in innovation activities [37]. This, in turn, helps to drive low-carbon technology innovation. Therefore, the following hypotheses are proposed:

H₂. 

The NICP policy drives low-carbon technology innovation through financial technology investment and talent aggregation.

H₃. 

The NICP policy drives low-carbon technology innovation through digital construction.

2.3. The Heterogeneity of the Effects of the NICP Policy on Low-Carbon Technology Innovation

During the nascent phase of the establishment of cities reliant on natural resources, there is a tendency to prioritize the development of resource-based industries due to the relatively abundant resource endowments. Subsequently, the continuous expansion of industries further promotes the gradual aggregation of factor resources towards resource-based industries, thus forming a resource-dependent development model. This model primarily hinges on the comprehensive utilization and extensive manipulation of natural resources on a grandiose scale, with a low resource utilization efficiency and significant room for improvement. In contrast, cities that are not reliant on natural resources possess multifaceted industrial frameworks and a relatively high efficiency in resource allocation such as capital and human resources. In this context, the NICP policy mainly relies on the allocation of innovative factors to promote innovation construction [38]. The higher the resource endowment of the city, and the lower the resource utilization efficiency, the more pronounced the impetus of the NICP policy on innovation.

Cities with more comprehensive environmental information disclosure and environmental regulations are conducive to strengthening corporate ecological and environmental responsibilities through social supervision, thereby promoting low-carbon technology innovation. However, when environmental information is not disclosed, it is difficult for environmental regulation to enhance low-carbon technology innovation by shaping the corporate green image [39]. The guidelines include indicators such as the all-encompassing energy utilization per 10,000 CNY of GDP and the intensity of carbon dioxide emissions per unit of GDP into the city’s assessment targets. This measure helps to strengthen government environmental regulations. The more comprehensive the environmental information disclosure, the more potent the impelling force for corporate low-carbon technology innovation under the duress of urban target evaluation and responsibility. Hence, the following hypotheses are proposed:

H₄. 

The driving effect of the NICP policy on low-carbon technology innovation is more pronounced for cities with higher resource endowments.

H₅. 

The driving effect of the NICP policy on low-carbon technology innovation is more significant for cities with higher levels of environmental information disclosure.

3. Research Design

3.1. Model Design

3.1.1. Time-Varying DID Model

In order to avoid endogeneity issues caused by quasi-natural experimental methods, this paper needs to carefully select control variables. Based on the existing literature, factors that affect low-carbon technology innovation are selected as control variables to solve endogeneity problems caused by omitted variables. Additionally, it is necessary to identify reverse causality, that is, whether there is a possibility of reverse causality between the NICP policy and low-carbon technology innovation. Considering the sample coverage of innovative pilot areas in this paper, the number of cities in the eastern, central-western, and northeastern regions are 37, 35, and 6, respectively, accounting for 47.4%, 44.9%, and 7.7% of the total number of innovative pilot cities. Therefore, the geographical distribution of pilot areas conforms to the “Guidelines for Building Innovative Cities”, which propose a coordinated layout of the eastern, central, and western regions to actively support and promote urban innovation development. Thus, the NICP policy does not exhibit bias towards high or low innovation bases in the pilot areas, avoiding the endogeneity problems caused by reverse causality and creating conditions for constructing quasi-natural experiments in this paper and using time-varying DID methods to identify the net effect of the NICP policy on low-carbon technology innovation. The baseline model is set as follows:

$$Y_{it}={\alpha }_0+{\alpha }_1{D}_{it}+{\alpha }_{2}X+{\theta }_i+{\eta }_t+{\lambda }_p\times {\gamma }_t+{\epsilon }_{it}$$

(1)

where $i$ refers to city, $p$ refers to province, $t$ refers to time. $Y_{it}$ refers to low-carbon technology innovation. $D_{it}$ refers to $Trea{t}_i\times Tim{e}_t$, $Treat$ is equal to 1 if an enterprise is included in the pilot regions of the ECRT system, and 0 otherwise. $Time$ is equal to 1 if the NICP policy is implemented, and 0 otherwise. $X$ refers to the control variables. ${\alpha }_0$ is the constant term, ${\alpha }_1$ represents the driving effect of the NICP policy on innovation, ${\alpha }_2$ refers to the estimation coefficient of controlled variables, ${\theta }_i$ refers to the city fixed effect, ${\eta }_t$ refers to the year fixed effect, ${\lambda }_p\times {\gamma }_t$ refers to the province-year fixed effect, ${\epsilon }_{it}$ is the stochastic error term.

3.1.2. Intermediary Effect Models

Based on the analysis of the impelling mechanisms, this manuscript additionally formulates models of mediating effects to scrutinize the impelling mechanisms of the NICP policy on low-carbon technology innovation. The intermediary effect models are set as follows:

$$M_{ijt}={\alpha }_0+{\beta }_1{D}_{it}+{\beta }_{2}X+{\theta }_i+{\eta }_t+{\lambda }_p\times {\gamma }_t+{\epsilon }_{it}$$

(2)

$$Y_{ijt}={\alpha }_0+{\lambda }_1{D}_{it}+{\lambda }_2{M}_{ijt}+{\lambda }_{3}X+{\theta }_i+{\eta }_t+{\lambda }_p\times {\gamma }_t+{\epsilon }_{it}$$

(3)

where $M_{it}$ refers to financial technology investment (SCI), human resource aggregation (HUM), and the digitalization of construction (INF), ${\beta }_1$ refers to the influence of the NICP policy on mediate variables, ${\lambda }_1$ refers to the direct influence of the NICP policy on low-carbon technology innovation, ${\lambda }_2$ refers to the influence of the mediate variables on low-carbon technology innovation. The residual variables bear identical connotations and notations to those in Equation (1).

3.2. Variables Explanations and Data Sources

In this article, the explanatory variable is defined as low-carbon technology innovation (Lnapply). Due to the stability, objectivity, and accessibility of patent data, it is closely related to innovation. Compared with utility model patents and design patents, invention patent data better reflects the quality of enterprise innovation than the quantity. Therefore, this article refers to the method proposed by Zhu et al. (2019) [40] to construct the urban low-carbon technology innovation index. This method mainly relies on the “Green Inventory” published by the World Intellectual Property Organization (WIPO) (https://www.wipo.int/classifications/ipc/green-inventory/home, accessed on 14 April 2023) to classify alternative energy production, transportation, energy conservation, carbon capture and storage, nuclear power generation, waste recycling, and administrative, regulatory, and design aspects related to climate mitigation into seven categories of low-carbon technology invention patents. The number of low-carbon invention patent applications at the city level published by the State Intellectual Property Office is collected according to the seven categories of codes in the Green Inventory. In order to ensure the stability of the data, the logarithm of the number of applications is taken. The core explanatory variable is the interaction term between the experimental group pilot cities and the pilot time, and the specific pilot cities and pilot time are shown in Appendix A.

In order to further ensure the robustness of the NICP policy’s driving effect, this article controls for the city characteristic variables described below which affect low-carbon technology innovation. (i) Urbanization level (URB): Urbanization promotes population concentration, which can provide human resource support for technology innovation [41]. This article uses the proportion of the non-agricultural population to the total urban population to measure the urbanization level. (ii) Financial leverage (FIN): Financial leverage can enable enterprises to obtain more financing for low-carbon technology innovation, and alleviate financing pressure faced by enterprises, but may also lead to operational risks [42]. This article employs the ratio of the aggregate amount of loans from financial institutions to the regional GDP as a gauge of financial leverage. (iii) Market consumption level (CS): Consumption level can reflect the scale of market demand, and market demand can lead to product innovation by enterprises [43]. This article uses the ratio of total retail sales of social consumer goods to regional GDP to measure the market consumption level. (iv) Public service level (PUB): On the one hand, the public service level can help to optimize government innovation services and improve patent approval efficiency. On the other hand, government public procurement that provides public services can encourage technology innovation by prioritizing the purchase and use of products and services that meet national green certification standards [44]. This article uses the proportion of employees in public management and social organizations to the total urban population to measure the public service level [45]. (v) Population density (POP): The influence of population density on technology innovation is twofold. On the one hand, areas with higher population density are more likely to produce inventors, and on the other hand, an elevated population density has the potential to diminish the per capita income. If the benefits of technology innovation are sensitive to income changes, a high population density may inhibit technology innovation [46]. This article uses the proportion of the total urban population to administrative area land area to represent population density. (vi) Level of openness to foreign investment (FDIL): Foreign direct investment may promote technology innovation in host countries through demonstration effects, spillover effects, competition effects, and talent mobility effects. This article employs the ratio of foreign capital utilization to the regional gross domestic product as a gauge of the degree of receptiveness towards foreign investment.

This article utilizes panel data from 274 Chinese cities spanning the years 2003–2020 as its research sample. The data are sourced from the annual “China Urban Statistical Yearbook”, the National Intellectual Property Office, and the CNRDS database, with price adjustments based on the 2003 baseline. Please refer to

Table 1. Descriptive statistics.

Variable	Variable Meaning	N	Mean	SD	Min	Max
Lnapply	Logarithm of low-carbon technology invention patent applications	4932	3.348	1.897	0	10.138
Lngrant	Logarithm of low-carbon technology authorized patent applications	4932	2.446	1.775	0	8.805
URB	Proportion of non agricultural population to the total urban population	4932	0.389	0.218	0.110	1.144
FIN	Proportion of loans from financial institutions to the regional GDP	4932	0.896	0.520	0.275	3.064
CS	Proportion of retail sales of social consumer goods to the regional GDP	4932	0.364	0.100	0.126	0.656
PUB	Proportion of public management practitioners to the total urban population	4932	0.012	0.004	0.005	0.027
POP	Proportion of the total urban population to the urban land area	4932	0.046	0.043	0.002	0.276
FDIL	Proportion of FDI to the regional GDP	4932	0.021	0.023	0.001	0.121

for variable explanations and descriptive statistics.

4. Empirical Analysis

4.1. Baseline Regression

Table 2. Baseline model results.

Variable	Lnapply
	(1)	(2)	(3)	(4)
$D_{it}$	0.420 ***	0.280 ***	0.232 ***	0.202 ***
	(10.871)	(6.877)	(4.594)	(3.849)
URB		−0.001		−0.062
		(−0.010)		(−0.238)
FIN		0.197 ***		0.086
		(4.390)		(1.298)
CS		−0.799 ***		0.140
		(−4.203)		(0.486)
PUB		0.797		17.022
		(0.122)		(1.633)
POP		17.012 ***		7.173 **
		(15.681)		(2.366)
FDIL		0.052		−2.026 **
		(0.089)		(−2.411)
Constant	1.155 ***	0.494 ***	1.761 ***	1.048 ***
	(15.892)	(3.826)	(42.367)	(3.779)
Control	N	Y	N	Y
Year FE	Y	Y	Y	Y
City FE	N	N	Y	Y
Province × Year FE	N	N	Y	Y
SE clustered at the city level	N	N	Y	Y
Observations	4932	4932	4932	4932
R-squared	0.817	0.812	0.884	0.881

Note: The value within the parentheses represents the t-value; *** and ** indicate statistical significance at the 1% and 5% levels, respectively.

showcases the outcomes of the baseline model with the number of low-carbon technology invention applications (Lnapply) as the explained variable. Columns (1) and (2) consider only time-fixed effects, while Columns (3) and (4) further incorporate city-fixed effects and province-year fixed effects, clustering standard errors at the city level. The findings indicate that the estimated coefficients exhibit a significant positive correlation at the 1% level, suggesting that the NICP policy can significantly drive low-carbon technology innovation in pilot cities, and that the NICP policy has led to a 20.2% increase in the amount of low-carbon technology innovation for cities. Hypothesis H₁ is thus confirmed.

4.2. Robustness Tests

4.2.1. Parallel Trend Test

The premise for the effectiveness of the time-varying DID model is the parallel trends assumption, which states that the experimental and control group cities should have the same trend in low-carbon technology innovation. The paper employs the time-varying parallel trend test method and constructs Equation (4) to test for parallel trends.

$$Y_{ijt}={\alpha }_0+{\displaystyle \sum _{k=-3}^4}{\beta }_k{D}_{tp+k}+{\alpha }_{1}X+{\theta }_i+{\eta }_t+{\varphi }_j+{\tau }_k+{\epsilon }_{ijkt}$$

(4)

where $t_p$ equals the pilot year of the NICP policy. The parallel trend graph is selected with the base year of 2003, $D_{tp+k}$ serves as the interaction term between the pre-pilot years, the pilot year, and the post-pilot years with the virtual variable; if $t-t_p=k$ ($k$ = −6, −5, −4, −3, −2, −1, 0, 1, 2, 3, 4, 5, 6), $D_{t_t+k}$ = 1, otherwise $D_{t_t+k}$ = 0, among which $k$ = −1 refers to the year before the implementation of the NICP policy, $k$ = 0 refers to the year of the NICP policy, $k$ = 1 refers to the second year of the implementation of the NICP policy. The connotation of the remaining k-values follows accordingly.

Figure 1 and

Table A1. Dynamic effects.

Variable	Lnapply
pre_6	−0.079
	(−1.048)
pre_5	0.034
	(0.482)
pre_4	0.002
	(0.032)
pre_3	0.101
	(1.516)
pre_2	0.116
	(1.645)
pre_1	0.106
	(1.576)
current	0.182 **
	(2.479)
after_1	0.129 *
	(1.953)
after_2	0.154 **
	(2.314)
after_3	0.253 ***
	(3.710)
after_4	0.260 ***
	(4.576)
after_5	0.254 ***
	(3.839)
after_6	0.209 ***
	(3.856)
urb_w	−0.048
	(−0.183)
fin_w	0.111
	(1.498)
cs_w	0.077
	(0.262)
pub_w	13.715
	(1.301)
pop_w	8.588 ***
	(2.862)
fdi_w	−2.089 **
	(−2.442)
Constant	0.832 ***
	(2.983)
Control	Y
Year FE	Y
City FE	Y
Province × Year FE	Y
SE clustered at the city level	Y
Observations	4932
R-squared	0.882

Note: The value within the parentheses represents the t-value; ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

present the simplified results, while the estimated results of the dynamic effects of Table A1 are available in the Appendix B. The results indicate that before the implementation of the NICP policy, the estimated coefficients of $D_{it}$ did not pass the significance test, indicating no significant difference in low-carbon technology innovation between pilot and non-pilot cities, satisfying the parallel trend hypothesis. After the implementation of the NICP policy, the significance level of the estimated coefficients of $D_{it}$ increase from 5% to 1%, and the estimated coefficient values show a fluctuating upward trend, indicating that the driving effect of the NICP policy on low-carbon technology innovation becomes increasingly evident with time.

4.2.2. Placebo Test

In order to eliminate interference from other factors in the pilot cities and confirm that the policy effect in this article is caused by the NICP policy, this article adopts the method of Chetty et al. (2009) [47] and uses a non-parametric permutation test for placebo testing. Specifically, this article conducts non-repetitive sampling of prefecture-level cities and policy time, with a sampling frequency of 1000 times, and selects 75 cities as the virtual experimental group and the remaining cities as the virtual control group to obtain 1000 estimated coefficients of $D_{it}$ for the virtual experimental group and virtual policy time interaction through the baseline model. The baseline model results shown in Figure 2 pass the placebo test.

4.2.3. PSM-DID Method

Owing to the plausible disparities in economic progression and resource allotments among urban specimens, the use of the time-varying DID method directly may suffer from selection bias. Therefore, this article utilizes the propensity score matching technique to equate the pilot and non-pilot cities, ensuring the homogeneity of the experimental and control group samples, and then uses the time-varying DID regression method. As shown in Column (1) of

Table 3. Robustness tests.

Variable	Lnapply		Lngrant
	PSM-DID	Eliminating Interference	OLS		FE
	(1)	(2)	(3)	(4)	(5)	(6)
$D_{it}$	0.254 ***	0.326 ***	0.420 ***	0.267 ***	0.214 ***	0.169 ***
	(3.557)	(3.328)	(9.547)	(5.996)	(3.694)	(2.724)
Constant	1.079 **	0.640 *	0.791 ***	−0.147	1.607 ***	1.197 ***
	(2.338)	(1.685)	(10.251)	(−1.006)	(31.540)	(3.404)
Control	Y	Y	N	Y	N	Y
Year FE	Y	Y	Y	Y	Y	Y
City FE	Y	Y	N	N	Y	Y
Province × Year FE	Y	Y	N	N	Y	Y
SE clustered at the city level	Y	Y	N	N	Y	Y
Observations	4932	4122	4932	4932	4932	4932
R-squared	0.937	0.870	0.636	0.643	0.729	0.738

Note: The value within the parentheses represents the t-value; ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

, the estimated coefficient $D_{it}$ is significantly positive at the 1% level, confirming the robustness of the baseline model results.

4.2.4. Eliminating Interference from Other Related Policies

Considering the synergistic policies of carbon emission trading policy, energy-consuming right trading policy, and low-carbon city pilot projects during the sample period, they may have collectively contributed to the advancement of low-carbon technology innovation [6,7], thus presenting potential cross-effects. Specifically, the carbon emission trading pilot policy implemented in 2013 has driven low-carbon technology innovation through the cost and benefit of carbon emission quotas. The energy-consuming right trading policy has promoted low-carbon technology innovation through the cost and benefit of energy quotas. The low-carbon city pilot projects have encouraged low-carbon technology innovation by tracking the latest low-carbon technology and increasing research and development investments in low-carbon technology. To address this, this article eliminates the sample of carbon emission trading pilot cities, energy-consuming right-trading pilot cities, and low-carbon pilot cities. The article then re-regresses using Equation (1) and finds that the estimated coefficient $D_{it}$ is positive and reaches a significant level of 1%, which is consistent with the benchmark regression results.

4.2.5. Substituting the Explained Variable

The number of patents granted is also an important indicator of technology innovation; however, it takes 2 to 3 years for a patent to be granted from the application, making it difficult to quickly evaluate the impact of some policies on green technology innovation [48]. In order to further fortify the strength of the estimation outcomes, this article replaces the explained variable in Equation (1) with the number of low-carbon technology invention patents granted, and lags the explained variable by two periods. The results, as shown in Columns (3)–(6) of Table 3, indicate that the estimated coefficients $D_{it}$ are significantly positive at the level of 1%, which confirms the benchmark regression results.

5. Analysis of the Mechanisms of Influence

5.1. Investment in Innovative Resources

Table 4. The driving mechanisms.

Variable	SCI	Lnapply	HUM	Lnapply	INF	Lnapply
	(1)	(2)	(3)	(4)	(5)	(6)
$D_{it}$	0.023 ***	0.184 ***	0.049 ***	0.184 ***	0.170 ***	0.193 ***
	(5.805)	(3.438)	(3.813)	(3.476)	(2.963)	(3.681)
SCI		0.804 *
		(1.717)
HUM				0.365 **
				(2.015)
INF						0.056 *
						(1.809)
Constant	0.126 ***	0.944 ***	−0.029	1.058 ***	−3.609 ***	1.248 ***
	(5.997)	(3.337)	(−0.600)	(3.824)	(−9.554)	(4.103)
Control	Y	Y	Y	Y	Y	Y
Year FE	Y	Y	Y	Y	Y	Y
City FE	Y	Y	Y	Y	Y	Y
Province × Year FE	Y	Y	Y	Y	Y	Y
SE clustered at the city level	Y	Y	Y	Y	Y	Y
Observations	4932	4932	4932	4932	4932	4932
R-squared	0.468	0.882	0.635	0.882	0.588	0.882

Note: The value within the parentheses represents the t-value; ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.

presents the results of the test for the mediating effects of investments in innovative resources. Columns (1) to (2) correspond to the mediating effects of financial technology investment on low-carbon technology innovation, while Columns (3) to (4) correspond to the mediating effects of talent aggregation on low-carbon technology innovation. In this study, the proportion of fiscal technology expenditure to regional GDP is used to measure financial technology investment, and the proportion of urban employment to the total urban population is used to measure talent aggregation. Among them, the fiscal expenditure on science and technology, urban employment, urban population, and regional GDP are all derived from the annual “China Urban Statistical Yearbook” over the years. The results show that the estimated coefficients $D_{it}$ are significantly positive at the 1% level with financial technology investment and talent aggregation as dependent variables, and the estimated coefficients $D_{it}$ are also positive and significant at the 1% level with low-carbon technology innovation as the dependent variable. Compared with the benchmark regression results, the introduction of financial investment and talent aggregation into the control variables in Columns (2) and (4) resulted in a decrease in the estimated coefficient values from 0.202 to 0.184, indicating that the NICP policy can drive low-carbon technology innovation through financial technology investment and talent aggregation. The NICP policy, through fiscal technological investment, directly augments low-carbon technology innovation investment. Additionally, it serves to broaden financing channels by transmitting signals. The NICP policy also encourages large enterprises to form innovative alliances with universities and research institutions, promoting the gathering of innovative talent and realizing the enhancement of low-carbon technology innovation capabilities. Hypothesis H₂ is thus validated.

5.2. Digital Construction

Table 4 showcases the outcomes of the digital construction mediating effects test, as demonstrated in Columns (5) to (6). This paper conducts a principal component analysis on four indicators, namely the number of internet broadband access users per hundred people, the proportion of computer software and software industry employees to urban unit employees, the per capita total amount of telecommunications services, and the number of mobile phone users per hundred people. Two main components are selected based on the criterion of a cumulative contribution rate exceeding 80%, and the scores of the main components are calculated. A comprehensive index of the digital infrastructure construction level is obtained through the weighted average of the scores of the main components. It is noteworthy that the aforementioned indicators, namely the number of internet broadband access users per hundred people, the proportion of computer software and software industry employees to urban unit employees, the per capita total amount of telecommunications services, and the number of mobile phone users per hundred people, are all derived from the annual “China Urban Statistical Yearbook”. The estimated coefficients $D_{it}$ are significantly positive at the 1% level with the level of digital construction as the dependent variable, as well as the estimated coefficients of $D_{it}$ low-carbon technology innovation as the dependent variable. Compared with the benchmark regression results, the introduction of digital construction into the control variables in Column (6) resulted in a decrease in the estimated coefficient values $D_{it}$ from 0.202 to 0.193, indicating that the pilot policy for innovative cities is being implemented to enhance the level of urban innovation governance through the construction of digital infrastructure. The construction of digital infrastructure can eliminate obstacles to the flow of innovative elements caused by information asymmetry, effectively improving the efficiency of innovative resource allocation, and thereby promoting low-carbon technology innovation. Hypothesis H₃ is thus validated.

6. Heterogeneity Analysis

6.1. The Heterogeneity of Resource Endowments

According to the “National Sustainable Development Plan for Resource-based Cities (2013–2020)”(http://www.gov.cn/zwgk/2013-12/03/content_2540070.htm, accessed on 14 April 2023), the sample has been bifurcated into cities that are dependent on resources (T-resource) and those that are not (non-resource), and the resource-based cities are further divided into high resource-endowed cities (growth-oriented resource cities), medium resource-endowed cities (mature resource cities, i.e., M-resource), relatively low resource-endowed cities (regenerative resource cities, i.e., R-resource), and resource-depleted cities (declining resource cities, i.e., D-resource) based on their resource endowment levels, followed by a heterogeneity analysis of resource endowments. As the scope of the NICP policy has not yet covered growth-oriented resource-based cities, this paper excludes growth-oriented resource-based cities from the sample in the evaluation. From

Table 5. Heterogeneity of city characteristics.

Variable	Lnapply
	T-Resource	M-Resource	R-Resource	D-Resource	Non-Resource	PITI	Non-PITI
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
$D_{it}$	0.313 ***	0.537 ***	0.245	−0.006	0.117 *	0.236 ***	−0.068
	(2.651)	(2.787)	(1.379)	(−0.021)	(1.897)	(3.727)	(−0.451)
Constant	0.090	−0.295	−1.906	1.147	1.477 ***	2.040 ***	2.120 ***
	(0.208)	(−0.525)	(−0.519)	(0.705)	(3.707)	(4.153)	(5.508)
Control	Y	Y	Y	Y	Y	Y	Y
Year FE	Y	Y	Y	Y	Y	Y	Y
City FE	Y	Y	Y	Y	Y	Y	Y
Province × Year FE	Y	Y	Y	Y	Y	Y	Y
SE clustered at the city level	Y	Y	Y	Y	Y	Y	Y
Observations	1595	888	233	317	3337	1941	2991
R-squared	0.874	0.918	0.978	0.896	0.906	0.929	0.862

Note: The value within the parentheses represents the t-value; *** and * indicate statistical significance at the 1% and 10% levels, respectively.

, it can be seen that the estimated coefficients of $D_{it}$ in resource-based cities are significantly higher than those of non-resource-based cities, regardless of the coefficient values or significance levels, as compared between Columns (1) and (5). Further distinguishing between various classifications of resource-based cities, the estimated coefficients of $D_{it}$ in mature resource-based cities with higher resource endowment levels are significantly positive at the 1% level, as compared between Columns (2) to (4), while the estimated coefficients of $D_{it}$ in regenerative resource-based cities and declining resource-based cities are not significant. These findings indicate that the higher the resource endowment level of a city, the more significant the driving effect of the NICP policy on low-carbon technology innovation. Possible reasons may lie in the fact that cities with higher resource endowments exhibit a greater reliance on resource-based industries, lower resource utilization efficiency, and a greater potential for innovation enhancement. Consequently, the driving effect of innovative pilot policies becomes more pronounced. Thus, hypothesis H₄ is validated.

6.2. Heterogeneity in Environmental Information Disclosure

Since 2009, the Institute of Public and Environmental Affairs (IPE) and the Natural Resources Defense Council (NRDC) have jointly released the Pollution Information Transparency Index (PITI) system, and to date, 120 cities have published their PITI scores. Cities that disclose their PITI scores have a higher level of environmental information disclosure, while cities that do not disclose their PITI scores have a lower level of environmental information disclosure. Therefore, this paper divides cities based on whether they have published their PITI scores and conducts a heterogeneity test of environmental information disclosure. From Table 5, it can be seen that the estimated coefficients of $D_{it}$ for cities that disclose their PITI scores are significantly positive at the 1% level, as compared between Columns (6) and (7), while the estimated coefficients of $D_{it}$ for cities that do not disclose their PITI scores are negative and not significant. This implies that as the level of environmental information disclosure increases, the impact of the NICP policy on low-carbon technology innovation becomes more pronounced. Conversely, a lower level of environmental information disclosure may inhibit the innovative effects of such a policy. A higher level of environmental information disclosure can aid in strengthening monitoring and regulatory efforts of assessment indicators, such as energy consumption and pollutant emissions, outlined in the evaluation targets for innovative pilot cities. Furthermore, this would foster low-carbon technology innovation, assuming that H₅ is validated.

7. Conclusions

This article utilizes panel data from 274 Chinese cities between 2003 and 2020 and employs the time-varying DID method to evaluate its ability to drive low-carbon technology innovation. In addition to the baseline model, this study also employs a parallel trend test, placebo test, and PSM-DID test, eliminates interference from other related policies, and replaces the explained variable to ensure the robustness of the estimation results. Based on these tests, the article further explores the path of the NICP policy driving low-carbon technology innovation through financial technology investment, talent aggregation, and digital construction, and examines the heterogeneity of the influence of the NICP policy based on city resource endowments and environmental information disclosure characteristics. Based on the research findings, this article proposes the following policy recommendations:

(1)

Given the background of the “dual carbon” target, the Chinese government should encourage local governments to actively participate in innovative city pilot projects, and steadily and rationally expand the coverage of this policy pilot. As can be seen from the conclusion of the benchmark model, China’s innovative city pilot policy can significantly drive low-carbon technology innovation, and the driving effect of this policy on low-carbon technology innovation shows a fluctuating upward trend with time. Therefore, it is necessary to continuously refine the assessment indicators for energy consumption and pollutant emissions of innovative pilot cities, strengthen the monitoring and supervision of assessment indicators, and use green and low-carbon indicators with strong constraints to continuously and effectively drive enterprise low-carbon technology innovation. Other countries can also learn from this national-level innovative city pilot policy.

(2)

Considering the “dual externalities” of green and low-carbon technology innovation, local governments should provide more investment and financing channels for such innovation. As the research findings suggest, the innovative city pilot policy can not only directly increase enterprise investment in low-carbon technology innovation through increased financial and technological inputs but also play a signaling role through financial and technological inputs, guiding social capital to provide financing support for low-carbon technology innovation.

(3)

Developing more lenient policies attracts greater talent. As the research findings suggest, the innovative city pilot policy can promote low-carbon technology innovation through talent aggregation. Therefore, to break through the bottleneck of low-carbon technology innovation, it is not only necessary to encourage large enterprises to form innovation consortia with universities and research institutions, and build collaborative innovation platforms for low-carbon technology, but also to provide good public services and innovation carriers for talent innovation.

(4)

We recommend the acceleration of the construction of digital infrastructure such as big data, the Internet of Things, and cloud computing. As the research findings suggest, the innovative city pilot policy can reduce information asymmetry in innovation activities through the construction of digital infrastructure in order to better stimulate low-carbon technology innovation. Therefore, strengthening the construction of digital infrastructure, and removing innovation element flow barriers caused by information asymmetry, can effectively improve the efficiency of innovation resource allocation.

(5)

Formulating innovative city pilot policies requires tailored and targeted measures based on local conditions. As the research findings suggest, the driving effect of innovative city pilot policies on low-carbon technology innovation varies depending on the characteristics of the city. Specifically, the higher the urban resource endowment, the greater the marginal contribution of innovative city pilot policies to low-carbon technology innovation. Among them, this policy has the greatest impact on low-carbon technology innovation in mature resource cities. Therefore, cities with relatively high resource endowments, such as growing resource cities with high resource endowments, should be included in the scope of innovative city pilots to better improve the allocation of innovation resources. Cities with higher levels of environmental information disclosure have a significant impact on low-carbon technology innovation through innovative city pilot policies, while for cities with lower levels of environmental information disclosure, innovative city pilot policies have not yet had an impact on low-carbon technology innovation. Therefore, a lower level of environmental information disclosure may inhibit the impact of innovative city pilot policies on low-carbon technology innovation. Therefore, all pilot cities should quickly and comprehensively disclose relevant environmental information indicators, such as PITI, and clarify the environmental protection goals and responsibilities of each environmental protection subject through social supervision.

This article examines the driving effects of the NICP policy on low-carbon technology innovation. Prospects for further in-depth research include: firstly, the need to examine the driving effects of the NICP policy on low-carbon technology innovation at the micro-enterprise level; and secondly, the need to differentiate low-carbon technology innovation types and examine the impact of the NICP policy on different types of low-carbon technology innovation.