Impact of renewable energy technology innovation on green industrial structure upgrading: evidence from 284 cities in China

Zhonghai Zheng; Boqiang Lin; Nan Ye; Chaofeng Zheng; Yongjing Xie

doi:10.1088/2515-7620/ad0b2a

1. Introduction

As one of the major emerging countries (Lin and Wang 2020), China has witnessed remarkable economic expansion in the past few decades (Chen and Zheng 2023). However, this growth has also been accompanied by significant environmental challenges (Su and Fan 2022), such as rapid resource depletion and severe environmental pollution (Chen et al 2022). China has established goals for green development to address these issues and tackle global environmental challenges like climate change (Zhu et al 2019). This goal aims to foster synergistic growth between economic expansion and environmental protection by facilitating green industrial structure upgrading (GISU) (Yang et al 2020; Song et al 2022). GISU is considered to be a fundamental approach to achieving sustainable development. GISU refers to the gradual transition of traditional high-energy and high-emission industries into a low-carbon and environmentally friendly industrial structure (Song et al 2021, Zhong et al 2023). This transformation involves promoting changes in production methods, technologies, and resource utilization to mitigate adverse environmental impacts and enhance resource efficiency.

Previous research suggested that green innovation is vital in advancing GISU (Xie and Teo 2022). As a critical component of green innovation (Ge et al 2022), renewable energy technology innovation (RETI) may potentially have a substantial impact on GISU. For example, the studies by Ge et al (2022) and Su and Fan (2022) highlighted that RETI plays a crucial role in promoting GISU. Some data also seems to support the viewpoint above. Figure 1 presents the scatterplot depicting the relationship between RETI and GISU in China from 2006 to 2020. A detailed description of the calculation processes for RETI and GISU is elaborated in section 3.2. Therefore, these data appear to indicate a certain level of correlation between RETI and GISU.

**Figure 1.** Scatterplot depicting the relationship between RETI and GISU in China from 2006 to 2020.
Download figure:
Standard image High-resolution image

With an increasing emphasis on the sustainable development potential of RETI in China, more and more studies concentrated on analyzing the implications of RETI on green development, energy poverty, renewable energy generation, and other relevant factors (Zheng et al 2021c, Lee et al 2022, Su and Fan 2022). In recent years, a few studies discussed the connection between RETI and GISU (Ge et al 2022, Su and Fan 2022). Nevertheless, these studies primarily focus on the provincial level and lack empirical evidence at the urban level. Cities serve as central hubs for economic activities and energy consumption (Liu et al 2022), and studying the application of RETI in urban contexts can directly reveal its role in practical economic operations. Additionally, cities possess higher levels of technological innovation and market aggregation capabilities (Marchesani et al 2022), making urban environments more conducive to promoting and applying RETI. Thus, understanding the implications of RETI on GISU at the urban level holds substantial significance.

Furthermore, existing literature has pointed out the significance of environmental regulations, financial development, and human capital as influential factors in determining the GISU (Wang and Wang 2021, Wu and Liu 2021, He and Zheng 2023). However, limited research integrates these factors with RETI and GISU within a unified research framework. Consequently, the potential influence of these factors on the correlation between RETI and GISU might have been underestimated.

To address these research gaps, our study conducts a comprehensive analysis of the effects of RETI on GISU, utilizing the data from 284 cities in China during the period 2006–2020. This article closely associates with the studies by Ge et al (2022) and Su and Fan (2022) and provides unique contributions that supplement their innovative discoveries. First, while existing studies primarily explore the effects of RETI on green development, energy poverty, and renewable energy generation, there is limited research on the impact of RETI on GISU, particularly at the city level. By adopting an urban perspective, this study reveals the positive influence of RETI on GISU, thus enriching the literature in this area. Second, while previous research emphasizes the significant effects of environmental regulations, financial development, and human capital on GISU separately (Wang and Wang 2021, Wu and Liu 2021, He and Zheng 2023), few studies integrate these factors within a unified research framework that includes RETI. Our study demonstrates that environmental regulations, green finance, and human capital enhance GISU's promotion through RETI. Lastly, compared to existing research, this study provides a more detailed exploration of heterogeneity among regions, contributing to a deeper understanding of the varying impacts of RETI on GISU in different types of cities. Our research findings indicate that in economically developed cities with favorable innovation environment and cities with higher pollution levels, the promotion effect of RETI on GISU is more pronounced.

The remaining sections are organized as follows: section 2 provides a literature review and outlines the research hypotheses. Section 3 elucidates the research design. Section 4 displays the empirical results and ensuing discussions. Section 5 presents our research conclusions. The last section summarizes the limitations of this study and presents future research directions.

2. Literature and hypotheses

2.1. Renewable energy technology innovation (RETI)

With increasing global concerns for environmental sustainability and limitations on fossil fuel energy (Engel-Cox and Chapman 2023), renewable energy is widely considered a substantial choice instead of conventional energy sources (Bui and Tseng 2021, Herran and Ashina 2023). China has actively encouraged the advancement of renewable energy and implemented various policies and measures to foster technological innovation in the sector (Yan et al 2020, Lin and Xie 2023b). Under the incentive policies, China's RETI level continues to improve (Lin and Xie 2023c). China has achieved globally advanced levels of technological development in areas such as wind power and photovoltaics.

With continuous innovation and widespread application, renewable energy technology promises to reduce dependence on finite resources and foster the advancement of clean energy (Huang et al 2012). Currently, RETI has garnered significant attention from both industry and academia. Relevant research primarily focuses on the determinants of RETI and its economic consequences. For example, the study by Su and Fan (2022) revealed that a one-unit increase in RETI results in a 0.3-unit improvement in the region's green development. Wang et al (2023a) unveiled that a 1-unit rise in RETI yields a 0.05-unit improvement in productivity. Likewise, Su et al (2023) also affirmed the favorable influence of RETI on productivity. Wang et al (2022c) demonstrated that an increase of one unit in RETI results in a 0.02-unit reduction in the energy poverty index. Similar research conclusions were reached by Lee et al (2022). Moreover, Zheng et al (2021c) discovered that RETI facilitates renewable energy generation.

2.2. Green industrial structure upgrading (GISU)

Researchers have widely explored the methods for achieving GISU. For instance, the study by Song et al (2021) suggested that increasing one unit in the level of environmental regulations promotes a 0.06-unit enhancement in GISU. Similarly, Du et al (2021) also discovered a positive impact of environmental regulations on GISU. However, a study by He and Zheng (2023) found that environmental regulations exhibit a nonlinear relationship with GISU. Ren et al (2023) highlighted that a one-unit rise in digital finance results in a 0.14-unit improvement in GISU. Similarly, Zhong et al (2023) also argued that digital finance significantly promotes GISU. The study by Wang and Wang (2021) revealed that green finance and human capital significantly drive GISU. Specifically, an increase of 1 unit in green finance and human capital levels leads to an improvement of 2.1 units and 0.7 units in GISU, respectively.

In summary, a growing body of research highlights the significance of RETI and GISU, respectively. Nonetheless, there is a lack of research that delves into their interrelationship, particularly at the city level, where empirical evidence is scarce. Furthermore, while previous research has examined factors such as environmental regulations, financial development, and human capital on GISU, a limited number of studies have integrated these factors within a comprehensive research framework alongside RETI and GISU. Therefore, further empirical investigations are needed to address the limitations of these studies.

2.3. Renewable energy technology innovation and green industrial structure upgrading

GISU is a complex and dynamic process (Zhou et al 2020) in which RETI plays a crucial role. The promotion of RETI contributes to GISU in multiple ways.

First, through continuous innovation, the production efficiency of renewable energy has been enhanced, and costs have been steadily reduced, gradually equipping renewable energy with a competitive advantage over traditional energy sources (Zheng et al 2021b, Lin and Xie 2023a, 2023b). This competitive advantage has propelled the rapid growth of the clean energy sector, thereby facilitating GISU. Second, traditional industries often rely on fossil fuels, which leads to environmental pollution (Lin and Wang 2012). RETI assists in decreasing dependence on fossil fuels and alleviating the adverse environmental consequences of their extraction and combustion. Industries can significantly reduce their carbon footprint by transitioning towards cleaner energy alternatives. This transformation helps to promote the development of industrial structures towards more environmentally friendly and sustainable directions (Ge et al 2022), thus enhancing GISU. Lastly, RETI involves multiple sectors, including solar energy, wind energy, bioenergy, etc, encompassing various aspects such as R&D, equipment manufacturing, project development, and operational management (Mastrocinque et al 2022). The added value, technological content, and economic benefits of related industries can be enhanced by introducing advanced renewable energy technologies and processes. It is possible to promote synergistic development among related industries, foster new industrial growth areas, and facilitate GISU (Su and Fan 2022). Therefore, we postulate the following hypothesis:

H1. Renewable energy technology innovation (RETI) significantly facilitates green industrial structure upgrading (GISU).

2.4. The moderating effects

2.4.1. The moderating effect of environmental regulations

Environmental regulations encompass government-issued environmental laws, emission reduction targets, subsidy policies, and environmental standards, including incentive-based and mandatory approaches (Liu et al 2021, Cheng and Kong 2022). Environmental regulations can provide economic incentives (Tsireme et al 2012) to the clean energy sector, including renewable energy, to encourage greater investment in technological innovation. These incentives can reduce the cost of RETI and enhance the return on investment for projects, attracting more capital and talent into the renewable energy field and driving accelerated development of RETI. Meanwhile, as the stringency of environmental regulations increases, traditional industries face more stringent pressure to reduce emissions (Cao et al 2020, Lu et al 2023). To comply with environmental laws, these industries are increasing investments in environmental projects, thereby stimulating the rapid expansion of the green industry. Against this backdrop, RETI can transform industrial structures from high-pollution and high-emission to low-carbon and clean directions (Ge et al 2022). Thus, we formulate the following research hypothesis:

H2(a). Environmental regulations positively moderate the relationship between RETI and GISU.

2.4.2. The moderating effect of green finance

Green finance entails incorporating environmental sustainability considerations within the financial system, aiming to foster sustainable development through the guidance of capital and investment channels (Du et al 2023, Feng et al 2023). Green finance offers financial support and investment opportunities for RETI (Ma et al 2023b). By implementing green finance mechanisms, increased capital inflow can be directed towards RETI. This, in turn, helps lower the costs associated with technological innovation, accelerating the progress and implementation of renewable energy technologies. Additionally, green financial institutions can provide enterprises meeting green standards with low-interest loans and favorable financing conditions, effectively reducing their financing costs (Xu and Li 2020). This offers opportunities for the rapid development of green and environmentally friendly enterprises, facilitating GISU. Consequently, green finance is beneficial to enhance the catalytic role of RETI in promoting GISU. Hence, we argue that:

H2(b). Green finance is a positive moderator in the relationship between RETI and GISU.

2.4.3. The moderating effect of human capital

Human capital denotes the accumulation and improvement of a nation or region's labor force regarding education, training, and skills (Çakar et al 2021). Enhancing human capital levels creates a conducive environment for innovation (Wang et al 2023b) and provides necessary talent support for RETI, thus enabling the commercialization and application of renewable energy technologies (He et al 2022). Moreover, the improvement in human capital levels fosters the development of a larger pool of specialized professionals capable of meeting the needs of various industries (Saqib et al 2023). This, in turn, drives the emergence of nascent industries and the upgrading and transformation of traditional industries, ultimately facilitating the optimization and adjustment of industrial structures (Ma et al 2023a). Consequently, in regions with higher levels of human capital, the promotion of RETI can better leverage its influence on GISU. This study presents the following research hypothesis:

H2(c). The relationship between RETI and GISU is positively influenced by human capital.

Therefore, our theoretical model is depicted in figure 2.

3. Research design

3.1. Data

Considering data availability, we exclude cities with significant missing data and retain data from 284 prefecture-level cities. Furthermore, rapid progress has been witnessed in China's renewable energy industry since the enactment of the Renewable Energy Law in 2006. Simultaneously, considering the patent granting period, the observation period for this study is selected as 2006 to 2020. The primary data sources for this study include the China National Intellectual Property Administration (CNIPA), the China City Statistical Yearbook, the CEIC database, and local statistical bureaus.

3.2. Selection of variables

3.2.1. Dependent variable

As emphasized by Du et al (2021) and Zhao et al (2022), the growth of the tertiary industry signifies environmentally friendly and sustainable production practices. Considering the availability of city-level subsector data, prior research has commonly utilized the percentage of the tertiary industry's GDP to the secondary industry's GDP as a measure of GISU (Du et al 2021, Zhao et al 2022, Wang et al 2022a). They argued that this metric could illuminate structural changes among various industries, thus offering a more effective measure of GISU. Therefore, we also utilize this indicator to measure GISU. The detailed calculation process has been summarized in appendix A.

3.2.2. Independent variable

Due to the availability of urban-level data, current research frequently employs patent data as a metric for assessing RETI (Yan et al 2020, Cao et al 2022, Wang et al 2022b, Wang et al 2023a). These studies assert that a substantial number of patent applications can indicate the allocation of limited research resources to specific technology fields.

We refer to the classification codes of renewable energy patents provided in existing research (Wang et al 2022b, Wang et al 2023a) and manually collect and organize data on authorized renewable energy patents from the CNIPA for 284 cities. To more accurately gauge the extent of technological innovation, previous research utilized the stock of authorized renewable energy patents to assess the RETI level (Yan et al 2020, Cao et al 2022, Wang et al 2023a). Consistent with these studies, the detailed calculation process is as outlined below:

$\begin{eqnarray}&&{{RETI}}_{i,t}=\displaystyle \sum _{k=0}^{t}{{REPA}}_{i,k}\exp \left[-{\delta }_{1}(t-k)\right]\bullet \left\{1-\exp \left[-{\delta }_{2}(t-k)\right]\right\}\end{eqnarray} \tag{ 1 }$

where ${REPA}$ represents the number of authorized patents in renewable energy. ${\delta }_{1}$ refers to the technological depreciation effect, and ${\delta }_{2}$ is the technology diffusion effect. Drawing upon prior research (Cao et al 2022, Wang et al 2023a), their values are determined to be 0.36 and 0.03, respectively. Furthermore, considering the accessibility of patent data, this article takes 1985 as the base year to gauge the stock of authorized patents in renewable energy for Chinese cities in each subsequent year.

3.2.3. Moderating variable

We incorporate three moderating variables: environmental regulations, green finance, and human capital. First, consistent with prior studies by Shi et al (2019) and Xu et al (2023), the proportion of environmentally-related terminology in local government annual reports serves as a metric to evaluate environmental regulations. These studies indicated that this metric can reflect the commitment and willingness of local governments to environmental protection and is closely associated with formulating environmental regulatory policies. Second, drawing on previous studies (He et al 2019, Chen et al 2023), a composite measure of green finance is established using the entropy methods. The comprehensive index serves as an indicator for evaluating the level of green finance. Lastly, in line with existing research (Peng et al 2021, Peng et al 2023), the number of university students is used to measure human capital.

3.2.4. Control variable

Following existing research (Zheng et al 2021a, Zhu 2022, He and Zheng 2023), the control variables of this research include: (1) Economic development level: The per capita GDP serves as a common metric used to gauge a region's economic performance. (2) Population density: It quantifies the number of individuals per unit area and reflects the spatial concentration of the population, which can affect the availability of resources and the competitiveness of industries. (3) Government support: This index measures the extent and effectiveness of government policies and support measures to promote industrial upgrading. (4) Financial development: It gauges the level of development and efficiency of the financial sector, which is essential in providing capital, investment, and risk management services to facilitate industrial upgrading. The measurement of these variables is summarized in appendix A.

3.3. Models

To comprehensively investigate the impact of RETI on GISU, we develop the following research model:

$\begin{eqnarray}&&{{GISU}}_{i,t}={\alpha }_{1}+{\beta }_{1}{{RETI}}_{i,t}+{{Controls}}_{i,t}+{\eta }_{i}+{\eta }_{t}+{\mu }_{i,t}\end{eqnarray} \tag{ 2 }$

As described in section 2, the impact of RETI on GISU may be influenced by environmental regulations, green finance, and human capital. To examine the relationships among these variables, we further incorporate them into our model and construct the following moderating effect models:

$\begin{eqnarray}&&{{GISU}}_{i,t}={\alpha }_{2}+{\beta }_{2}{{RETI}}_{i,t}+{\beta }_{3}{{{ER}}_{i,t}\times {RETI}}_{i,t}+{\beta }_{4}{{ER}}_{i,t}+{{Controls}}_{i,t}+{\eta }_{i}+{\eta }_{t}+{\mu }_{i,t}\end{eqnarray} \tag{ 3 }$

$\begin{eqnarray}&&{{GISU}}_{i,t}={\alpha }_{3}+{\beta }_{5}{{RETI}}_{i,t}+{\beta }_{6}{{{GF}}_{p,t}\times {RETI}}_{i,t}+{\beta }_{7}{{GF}}_{p,t}+{{Controls}}_{i,t}+{\eta }_{i}+{\eta }_{t}+{\mu }_{i,t}\end{eqnarray} \tag{ 4 }$

$\begin{eqnarray}&&{{GISU}}_{i,t}={\alpha }_{4}+{\beta }_{8}{{RETI}}_{i,t}+{\beta }_{9}{{{HC}}_{i,t}\times {RETI}}_{i,t}+{\beta }_{10}{{HC}}_{i,t}+{{Controls}}_{i,t}+{\eta }_{i}+{\eta }_{t}+{\mu }_{i,t}\end{eqnarray} \tag{ 5 }$

where $i,$ $p,$ and $t$ represent the city, the province, and the year, respectively. ${GISU}$ is the dependent variable, and ${RETI}$ is the independent variable. ${ER},$ ${GF},$ and ${HC}$ are the moderating variables. ${Controls}$ represent the control variables. ${\eta }_{i}$ and ${\eta }_{t}$ represent city-fixed effects and time-fixed effects, respectively. ${\mu }_{i,t}$ is the error term.

4. Empirical results and discussion

4.1. Descriptive statistics and correlation analysis

As shown in table 1, the values of ${GISU}$ vary between 9.426 and 535.000, with a mean value of 98.455. This indicates significant variations in industrial structure among Chinese cities. For ${RETI},$ its values exhibit a range from 0.000 to 714.625, with a mean value of 8.871. This result suggests substantial disparities in RETI among these cities and indicates significant room for enhancing technological innovation within the renewable energy sector. In addition, the results of the correlation analysis for the main variables are presented in table 2. The average variance inflation factor (VIF) for the main variables is 1.41 ( $\lt 10$ ), and the maximum value of VIF is 1.45 ( $\lt$ 3.3) (Kock and Lynn 2012, Cheng et al 2021, Lin and Xie 2023a), suggesting no issue of multicollinearity among these variables.

Table 1. Descriptive statistics.

Variable	Obs.	Mean	Median	Std. Dev.	Min.	Max.	Unit
${GISU}$	4260	98.455	86.154	55.103	9.426	535.000	%
${RETI}$	4260	8.871	1.598	31.226	0.000	714.625	—
${ER}$	4260	0.326	0.308	0.144	0.018	1.239	%
${GF}$	4260	0.264	0.249	0.096	0.091	0.804	—
${HC}$	4260	8.825	3.274	16.052	0.000	130.714	10,000 persons
${ED}$	4260	10.426	10.391	0.808	7.923	13.228	—
${PD}$	4260	431.937	357.709	337.666	4.700	2926.602	${people}/{{km}}^{2}$
${GI}$	4260	15.407	13.137	8.925	3.512	145.013	%
${FD}$	4260	230.662	198.983	115.684	56.001	2129.694	%

Table 2. Correlation analysis.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
(1) ${GISU}$	1.000
(2) ${RETI}$	0.410	1.000
(3) ${ER}$	0.022	0.046	1.000
(4) ${GF}$	0.119	0.218	0.067	1.000
(5) ${HC}$	0.313	0.493	0.042	0.063	1.000
(6) ${ED}$	0.106	0.414	0.191	0.170	0.402	1.000
(7) ${PD}$	0.004	0.330	−0.054	−0.045	0.356	0.275	1.000
(8) ${GI}$	0.380	−0.068	0.083	0.100	−0.181	−0.273	−0.357	1.000
(9) ${FD}$	0.592	0.388	0.125	0.126	0.500	0.343	0.116	0.244	1.000

4.2. Baseline regression

The regression outcomes concerning the influence of RETI on GISU are displayed in table 3. Initially, we include only the independent variable, as shown in the first column. The results indicate that RETI significantly promotes GISU at a 1% significance level. Next, we gradually introduce control variables to observe changes in the coefficient of ${RETI}.$ The results in columns (2) to (3) also demonstrate that RETI facilitates GISU. In the last column, we re-examine the above results using a random effects model, which yields consistent conclusions. The results of the Hausman test support the validity of employing the fixed effects model (Xie et al 2021).

Table 3. Baseline regression.

Variable	(1) ${GISU}$	(2) ${GISU}$	(3) ${GISU}$	(4) ${GISU}$
${RETI}$	0.1645***	0.1862***	0.1939***	0.5331***
	(8.52)	(9.95)	(10.50)	(22.96)
${ED}$		−56.4406***	−41.6659***	−6.4864***
		(−25.19)	(−16.76)	(−7.11)
${PD}$		−0.0350***	−0.0309***	−0.0050**
		(−4.86)	(−4.36)	(−2.42)
${GI}$			0.2995***	1.5717***
			(3.56)	(19.26)
${FD}$			0.0769***	0.2138***
			(10.91)	(33.61)
Constant	96.9960***	700.3751***	522.1248***	89.9878***
	(257.69)	(29.30)	(19.10)	(9.41)
City FE	YES	YES	YES
Year FE	YES	YES	YES
Hausman test			${\chi }^{2}$ =355.03, p=0.00
${R}^{2}$	0.853	0.874	0.879	0.476
Obs.	4,260	4,260	4,260	4,260

Note: *** p < 0.01, ** p < 0.05, * p < 0.1, the same as below.

In conclusion, RETI has been found to promote GISU, thus supporting the validity of H1. This finding demonstrates that RETI has emerged as a catalyst for transforming traditional industries and driving them towards a more sustainable and efficient future. The possible reason for this is that RETI can enhance energy conversion efficiency and reduce production costs (Zheng et al 2021b, Lin and Xie 2023a, 2023b), promoting the low-carbon transformation of high-energy-consuming and heavily polluting industries (Ge et al 2022), and facilitating the green and collaborative development of related industries (Su and Fan 2022), thereby promoting GISU.

Similarly, a recent study by Ge et al (2022) also obtained similar research findings. Using panel data from Chinese provinces, they observed a positive effect of RETI on GISU. However, our study differs in its focus on the city level, providing more micro-level empirical evidence and complementing the existing research literature.

4.3. Robustness test

(1) To provide a more precise depiction of the regional RETI level, we utilize the stock of granted invention patents to measure RETI in this study. The outcomes are presented in the first column of appendix B. (2) Drawing on the research by Zhou et al (2020), we utilize the share of total output value contributed by heavy-polluting industries as a proxy for the inverse measurement of GISU. Due to the lack of data on the output value of various industries at the city level, we adopt a similar approach to existing research (Yin et al 2023) by aggregating enterprise-level data to obtain city-level data. The outcomes can be found in column (2) of appendix B. (3) Considering the persistence and lagged responses in the industrial structure adjustments, we further employ the GMM model for robustness testing. The results are presented in the third column of appendix B. (4) To mitigate the adverse effects caused by outlier data, we normalize the primary variables. The re-regression outcomes are presented in the fourth column of appendix B. (5) To mitigate the impact of specific sample selection on the study results, we utilize a random sampling method, splitting the research sample into two groups with an 80% and 20% allocation. The regression results for these two groups are presented in the fifth and sixth columns of appendix B.

4.4. Endogenous problems

4.4.1. Reverse causality issue

To tackle the issue of reverse causality, the independent variable is lagged by one period, and the results are showcased in the first column of appendix C.

4.4.2. Omitted variable problem

Previous studies have indicated that environmental regulations, green finance, and human capital have potentially significant impacts on GISU (Wang and Wang 2021, Wu and Liu 2021, He and Zheng 2023). Therefore, Environmental regulations, green finance, and human capital are also included as control variables. The results of the re-estimated regression with these control variables are shown in the second column.

4.4.3. Instrumental variables

In the existing body of research (Goldsmith-Pinkham et al 2020, Hasan et al 2020), the Bartik instrumental variable (IV) is extensively utilized to address endogeneity concerns. Referring to these studies, Bartik IV is constructed by taking the first difference over time of the average RETI levels in each province's prefecture-level cities and multiplying it by the RETI level of the same city in the previous year. For a specific city at the prefecture level, the temporal changes in the average RETI levels in each province's prefecture-level cities are considered exogenous. Simultaneously, a correlation exists between the city's RETI and the previous year's levels. Thus, selecting this IV is appropriate. The third and fourth columns of appendix C present the regression results for the first and second stages, respectively.

4.5. Moderating effect analysis

Section 4.2 has revealed the positive impact of RETI on GISU. Based on this, this section will further discuss the influences of environmental regulations, green finance, and human capital on their relationship. Therefore, we construct interaction terms ${RETI}\times {ER},$ ${RETI}\times {GF},$ and ${RETI}\times {HC},$ respectively, and introduce them into our research model as shown in models (3)-(5). The estimated results for these variables are presented in table 4.

Table 4. Moderating effect analysis.

Variable	(1) ${GISU}$	(2) ${GISU}$	(3) ${GISU}$	(4) ${GISU}$
${RETI}$	0.1169***	0.0634	−0.0935**	−0.2784***
	(3.10)	(1.37)	(−2.10)	(−4.29)
${RETI}\times {ER}$	0.2064**			0.2223***
	(2.46)			(2.59)
${RETI}\times {GF}$		0.2492***		0.2659***
		(3.24)		(2.97)
${RETI}\times {HC}$			0.0049***	0.0040***
			(6.62)	(5.20)
Constant	523.2468***	536.7177***	534.6931***	550.3464***
	(19.15)	(19.59)	(19.66)	(20.22)
Control	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
${R}^{2}$	0.879	0.880	0.881	0.882
Obs.	4,260	4,260	4,260	4,260

4.5.1. Environmental regulations

The findings in the first column of table 4 reveal that the coefficient of ${RETI}\times {ER}$ is significantly positive, suggesting that environmental regulations effectively enhance the interaction between RETI and GISU. Consequently, H2 (a) is confirmed. One reasonable explanation is that implementing environmental regulations can contribute to creating a stable policy environment for the development of RETI and the transition of industrial structure towards a greener and more sustainable direction, thereby enhancing the role of RETI in promoting GISU. Similarly, in the studies by Du et al (2021) and Song et al (2021), the positive influence of environmental regulations on GISU was underscored.

4.5.2. Green finance

The results in column (2) of table 4 demonstrate a significantly positive coefficient for ${RETI}\times {GF},$ signifying that green finance positively moderates the relationship between RETI and GISU. Therefore, H2 (b) is supported. One plausible explanation is that the advancement of green finance provides financial resources, lowers innovation costs, and offers favorable financing conditions for RETI and the rapid development of the low-carbon industry, thereby promoting GISU. A recent study by Ge et al (2022) has reached similar research conclusions.

4.5.3. Human capital

In column (3) of table 4, the outcomes indicate a significantly positive coefficient for ${RETI}\times {HC},$ indicating that human capital serves as a positive moderator in the association between RETI and GISU. As a result, H2 (c) is validated. A plausible explanation is that the enhancement of human capital implies an increase in individuals with higher technical and innovative capabilities who can drive RETI and promote industrial structure development towards a more competitive and high value-added direction. Previous studies have also found a positive impact of human capital on GISU (Wang and Wang 2021, Xiang and Zhou 2022).

Lastly, we simultaneously include all three interaction terms in the model, and the results are shown in the fourth column of table 4. The coefficients of ${RETI}\times {ER},$ ${RETI}\times {GF},$ and ${RETI}\times {HC}$ remain significantly positive. Therefore, the effectiveness of H2(a)-(c) has been re-validated.

4.6. Heterogeneity assessment

4.6.1. Economic development level

The impact of RETI on GISU may exhibit heterogeneity across different regional economic levels. Based on the mean per capita GDP, the samples in this study are divided into two groups: economically developed cities (EDC) and economically underdeveloped cities (EUDC). The regression outcomes for each group are demonstrated in columns (1) and (2) of table 5. The coefficient of ${RETI}$ in the first column is significantly positive, while in column (2), its coefficient is positive but not statistically significant. This implies that in EDC, the influence of RETI on GISU is more notable.

Table 5. Heterogeneity assessment.

	(1)	(2)	(3)	(4)	(5)	(6)
Variable	EDC	EUDC	FIEC	PIEC	HPC	LPC
${RETI}$	0.2396***	0.0865	0.2361***	0.0002	0.2373***	−0.1261
	(13.12)	(0.79)	(14.33)	(0.00)	(18.19)	(−1.31)
Constant	20.5891	792.0067***	258.1994***	539.3369***	283.3091***	601.8121***
	(0.44)	(23.94)	(5.57)	(15.84)	(7.95)	(16.47)
Control	Yes	Yes	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
${R}^{2}$	0.938	0.851	0.928	0.859	0.946	0.852
Obs.	1,320	2,940	1,380	2,880	1,560	2,700

Possible reasons for this disparity may be that regions with higher economic development may have better infrastructure, financial resources, and technological capabilities, which enable them to facilitate the adoption and integration of renewable energy technologies into their industrial sectors. As a result, the positive effects of RETI on GISU may be more pronounced in these economically advanced regions. In contrast, regions with lower economic development may face challenges such as limited financial resources, inadequate infrastructure, and weaker technological capabilities (Yan et al 2020), which could hinder the effective implementation and adoption of renewable energy technologies. Consequently, the impact of RETI on GISU may be relatively weaker in these regions.

4.6.2. Innovation environment

The impact of the innovation environment on the association between RETI and GISU can contribute to heterogeneity. In accordance with prior research conducted by Lin and Ma (2022), this study assesses the innovation environment of regions using the proportion of technology expenditure as a measurement. The samples are then divided into two groups: cities with favorable innovation environment (FIEC) and cities with poor innovation environment (PIEC). The regression outcomes for each subgroup are displayed in columns (3) and (4) of table 5. The findings reveal that only the coefficient of ${RETI}$ in the third column is significantly positive, suggesting that in FIEC, the incentive effect of RETI on GISU is more significant.

One reasonable explanation for this finding is that regions with favorable innovation environment, characterized by strong R&D institutions, supportive policies, and innovation culture, are more likely to foster the successful development and commercialization of renewable energy technologies. In such environments, the positive impact of RETI on GISU is expected to be more significant. Conversely, regions with a less conducive innovation environment, lacking adequate R&D capabilities, weak policy support, and limited innovation culture, may face obstacles in effectively utilizing renewable energy technologies to upgrade their industrial structure. Consequently, the impact of RETI on GISU may be relatively limited in these regions. Similarly, Xu et al (2022) also emphasized the positive influence of the innovation environment on GISU. Their study highlighted that the improved cluster innovation capabilities of the region help to increase the proportion of high-tech industries.

4.6.3. Regional pollution levels

The heterogeneity in the impact of RETI on GISU can also arise from varying pollution levels across regions. Based on data on industrial wastewater and ${{\rm{SO}}}_{2}$ emissions for each city, we construct a comprehensive evaluation index for regional pollution levels using the entropy method (Du et al 2021). Drawing on previous studies (Zhang et al 2020, Pan et al 2021), the samples are divided into high-pollution cities (HPC) and low-pollution cities (LPC) based on the average pollution levels of these regions. The results indicate that the coefficient for ${RETI}$ in the fifth column of table 5 is significantly positive, while it is not significant in the sixth column, suggesting that in HPC, the promotion effect of RETI on GISU is more significant.

One potential rationale for this finding could be that areas with high pollution levels may encounter increased pressure and regulatory requirements to reduce their environmental footprint and transition towards cleaner and sustainable industries (Pan et al 2021). In this context, RETI offers opportunities for these regions to mitigate pollution and upgrade their industrial structure towards greener alternatives. Therefore, the positive impact of RETI on GISU may be more pronounced in areas with higher pollution levels. Conversely, areas with lower pollution levels may have relatively lower urgency or regulatory incentives to adopt renewable energy technologies for GISU. As a result, the impact of RETI on GISU may be comparatively weaker in these regions.

5. Conclusions

RETI and GISU are crucial for achieving sustainable economic development. Currently, there is a shortage of research investigating the link between RETI and GISU, particularly at the urban level. The primary discoveries from this research can be summarized as follows, and a comparison of these results with relevant literature is provided in appendix D.

First, our study reveals a significant and positive association between RETI and GISU, which supports the transformative potential of RETI in driving sustainable development. Similar findings were obtained by Ge et al (2022). The positive relationship between RETI and GISU underscores the transformative potential of RETI in shaping sustainable development. Therefore, to encourage industrial restructuring, it is advisable for the government to actively support and foster innovation in the realm of renewable energy technologies. Specifically, greater financial support for the R&D activities of renewable energy enterprises can incentivize increased investments in RETI.

Second, our research also underscores the importance of crucial moderating factors in strengthening the relationship between RETI and GISU. Specifically, environmental regulation, green finance, and human capital strengthen the positive impact of RETI on GISU. These factors act as enhancing mechanisms that facilitate the integration and diffusion of renewable energy technologies within the industrial sector, leading to more substantial advancements in GISU. Consequently, strengthening environmental regulations can effectively encourage companies to invest in RETI, thereby facilitating GISU. Moreover, establishing green finance markets and providing more investment opportunities can attract more significant funding into the renewable energy sector. Lastly, intensified efforts in talent development in relevant fields by both the government and companies can leverage the role of RETI in advancing industrial restructuring.

Lastly, the heterogeneity analysis outcomes uncover that RETI's promotion effect on GISU is particularly significant in EDC, FIEC, and HPC. This suggests that the context and characteristics of each city play a vital role in determining the magnitude and significance of the relationship between RETI and GISU. To more effectively harness the positive impact of RETI on GISU, the government can tailor their policies to account for the economic development level, innovation environment, and pollution levels of cities. For EDC, it is advisable for the government to offer additional support for innovative renewable energy technology, encouraging these cities to play a more substantial role in industrial restructuring. For FIEC, establishing more flexible and open mechanisms for scientific cooperation can promote inter-regional collaboration in RETI. In HPC, stronger environmental governance measures can be implemented alongside increased support for RETI to promote RETI and GISU.

5.1. Limitation and future research

First, due to the lack of city-level GDP data for various subsectors, this study employs the percentage of the tertiary industry's GDP to the secondary industry's GDP to measure GISU. However, not all subsectors of the tertiary sector are entirely low-carbon industries, and similarly, not all subsectors of the secondary sector are entirely high-carbon industries. Employing this measurement method could result in some measurement errors. Hence, a crucial future research direction is to seek more precise ways to gauge GISU. Second, our study primarily centers on China's cities, which may constrain the applicability of the results to other countries or regions characterized by distinct contexts and characteristics. Subsequent research endeavors could broaden the scope beyond China by encompassing a wider range of countries or regions, thereby offering a more comprehensive comprehension of the correlation between RETI and GISU. Lastly, this study focuses on discussing the impact of RETI on GISU. Building upon this research, future studies can delve further into the effects of RETI on urban green development, such as green total factor productivity.

Acknowledgments

This work was supported by consulting project of State Grid Fujian Electric Power Cord. Ltd (No. B3130N220021).

Data availability statement

The data cannot be made publicly available upon publication because they contain commercially sensitive information. The data that support the findings of this study are available upon reasonable request from the authors.

Appendix A.: Variable definitions

Type	Variable names	Abbreviations	Measures
Dependent variable	Green industrial structure upgrading	${GISU}$	(Tertiary industry's GDP to secondary industry's GDP) ×100
Independent variable	Renewable energy technology innovation	${RETI}$	The stock of authorized patents in renewable energy
Moderating variable	Environmental regulations	${ER}$	The percentage of environmental vocabulary frequency in prefecture-level government reports
	Green finance	${GF}$	The comprehensive index of green finance
	Human capital	${HC}$	The number of university students
Control variables	Economic development level	${ED}$	Ln (GDP/ the total population)
	Population density	${PD}$	The number of individuals per unit area
	Government intervention	${GI}$	Local government fiscal expenditure, excluding spending on science and education/GDP×100
	Financial development	${FD}$	(The total amount of bank deposits and loans) /GDP×100

Appendix B.: Robustness test

	(1)	(2)	(3)	(4)	(5)	(6)
Variable	${GISU}$	${GISU}$	${GISU}$	${GISU}$	${GISU}$	${GISU}$
${RETI}$	0.4939***	−0.0720***	0.1248***	0.2636***	0.1967***	0.1853***
	(12.20)	(−3.73)	(2.87)	(10.50)	(9.55)	(4.08)
$L.{GISU}$			1.0143***
			(69.02)
Constant	514.2903***	−30.3465	16.8071*	0.3573***	523.9412***	423.1844***
	(18.70)	(−1.06)	(1.96)	(23.74)	(17.15)	(5.53)
Control	Yes	Yes	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
${R}^{2}$	0.880	0.756		0.879	0.875	0.927
Obs.	4,260	4,260	3,976	4,260	3,408	852

Appendix C.: Endogeneity test

	(1)	(2)	(3)	(4)
Variable	${GISU}$	${GISU}$	${RETI}$	${GISU}$
${RETI}$		0.1836***		0.2011***
		(9.59)		(8.78)
$L.{RETI}$	0.1867***
	(8.93)
Bartik			0.0174***
			(82.32)
Constant	519.7246***	545.5191***	23.0693
	(18.16)	(19.92)	(1.24)
Control	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Anderson LM				2578.310***
Cragg-Donald Wald F				6777.405
${R}^{2}$	0.886	0.880		0.8860
Obs.	3,976	4,260	3976	3976

Appendix D.: Comparison of research findings with relevant literature

This study's conclusions	Related studies' conclusions
RETI significantly promotes the GISU of cities in China	Ge et al (2022) discovered that RETI significantly enhances the GISU of Chinese provinces
Environmental regulations, green finance, and human capital bolster the positive impact of RETI on the GISU of cities in China	Du et al (2021) and Song et al (2021) highlighted the positive influence of environmental regulations on GISU.
	Ge et al (2022) found that green finance positively moderates the relationship between RETI and GISU of Chinese provinces
	Wang and Wang (2021) and Xiang and Zhou (2022) suggested that human capital has a positive impact on GISU.
The beneficial impact of RETI on GISU is particularly pronounced in economically developed cities with a conducive innovation environment and those facing severe pollution challenges.	Yan et al (2020) argued that the impact of RETI on green productivity is significant only in economically developed regions
	Xu et al (2022) underscored the beneficial influence of the innovation environment on GISU
	Pan et al (2021) suggested that regions characterized by elevated pollution levels are more likely to experience pressure and regulatory requirements