Industrial Green Innovation Efficiency: Spatial Patterns, Evolution, and Convergence in the Yangtze River Economic Belt

Abstract

This study examines the relationship between technological innovation and economic development in the Yangtze River economic belt context. Specifically, the study employs the SBM-GML model to assess the efficiency of industrial green technology innovation across 110 prefecture-level cities between 2006 and 2022. The study also employs exploratory spatial data analysis (ESDA) and the Spatio-temporal transition method to analyze the spatial evolution pattern of the GML index of industrial green technology innovation. In addition, the study investigates the convergence mechanism using absolute and conditional β convergence models. The findings reveal that the GML index of industrial green technology innovation in the Yangtze River Economic Belt exhibits an upward trend, and technological progress is a key driver. Moreover, the spatial and temporal transition of the GML index of industrial green technology innovation shows substantial spatial dependence and solid spatial stability. The study also finds regional heterogeneity in the absolute and conditional β convergence characteristics and their influencing factors. Considering regional differences, the results suggest differentiated policy recommendations to promote the coordinated development of industrial green technological innovation efficiency in the Yangtze River Economic Belt. The study contributes to the literature on the relationship between technological innovation and economic development, highlighting the importance of spatial considerations and regional heterogeneity in promoting sustainable economic growth.

1. Introduction

Currently, the sustainable development of the global economy has shifted toward the “green” direction [1]. The National Development and Reform Commission and the Ministry of Science and Technology have jointly issued the “Guidance on the construction of a market-oriented green technology innovation system” to guide green technology innovation development. Green technology innovation is becoming an emerging field in the new round of the global industrial revolution and scientific and technological competition [2]. However, with the continuous advancement of China’s industrialization process over the past 40 years, the traditional extensive economic growth model has led to environmental pollution that needs to be effectively curbed [3], which is a key constraint to the high-quality development of the economy [4,5]. The Yangtze River economic belt, known as the “backbone of China’s economy”, faces a dilemma between economic development and the protection of resources and the environment due to its complex natural environment and potential ecological security problems [6]. Industrial green technology innovation is the endogenous driving force of green industrial development [7]. To transform the Yangtze River Economic Belt from a “catch-up orientation” to an “efficiency orientation” and promote the high-quality development of a green economy, it is crucial to abandon the traditional industrial development mode of “high pollution, high emission, and high energy consumption” [8] and utilize green technology innovation as a leading force in industrial economic development. Therefore, it is important to systematically and scientifically analyze the temporal and spatial evolution of green industrial innovation in the Yangtze River Economic Belt, explore the spatial convergence mechanism, and promote the market-oriented industrial green technology innovation system. This will help coordinate social, economic, and ecological benefits in the Yangtze River Economic Belt, with strategic significance.

Green technology innovation is a promising approach to promoting high-quality economic development by effectively coordinating economic growth with resources and the environment [9]. As Braun and Wield [10] pointed out, assessing the level of innovation in this field is a critical issue in academic research. Two widely used methods for evaluating innovation efficiency are stochastic frontier analysis (SFA), a parametric method suitable for measuring the efficiency of multiple inputs and a single output, and data envelopment analysis (DEA) [11]. This nonparametric method can deal with the efficiency evaluation of multi-input and multi-output and incorporate negative and undesirable outputs into the model. While SFA has limitations in measuring innovation efficiency when various expected and undesirable outputs accompany the output variable [12], DEA can effectively address this issue [13]. Some studies have applied DEA to evaluate the efficiency of green technology innovation in different contexts [14]. For example, some scholars have combined DEA with other methods to examine the relationship between financial development and green technology innovation [11,15,16]. In contrast, others have proposed effective improvement measures for renewable energy enterprises. Additionally, scholars have increasingly focused on the spatial characteristics of green technology innovation evolution, examining the spatial trends of green innovation efficiency across different regions in China [17]. Such research has employed a variety of models, including Super-SBM and nonlinear time-varying factor models, to explore the temporal and spatial evolution patterns of green technology innovation efficiency [2,18,19]. Overall, this body of research offers valuable insights into the challenges and opportunities associated with promoting sustainable economic development through green technology innovation [20].

The scientific community has extensively explored the efficacy of green technology innovation and its spatial development through theoretical and empirical analysis. This existing literature serves as a foundation for the present study, identifying two gaps in the current research. Firstly, assessing green technology innovation efficiency typically focuses on static evaluation and needs to incorporate dynamic evolution trend analysis. Secondly, although provincial and regional levels have been extensively studied, there needs to be more research on the spatial evolution characteristics of green technology innovation at the urban scale [7]. As research indicates that spatial distance impacts the relationship between factors, this study focuses on 110 prefecture-level cities and above in the Yangtze River Economic Belt from 2006 to 2022. Using the non-radial and non-angle SBM-GML model, the authors measure the efficiency of industrial green technology innovation and systematically analyze its dynamic evolution trend. Additionally, by drawing on Ma’s [21] research, the authors explore the transfer types and paths of regional innovation efficiency using the method of space-time transition. Lastly, this study incorporates spatial effects into the traditional convergence model to reveal the spatial convergence characteristics of industrial green innovation efficiency in the Yangtze River Economic Belt. These findings provide theoretical references and suggestions for achieving green and sustainable development in the region during the new era.

Based on existing research, this study makes the following specific contributions: (1) This study innovatively applies the SBM-GML model considering unexpected output to measure the efficiency of industrial green technology innovation in the Yangtze River Economic Belt. Compared with traditional efficiency measurement methods, the SBM-GML model can more accurately capture the environmental impact of technological innovation, providing a new perspective and tool for understanding and evaluating the efficiency of green technology innovation. (2) Through exploratory spatial data analysis (ESDA) and spatiotemporal transition methods, the spatiotemporal evolution characteristics of industrial green technology innovation efficiency in the Yangtze River Economic Belt were deeply explored. Identifying the differences and evolving trends in technological innovation efficiency among different regions is of great significance for understanding technological innovation’s synergistic effects and competitive relationships between areas. (3) Different policy recommendations have been proposed by emphasizing regional collaboration and cooperation, providing practical and feasible strategies for achieving green and sustainable development of the Yangtze River Economic Belt. This is an essential guiding value for policymakers and practitioners.

The rest of this article is structured as follows. The second part elaborates on the research methods and data explanation. The third part is the empirical results of this article. In the fourth part, the conclusions and inspirations were introduced, as well as the shortcomings of this study and its prospects for future research.

2. Research Methods and Data Description

2.1. SBM-GML Model Considering Unexpected Output

The concept of green technology innovation is a crucial driver for promoting high-quality development and the construction of ecological civilization, exemplifying the combination of “green” and “innovation”. In evaluating the efficiency of green technology innovation, it is crucial to consider factors such as slack variables, distinguishable decision units, and different radial and angle measures to avoid potential bias. Building on prior research [22,23,24], this study employs the Global Malmquist-Luenberger (GML) index to measure the efficiency of green industrial innovation in the Yangtze River Economic Belt. Specifically, the authors construct a non-radial and non-angle SBM model to capture the unique features of this context and provide a robust measure of efficiency.

The SBM model considering undesirable output is

$$\begin{array}{l}{\rho }^{*}=\min {\displaystyle \frac{1-\frac{1}{m}{\displaystyle \sum _{i=1}^m\frac{s_{x_{i0}}^{-}}{}}}{1+\frac{1}{s_1+s_2}\left({\displaystyle \sum _{r=1}^{s_1}\frac{s_r^g}{y_{r0}^g}+{\displaystyle \sum _{r=1}^{s_2}\frac{s_r^b}{y_{r0}^b}}}\right)}}\\ s.t.\left\{\begin{array}{l}{x}^0=x\lambda +s^{-}\\ y_0^g=y^g\lambda -s^g\\ y_0^b=y^b\lambda +s^b\\ s^{-}\ge 0,s^g\ge 0,s^b\ge 0,\lambda \ge 0\end{array}\right.\end{array}$$

(1)

where $s^{-}, s^g, s^b$ denote the slack of input variable, expected output, and unexpected output, respectively. ${\rho }^{*}$ is the objective function, a value range is [0, 1], when ${\rho }^{*}=1$, $s^{-}, s^g, s^b$ are 0, the decision-making unit is entirely efficient; when ${\rho }^{*}<1$ there are elements of redundancy in the decision-making unit, which can improve efficiency by optimizing configuration.

Building on the preceding analysis, it is evident that the Global Malmquist-Luenberger (GML) index represents a noteworthy improvement over the traditional Malmquist-Luenberger (ML) index and enables intertemporal comparison of green technology innovation efficiency [25]. Specifically, the GML index for period t to period $t+1$ is the result of ongoing efforts to refine and enhance measures of green technology innovation efficiency:

$$\begin{array}{l}GM{L}_{t+1}^t={\displaystyle \frac{1+{\theta }^G\left(x^t,y^t,a^t\right)}{1+{\theta }^G\left(x^{t+1},y^{t+1},a^{t+1}\right)}}={\displaystyle \frac{1+{\theta }^t\left(x^t,y^t,a^t\right)}{1+{\theta }^{t+1}\left(x^{t+1},y^{t+1},a^{t+1}\right)}}\times \\ \left[{\displaystyle \frac{1+{\theta }^G\left(x^t,y^t,a^t\right)}{1+{\theta }^G\left(x^t,y^t,a^t\right)}}\times {\displaystyle \frac{1+{\theta }^{t+1}\left(x^{t+1},y^{t+1},a^{t+1}\right)}{1+{\theta }^G\left(x^{t+1},y^{t+1},a^{t+1}\right)}}\right]=GE{C}_{t+1}^t\times GT{C}_{t+1}^t\end{array}$$

(2)

When $GM{L}_{t+1}^t>1$ indicates from t to t + 1 period of green technology innovation efficiency showed an upward trend; when $GM{L}_{t+1}^t<1$ indicates from t to t + 1 period of green technology innovation efficiency showed a downward trend; $GE{C}_{t+1}^t, GT{C}_{t+1}^t$, respectively, represent the changes in technical efficiency and technological progress from the t period to the t + 1 period.

2.2. Exploratory Spatial Data Analysis (ESDA)

The spatial autocorrelation test represents a crucial foundation for investigating the spatial-temporal transition and spatial convergence of industrial green technology innovation efficiency in the Yangtze River Economic Belt. To this end, exploratory spatial data analysis (ESDA) is applied, which draws on the first law of geography and places spatial correlation measurement at the forefront of its methodology. By describing the spatial distribution patterns of objects and revealing the spatial connections and mechanisms of interaction between them, ESDA enables the examination of spatial dependence and heterogeneity [26,27]. Global spatial autocorrelation and local spatial autocorrelation are generally employed to characterize the spatial correlation features of interest. Global spatial autocorrelation primarily reflects the spatial characteristics of the entire study area. It is typically measured using the global Moran’s I index, which can be calculated using the following formula:

$$I={\displaystyle \frac{m{\displaystyle \sum _{i=1}^m{\displaystyle \sum _{j=1}^m{\omega }_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}}}{S^{2}m{\displaystyle \sum _{i=1}^m{\displaystyle \sum _{j\ne 1}^m{\omega }_{ij}}}}}$$

(3)

$m$ represents the number of cities in the study area, $x_i, x_j$ represents the GML index of industrial green technology innovation in city $i$ and city $j$, respectively, $\overline{x}$ is the average value of the GML index of industrial green technology innovation in each city, $S^2={\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$m$}\right.}{\displaystyle \sum _{i=1}^m{\left(x_i-\overline{x}\right)}^2}$ represents the sample variance, ${\omega }_{ij}$ representation of space Adjacent weight matrix (When two cities are adjacent, ${\omega }_{ij}=1$, else ${\omega }_{ij}$ = 0), $I$. The value range of $I$ is [−1, 1], when $I>0$ represents a positive spatial correlation, when $I<0$ represents a negative spatial correlation, and when $I=0$, which means that the research samples are randomly distributed, and there is no spatial correlation.

In contrast to global spatial autocorrelation, local spatial autocorrelation is better suited to capturing the correlation between proximate spatial units and illuminating the heterogeneity of GML index spatial distribution concerning industrial green technology innovation efficiency in the region [28]. This approach is typically represented by the Local Indicators of Spatial Association (LISA), and Moran scatter diagrams. The formula for local spatial autocorrelation is as follows:

$$L_i={\displaystyle \frac{\left(x_i-\overline{x}\right)}{S^2}}{\displaystyle \sum _j{\omega }_{ij}\left(x_j-\overline{x}\right)}$$

(4)

$L_i$ represents local spatial autocorrelation index, the value range is [−1, 1] when $L_i>0$ it shows high or low agglomeration between the region and adjacent regions. There is a spatial agglomeration effect between cities with a high (low) GML index of green technology innovation. When $L_i<0$ it shows high and low agglomerations or low and high agglomerations between the region and adjacent regions. Cities with a high (low) GML index of green technology innovation are surrounded by towns with low (high) levels. When $L_i=0$ it shows that the spatial distribution of green technology innovation is independent of each other.

2.3. Spatial Convergence Model

(1)

Traditional Convergence Test Model

The initial convergence test has typically been employed in research on economic growth. This test indicates a negative correlation between static indicators of different economies at the outset of development under closed economic conditions. Specifically, backward regions experience higher economic growth rates than developed regions, leading to a gradual narrowing and eventual disappearance of the gap between different economies [29]. Following the principles of neoclassical economics [30], α convergence and β convergence are commonly used to test for convergence. ɑ convergence indicates that the degree of dispersion among different economies decreases over time [31]. β convergence, on the other hand, refers to a higher growth rate in regions with lower initial levels relative to economies with higher initial levels, leading to a narrowing of the gap through a “catch-up effect” until a unified steady state is achieved. β convergence can be categorized into absolute β convergence and conditional β convergence depending on the assumptions made [32]. The formula for β convergence is as follows:

$${\displaystyle \frac{1}{T}}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)=\alpha +\beta \ln GTI{E}_{it}+{\mu }_{it}$$

(5)

$${\displaystyle \frac{1}{T}}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)=\alpha +\beta \ln GTI{E}_{it}+\gamma X_{it}+{\mu }_{it}$$

(6)

Equations (5) and (6) are absolute and conditional $\beta$ convergence expressions, respectively. $GTI{E}_{it}$ represents the GML index of industrial green technology innovation in the phase $t$ of the $i$ region. $T$ is the period, $\alpha$ is a constant term, ${\mu }_{it}$ is a random error term, $\gamma$ is the coefficient of control variables, $X_{it}$ is control variable, this paper selects government intervention ($GOV$) [33]: Measured by the proportion of government fiscal technology expenditure to regional GDP, this indicator reflects the government’s support for technological innovation activities. It indirectly reflects the degree of intervention in industrial green technology innovation. Industrial structure ($IS$) [34]: Expressed as the proportion of the added value of the secondary industry to the regional gross domestic product, it is used to measure the position and structural status of the industry in the overall economy of the region environmental regulation ($ER$) [35]: Selecting the proportion of completed investment in industrial pollution control to industrial added value as a proxy variable for environmental regulation intensity, the higher the proportion, the higher the emphasis on industrial pollution control in the region, and the stricter the environmental regulation, Foreign Direct Investment ($FDI$) [36]: Measured by the proportion of actual utilization of foreign investment (converted into RMB at the current exchange rate) to the regional gross domestic product, it reflects the degree of regional economic openness to the outside world and the depth and breadth of participation in international economic cooperation. As control variables according to relevant literature. $\beta$ is the convergence coefficient, when $\beta <0$ and passes the dominance test, it indicates absolute (conditional) $\beta$ convergence. On the contrary, various regions’ GML industrial green technology innovation index shows divergent characteristics.

(2)

Spatial Convergence Test Model

The conventional convergence test is predicated on the assumption of spatial independence amongst individuals [37], which overlooks the spatial correlations resulting from the flow of elements in neighboring areas. The New Economic Geography, on the other hand, posits that economic development is influenced by its historical level of development in time and the economic development level of the adjoining regions in space [38]. Consequently, this study examines the significance of spatial factors in the conventional convergence test by constructing the spatial lag model (SLM) and the spatial error model (SEM) for the β convergence test. The expressions for the spatial lag model and the spatial error model of absolute β convergence are presented as follows:

$${\displaystyle \frac{1}{T}}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)=\alpha +\rho {\omega }_{ij}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)+\beta \ln GTI{E}_{it}+{\mu }_{it}$$

(7)

$${\displaystyle \frac{1}{T}}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)=\alpha +\beta \ln GTI{E}_{it}+{\left(\mathrm{I}-\lambda {\omega }_{ij}\right)}^{-1}{\mu }_{it}$$

(8)

The spatial lag model and spatial error model of conditional $\beta$ convergence are as follows:

$${\displaystyle \frac{1}{T}}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)=\alpha +\rho {\omega }_{ij}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)+\beta \ln GTI{E}_{it}+\gamma X_{it}+{\mu }_{it}$$

(9)

$${\displaystyle \frac{1}{T}}\ln \left({\displaystyle \frac{GTI{E}_{i,t+T}}{GTI{E}_{it}}}\right)=\alpha +\beta \ln GTI{E}_{it}+\gamma X_{it}+{\left(I-\lambda {\omega }_{ij}\right)}^{-1}{\mu }_{it}$$

(10)

Equations (7)–(10), $\rho$ and $\lambda$ denote the spatial lag and spatial error coefficients.

2.4. Indicator Selection and Data Source

The precise selection of variables is of utmost importance for accurately evaluating the efficiency of industrial green technological innovation in the Yangtze River Economic Belt. In previous studies, R&D personnel investment and R&D funding have been utilized as input variables, given their ability to fully capture the scale and potential of innovation in real-world settings [39,40]. However, when assessing the efficiency of green technology innovation, it is imperative to consider the “green” elements, and consequently, industrial energy consumption is incorporated into the input variables. The output variables consist of anticipated output and undesired output. To gauge the anticipated output, we have selected the number of effective invention patents, new product projects, and product sales income of large-scale industrial enterprises. Furthermore, following Wang Y’s [41] research, industrial wastes (such as wastewater, exhaust gas, and solid waste) are regarded as undesirable output in the index system.

This study focuses on analyzing 110 cities situated in the Yangtze River Economic Belt over a 15-year period ranging from 2006 to 2022. The input and output variables utilized in this research were obtained from several credible sources including the China Science and Technology Statistical Yearbook, the China Energy Statistical Yearbook, the China Industrial Statistical Yearbook, as well as the EPS. In this study, we used interpolation to fill in the data due to missing data in some cities, such as Bijie and Tongren. Specifically, linear interpolation is used. Linear interpolation is based on two known data points and estimates the value of missing data by constructing a linear function. This method assumes that the data changes before and after missing data points show a linear trend and uses this as a basis for filling in the data.

3. Empirical Results Analysis

3.1. Dynamic Evolution Analysis

This study employed the non-radial and non-angle SBM-GML model to assess the industrial green technological innovation efficiency of 110 cities in the Yangtze River Economic Belt. The MaxDEA Ultra 8.0 software was utilized to facilitate the measurement above. The obtained results, depicted in Figure 1, showcase the dynamic evolution trend of said efficiency.

During the study period, the average index for green technological innovation efficiency in the industrial sector of the Yangtze River Economic Belt, as measured by the GML index, was 1.054. As presented in Figure 1, this value was greater than 1, indicating an overall upward trend in the efficiency of green technological innovation during the study period, except for the periods of 2007–2008, 2008–2009, and 2015–2016, where the index was less than 1. The decline in green technological innovation efficiency during these periods could be attributed to the effects of the international financial crisis, which increased trade barriers and slowed the development of strategic emerging industries. Additionally, the lack of effective transformation of innovation resulted in a decline in green technology innovation during this time.

In response China has proposed measures to enhance the innovation and application capabilities of green technology, formulate a national action plan for green and low-carbon development innovation, improve the level of green technology re-search and development, and promote the industrialization support capacity of tech-nological innovation for green growth. However, the transformation of innovation achievements can only be achieved after a period of time, and enterprises need time to digest and absorb new technologies and management experience. Thus, there was a temporary decline in industrial green technology innovation efficiency in the Yangtze River Economic Belt.

Decomposition results showed that technological progress played a crucial role in driving the improvement of green technical innovation efficiency, as GTC was larger than GEC in every period except for 2015–2016, where the opposite was observed. The occurrence of technological progress benefits from multiple factors. Research and development investment is a necessary support. In the Yangtze River Economic Belt, enterprises and research institutions in Shanghai, Nanjing, and other places have invested significant funds in green technology research and development, spurring innovative achievements and effectively promoting technological progress. Talent mobility is equally indispensable. Taking Wuhan as an example, many relevant professional talents trained by universities flow to local enterprises or surrounding areas, promoting the diffusion of knowledge and technology and injecting vitality into green technology innovation for enterprises. In addition, technological innovation is disseminated between regions through various channels. Industrial interdependence plays an essential role in the chemical industry, where upstream enterprises innovate green production technologies, prompting downstream enterprises to adopt new technologies to ensure product quality and production efficiency and achieve technology transmission in the industrial chain. Regional economic cooperation is also a key channel for technology dissemination. The Yangtze River Delta region has promoted the diffusion of advanced green technologies from developed to relatively underdeveloped areas by jointly building cross-regional industrial parks, such as the Jiangsu Anhui Cooperation Demonstration Zone, which has improved the overall technological level of the region.

3.2. Spatial Correlation Testing

To investigate the spatial distribution characteristics of the GML index of industrial green technology innovation in the Yangtze River Economic Belt, this study utilizes global spatial correlation analysis and local spatial correlation analysis of exploratory spatial data analysis (ESDA). This methodology is employed to examine the spatial agglomeration characteristics of the GML index.

The present study utilizes the global Moran’s I value and the 1% aboriginality level test, as presented in

Table 1. Global Moran’s I and its test results.

Year	2007	2008	2009	2010	2011	2012	2013	2014
Moran’s I	0.19	0.20	0.21	0.23	0.23	0.22	0.24	0.24
Z	0.40	3.54	3.69	3.95	3.96	3.78	4.08	4.14
P	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Year	2015	2016	2017	2018	2019	2020	2021	2022
Moran’s I	0.22	0.23	0.22	0.23	0.25	0.25	0.21	0.24
Z	3.89	3.98	3.78	3.97	4.29	4.24	3.95	4.13
P	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00

, to analyze the spatial distribution characteristics of the GML index of industrial green technology innovation in the Yangtze River Economic Belt. The results show a positive spatial correlation between the efficiency of industrial green technology innovation in the region, suggesting a non-random distribution in space. The spatial correlation of the GML index fluctuates between 0.19 and 0.25 over the study period, indicating strengthened exchanges and cooperation between regions. Moreover, the spatial convergence effect of the industrial green technology innovation efficiency in the Yangtze River Economic Belt cities is gradually strengthened, as demonstrated by the four spatial distribution patterns (H-H agglomeration, L-H agglomeration, L-L agglomeration, and H-L agglomeration) shown in Figure 2. The analysis reveals that 69 cities belong to H-H agglomeration and L-L agglomeration, accounting for 62.7% of the total, while H-L agglomeration and L-H agglomeration are relatively sparse. These findings provide further verification of the spatial agglomeration characteristics of the GML index of industrial green technology innovation efficiency in the Yangtze River Economic Belt.

3.3. Analysis of Temporal and Spatial Dynamic Evolution of LISA

(1)

LISA Time Path

The LISA time path, a novel approach to traditional static LISA analysis, incorporates time dimension and visualizes the research object’s observation value and spatial lag effect. It can be regarded as a type of spatial Markov transfer, providing insights into the spatial-temporal collaborative change in the GML index of industrial green technology innovation across different cities at the regional level, as well as local spatial variations and their temporal dynamics [42]. Two key components of the LISA time path are its length and curvature, wherein the former reflects the dynamic nature of local spatial structure [43,44]. The formula for computing the LISA time path is as follows:

$$L={\displaystyle \frac{M{\displaystyle \sum _{t=1}^{T-1}d\left(L_{i,t},L_{i,t+1}\right)}}{{\displaystyle \sum _{i=1}^M{\displaystyle \sum _{t=1}^{T-1}d\left(L_{i,t},L_{i,t+1}\right)}}}}$$

(11)

LISA time path curvature reflects the fluctuation characteristics of local spatial structure. The formula is as follows:

$$\delta ={\displaystyle \frac{{\displaystyle \sum _{t+1}^{T-1}d\left(L_{i,t},L_{i,t+1}\right)}}{d\left(L_{i,1},L_{i,T}\right)}}$$

(12)

In Equations (11) and (12) $M=110$, is the number of research units, $T$ is a time interval, $L_{i,t}$ is the LISA coordinates of the study unit $i$ in a year $t$, and $d$ is the moving distance of the research unit in two periods. The larger the $L$ is, the stronger the local spatial dynamics are when $L>1$ the moving distance of the research unit $i$ is greater than the average urban moving distance. The larger the $\delta$ is, the greater the volatility of local spatial structure is, and the more curved the LISA time path is. When $\delta >1$ the moving bending degree of the research unit $i$ is greater than the average urban moving bending degree, in this paper, the relative length and curvature of the LISA time path are divided into four grades by using the natural breakpoint method: low relative length (curvature), low relative length (curvature), and high relative length (curvature) and high relative length (curvature).

The LISA time path, which integrates the time element into the traditional static LISA analysis and visualizes the observation value and spatial lag effect of the research object, is a useful tool for understanding the spatial-temporal collaborative change in the GML index of industrial green technology innovation at both the regional and local spatial levels. This tool includes both the length and curvature of the LISA time path, with the length reflecting the dynamic characteristics of local spatial structure(Figure 3). Our analysis shows that 29 cities, accounting for 26.4% of the total, have a high relative path length, indicating that the overall spatial pattern of the GML index of industrial green technology innovation efficiency in the Yangtze River Economic Belt is relatively stable. Cities with a considerable relative length of time path are primarily located in the upper reaches of the Yangtze River Economic Belt in areas such as Yunnan, Guizhou, and Sichuan, while cities with a small relative length of time path are mainly distributed in the coastal regions of the lower reaches of the Yangtze River and central cities such as Chongqing, Chengdu, Changsha, and Wuhan. The stability of the local spatial system in these areas is due to their favorable economic foundations and industrial structures, which provide an excellent innovation environment for the development of industrial enterprises.

Additionally, 23 cities, accounting for 20.9% of the total, have a high curvature from the LISA time path curvature, and these cities are evenly distributed in the three regions, where the industrial structure is relatively simple. For example, Hangzhou is dominated by the internet, Changde is dominated by equipment manufacturing, and Yunnan and Guizhou are dominated by agriculture. These cities have significant volatility in spatial dependence direction and have non-dynamic changes with neighboring cities. Moreover, there are 41 cities in the positive coordinated transition (0°−90°) and 28 cities in the negative collaborative change (270°−360°), accounting for 62.6% of the total cities, indicating that the spatial pattern evolution of the GML index of industrial green technological innovation efficiency in the Yangtze River Economic Belt has a solid spatial integration.

(2)

LISA Spatial-time Transitions

The present study delves into the spatiotemporal transition phenomenon that characterizes the migration of diverse local correlation types in the local Moran’s I scatter diagram. Furthermore, it elucidates the spatial and temporal variations in the spatial structure of the GML index concerning industrial green technology innovation in the Yangtze River Economic Belt. Drawing on the space-time transition theory postulated by Rey, the current investigation classifies the space-time transition into four distinct types (refer to

Table 2. LISA space-time transition types.

Type	Meaning	Symbolic Expression
I	Self-transition, neighborhood stability	HHt → LHt+1, HLt → LLt+1, LHt → HHt+1, LLt → HLt+1
II	Self-stability, neighborhood transition	HHt → HLt+1, HLt → HHt+1, LHt → LLt+1, LLt → LHt+1
III	Self-transition, neighborhood transition	HHt → LLt+1, HLt → LHt+1, LLt → HHt+1, LH t → HLt+1
IV	Self-stability, neighborhood stability	HHt → HHt+1, HLt → HLt+1, LLt → LLt+1, LHt → LHt+1

). The local Moran’s I space cohesion is computed using the following formula:

$$C={\displaystyle \frac{F_{0,t}}{n}}$$

(13)

$C$ is the spatial cohesion, the value range [0, 1], the greater the value, indicating that the higher the spatial cohesion, the greater the obstacle to the transition. $F_{0,t}$ is the number of type IV transitions in $t$ time. $n=\left(2022-2007\right)\times 110=1430$ is the number of research units for all possible transitions.

The findings from the study reveal that during the research period, the probability of type IV spatial-temporal transition of the GML of industrial green technology innovation efficiency in the Yangtze River Economic Belt is 74.7%, as per the spatial probability transfer matrix presented in

Table 3. Local Moran’s I transition probability matrix.

	HHt+1	HLt+1	LLt+1	LHt+1
HHt	IV (0.371)	II (0.034)	III (0.021)	I (0.014)
HLt	II (0.019)	IV (0.105)	I (0.028)	III (0.025)
LLt	III (0.006)	I (0.019)	IV (0.113)	II (0.036)
LHt	I (0.003)	III (0.025)	II (0.023)	IV (0.158)

. Additionally, the high path-locking characteristics of spatial cohesion indicate that the spatial structure of industrial green innovation efficiency in the Yangtze River Economic Belt remains relatively stable with minimal transitions between different types. Furthermore, there is evidence of transition inertia among different urban types. Among other changes, the LL_t → LH_{t + 1} probability transfer records the highest value at 0.036. This observation indirectly indicates that the spatial structure of industrial green technological innovation efficiency in the Yangtze River Economic Belt is inclined towards stability.

The investigation yields that the efficiency of industrial green technology innovation in the Yangtze River Economic Belt evinces pronounced spatial stability, primarily in the coastal cities of the Yangtze River Delta and the metropolitan regions centered on Wuhan, Changsha, and Chongqing. These areas serve a critical function in the propagation of the economy and resources in the surrounding regions, with a relatively short length and curvature of the time path and a cooperative, affirmative transition mode. Notably, the development of metropolitan areas has partially dismantled administrative barriers over an extended period, fostered the logical flow of factors across regions, and reinforced regional spatial linkage development. Under the stimulation of the economic belt, metropolitan area, and growth pole, underdeveloped cities can surmount their developmental impasses and realize the synchronized development of regional green innovation efficiency. Particularly in the context of the “joint efforts to protect and not to develop”, the significant urban agglomerations in the Yangtze River Economic Belt have taken the lead in coordinating and promoting scientific and technological progress, institutional innovation, industrial structure upgrading, and green development. The outcome has strengthened the regional spatial agglomeration effect, resulting in a relatively stable spatial dependence of the GML index of industrial green technical innovation efficiency in the Yangtze River Economic Belt, demonstrating convergence characteristics.

3.4. Spatial Convergence

Based on the spatial correlation test, the GML index of industrial green technology innovation in the Yangtze River Economic Belt exhibits salient spatial correlation features. However, conventional β convergence models may not fit the sample data well. As such, this study employs a spatial econometric model based on the traditional β convergence model analysis to examine the spatial convergence characteristics of the GML index of industrial green technology innovation in the Yangtze River Economic Belt. According to the judgment principle proposed by Anselin [45], the Lagrange multipliers LM-log and LM-err of both the spatial lag model and the spatial error model pass the 1% level of dominance test, making it difficult to determine which model to choose. Nonetheless, the spatial error model is selected after comparing the dominance of Robust LM-lag and Robust LM-err. Moreover, through the Hausman test, both the spatial lag model and the spatial error model fail to pass the explicit level test, and the original hypothesis of random effect cannot be rejected. Therefore, the spatial error model of random effect is employed to test the spatial convergence of the GML index of industrial green technology innovation in the Yangtze River Economic Belt.

(1)

Absolute β convergence test

Based on the findings in

Table 4. Absolute $\beta$ convergence results of GML index of green industrial innovation in the Yangtze River Economic Belt.

Variables	Absolute $\mathit{\beta }$ Convergence (Model 5)				Spatial $\mathit{\beta }$ Convergence (Model 8)
	Mean Value	Upper	Middle	Lower	Mean Value	Upper	Middle	Lower
$\beta$	−0.013 *** (−2.74)	−0.021 *** (−3.17)	−0.062 *** (−4.85)	−0.003 *** (−2.31)	−0.025 *** (−3.62)	−0.031 *** (−3.81)	0.004 *** (2.53)	−0.016 *** (−3.02)
$\alpha$	0.003 *** (2.26)	0.002 *** (2.03)	0.136 *** (3.95)	0.003 *** (2.27)
$\lambda$					0.315 *** (5.52)	0.217 *** (4.18)	0.235 *** (4.62)	0.308 *** (5.35)
$s$ (%)	0.062	0.182	0.372	0.014	0.075	0.064		0.037
$\tau$	8.34	5.3	2.5	27.2	7.53	5.2		21.6
$R^2$	0.1081	0.4165	0.3082	0.0932	0.2364	0.0825	0.2841	0.5031
$\log L$	321.516	305.264	276.012	230.926	474.283	405.304	362.307	307.136

Notes: *** represent p < 0.01, in parentheses are t statistics.

, both the traditional β convergence model and the spatial β convergence model show negative β coefficients at a significance level of 1%. This indicates that the GML index of industrial green technology innovation in the Yangtze River Economic Belt exhibits significant absolute β convergence and specific spatial dependence. It demonstrates a “catching up later” trend in backward areas. The green technology innovation efficiency in a specific region can affect the surrounding areas’ efficiency level through spatial effect, eventually converging to a unified steady state. Through a comparison of the convergence rate (s) and half-life cycle (τ) of the traditional β convergence model and the spatial β convergence model, the study found that after incorporating the spatial error model, the convergence rate increased from 0.062% to 0.075%, and the half-life cycle decreased from 8.34 to 7.53. These results demonstrate that the spatial spillover effect strengthens the flow of elements and technical cooperation between adjacent regions, promotes the coordinated development of cities, reduces the convergence time, and verifies the introduction of the spatial effect’s applicability.

In terms of subregional analysis, this study found that the GML index of industrial green technology innovation in the Yangtze River Economic Belt regions exhibits absolute β convergence without considering the spatial effect. Specifically, the coefficients of the upper, middle, and lower reaches estimated by the traditional β convergence model are all less than 0 at the 1% significance level, suggesting a trend of “catching up later” in backward areas. However, after introducing the spatial effect in the spatial β convergence model, the downstream and upstream regions of the Yangtze River Economic Belt maintain good convergence characteristics, with their β coefficients all less than 0 at the 1% significance level, indicating that they converge to a steady-state level for regional coordinated development. The convergence rate of the downstream is significantly higher than that of the upstream, which may be attributed to the strong government support for green technology innovation and the implementation of the Western Development Strategy. Meanwhile, the coefficient of spatial β convergence in the middle reaches of the Yangtze River Economic Belt is positive, indicating a lack of β convergence characteristics. This may be attributed to the spatial non-equilibrium features of “core-periphery” and the “siphon effect” of the central city, which attract more vital elements to the local area, thus widening the difference between regions and causing the GML index of industrial green technology innovation in the middle reaches of the Yangtze River Economic Belt to exhibit diffusion characteristics. The convergence speed also shows the characteristics of “middle > upper > lower”.

(2)

Condition β Convergence

Table 5. Condition $\beta$ convergence results of GML index of green industrial innovation in the Yangtze River Economic Belt.

Variables	Absolute $\mathit{\beta }$ Convergence (Model 6)				Spatial $\mathit{\beta }$ Convergence (Model 10)
	Mean Value	Upper	Middle	Lower	Mean Value	Upper	Middle	Lower
$\beta$	−0.035 *** (−4.17)	−0.031 *** (−3.82)	−0.157 *** (−5.26)	−0.045 *** (−4.58)	−0.042 *** (−4.35)	−0.102 *** (−4.93)	0.036 *** (2.75)	−0.081 *** (−4.87)
$\alpha$	0.014 *** (2.73)	0.024 *** (3.15)	0.162 *** (4.25)	0.027 *** (3.23)	-	-	-	-
$\lambda$	-	-	-	-	0.317 *** (5.71)	0.146 *** (3.32)	0.105 *** (3.27)	0.231 *** (4.31)
$GOV$	0.015 *** (2.80)	0.043 *** (3.25)	0.015 *** (2.81)	0.104 *** (3.18)	0.036 *** (3.08)	0.023 *** (2.97)	-	0.236 *** (4.67)
$IS$	0.006 (1.32)	−0.006 ** (−2.43)	−0.013 *** (−2.83)	0.184 (1.61)	0.217 (1.72)	−0.003 *** (−2.32)	-	0.046 (1.52)
$ER$	0.046 (1.51)	0.142 *** (3.68)	0.007 *** (2.73)	0.163 *** (4.05)	0.285 *** (5.36)	0.075 *** (3.32)	-	0.148 *** (3.46)
$FDI$	0.032 ** (3.26)	0.027 (1.02)	0.012 *** (2.53)	0.421 *** (6.32)	0.051 * (2.03)	0.004 (1.27)	-	0.537 *** (6.86)
$s$ (%)	0.083	0.182	0.372	0.014	0.097	0.064	-	0.037
$\tau$	7.05	4.7	2.1	23.5	6.48	4.2		13.4
$R^2$	0.3725	0.4632	0.4075	0.2106	0.4016	0.2572	0.3753	0.5706
$\log L$	375.236	362.107	331.274	279.072	573.106	446.385	402.636	361.532

Notes: ***, **, and * represent p < 0.01, p < 0.05, and p < 0.1, respectively, in parentheses are t statistics.

presents the estimation results of the conditional β convergence of industrial green innovation GML index in the Yangtze River Economic Belt, which was tested at a 1% aboriginality level. The β coefficient is negative, indicating that the study sample will eventually converge to a steady state under the influence of external factors. By comparing the convergence speed and half-life cycle of conditional β convergence and absolute β convergence, it is found that after controlling for other variables, the rate of conditional convergence is significantly higher than that of absolute convergence, and the time to reach a steady state is shorter than that of absolute convergence. This is because the conditional β convergence model considers the heterogeneity of economic development levels, industrial structures, and environmental regulations among different regions in the Yangtze River Economic Belt, which improves the convergence rate and makes the convergence result more reliable.

The results show that the GML index of industrial green technology innovation in the upper, middle, and lower reaches of the Yangtze River Economic Belt has conditional β convergence characteristics. Specifically, the estimation results of the traditional β convergence model indicate that the β coefficients of the three regions are significantly negative. However, after adding the spatial effect, the β coefficient of the upper and lower reaches of the Yangtze River Economic Belt is significantly negative, indicating the existence of conditional β convergence. In contrast, the middle reaches of the Yangtze River Economic Belt do not exhibit conditional β convergence, as they fail to meet the level of visibility. The regression results are similar to the absolute β convergence model. Compared with the absolute β convergence, the log-likelihood test value log-L and the goodness of fit R² of the Yangtze River Economic Belt and the three regions are improved, suggesting that the conditional β convergence model has good explanatory power. Therefore, the conditional β convergence model is recommended to test the trend of the GML index of industrial green technology innovation in the Yangtze River Economic Belt.

Within the Yangtze River Economic Belt, the GML index of green technology innovation is observed to converge due to government intervention in the control variables, primarily through strong support for high-tech industries. This is achieved through various financial support and incentive policies for independent research and development of enterprises, leading to improved technological innovation efficiency and quicker convergence. However, the convergence of industrial structure on the GML index of industrial green technology innovation needs to pass the visibility test, potentially due to differences in economic development and industrial structure levels across the region. Upgrading the industrial structure may promote regional innovation efficiency but may hinder the rational allocation of production factors, inhibiting progress in innovation efficiency.

In the traditional β convergence model, environmental regulation does not pass the test of explicitness level. However, the spatial β convergence model is positive, indicating that ecological law promotes the convergence of the GML index of industrial green technology innovation in the Yangtze River Economic Belt under spatial effect. This supports Porter’s [46] hypothesis that appropriate environmental regulation policies can compensate for additional costs caused by ecological regulation and form an “innovation compensation effect”.

Positive coefficients of openness suggest that opening to the outside world can introduce advanced technical knowledge and management experience, thereby improving innovation efficiency through technology learning and resource sharing and promoting convergence of the GML index of green technology innovation.

In the upstream region of the three Yangtze River Economic Belt regions, government intervention and environmental regulation play a positive role in promoting convergence, while industrial structure inhibits it. The downstream region has significant positive coefficients of government intervention, environmental regulation, and opening-up, indicating that the three promote convergence of the GML index of industrial green technology innovation. The downstream area’s unique geographical advantages and rapid economic development are conducive to attracting foreign investment, and government support can improve green technology innovation efficiency, reduce differences between cities, and promote convergence of the GML index of green technology innovation.

4. Discussion

In terms of research methodology, this study uses the SBM-GML model to consider unexpected outputs, thereby accurately measuring the efficiency of industrial green technology innovation. Compared with traditional research methods, this model can more effectively assess the environmental impact of technological innovation [47,48], providing a new perspective and powerful tool for studying the efficiency of green technological innovation. Although some previous studies have used DEA and other methods to evaluate the efficiency of green technology innovation, handling unexpected outputs is not comprehensive enough to fully reflect the actual efficiency of technological innovation [49,50]. In addition, this study comprehensively utilizes exploratory spatial data analysis (ESDA) and spatiotemporal transformation methods to deeply explore the spatiotemporal evolution characteristics of industrial green technology innovation efficiency, which is rare in previous research on urban scales. Most existing studies have focused on the provincial or regional level [51], with insufficient research on the spatial evolution characteristics at the metropolitan scale.

From the research results, this study found that the overall efficiency of industrial green technology innovation in the Yangtze River Economic Belt is on the rise, and technological progress is the key driving force. This is consistent with some existing research findings that technological progress is essential in promoting the innovative development of green technologies [52,53]. However, this study further revealed that the efficiency exhibits significant spatial dependence and stability in spatial and temporal conversion, enriching our understanding of the spatial characteristics of green technology innovation efficiency [54]. Regarding spatial convergence, research has shown that the GML index for industrial green technology innovation in the Yangtze River Economic Belt exhibits absolute β and conditional β convergence but shows significant heterogeneity in different regions. Among them, the midstream region exhibits diffusion characteristics in both absolute β convergence and conditional β convergence due to the spatial imbalance between the “core edge” and the “siphon effect” of the central city, which is in sharp contrast to the upstream and downstream regions. Previous studies have rarely conducted such detailed analyses on the differences in spatial convergence characteristics among different Yangtze River Economic Belt regions.

In addition, this study considers spatial dependence when analyzing the regional differences in industrial green technology innovation efficiency in the Yangtze River Economic Belt. However, further exploration of its fundamental reasons, especially the influence of local political, cultural, institutional, and other factors, is needed. Politically, different levels of government emphasis and policy support for green development can lead to differences in the innovation environment for enterprises and affect innovation efficiency. In terms of culture, the innovative cultural atmosphere and public recognition of green products can promote or hinder the innovation of green technology in enterprises. Institutional factors are equally crucial. A sound intellectual property protection and efficient system for the transformation of scientific and technological achievements can promote the improvement of innovation efficiency, while the opposite will constrain it. These factors play a significant role in regional differences, and future research should comprehensively consider their interactions, construct more comprehensive theoretical and analytical frameworks, and help promote the coordinated development of green technology innovation in the Yangtze River Economic Belt. Next are key stakeholders such as businesses and communities. As the main innovation body, large enterprises actively invest in green technology research and development with their financial and technological advantages, such as automobile manufacturing enterprises researching and developing new energy technologies. Small and medium-sized enterprises are more cautious about innovation due to a shortage of funds and talent. At the community level, communities with strong environmental awareness will encourage enterprises to accelerate the application of green technologies, such as chemical enterprises increasing environmental protection investment under the supervision of residents. Some communities that do not understand or are concerned about the impact of green technology innovation on employment and the economy will hinder corporate innovation.

5. Conclusions and Implications

This paper utilizes a panel dataset of 110 prefecture-level cities in the Yangtze River Economic Belt from 2006 to 2022 to assess the efficiency of industrial green technology innovation. The study constructs the SBM-GML model, which considers unexpected output, to measure the efficiency of industrial green technology innovation and analyzes its dynamic evolution characteristics. Spatial pattern and spatio-temporal transition characteristics of the GML index of industrial green technology innovation are explored using the ESDA analysis and Spatio-temporal transition method. The study then applies the spatial convergence model to analyze the absolute β convergence and conditional β convergence of the GML index of industrial green and innovation in the Yangtze River Economic Belt.

The study finds that the overall level of innovation efficiency has improved, and technological progress is the key factor driving green technology innovation efficiency improvement. The GML index of industrial green technology innovation in the Yangtze River Economic Belt shows strong spatial stability and integration by analyzing the spatial and temporal transition types. The convergence rates in different regions show “middle > upper > lower” characteristics. After adding the spatial effect, only the upstream and downstream regions show indigenous convergence, but the convergence rate is higher than in the traditional convergence model.

Moreover, the study discovers that the GML index of industrial green technology innovation in the middle reaches of the Yangtze River Economic Belt shows diffusion characteristics due to the region’s spatial non-equilibrium characteristics of “core-periphery”. For the control variables, external factors have significant differences in the convergence of the GML index of industrial green technology innovation in different regions because of the differences in geographical location, resource endowment, and industrial structure in the three major regions of the Yangtze River Economic Belt. Overall, this study offers insights into the efficiency of industrial green technology innovation and its spatial convergence characteristics in the Yangtze River Economic Belt.

Drawing on the research findings, this paper suggests several policy implications for promoting industrial green technology innovation in the Yangtze River Economic Belt. Firstly, breaking administrative barriers and prioritizing the spatial linkage of regional green technology innovation is essential. Cross-regional green technology innovation learning platforms should be constructed to facilitate regional technology exchange and innovation cooperation, and to establish mechanisms for resource sharing to achieve coordinated urban green technology innovation development. Coastal and central cities in the Yangtze River Economic Belt should be given special attention for their role in promoting green technology innovation efficiency in surrounding areas, and to enhance the “catch-up effect” of “efficiency depression”. For example, coastal cities in the Yangtze River Delta, such as Shanghai, Nanjing, Hangzhou, etc., have strong green technology innovation capabilities and resource advantages. They should lead and drive surrounding cities to improve the efficiency of green technology innovation through technology output, industrial cooperation, and other means.

Secondly, differentiated development strategies should be implemented based on each region’s specific characteristics and needs. Given the differences in resource endowments and economic development, each region should develop green technology innovation measures tailored to their resource advantages. The policy of industrial restructuring also faces challenges in promoting industrial green technology innovation. When promoting the green transformation of traditional industries, enterprises often face technical difficulties and funding bottlenecks. Due to the lack of technological accumulation and professional talents, traditional industrial enterprises struggle to achieve green upgrading smoothly. The shortage of funds also limits the introduction of advanced technology and equipment and the development of research and development activities by enterprises, resulting in obstacles to the process of industrial structural adjustment and difficulty in fully reflecting the promotion of industrial green technology innovation. For instance, downstream regions should actively undertake industrial transfer. In contrast, upstream and midstream regions should focus on strengthening the research and development of independent industries, eliminating backward production capacity, and promoting green technological innovation achievements. Upstream regions such as Yunnan and Guizhou have abundant natural resources, but their economic development level is relatively low, and their industrial foundation is relatively weak. We should rely on local resource advantages and increase support for green technology research and development in characteristic industries. For example, Yunnan has unique advantages in the flower industry, which can focus on the research and development of green cultivation techniques and flower preservation technologies in the flower planting process, reduce the use of pesticides and fertilizers, lower the impact on the environment, and improve the quality and yield of flowers, promoting the green upgrading of the flower industry and in addition, phasing out outdated production capacity, strengthening ecological environment protection, and achieving green development. With its geographical advantages, the downstream region should shift from “attracting investment” to “attracting investment and selecting capital” to improve its green technology innovation level and national competitiveness. Taking Suzhou as an example, in recent years, Suzhou has actively introduced high-tech, green, and environmentally friendly enterprises and rejected high pollution and high-energy consumption projects.

Thirdly, for policymakers to further promote industrial green technology innovation in the Yangtze River Economic Belt and achieve adequate control of energy consumption and emissions, the following suggestions are proposed: the government should fully play the role of macroeconomic regulation and formulate differentiated policies. On the one hand, differentiated energy consumption and emission standards and environmental regulatory policies should be formulated based on the industrial structure and development level of different regions. Implementing strict energy consumption standards and environmental supervision in areas with concentrated high energy-consuming industries forcing enterprises to upgrade their technology. For regions with rapidly developing emerging industries and developed economies, appropriate optimization of standards and regulatory flexibility should be implemented to encourage innovation while ensuring environmental quality. For economically underdeveloped areas, while ensuring the ecological bottom line, policy support and technical guidance should be provided to reduce the cost of green transformation for enterprises. On the other hand, we will increase support for the research and application of green energy, establish special funds, encourage enterprises and research institutions to study and apply clean energy technologies such as solar energy, wind energy, and hydropower, and optimize the energy structure. In addition, establish a carbon emissions trading market and improve regulatory mechanisms, reward enterprises with lower carbon emissions through market means, punish enterprises that exceed standards, and encourage enterprises to save energy and reduce emissions actively. At the same time, we can learn from the experience of establishing a special fund for green technology innovation in Chongqing, providing financial subsidies and tax incentives for enterprises, encouraging enterprises to increase innovation investment in energy conservation, emission reduction, resource recycling, and other fields, promote the dual wheel drive of “green development” and “innovative development” in the Yangtze River Economic Belt, and achieve the coordinated progress of industrial green technology innovation and sustainable development.

6. The Shortcomings and Prospects of This Study Are as Follows

The present study examines a sample of 110 prefecture-level cities in the Yangtze River Economic Belt from 2006 to 2022. This study used the interpolation method to supplement data for cities such as Bijie and Tongren, where there was missing data. Although this is a common approach to addressing data loss, its impact on research results still needs further exploration to assess potential biases accurately. In dynamic evolution analysis, the lack of data in these cities may lead to overestimation or underestimation of the efficiency value of industrial green technology innovation in specific years, affecting the accurate judgment of the reasons for efficiency changes; In terms of spatial correlation testing, it can interfere with the judgment of real spatial correlations between cities, causing deviations in spatial autocorrelation analysis results and affecting the evaluation of spatial agglomeration effects; During the analysis of spatiotemporal dynamic evolution, errors may occur in the results of LISA time path and spatiotemporal transformation analysis due to missing data, which can alter the judgment of spatial pattern stability and collaborative changes between cities; In spatial convergence analysis, data loss can affect the judgment of convergence characteristics and regional differences, leading to estimation biases in convergence speed, characteristics, and regional differences. Although interpolation has a specific effect, to reduce potential bias, future research can use various interpolation methods for comparison or combine them with other supplementary data sources for comprehensive judgment to improve research conclusions’ reliability.

In terms of research content, this study lacks sufficient exploration of the effectiveness of green technology innovation in achieving environmental and social goals. In the future, a multi-objective evaluation system can be established to comprehensively evaluate the contributions of industrial green technology innovation in energy conservation and emission reduction, resource utilization efficiency improvement, employment creation, social equity, and other aspects. Furthermore, this study did not delve into the differences and collaborative mechanisms among enterprises of different scales in industrial green technology innovation. Subsequent research can compare the differences in innovation input, output, obstacles faced, and response strategies among large, medium, and small enterprises and explore effective models for promoting innovation collaboration among enterprises.