Dynamic Interactive Effects of Technological Innovation, Transportation Industry Development, and CO₂ Emissions

Abstract

This paper aims to clarify the intricate relationships between technological innovation, transportation industry development, and CO₂ emissions to facilitate a positive synergy among technology, the economy, and climate, advancing the fulfillment of the ‘double carbon’ goal. Utilizing panel data from 30 provinces in China from 2005 to 2020, we employ the panel vector autoregressive model using a generalized method of moments to empirically examine the dynamic interactive effects between these participants. The findings reveal that the transportation industry significantly promoted the inhibitory impact of technological innovation on CO₂ emissions. However, such reductions cannot counterbalance the rise in emissions from the transportation industry. Moreover, its effects varied significantly across regions. Specifically, transportation industry development within eastern China contributed to a shift in the local carbon emission effects from positive to negative under the positive influence of technological innovation. In the northeast, the transportation industry enhanced the inhibitory effect of technological innovation on CO₂ emissions. In contrast, across the western region, industrial development in transportation intensified the role of technological innovation in promoting CO₂ emissions. Furthermore, this work found that CO₂ emissions notably diminished the CO₂ reduction performance of technological innovation in the eastern part and enhanced this performance in the northeastern region. These findings further revealed the complex interplay between technological innovation, the transportation industry, and CO₂ emissions. They offer insights for policymakers to tailor region-specific technologies to bolster the ‘dual carbon’ goal and sustainable transportation development strategies, thereby achieving CO₂ reduction.

1. Introduction

In the 21st century, climate change and resource scarcity have emerged as dual challenges worldwide. China, as a responsible member of the international community, bears significant responsibility for spearheading the global ‘green economic recovery’ [1]. The International Energy Agency reports that China’s transportation industry ranks as the third largest contributor to national greenhouse gas emissions, following the industrial and construction sectors. Reforming transportation infrastructure, network design, and service quality can impact 80% of direct and indirect CO₂ emissions, playing a pivotal role in achieving China’s ‘dual carbon’ goal [2,3]. Furthermore, technological innovation-driven progress is vital in addressing environmental pollution and mitigating CO₂ emissions [4]. Prioritizing technological innovation has become a key strategy for China to foster a low-carbon economy, establish a resource-efficient and eco-friendly society, and enhance national competitiveness [5]. In sum, transportation industry development and technological innovation are integral to CO₂ reduction. In this sense, exploring their incorporation into the same framework and dynamic interactive effects with CO₂ emissions is necessary, thereby guiding the endeavors targeting addressing the climate challenge of high CO₂ emissions in China.

For the above purpose, we reviewed the existing literature on this topic. Regarding research on technological innovation and CO₂ emissions, much attention is placed on decarbonization technologies, such as renewable energy development [6], negative emission technology [7], and CO₂ capture and storage [8]. In addition, an increasing number of researchers are exhibiting interest in the impact of technological innovation on the first-level management efficiency and energy consumption of sectors, i.e., the construction industry [9] and the transportation industry [10]. In studies on the transportation industry and CO₂ emissions, CO₂ emission measurement, influencing factors, and reduction strategies within one sector are classic directions. These studies typically focus on individual development elements, including the transport infrastructure scale [11], transport capacity [12], and transport service quality [13]. Furthermore, a limited number of studies have also explored the effect of these individual development elements within the transportation industry on urban CO₂ emissions [14]. In the realm of innovation and transportation industry development, factors such as transportation infrastructure [15] and transportation capacity [16] are essential for attracting innovation resources (i.e., urban technological talents and R&D institutions). Technological innovation and management innovation are viewed as fundamental pillars for the research, development, and implementation of intelligent transportation systems, contributing to efficient traffic management and the transformation of transportation energy [17].

Although previous research has extensively examined the relationships between technological innovation, transportation industry development, and CO₂ emissions, several areas remain underexplored. Firstly, the published works do not consider the interactive effects of technological innovation and transportation industry development on CO₂ emissions. Accordingly, the potential synergy between the two factors on CO₂ emissions is understudied, leading to deviations in policy formulation and implementation effects [18]. Secondly, due to neglecting the potential boomerang effect of CO₂ emissions on technological innovation and the transportation industry, discovering more effective solutions to CO₂ emission reductions is hindered [19]. Furthermore, a research gap exists between the development of the overall transportation industry and urban CO₂ emissions. Analyzing this relationship solely relying on a single development factor may fall short of unveiling the transportation industry’s effect on CO₂ emissions since this method may underestimate the positive externalities of transportation in reducing urban CO₂ emissions [20]. Moreover, previous research predominantly focuses on global, national, or urban single samples, neglecting regional heterogeneity. Given China’s vast territory and diverse natural resources, geographical features, and social and economic statuses across regions, overlooking these differences can impede the effective application of macro policies at a national level [21].

Based on the above analysis, three open questions remain unaddressed in the existing literature. First, what are the dynamic relationships between technological innovation, transportation industry development, and CO₂ emissions across China and its regions? Does any regional heterogeneity exist? Second, how does the interaction of technological innovation and transportation industry development affect CO₂ emissions—synergistically or antagonistically? Third, how can the Chinese government implement scientifically based subsidy policies for technological innovation and strategies for high-quality development of the transportation industry in light of regional realities to achieve the ‘dual-carbon’ goal effectively? These issues present urgent challenges for China and are key innovations of this research.

Our study utilized panel data from 30 provinces from 2005 to 2020 to advance relevant research and address the aforementioned questions. Briefly, by applying the panel vector autoregressive (PVAR) model estimated by the generalized method of moments (GMM), we analyzed the dynamic impacts and regional variations among technological innovation, transportation industry development, and CO₂ emissions. The objectives are to uncover the interaction mechanism between technological innovation, transportation industry development, and CO₂ emissions across China and its regions; to examine whether the two sectors have a synergistic effect on CO₂ reduction; to provide valuable insights for interdisciplinary research on the ‘dual carbon’ goal; to offer a theoretical framework for understanding the interactions between technology, the economy, and climate; and to present empirical evidence to guide national policymakers and regional governments in developing effective strategies for CO₂ reduction through coordinated efforts in technological innovation and transportation industry development.

This study presents three primary contributions. To our knowledge, it is the first investigation into the dynamic relationship between technological innovation, transportation industry development, and CO₂ emissions in the Chinese context. Additionally, this work comprehensively evaluates the overall development of the transportation industry from multiple dimensions, more accurately reflecting the relationship between the transportation industry, technological innovation, and CO₂ emissions. Furthermore, it systematically explores and quantifies the regional heterogeneity of the triadic relationships among technological innovation, transportation industry development, and CO₂ emissions, laying a foundation for policy formulation.

This manuscript is structured into five sections. Section 2 presents the literature review and hypothesis development. Section 3 details the data and methodology employed. Section 4 summarizes and discusses the results. Lastly, Section 5 encapsulates the main conclusions, illustrates policy implications, and offers insights into limitations and future research directions.

2. Literature Review and Hypothesis Development

2.1. Nexus between Technological Innovation and CO₂ Emissions

According to Schumpeter’s theory of ‘creative destruction’, the impact of technological innovation on environmental quality, particularly in terms of CO₂ reduction, is uncertain [22]. Numerous studies have reported that technological innovation favorably reduced CO₂ emissions [23,24]. According to Ostadzad et al. [25], technological innovation promoted CO₂ reduction by increasing production efficiency and developing renewable energy sources. Conversely, some conflicting views doubt whether technological advancements always decrease CO₂ emissions [26,27]. Khan et al. [28] argued that despite the upgrade of energy efficiency, technological innovation might be insufficient to significantly reduce CO₂ emissions without additional cost-effective measures such as new advancements, technologies, and equipment. Su et al. [29] found in their research using Brazilian quarterly data from 1990 to 2018 that technological innovation could potentially elevate CO₂ emissions. Furthermore, Dauda et al. [30] examined the panel data from 18 developed and developing economies from 1990 to 2016, revealing that technological innovation reduced CO₂ emissions in BRICS countries while increasing releases in the Middle East and North African countries. It is worth noting that stringent environmental regulations targeting high CO₂ emissions could impact technological innovation through mechanisms such as the ‘conduction effect’ and ‘crowding-out effect’ [31,32]. Zhao et al. [33] found that such regulations had a ‘conduction effect’, which could drive regional technological innovation through industrial restructuring and encouraging industrial agglomeration. In contrast, Herman et al. [34] identified the ‘crowding-out effect’, suggesting that strict regulations might hinder regional innovation due to the reduced availability of human resources and capital investment. Furthermore, increased costs for pollution control under these regulations could scale down capital input in technology and production, impeding technological progress and innovation. Similarly, Li et al. [35] discovered that between 2003 and 2017, environmental regulations on pollution-intensive enterprises in eastern and central China notably facilitated technological innovation, and such a promotion effect was insignificant in the western region. The diversity of research findings indicates that a single perspective cannot accurately explain the relationship between technological innovation and CO₂ emissions in China.

We propose the following hypothesis based on the above analysis:

H1: 

Technological innovation significantly influences CO₂ emissions, and vice versa, in a regionally heterogeneous manner.

2.2. Nexus between Transportation Industry Development and CO₂ Emissions

Existing research primarily examines the one-way influence exerted by individual indicators of transportation industry development on CO₂ emissions [11,36]. Awaworyi Churchill et al. [37] analyzed approximately 150 years of Economic Cooperation and Development (OECD) data, revealing that a 1% rise in transportation infrastructure stock resulted in a roughly 0.4% increase in CO₂ emissions. Umar et al. [38] confirmed that transportation passenger volume played a crucial role in CO₂ emission growth based on the panel data in China from 2000 to 2015. Li and Zhang [39] suggested that intervening in transportation mode selection through operation management could reduce CO₂ emissions. Notably, spatial variations were found in the mentioned impact. Xie et al. [40] examined data from 283 Chinese cities between 2003 and 2013, demonstrating that transportation infrastructure construction elevated CO₂ emissions in large and medium-sized cities rather than in small cities. Dzator et al. [41] observed that air transport infrastructure increased CO₂ emissions in 26 OECD countries from 1960 to 2018, and railway infrastructure delivered no significant effects. A limited volume of studies has explored the impact of CO₂ emissions on transportation development. Han et al. [42] illustrated that China’s CO₂ trading pilot policy encouraged pilot areas to improve their road transportation infrastructure. Similarly, Raux et al. [43] discovered that the French CO₂ trading policy could temporarily influence residents’ travel behaviors. These findings complicate the interplay between the transportation industry and CO₂ emissions in China.

On this basis, we derive a second hypothesis:

H2: 

Transportation industry development significantly influences CO₂ emissions, and vice versa, and this two-way influence displays regional heterogeneity.

2.3. Nexus between Technological Innovation and Transportation Industry Development

The transportation industry serves as an important engine driving technological innovation, as evidenced by prior studies [44,45]. Agrawal et al. [46] analyzed historical data on interstate transportation infrastructure in the United States and found that a 10% increase in highway stock contributed to a 1.7% rise in patent applications within five years. Similarly, Yang et al. [47] utilized an endogenous growth model that focused on high-speed rail fares and speeds and demonstrated that high-speed rail largely fostered innovation growth and convergence in China, with impact values of 14.73% and 5.91%, respectively. This impact also varies across cities and distances [48]. The impact of technological innovation on the transportation industry has also attracted scholarly attention [49,50]. Koukaki and Tei [51] suggested that adopting advanced information technology and new equipment played a crucial role in enhancing the efficiency of air transport. Mouratidis et al. [52] studied the different impacts of remote activities and sharing technologies on residents’ travel patterns. The results showed that teleconferencing reduced public transportation passenger volumes; in contrast, emerging models, such as car-sharing and air traffic, were expected to increase passenger traffic. Additionally, some literature focuses on the effects of the synergy between technological innovation and transportation industry development on CO₂ emissions [53,54]. Using the dynamic system-generalized method of moments estimation, Zhu et al. [55] revealed that the interaction between technological innovation and freight and railway infrastructure effectively reduced energy consumption and CO₂ emissions by improving energy efficiency and transport patterns. By employing spatial Durbin models and mediation effect models, Li et al. [56] revealed the indirect impact of different transportation infrastructures on CO₂ emissions in Chinese cities through technological progress; however, this effect varied across regions. Bieser et al. [54] found that telecommuting activities reduced long-distance commuting demand and thus lowered CO₂ emissions in the transportation sector. However, these activities also increased household energy use, highlighting a necessity to further explore the net CO₂ reduction benefits.

There is a scarcity of direct evidence supporting the CO₂ emissions’ reverse influence on the synergy between technological innovation and the development of the transportation industry. However, it can be speculated that CO₂ emissions promote such synergy based on practices in carbon trading markets and low-carbon tax policies. Notably, whether this promotion effect suppresses or stimulates the reduction in CO₂ emissions still warrants further verification. To sum up, a two-way interaction exists between technological innovation and transportation industry development, characterized by regional heterogeneity in effects. For this reason, further investigation incorporating specific regions is needed to clarify their relationship and influence mechanism on CO₂ emissions.

Based on these findings, this work develops the following hypotheses:

H3a: 

Technological innovation interacts with transportation industry development in a regionally heterogeneous pattern.

H3b: 

This interaction affects CO₂ emissions, and such an influence also has regional variations.

The theoretical framework of this study is illustrated in Figure 1.

3. Data and Methods

3.1. Data Sources and Variables

The provinces and analysis period were chosen based on data availability and integrity. Given the insufficient data for Tibet, Hong Kong, Macao, and Taiwan, this study relies on panel data from 30 Chinese provinces between 2005 and 2020. Furthermore, according to the National Bureau of Statistics of the People’s Republic of China, these provinces are divided into eastern, central, western, and northeastern regions. The data are mainly from the ‘China City Statistical Yearbook’, ‘China Science and Technology Statistical Yearbook’, and ‘China Energy Statistical Yearbook’. The variables are as follows.

CO₂ emissions (CO₂). To mitigate the impact of sparse population, this study calculates the average CO₂ emission weighted by urban population density, referring to Wu et al. [57]. Population density is defined as the ratio of permanent population to built-up area [58]. The quantification of the total CO₂ emissions follows the guidelines provided by the Intergovernmental Panel on Climate Change (2006) for national greenhouse gas inventories. The calculation formula is as follows:

$$C{E}_2={\displaystyle \sum _{i=1}^8{Q}_i}\times H_i\times C_i\times F_i\times {\displaystyle \frac{44}{12}}$$

(1)

where CE₂ is the total CO₂ emissions, Q is the annual consumption of energy, i is the energy (i.e., coal, coke, crude oil, gasoline, kerosene, diesel, fuel oil, and natural gas); H is the average low calorific value; C is the CO₂ emissions coefficient; and F is the carbon oxidation factor.

Technological Innovation (Uric). The number of patents is typically applied to measure technological advancement, encompassing both patent grants and filings [59]. Invention patents are more technically advanced and innovative than utility models and designs, making them a popular indicator for assessing technological innovation capacity [60]. In line with Ruiz-Ortega et al. [61], the proportion of invention patent applications to total patent applications is measured as a key metric for evaluating technological innovation.

Transportation Industry Development (Eotd). The evaluation of transportation industry development involves three dimensions: transportation infrastructure scale, transportation capacity, and service efficiency index. This category aligns with the studies by Galkin et al. [13]. The weight of each index item within these dimensions is obtained using Shannon’s (1948) entropy method. Considering the widespread use of this approach, this article omits the measurement process [62]. The specific results are detailed in

Table 1. Measuring index system and weight of transportation industry development.

Construction Dimension	Evaluation Index	Unit	Attribute	Weight
Infrastructure scale	Highway network density	km/km2	+	0.087
	Railway network density	km/km2	+	0.070
	Public transport network density	km/km2	+	0.050
	Inland waterway density	km/km2	+	0.048
	Urban road area per capita	km2	+	0.065
Transportation capacity	Road passenger turnover	billion person-km	+	0.056
	Road freight turnover	billion ton-km	+	0.072
	Railway passenger turnover	billion person-km	+	0.064
	Rail freight turnover	billion ton-km	+	0.084
	Water transport passenger turnover	billion person-km	+	0.043
	Water freight turnover	billion ton-km	+	0.078
	Public transport passenger turnover	million person-km	+	0.044
	Added value of the transportation industry	CNY 100 million	+	0.067
Service efficiency	Traffic accidents	number	−	0.049
	The per capita travel cost of urban residents	yuan	−	0.062
	The per capita travel cost of rural residents	yuan	−	0.027
	The proportion of coal energy consumption	%	−	0.034

Note: + denote the index is positive, and − denotes the index is negative.

.

In order to address the endogeneity issue, instrumental variables are introduced to mitigate the bias in estimation, including the following:

Economic growth (GDP). GDP is a key indicator of economic activity and greatly impacts the factors closely related to CO₂ emissions, including the total energy consumption and its structure. Kirikkaleli et al. [63] found that the per capita GDP in Denmark remarkably increased CO₂ emissions. For this consideration, per capita GDP is selected to measure economic growth.

Urbanization rate (Urra). The urbanization rate is an important indicator of the urbanization process and nonlinearly influences CO₂ emissions [64]. Referring to the research of Vo et al. [65], we compute the proportion of the urban population to the total population to measure the urbanization rate.

Environmental regulation (REGU). As a critical tool for government intervention in CO₂ reduction, environmental regulation controls the emissions through adjusting industrial structure and technological progress [66]. This article refers to Zhao et al. [33] and selects the weighted value of pollutant emissions to assess the intensity of environmental regulations.

Energy consumption structure (Ecs). The energy consumption structure represents the volume of energy consumed from each energy source and its relative proportion in the total energy consumption. The noticeable role of coal consumption in escalating CO₂ emissions has been highlighted by Zhang et al. [67]. Based on the work of Liu et al. [68], the current work defines the energy consumption structure as the ratio of coal consumption to total energy consumption.

Table 2. Descriptive statistics of the variables.

Region	Variable	Sample	Mean	Std. Dev	Min	Max
All	CO2	450	12.808	12.102	0.570	74.040
	Uric	450	31.623	10.916	9.240	72.720
	Eotd	450	0.303	0.073	0.177	0.528
Eastern	CO2	150	16.943	14.850	0.570	55.900
	Uric	150	31.107	12.786	9.240	65.260
	Eotd	150	0.352	0.082	0.219	0.528
Central	CO2	90	10.019	3.701	2.390	16.880
	Uric	90	30.690	9.169	13.560	55.610
	Eotd	90	0.331	0.047	0.236	0.426
Western	CO2	165	10.510	12.120	0.878	74.040
	Uric	165	31.244	10.492	13.370	72.720
	Eotd	165	0.254	0.038	0.177	0.348
Northeastern	CO2	45	13.027	8.862	3.734	29.286
	Uric	45	36.593	7.307	21.610	54.990
	Eotd	45	0.267	0.050	0.212	0.356

provides descriptive statistics for the entire country and its eastern, central, western, and northeastern regions.

3.2. Methods

The GMM-PVAR method is adopted for two primary reasons. First, PVAR addresses the cross-sectional dynamic heterogeneity by formulating equations for endogenous variables and incorporating time effects and fixed effects. It also uses the impulse response function to analyze long- and short-term interactions between variables, visually representing their immediate response to changes [69]. Second, the two-step GMM estimation method enhances robustness by comparing static and dynamic model results, effectively dealing with complex data distributions and potential endogeneity issues, thereby increasing the reliability of model estimation and the credibility of conclusions [70]. Additionally, the GMM-PVAR model has been applied in other multivariate dynamic studies [71,72]. Subsequently, we need to construct a new PVAR:

$$\left[\begin{array}{l}U{\mathrm{ric}}_{it}\\ Eot{d}_{it}\\ C{O}_{2it}\end{array}\right]={\alpha }_0+{\displaystyle {\sum }_{j=1}^m{A}_j}\left[\begin{array}{l}U{\mathrm{ric}}_{i,t-j}\\ Eot{d}_{i,t-j}\\ C{O}_{2i,t-j}\end{array}\right]+{\beta }_i+{\theta }_t+{\epsilon }_{it}$$

(2)

Here, i = (1, 2,⋯, 30), t = (1, 2,⋯, 15) denotes the provinces and time period, respectively; j represents the lag order of the variables; α₀ is the intercept term vector; A_j is the regression coefficient matrix; β_i, θ_t, and ε_it are the fixed effect, time effect, and random disturbance term, respectively.

The PVAR coefficient has been reported to be unreliable [69]. To address this issue, by taking the logarithm of the variable in Formula (2) and incorporating the control variable and the explanatory variable’s interaction term, we establish the GMM model. The specific model is as follows:

$$lnC{O}_{2it}={\alpha }_0+{\displaystyle {\sum }_{j=1}^m{A}_j}\left[\begin{array}{l}lnU{\mathrm{ric}}_{i,t-j}\\ lnEot{d}_{i,t-j}\\ lnU{\mathrm{ric}}_{i,t-j}\times Eot{d}_{it,t-j}\\ lnC{O}_{2i,t-j}\end{array}\right]+{\displaystyle {\sum }_{j=1}^m{\gamma }_j}Cont{r}_{i,t-j}+{\beta }_i+{\theta }_t+{\epsilon }_{it}$$

(3)

Here, Uric_i,t−j × Eotd_i,t−j represents the interaction effect of technological innovation and transportation industry development; Contr_i,t−j is a collection of control variables, including Ecgr, Urra, REGU, and Ecs.

4. Results and Discussion

4.1. Unit Root Test

Prior to the empirical analysis of panel data, it is crucial to assess the stability of time series variables, prevent spurious regression, and confirm the outcomes’ predictability. This study employs the Augmented Dickey–Fuller (ADF) to examine the stationary properties of these variables [73]. The first-order backward difference term is introduced in the lag period. This incorporation enables the adopted method to effectively evaluate the autocorrelation and heteroscedasticity in time series data, thereby enhancing the results’ reliability. It has been widely used in economics and social sciences [74,75]. The findings are presented in

Table 3. Results of the unit root test for the variables.

Region	Variable	Inverse Chi-Squared (8)	Inverse Normal	Inverse Logit t (24)	Modified Inv. Chi-Squared
All	CO2	18.404 **	−2.516 ***	−2.453 ***	2.601 ***
	Uric	16.904 **	−2.185 ***	−2.151 **	2.226 ***
	Eotd	26.120 ***	−3.462 ***	−3.617 ***	4.530 ***
Eastern	CO2	10.872 ***	−2.623 ***	−3.201 ***	4.436 ***
	Uric	6.632 **	−1.795 **	−1.933 **	2.316 ***
	Eotd	11.210 ***	−2.680 ***	−3.302 ***	4.605 ***
Central	CO2	8.935 ***	−2.274 ***	−2.626 ***	3.468 ***
	Uric	10.692 ***	−2.592 ***	−3.148 ***	4.346 ***
	Eotd	12.118 ***	−2.829 ***	−3.570 ***	5.059 ***
Western	CO2	8.1651 **	−2.123 **	−2.396 **	3.083 ***
	Uric	6.6564 **	−1.801 **	−1.940 **	2.328 ***
	Eotd	10.101 ***	−2.489 ***	−2.973 ***	4.050 ***
Northeastern	CO2	21.418 ***	−4.0818 ***	−6.3117 ***	9.709 ***
	Uric	7.779 **	−2.045 **	−2.280 **	2.890 ***
	Eotd	11.215 ***	−2.681 ***	−3.303 ***	4.607 ***

Notes: ***, **, and * show significance at a 1%, 5%, and 10% level, respectively.

. All variables reject the null hypothesis of ‘non-stationarity’ at a significant level after first differencing, indicating the absence of a unit root in each variable.

4.2. Impulse Response Function (IRF) and Variance Decomposition

The impulse response diagram visually reflects the short-term, unilateral impact of one variable on the other after applying an initial impact. Variance decomposition breaks down the fluctuations of the endogenous variable into components related to the disturbance term of the equation. By measuring the disturbance term’s contribution to the mean square error of model prediction, the long-term dynamic relationship between variables can be assessed [76]. Ensuring the stability of the model is crucial for the validity of impulse response analysis and variance results. This study, inspired by Dogan et al. [77], utilizes AR characteristic polynomial roots for PVAR stability testing. The findings indicate that all eigenvalues fall within the unit circle, suggesting a range less than ‘1’, thus passing the stability check.

4.2.1. Analysis of Impulse Response Function Results

In this study, we utilize 1000 Monte Carlo simulations to analyze the impulse response functions of the three variables at the national and regional (eastern, central, western, and northeastern) scales over a 0–10 period under standardized shocks from each other. The results are illustrated in Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6.

The simulation results on the national level are illustrated as follows. From Figure 2a,e,i, the positive impact of CO₂, Uric, and Eotd on their future trends is mitigated. As shown in Figure 2b,d, Uric undergoes a transition from a positive to a negative response when influenced by CO₂; following the impact of Uric, CO₂ exhibits a negative response, resulting in decreases to 0 by Phase 6. According to Figure 2c,g, positive influences are observed in Eotd and CO₂ due to mutual impact. Figure 2f,h indicate a positive response of Uric following the impact of Eotd, displaying a consistent trend with that of the Uric-impacted CO₂. In contrast, Eotd evolves from a positive to a negative response after the impact of Uric, ultimately approaching 0. These findings indicate that Eotd drives Uric and increases CO₂. This highlights that despite Eotd’s potential to enhance national Uric, this process still relies heavily on energy, resulting in substantial pressure for CO₂ reduction. In this sense, the Chinese government should fully leverage the driving role of green and low-CO₂ transport in urban innovation, increasing the output and speed of transforming technological innovations for CO₂ reduction to achieve the CO₂ peak at the earliest opportunity.

The following discussion unfolds on a regional scale. According to Figure 3a,e,i, Figure 4e,i, Figure 5a,e,i and Figure 6e,i, the impulse responses of regional Uric and Eotd to themselves are consistent with those at the national scale, as well as the impulse responses of CO₂ in the eastern and western regions. Notably, the results in the central and northeastern regions are relatively complex (see Figure 4a and Figure 6a). As shown in Figure 3b, Figure 4b, Figure 5b and Figure 6b, the impulse response path of CO₂ in the eastern, central, and northeastern regions is consistent with that of the entire country after the impact of Uric. Despite such high consistency, the western area shows an opposite trend, suggesting that the rebound effect of local Uric may increase CO₂. As shown in Figure 3d, Figure 4d, Figure 5d and Figure 6d, the impulse response paths of Uric in the four regions impacted by CO₂ are significantly heterogeneous. As illustrated in Figure 3c,g, Figure 4c,g, Figure 5c,g and Figure 6c,g, the impact response path between CO₂ and Eotd in the four regions is consistent with that of the whole nation. In light of Figure 3f, Figure 4f, Figure 5f and Figure 6f the impact response path of Uric in the western and northeastern areas exhibits good agreement with that of the whole country following the Eotd impact. By comparison, the eastern and central regions experience a transition from a positive to a negative response, displaying a trend approaching 0, which indicates a more complex impact of Eotd on Uric in these regions. According to Figure 3h, Figure 4h, Figure 5h and Figure 6h, the response path of Eotd to the impact of Uric in northeastern region is consistent with that of the entire nation. The central and western regions exhibit a positive response that diminishes gradually; by comparison, a negative response with a similar decreasing trend is found in the east. These findings suggest that Uric in the central and western regions positively influences Eotd, while it delivers an inhibitory effect in the eastern part.

4.2.2. Variance Decomposition

As shown in

Table 4. Variance decomposition result.

Region		S	D_CO2	D_Uric	D_Eotd
All	D_CO2	10	0.976	0.001	0.022
	D_Uric	10	0.001	0.993	0.006
	D_Eotd	10	0.078	0.000	0.922
	D_CO2	20	0.976	0.001	0.022
	D_Uric	20	0.001	0.993	0.006
	D_Eotd	20	0.078	0.000	0.922
Eastern	D_CO2	10	0.956	0.002	0.041
	D_Uric	10	0.066	0.931	0.002
	D_Eotd	10	0.203	0.002	0.795
	D_CO2	20	0.956	0.002	0.041
	D_Uric	20	0.066	0.931	0.002
	D_Eotd	20	0.203	0.002	0.795
Central	D_CO2	10	0.891	0.061	0.048
	D_Uric	10	0.065	0.904	0.031
	D_Eotd	10	0.088	0.006	0.906
	D_CO2	20	0.891	0.061	0.048
	D_Uric	20	0.065	0.904	0.031
	D_Eotd	20	0.088	0.006	0.906
Western	D_CO2	10	0.998	0.001	0.000
	D_Uric	10	0.012	0.917	0.071
	D_Eotd	10	0.028	0.000	0.972
	D_CO2	20	0.998	0.001	0.000
	D_Uric	20	0.012	0.917	0.071
	D_Eotd	20	0.028	0.000	0.972
Northeastern	D_CO2	10	0.961	0.021	0.018
	D_Uric	10	0.182	0.745	0.073
	D_Eotd	10	0.049	0.024	0.927
	D_CO2	20	0.961	0.021	0.018
	D_Uric	20	0.182	0.745	0.073
	D_Eotd	20	0.049	0.024	0.927

, the variance decomposition results of the 10th and 20th prediction periods are basically consistent, indicating that the ability of each shock to explain variable changes remains stable after ten prediction periods.

On the national level, CO₂, Uric, and Eotd contributed significantly to themselves, with each exceeding 90%. Notably, the long-term relationship between Uric and CO₂ showed a minimal impact of 0.1%. This negligible impact may be due to the fact that Uric’s potential to reduce CO₂ has not been fully tapped. Based on these findings, the government should strengthen policies that leverage Uric to achieve a CO₂ peak. Furthermore, Eotd had a 2.2% long-term contribution to CO₂, and CO₂ contributed 7.8% to Eotd. These results suggested a mutual promotion relationship, typically causing undesirable consequences. In this sense, the government needs to prioritize the adoption of clean energy and facilitate the energy transition within the transportation sector. Eotd delivered a 0.6% long-term contribution rate to Uric; conversely, Uric contributed a negligible 0.0% to Eotd. This implied that Eotd influenced Uric, while the reverse relationship was not significant. In light of the above analysis, the Chinese government needs to address the challenges associated with integrating Uric within the Eotd to foster a positive synergy between the two elements.

At the regional scale, this research revealed that in the eastern, central, western, and northeastern regions of China, the long-term contribution rates of Uric to CO₂ varied, representing 0.2%, 6.1%, 0.1%, and 2.1%, respectively. In contrast, the long-term contribution rates of CO₂ to Uric accounted for 6.6%, 6.5%, 1.2%, and 18.2%, respectively. These findings indicated that Uric in the central and northeastern regions engage in CO₂ reductions. Additionally, environmental regulations targeting CO₂ reductions across the whole nation significantly influenced Uric to varying degrees. Eotd in the four regions of interest contributed to CO₂ at different rates, representing 4.1%, 4.8%, 2.8%, and 0%, respectively. Furthermore, the long-term contribution rates of CO₂ to Eotd were 20.3%, 8.8%, 2.8%, and 4.9%, respectively. These results suggested that the eastern regions were characterized by developed economies, high transportation activity, and great reliance on fossil energy, thus encountering more pressure to reduce transportation CO₂. The relatively low contribution in the west reflects the region’s less developed transportation, while the resulting values in the northeast may be linked to industrial restructuring and energy transformation. Local governments should implement measures to decrease CO₂ release from the development of the transportation industry in high-emission areas. Across the eastern, central, western, and northeastern regions, the long-term contribution rates of Eotd to Uric occupied 0.2%, 3.1%, 7.1%, and 7.3%, respectively. In addition, the long-term contribution rates of Uric to Eotd comprised 0.2%, 0.6%, 0.0%, and 2.4%, respectively. These figures inferred that fostering high-quality Eotd of the western and northeastern regions could drive Uric, thereby accelerating the research, development, and application of new technologies. Notably, Uric in northeast China is predominant in advancing Eotd; in contrast, the western part may encounter challenges in achieving the synergic integration of Uric and Eotd.

4.3. GMM-PVAR Estimation

The model’s stability was initially assessed using the previous AR characteristic polynomial root. The results demonstrated that all moduli of feature values were less than 1, confirming that the model passed the stability test. Subsequently, overidentification and sequence correlation tests were conducted on the tool variables (Ecgr, Urra, REGU, and Ecs), and the results are presented in

Table 5. Estimation results of the GMM-PVAR model.

Variable	All		Eastern		Central		Western		Northeastern
	SYS-GMM	FE	SYS-GMM	FE	SYS-GMM	FE	SYS-GMM	FE	SYS-GMM	FE
lnUric	−0.727 ***	−0.932 ***	−0.727 ***	−0.858 **	6.417 ***	1.225	3.156 ***	−0.764 *	−2.670 ***	2.445
lnEotd	1.802 ***	2.087 ***	1.802 ***	1.754 *	−1.586 ***	−4.134	−7.619 ***	1.012	13.207 ***	−4.623 **
lnUric × lnEotd	0.332 **	−0.690 ***	−0.332 ***	−0.540 *	−1.089 ***	1.316 *	2.528 ***	−0.574	−2.673 ***	1.732 *
L1.CO2	0.771 ***	0.617 ***	0.979 ***	0.622 ***	0.949 ***	0.769 ***	0.945 ***	0.589 ***	0.972 ***	0.763 ***
Constant term	7.550 **	4.626 ***	7.550 ***	4.862 ***	−0.817 ***	−1.820	−2.951 *	2.619	0.337 **	−4.342 *
AR (1)	0.011		0.010		0.037		0.012		0.006
AR (2)	0.359		0.533		0.219		0.460		0.740
Sargan	0.275		0.517		0.441		0.596		0.429

Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

. The p value of the Sargan statistics exceeded 0.1, suggesting the validity of these tool variables and no overidentification issues. Additionally, the AR (1) and AR (2) statistics revealed no sequence-related disturbances. Based on these outcomes, it can be concluded that the developed model is suitable, and the estimates are reliable and consistent. The estimated coefficients of the model are given in Table 5.

On the national level, a 1-unit increase in Uric only resulted in a 0.0727-unit CO₂ reduction with a significance level of 1%. For each unit increase in the Eotd, CO₂ increased by 1.802 units at a significance level of 1%. Moreover, in the case of every 5-unit increment of Uric × Eotd, a 0.0332-unit increase was seen in CO₂. These results indicated that Uric could reduce CO₂, and Eotd amplified this positive effect, yet the reductions are inadequate to offset the rise in transportation CO₂ fully. The potential reasons for these findings are summarized as follows. First, although technological innovation can reduce CO₂ emissions, its potential has not been fully utilized. Second, China’s rapid urbanization has driven the expansion of transport infrastructure and capacity, rendering CO₂ a primary negative environmental consequence, which aligns with the findings of Wang et al. [78]. Third, although the expansion of transportation networks has facilitated the flow of innovative resources, the integration and practical application of green innovations remain underdeveloped. This is consistent with the finding by Acheampong et al. [14], which showed that the interplay between transport infrastructure and innovation could decrease CO₂ levels in EU countries. However, the investigation into 14 G20 countries by Erdogan et al. [79] presented inconsistent results. This can be attributed to the varying levels of economic development, policy frameworks, technological adoption, and social contexts across regions, highlighting the importance of investigating regional disparities in this paper.

Regionally, an increase of 1 unit in Uric in the eastern and northeastern regions contributed to reductions of 0.727 and 2.670 units of CO₂, respectively. In contrast, for the central and western regions, each unit rise of Uric resulted in increases of 6.417 and 3.156 units of CO₂, respectively. Possible reasons for this include the following: First, Uric in the central and western regions exerts a threshold effect on reducing environmental pollution. Additionally, innovation in these areas progresses slower than in the other two areas. The potential of Uric in CO₂ reduction necessitates further exploration. Upon surpassing the threshold value, the inhibitory effect of Uric on CO₂ can be visible. This finding aligns with the research by Wang and Dong [80] on OECD countries. Second, the central and western regions experiencing relatively slow development encounter an urgent need for development. Uric involves uncertainties and risks, and its rebound effect increases productivity and economic growth benefits at the potential cost of inadvertently elevating CO₂ pressure levels. This observation is consistent with the findings of Adebayo and Kirikkaleli [24] for Japan and Su et al. [29] for Brazil. In the eastern and northeastern regions, each unit increase in Eotd contributed to respective increases of 1.802 and 13.207 units of CO₂. By comparison, the same increase for the central and western regions resulted in declines of 1.586 and 7.619 units of CO₂, respectively. Our research innovatively observed an inhibitory effect of Eotd on CO₂. This result may be attributed to the pronounced regional disparities in economic conditions, geographical characteristics, and transportation development levels across China. In the northeast, industrial production heavily relies on traditional fuel-powered vehicles, consuming substantial energy. In the east, the advanced economy and high population density drive frequent transportation activities; however, the limited adoption of green transportation technologies cannot effectively mitigate the resulting high energy use. In contrast, the vast territories and low population density of the central and western regions hindered efficient, long-distance transportation. The improved transportation networks have recently addressed these challenges, greatly reducing energy consumption. Despite the lag in the local transportation industry development, more modern energy-efficient technologies have been incorporated into the later-built infrastructure, further decreasing energy usage. In the eastern, central, and northeastern regions, every 1-unit increase in Uric × Eotd reduced CO₂ by 0.332, 1.089, and 2.673 units, respectively. By comparison, the western region experienced a 2.528-unit elevation in CO₂ with each unit increase in Uric × Eotd. These findings indicated that the Uric × Eotd in eastern, central, and northeastern China synergistically inhibited the regional CO₂, while the opposite effect was observed in the western part. Notably, the influence mechanisms of these factors vary. Specifically, in the eastern region, Uric reduced more CO₂ by limiting Eotd. The Eotd decreased CO₂ through enhancing Uric in Northeast China. In the western region, Eotd increased CO₂ despite the enhancement in Uric. The influence mechanism in central China was complex. These differences may be attributed to several factors. In the eastern region, innovations in information and network technology have adjusted travel demand and structure, leading to diminished passenger turnover and energy consumption. In the northeast, the transportation industry has accelerated the advancement and application of low-CO₂ technologies, optimizing energy efficiency. In the western part, despite the technological innovation driven by the transportation industry, the resulting scale effect has increased energy consumption due to the lack of green and low-CO₂ innovations. In the central region, where the economic level and transportation industry development lie between those of the east and west, both synergistic and contradictory effects are found. The coefficient of the explanatory variables with one period lagged (L1. CO₂) in China’s eastern, central, western, and northeastern regions exhibited a positive impact at a 1% significance level. This observation suggested that the historical CO₂ delivered a promotional effect on the current period. This relationship was influenced by the interaction of various factors affecting CO₂ as well as temporal and spatial dependencies.

5. Conclusions and Policy Recommendations

Based on the comprehensive analysis of 30 inter-provincial panel data in China from 2000 to 2020, this study utilized the GMM-PVAR model to investigate the dynamic interaction effects and regional heterogeneity of Uric, Eotd, and CO₂. The following conclusions and policy recommendations are drawn:

On the national level, Uric promotes CO₂ mitigation, and mutual enhancement effects exist between Eotd and CO₂. The Eotd can reduce CO₂ by fostering Uric. However, the increased emissions in this industrial sector exceeded those offset by Uric-driven reductions. Regional heterogeneity was also illustrated as follows. In the eastern region, CO₂ and Uric interacted in a mutually suppressive manner, while CO₂ and Eotd possessed a mutual promotion relationship. Moreover, the role of Uric in reducing CO₂ could be enhanced by inhibiting Eotd. In Northeast China, Uric suppressed CO₂, while CO₂ drove the progress in Uric. The interaction between Eotd and CO₂ was characterized by mutual reinforcement. Eotd decreased CO₂ by enhancing Uric. Across the central region, the interplay between the three factors was intricate. In the west of China, except for the negative effect of Eotd on CO₂, the remaining relationships positively influenced the factors involved. Additionally, unfavorable interactions were found between Uric and Eotd. The above phenomena can be exemplified by the fact that Uric-fueled Eotd increased CO₂.

Based on the above conclusions, this research summarized several policy recommendations. Firstly, the Chinese government should develop and implement technology-supported CO₂ peak strategies and transportation industry development plans tailored to regional conditions. Additionally, it is necessary to establish a CO₂ reduction sharing mechanism that encompasses green transportation infrastructure, low-CO₂ transportation means, and green, efficient operational services. Moreover, shifting the impact of CO₂ emission on the entire transportation industry chain from positive to negative should be considered. Additionally, further integration of the low-CO₂ transportation industry and green technological innovation should be highlighted. Enhancing the accessibility of the low-CO₂ transportation network and improving the quality and efficiency of transportation services are crucial for promoting the diffusion and aggregation of green innovations. Secondly, the eastern region should prioritize the transformation and utilization of technological innovations, such as big data, artificial intelligence, and virtual reality in multiple industrial sectors, to upgrade social operational efficiency and reduce reliance on private vehicles, thereby achieving dual CO₂ reductions within the transportation industry and society as a whole. It is also important to remain vigilant about the potential negative effects of environmental regulations on technological innovations to avoid compromises of regional CO₂ reduction efforts. Thirdly, in the northeast of China, strict CO₂ accounting standards and control systems should be established by prioritizing low-CO₂ solutions within enterprises. Enhancing the connectivity of green transportation networks is vital for reducing transportation costs and increasing efficiency, potentially encouraging cross-regional collaboration among researchers and businesses. Fourthly, governments in the western region should focus on a more efficient and accessible transportation network to reduce CO₂ emissions associated with inefficient transportation and redundant activities. In addition, efforts should be made to guide the transformation of green technological innovations and manage potential risks, such as uncertainties and ‘CO₂ leakage’, to prevent adverse effects on CO₂ reduction. Finally, the authorities in the central region should adopt successful practices from other regions in high-quality transportation management and green technology innovations. Tailoring strategies to unique regional conditions should be emphasized to maximize the synergistic effects of transportation and technological innovations, thereby maximizing CO₂ reduction.

Although this article holds theoretical and practical significance, it also has limitations that necessitate further improvements. The measurement of the transportation industry’s development is constrained by data source limitations, particularly the lack of detailed sub-items for railway and air transportation systems. Moreover, the study’s temporal and spatial scope, focusing on the provincial scale over fifteen years from 2005 to 2020, are influenced by geographical, economic disparities, and the impact of the COVID-19 pandemic. Future research should consider a prolonged time frame and reduced spatial units for a more comprehensive analysis and comparison. Additionally, the uncertain impact of various factors on regional heterogeneity in CO₂ reductions requires ongoing attention from the academic community to provide more insightful findings and effective policy recommendations for achieving the CO₂ peak.