Urban Technology Transfer, Spatial Spillover Effects, and Carbon Emissions in China

Abstract

Technology transfer is essential for addressing technological disparities in urban areas and reducing carbon emissions. This study analyzes inter-city technology patent transfers and carbon emission data from China (2010–2019) using a spatial Durbin model to assess the effects of technology inflow and outflow on carbon emissions and their spatial spillover effects. Key findings include the following: ① a significant spatial correlation exists between technology inflow, outflow, and carbon emissions, with notable spillover effects; ② both technology inflow and outflow reduce intra-regional carbon emissions, but only outflow significantly reduces inter-regional emissions; ③ regional differences are evident, with the eastern and central regions showing significant reductions due to technology transfers, while the western and northeastern regions do not. In the northeastern region, technology transfer significantly aids neighboring cities in reducing emissions. However, the central region lacks spillover effects from outflow, and technology inflow and outflow in the western region hinder reductions. This paper provides policy recommendations to improve urban technology transfer and carbon emissions mitigation.

1. Introduction

Currently, global climate change is intensifying, underscoring the critical importance of regulating greenhouse gas emissions, particularly carbon dioxide (CO₂). The pursuit of green and low-carbon development constitutes a collective objective for human society in the 21st century and serves as an essential pathway for achieving high-quality development in China. In alignment with global climate goals, the Chinese government has committed to reaching a carbon peak before 2030 and attaining carbon neutrality by 2060. Consequently, the imperative of “actively and steadily promoting carbon peak and carbon neutrality” has emerged as a significant developmental challenge for China at this juncture. However, the “Carbon Emissions Report 2023” published by the International Energy Agency (IEA), indicates that global carbon emissions reached a record high of 37.4 billion tons in 2023, with emissions from China contributing 12.6 billion tons to this total. These emission levels are considerably elevated and remain well above the established global targets for emission reduction. Since the ratification of the United Nations Framework Convention on Climate Change, both academic and political spheres have recognized the pivotal role of technological innovation in achieving the objectives of carbon peak and carbon neutrality. Nevertheless, technological innovation is characterized by substantial investment requirements, extended development cycles, and elevated risks, reflecting the typical features of “dual externalities” associated with environmental and innovative factors. These characteristics contribute to market failures and a lack of sufficient incentives for innovation, thereby exacerbating disparities in regional green innovation and carbon emission reduction efforts. As a result, addressing the inadequacies in research and development incentives and bridging technological gaps between cities through technology transfer mechanisms emerges as a vital strategy for enhancing the carbon emission reduction capabilities and demands of various urban areas. According to the “China Science and Technology Achievement Transformation Annual Report 2022”, the transaction volume of technology contracts in China increased from CNY 17.7 trillion in 2018 to CNY 47.8 trillion in 2020. This growth reflects the ongoing vitality of China’s technology market, which is playing an increasingly significant role in the pursuit of sustainable development. Currently, various cities across China have implemented numerous measures to actively promote cooperation in technological innovation. However, the transfer of technology has not been effectively synchronized with efforts to reduce carbon emissions. The disparity in the development of carbon emission reduction among cities is becoming more pronounced, and the lagging status of low-carbon technology in certain western cities remains unaddressed. In these areas, carbon emissions continue to exhibit an upward trend, and the pressure to achieve a carbon peak remains substantial. Therefore, enhancing the efficiency of technology transfer and its supportive role in reducing carbon emissions has emerged as a critical factor in attaining China’s overarching objectives of reaching a carbon peak and achieving carbon neutrality.

Technology transfer can be understood as a dynamic flow of various factors. According to Schumpeter’s theory of innovation economics, the value of technological advancements is only realized when they are disseminated and applied [1]. The movement of technological factors, including labor, capital, and information, plays a crucial role in enhancing resource allocation, fostering the accumulation of innovation, stimulating innovative vitality, and elevating the overall level of science and technology. Furthermore, technology transfer is a pivotal element in facilitating industrial upgrading [2] and supporting the green transformation of economy [3]. On one hand, the inflow of technology can mitigate the high risks associated with research and development (R&D) while securing patent rights in the short term, thereby rapidly advancing technological capabilities and directing factor resources toward more advanced and environmentally sustainable industries. Conversely, the outflow of technology can enhance the stock of R&D capital, invigorate researchers’ enthusiasm for R&D, attract additional research talent, and cultivate a conducive endogenous environment for research and innovation. However, the academic discourse has predominantly focused on the low-carbon effects of technological innovation and carbon emissions, as well as the relationship between technological progress and carbon emissions. Comparatively fewer scholars have investigated the direct and overall impact of technology transfer on carbon emissions from a purely transfer-oriented perspective. The existing literature provides a valuable reference for this study, as technology transfer and technological innovation represent two distinct pathways of technological advancement. Technology transfer serves as the transmission mechanism for technological innovation, while technological innovation constitutes the fundamental premise for technology transfer. These three elements are intricately linked and interdependent, and theoretically, technology transfer can yield various benefits related to green and low-carbon initiatives. Nevertheless, due to the uneven development across different regions of China, the low-carbon impacts and spillover effects of technology transfer cannot be generalized. This raises several critical questions: Is there a significant spatial correlation between technology transfer and carbon emissions? Can technology transfer engender low-carbon effects, and how do these effects manifest under spatial interactive influences? Additionally, are there variations in this impact across different developmental regions?

In light of the aforementioned considerations, the primary objective of this research paper is to examine the spatial correlation between technology transfer and carbon emissions, to empirically investigate the spatial interactive effects of technology transfer on carbon emissions, and to explore the heterogeneous characteristics associated with these phenomena. The findings indicate that both technology inflow and technology outflow exhibit significant spatial correlation with carbon emissions; specifically, technology inflow and outflow have notable effects on carbon reduction within regions. However, technology inflow does not significantly influence inter-regional carbon reduction, whereas technology outflow demonstrates a significant impact on inter-regional carbon emissions. Furthermore, the direct and indirect effects identified exhibit considerable spatial heterogeneity. The potential marginal contribution of this study is twofold: First, from the perspective of the research subject, it constructs a comprehensive database of urban technology transfer in China from 2010 to 2019 through the processes of big data mining, cleaning, and extraction, categorizing domestic technology transfer into technology inflow and outflow. This approach allows for a more nuanced understanding of the dynamic characteristics of urban technology flows and their relationship with carbon emissions. Second, from the perspective of empirical strategy, this study elucidates the impact and spillover effects of technology transfer on carbon emissions across different regions, thereby providing a theoretical and practical foundation for enhancing the construction of technology transfer systems and promoting low-carbon and green development.

The structure of this paper is outlined as follows: Section 2 includes a literature review and the development of research hypotheses; Section 3 elaborates on research design; Section 4 provides an empirical analysis of the spatial correlation and interactive effects between technology transfer and carbon emissions; and Section 5 concludes with a discussion of the findings and their implications.

2. Literature Review and Research Hypotheses

Technology transfer, acknowledged as a vital mechanism for facilitating knowledge spillovers, has garnered significant attention in academic research due to its economic benefits and its role in enhancing innovation performance [4,5]. Contemporary studies on technology transfer reveal four predominant characteristics. First, the primary focus of investigation is on micro-level technology innovation entities, such as enterprises and universities (including research institutions) [6,7,8], as well as on more granular entities, including individual researchers, brokers, and entrepreneurs [9,10]. Second, the research predominantly emphasizes the micro-level organizational context [11,12], thereby addressing technology transfer issues at the organizational level [13]. Third, qualitative research methodologies are primarily employed, utilizing techniques such as in-depth interviews, semi-structured interviews, questionnaire surveys, and telephone interviews to collect primary research data [14,15]. Fourth, the prevailing research perspective tends to concentrate on the scale and speed of technology transfer [16,17]. While some studies have examined technology transfer within specific technical domains and inter-regional contexts, there remains a significant gap in the investigation of the multidimensionality of technology transfer from both theoretical and practical perspectives. These four characteristics also highlight four notable deficiencies in current technology transfer research: the rigidity of research subjects, the generalization of research content, the simplification of research methodologies, and the diminishing emphasis on spatial considerations. In light of the trends of knowledge proliferation and globalization, cities have emerged as critical spatial entities in the global landscape of scientific and technological competition. Therefore, urban and regional innovation issues, with a focus on cities as research subjects, should be prioritized as frontier topics within the fields of innovation economics and innovation geography. Urban technology transfer encompasses the technological exchanges occurring between different cities and among various entities within the same city (i.e., spatial flow). This fluidity also affects the differential vectors of technology inflow and outflow in urban technology transfer, a factor that is often overlooked in the existing literature. Furthermore, as a unique form of element flow, urban technology transfer exhibits spatial interactions influenced by knowledge spillover and technology diffusion. The current body of literature has rarely approached technology transfer from the perspectives of spatial interaction and vector characteristics, which represents a critical area that this paper seeks to explore.

The academic community has generated a considerable volume of research addressing the issue of carbon emissions. Initially, a theoretical framework has been established, encompassing concepts such as the Environmental Kuznets Curve, the Porter Hypothesis, and the “Pollution Haven” Hypothesis [18,19]. These theories investigate the potential impacts of carbon dioxide emissions across environmental, corporate, and socio-economic dimensions. Furthermore, significant progress has been made in understanding the factors influencing carbon emissions. For instance, some scholars contend that elements such as the upgrading of industrial structure, technological advancements, green finance, the degree of digitalization, energy efficiency, government intervention, and foreign direct investment (FDI) play a role in mitigating carbon emissions [20,21,22,23,24]. In contrast, factors such as economic scale, population growth, trade openness, capital investment, and energy structure are posited to have a positive correlation with increased carbon emissions [25,26,27]. The discrepancies in findings can be attributed to variations in research subjects and methodologies employed. Lastly, concerning the spillover effects of carbon emissions, numerous scholars have undertaken both macro and micro-level analyses of spatial spillover effects utilizing nighttime light data, economic statistics, and energy consumption data, in conjunction with index construction and model estimation. It is widely acknowledged that mechanisms such as regional trade cooperation, economic growth, levels of technological innovation, population mobility, and industrial transfer act as conduits for regional associations and influences on carbon emissions [28,29,30], demonstrating significant spatial spillover effects.

Research in the field of technology-related emission reductions has primarily focused on the overarching themes of technological advancement and innovation [31,32,33]. Discussions specifically addressing carbon emissions associated with technology transfer—a distinct aspect of technological evolution—have largely emerged within the context of promoting south–south cooperation and north–south interactions [34]. Scholars have employed qualitative analyses, mathematical simulations, and various methodologies to elucidate the significant role that international technology transfer can play in mitigating emissions and regulating temperature. It has been emphasized that the transfer of green technology from developed nations to developing countries constitutes a vital strategy for alleviating global environmental pressures [35] and serves as an effective mechanism for fostering low-carbon development in these regions. Some researchers have posited that the comprehensive sharing of low-carbon technologies among nations could potentially lead to a reduction of approximately 40% in global cumulative carbon emissions [36]. Within the Chinese academic discourse regarding the impact of technology transfer among cities, the types of technology and their ecological implications are diverse, with a predominant focus on the transfer of green and low-carbon technologies [37], or solely on the environmental consequences of unidirectional technology flows, such as technology spillovers or introductions [38,39]. There exists a paucity of studies that specifically examine the relationship between “technology transfer” and “carbon emissions”. In the limited literature pertinent to this study, scholars such as Huang, R. et al. (2020) [40] and Khan, Y. et al. (2022) [41] have utilized import and export trade as proxies for technology transfer, concluding that the spillover effects of such transfers significantly enhance carbon emission reductions. Conversely, some researchers, including Hao, Y. et al. (2021) [42], argue that the pathways of foreign direct investment (FDI) and outward foreign direct investment (OFDI) have resulted in increased carbon emissions. The aforementioned studies provide a robust foundation for this paper; however, a consensus regarding the impact of technology transfer on carbon emissions has yet to be achieved, indicating a need for further investigation. Consequently, this paper incorporates inter-city technology patent transfer data and employs spatial econometric analysis to empirically assess the influence of urban technology transfer on carbon emissions, considering the directional vectors of technology transfer, spatial spillover effects, and regional heterogeneity. This exploration aims to identify effective strategies for enhancing carbon emission reduction capabilities across different regions through technology transfer. In conclusion, this paper presents the following research hypotheses.

Hypothesis 1.

A significant spatial correlation exists among technology inflow, technology outflow, and carbon emissions.

Hypothesis 2.

Carbon emissions demonstrate a significant spatial spillover effect.

Hypothesis 3.

The inflow of technology has a significant direct impact on the reduction of carbon emissions.

Hypothesis 4.

The inflow of technology has a significant indirect impact on the reduction of carbon emissions.

Hypothesis 5.

The outflow of technology has a significant direct impact on the reduction of carbon emissions.

Hypothesis 6.

The outflow of technology has a significant indirect impact on the reduction of carbon emissions.

3. Research Design

3.1. Sample and Data Sources

The temporal scope of this study spans the years 2010 to 2019. The dependent variable, carbon emissions, is sourced from the China Carbon Accounting Database. Various economic indicators, which serve as control variables, are extracted from the annual “China City Statistical Yearbook”. The explanatory variable, technology transfer, is obtained from the Patent Search and Analysis platform of the China National Intellectual Property Administration. The primary methodology for data acquisition involves the application of web scraping techniques to extract patent transaction information from the platform. This information encompasses patent numbers, changes in ownership, classification numbers, details of pre- and post-change patent owners, their addresses, and authorization announcement dates. Python 3.10 programming is utilized to identify the prefecture-level cities associated with the patent rights before and after the ownership changes. By employing the social network data analysis capabilities of GEPHI, the weights and sums of all arcs originating from and terminating at city nodes are computed, resulting in the quantification of patents transferred out and transferred in for each city (for detailed information on the data source and acquisition process related to the technology transfer discussed in this paper, please refer to the Supplementary Materials of the paper). The data cleaning process involves the exclusion of data from Hong Kong, Macao, and Taiwan, as well as the removal of invalid entries. Furthermore, interpolation methods are applied to address missing values for certain cities. Ultimately, a balanced panel dataset comprising 2690 observations over the decade, focusing on the transfer of invention patents and utility model patents across 269 prefecture-level cities, is established.

3.2. Research Methods

3.2.1. Spatial Autocorrelation

To address the potential bias in regression outcomes that may arise from the neglect of spatial autocorrelation in traditional panel data models, this paper references the study conducted by Deng, Y. et al. (2022) [43]. Initially, an exploratory spatial data analysis is conducted on urban technology transfer and urban carbon emission data to determine the preliminary spatial correlation of the datasets. This study utilizes the widely recognized Moran’s I statistic to perform the spatial autocorrelation test. The definition of the global Moran’s I statistic is provided in Equation (1).

$$Moran’sI={\displaystyle \frac{N}{S0}}{\displaystyle \frac{{\displaystyle {\sum }_{i=1}^N{\displaystyle {\sum }_{j=1}^{N}wij(yi-\overline{yi})(yj-\overline{yi})}}}{{\displaystyle {\sum }_{i=1}^N{{\displaystyle (yi-\overline{y})}}^2}}}$$

(1)

In the equation, the variables i and j denote distinct cities; y_i signifies the observed value of variable y in spatial unit i; $\overline{yi}$ represents the mean value of y_i; and W_ij indicates the spatial weight matrix, which reflects the adjacency relationship between city i and city j. This study employs a geographical proximity weight matrix for calculations, where, if two cities are geographically adjacent, W_ij = 1, otherwise W_ij = 0; N denotes the number of spatial units; S₀ is the set of all spatial weights. When Moran’I = 0, it signifies spatial independence; when Moran’I > 0, it indicates positive spatial autocorrelation; and when Moran’I < 0, it denotes negative spatial autocorrelation.

3.2.2. Spatial Econometric Model

Given the potential spatial dependency of carbon emissions across various regions, coupled with the mobility and diffusion of technology or knowledge during the technology transfer process, it is challenging for each sample area to meet the assumption of random independence. Consequently, there is a necessity for a spatial econometric model that accounts for the effects of spatial interaction in order to analyze the influencing factors. Spatial econometric models encompass several types, including the spatial lag model, spatial error model, and spatial Durbin model. LeSage and Pace (2009) [44] argue that the spatial Durbin model serves as an optimal starting point. This model effectively mitigates the regression risks associated with the omission of spatially lagged terms for both dependent and independent variables, while also facilitating the analysis of model degradation. Therefore, this paper employs the spatial Durbin model for its analysis, with the intention of conducting a more comprehensive model suitability test at a later stage. Acknowledging that technology transfer possesses a vector nature, characterized by both technology inflow and outflow, this paper analyzes these two scenarios separately using the spatial Durbin model.

This study considers three key critical factors related to technology transfer. First, technology transfer is characterized by a vector nature, which necessitates its classification into two distinct categories: technology inflow and technology outflow. Second, to account for the potential mixed effects of both inflow and outflow occurring simultaneously, this paper integrates both variables within a single analytical model. Third, the impact of technology transfer on carbon emissions is influenced by a temporal lag; consequently, this study analyzes carbon emissions with a one-period lag. The spatial Durbin model (SDM) developed in this paper is delineated as follows:

$$\begin{array}{l}\ln C{O}_2\_ngd{p}_{it+1}={\beta }_0+\rho W\ln C{O}_2\_ngd{p}_{it+1}+{\beta }_{1}Inz{c}_{it}+{\beta }_{2}WInz{c}_{it}+{\beta }_{3}Inz{r}_{it}+\\ {\beta }_{4}WInz{r}_{it}+{\beta }_{5}Control{s}_{it}+u_i+{\lambda }_t+{\epsilon }_{it}\end{array}$$

(2)

In the equation, ln CO₂_ngdp_it+1 represents the logarithm of carbon emission intensity. Inzc_it denotes the logarithm of the number of patents that have flowed out, while Inzr_it indicates the logarithm of the number of patents that have flowed in. Controls_it refers to a series of control variables. W is the spatial weight matrix, which employs the geographical adjacency matrix. u_i represents individual fixed effects, λ_t is the time fixed effect, and ε_it is the stochastic disturbance term. β₀ is the constant term, and ρ is the spatial lag coefficient of the carbon emission intensity variable, reflecting the degree of influence that the dependent variable in neighboring cities has on the dependent variable in the current city. β₁ and β₃ represent the impact coefficients of the current period’s technology outflow and technology inflow, respectively, on the city’s carbon emissions in the subsequent period. Conversely, β₂ and β₄ represent the impact coefficients of the current period’s technology outflow and technology inflow from surrounding cities on the city’s carbon emissions in the next period, respectively.

3.3. Variable Selection

Dependent Variable: Carbon Emissions. The primary methodologies for quantifying carbon emission indicators encompass carbon emissions per unit of gross domestic product (GDP), carbon emissions per unit area, and carbon emissions per capita. This study employs the carbon emissions per unit GDP approach. According to SHANG Yongmin et al. (2023) [37], the carbon emissions per unit GDP method effectively assesses the relationship between input costs associated with a specific economic output and carbon dioxide emissions, provided that labor, capital, and energy inputs remain constant. The urban carbon emissions data utilized in this analysis is sourced from Carbon Emission Accounting and Datasets (CEADs), accessible at https://www.ceads.net/data/county/, accessed on 24 September 2024. The total carbon emissions reported by CEADs are derived from mass balance theory, which calculates emissions based on the consumption of various fossil fuel types across different industries in China, multiplied by their respective emission factors [45]. This dataset is characterized by consistent statistical standards and robust continuity, making it a widely utilized resource for carbon emission data in recent years.

Core Independent Variable: Technology Transfer. Patent technology transfer is often conceptualized as the transfer of explicit knowledge, which serves as a direct manifestation and primary vehicle for technology transfer. This form of data is relatively accessible and more comprehensive compared to other modalities of technology transfer, such as technology outflow, collaborative research publications, talent mobility, and technology transaction volumes. Patent data delineates the pathways of technology transfer from a micro perspective, thereby providing a more accurate representation of the flow of technological elements. As referenced in the relevant literature [46], this study employs the number of patent transfers as a metric for assessing the level of technology transfer. The data utilized is primarily sourced from the patent search platform of the China National Intellectual Property Administration and is acquired through advanced big data mining techniques and geographic information coding technologies. This methodology facilitates the collection of detailed information regarding the transfer of invention patents and utility model patents across various Chinese cities. Given that technology transfer encompasses the vector interaction of inflow and outflow, the analysis is bifurcated into two components: technology inflow and technology outflow. Technology inflow is defined as the number of patents transferred into a city from other domestic cities, while technology outflow refers to the number of patents transferred out of a city to other domestic cities. The respective counts of inflow and outflow patents serve as indicators of a city’s capacity to aggregate and disseminate patents from other domestic locations.

In the context of control variables and with reference to the established STIRPAT model [47,48], this study identifies five factors that influence urban carbon emissions as control variables: 1. Level of Economic Development (lnperngdp): this factor is represented by the logarithmic transformation of per capita regional gross domestic product (GDP). 2. Transportation Infrastructure (lnroad): This variable is measured by the area of urban roads. Enhanced urban accessibility is associated with a reduction in environmental pollution and greenhouse gas emissions resulting from the movement of individuals and the transportation of goods. 3. Scale of Financial Development (lnfin): This is represented by the total balance of various loans from financial institutions at the end of the year. Numerous studies indicate that the relationship between financial development and carbon dioxide emissions is complex and ambiguous [49]. For example, financial development may promote green technology innovation, reduce energy consumption, and contribute to carbon emission reductions. Conversely, it may also facilitate industrial financing, expand production capacities, and consequently increase carbon emissions. This study posits that a higher level of financial development correlates with increased carbon emission intensity. 4. Level of Openness to Foreign Markets (fdi): This is quantified by the ratio of foreign direct investment to GDP. A greater degree of openness to foreign markets is anticipated to enhance the spillover effects of foreign capital and technology, which can foster technological innovation and, to some extent, mitigate carbon emissions. 5. Industrial Structure (induss): This is represented by the ratio of GDP from the secondary and tertiary industries. The industrial structure reflects shifts in regional economic growth and development models, serving as a significant determinant of total energy consumption and intensity within the region [50]. Consequently, this study hypothesizes that a higher proportion of the secondary industry correlates with increased urban carbon emission intensity. The relevant data for this analysis is sourced from the “China City Statistical Yearbook” (2008–2020), with missing values addressed through smoothing techniques. Additionally, using 2010 as the base year, the corresponding provincial Consumer Price Index (CPI) is employed to deflate the relevant variables.

Variable definitions and descriptive statistical results are shown in

Table 1. Variable definitions and descriptive statistical results.

Var. Name	Definition	Obs.	Mean	SD	Min.	Max.
co2_ngdp	Carbon emission to GDP ratio	2690	3.434	4.867	0.001	44.939
lnzc	The logarithm of urban technology outflow	2690	4.036	1.921	0	9.590
lnzr	The logarithm of urban technology inflow	2690	3.896	1.988	0	9.747
lnperngdp	The logarithm of per capita GDP	2690	10.690	0.592	8.576	13.571
lnroad	The logarithm of urban road area	2690	17.080	7.289	1.37	66.493
lnfin	The logarithm of the year-end balance of various loans by financial institutions	2690	0.950	0.572	0.165	9.622
fdi	Foreign direct investment (FDI) to GDP ratio	2690	0.019	0.036	0.001	0.599
induss	The ratio of GDP from the secondary sector to the tertiary sector	2690	1.309	0.688	0.058	9.196

.

4. Results and Analysis

4.1. Spatial Correlation Analysis

This paper utilizes the global Moran’s I index statistic to evaluate spatial autocorrelation. Figure 1a–c provides a visual trend analysis of the Moran’s I index in relation to technology outflow, technology inflow, and carbon emission intensity across 269 cities in China from 2010 to 2019, assessed at a 95% confidence interval. The results indicate that the Moran’s I index for domestic technology transfer and carbon emissions in these cities during the specified period passed the significance test and yielded positive values. This finding suggests that urban technology transfer and carbon emissions are not isolated phenomena nor randomly distributed; rather, they exhibit significant spatial correlation, thereby supporting the establishment of Hypothesis 1. The unique geographical proximity associated with knowledge spillover contributes to the spatial agglomeration characteristics of scientific and technological innovation activities, which in turn subsequently influence the corresponding spatial agglomeration of technology inflow and outflow. Furthermore, the pronounced spatial agglomeration characteristics of urban carbon emission intensity imply the persistence of unbalanced development issues in urban carbon emissions. Consequently, neglecting spatial factors may lead to biased estimations, highlighting the necessity of selecting a spatial econometric model that accounts for spatial correlation factors.

4.2. Spatial Econometric Model Suitability Test

Although the spatial Durbin double fixed model serves as an initial framework, its applicability necessitates further validation. This study incorporates the Wald and Likelihood Ratio (LR) degeneration tests to evaluate whether the spatial Durbin model (SDM) can be reduced to the spatial autoregressive model (SAR) and the spatial error model (SEM). The results, as presented in

Table 2. Spatial econometric model suitability test.

Test Method	Test Purpose	Urban Technology Flow
		Statistical Measure	p-Value
Wald-lag	Whether to Degenerate to SAR Model	51.13	0.0000
Wald-err	Whether to Degenerate to SEM Model	51.75	0.0000
LR-lag	Whether to Degenerate to SAR Model	50.61	0.0000
LR-err	Whether to Degenerate to SEM Model	52.20	0.0000

, reveal that the p-values for the corresponding statistics of the LR and Wald tests are all below 0.01, indicating a statistically significant rejection of the null hypothesis concerning the degeneration to SAR and SEM models. Therefore, it can be concluded that the SDM model cannot be simplified to either the SAR or SEM models. Ultimately, this paper employs the spatial Durbin spatiotemporal double fixed model as the benchmark for analyzing empirical results.

4.3. Baseline Regression Results and Analysis

In order to facilitate a comparison of the reliability of regression outcomes, this study conducts regressions utilizing individual fixed effects, time fixed effects, and a combination of both individual and time fixed effects. The results from the Log-Likelihood (Log-L) analysis indicate that the spatial Durbin model (SDM) employing double fixed effects (Log-L: −4428.0189) demonstrates greater robustness than the models utilizing only individual fixed effects (Log-L: −4431.3614) or time fixed effects (Log-L: −7739.5594). As presented in

Table 3. Spatial econometric model basic regression results and decomposition of the effects of independent variables.

	Urban Technology Flow
	Direct Effect	Indirect Effect	Total Effect
Inzc	−0.0577 *** (0.0595)	0.2158 ** (0.1084)	0.1581 (0.1263)
Inzr	−0.0707 *** (0.0534)	−0.0122 (0.0943)	−0.0829 (0.1189)
lnpergdp	−1.1938 *** (0.1455)	−0.6578 *** (0.1932)	−1.8516 *** (0.1707)
lnroad	0.0103 (0.0091)	0.0457 *** (0.0178)	0.0560 ** (0.0200)
lnfin	0.0181 (0.1112)	−0.5788 *** (0.1910)	−0.5608 *** (0.2058)
fdi	0.1819 (0.9489)	−6.8198 *** (2.2341)	−6.6379 *** (2.4647)
induss	−0.1904 *** (0.0700)	0.7083 *** (0.1451)	0.5179 *** (0.1583)
Individual Effect	Y
Time Effect	Y
ρ	0.1178 *** (0.0257)
R2	0.1695
N	2690

The numbers in parentheses are standard errors; *** and **, respectively, indicate that the variable is significant at the 1% and 5% significance levels.

, the influence coefficient of the spatial lag term of the dependent variable indicates that the regression outcome for urban carbon dioxide emissions under the double fixed effects model significantly enhances the carbon emission levels of surrounding cities at the 1% significance level, with an influence coefficient of 0.1178. This finding suggests a notable spatial spillover effect of carbon emissions between regions, thereby substantiating the validity of Hypothesis 2. Furthermore, the efficiency of resource allocation, the development of industrial structures, and the degree of openness to external influences among neighboring cities exhibit a certain degree of spatial interaction. Local enterprises tend to align their energy consumption methods and levels with those of neighboring cities through processes of mutual imitation and reference, resulting in a characteristic interdependence and spatial dependence regarding carbon emissions between local and adjacent regions. This underscores the notion that carbon reduction is not merely a localized environmental concern, but rather a collective responsibility shared among regions. Consequently, it is imperative for regions to transcend administrative boundaries, engage in collaborative governance, coordinate emissions reduction efforts, and advance the attainment of “dual carbon” objectives from the perspective of a community with a shared future for humanity.

The spatial Durbin model introduces complexities in the interpretation of the coefficients associated with independent variables, primarily due to the significant presence of spatial feedback effects among regions. Existing literature often interprets the regression coefficients of influencing factor variables and their spatial lag terms, as illustrated in Table 3, as indicative of the corresponding influencing mechanisms and spatial spillover effects. LeSage and Pace (2009) [44] contend that a more nuanced interpretation is necessary, achieved through the analysis of direct and indirect effects derived from partial differentiation. The direct effect, commonly referred to as the within-area effect, signifies the average impact of the independent variable on the dependent variable within a specific region. In contrast, the indirect effect, also known as the between-area effect or spatial spillover effect, reflects the average impact of the independent variable on the dependent variable in adjacent regions. The results presented in the effect decomposition of Table 3 indicate that the direct effect coefficient of urban technology outflow is −0.0577, with a significance level of 0.01, while the indirect effect coefficient is 0.2158, with a significance level of 0.1084. Conversely, the direct effect coefficient of urban technology inflow is −0.0707, which meets the 1% significance level test, whereas the indirect effect is not statistically significant. These findings suggest several implications: 1. Both urban technology outflow and urban technology inflow exhibit significant direct effects on carbon emission reduction. This highlights the critical role of technology transfer as a critical factor in enhancing the ecological environment of a region. Technology transfer may facilitate the dissemination of green and low-carbon technologies, encouraging enterprises to improve production processes, adopt sustainable technologies, and optimize regional energy consumption structures during industrial transformation and upgrading. Additionally, technology transfer may stimulate research and development (R&D) efforts and enhance the capacity for the absorption and assimilation of green technologies through effective talent allocation and capital aggregation, thereby strengthening the independent R&D capabilities for green and low-carbon technologies and creating a scale effect that contributes to carbon emission reduction within the region. 2. Urban technology outflow demonstrates a significant spillover effect that inhibits carbon emission reduction in surrounding cities. This phenomenon may be attributed to the fact that cities engaged in active technology outflow typically possess a high level of scientific and technological R&D and are often regional central cities or provincial capitals with established channels for scientific and technological collaboration. The implementation of urban technology outflow may further reinforce the competitive advantages of these central cities, exerting a stronger siphon effect on neighboring cities and hindering the advancement of green and low-carbon technologies in those areas, thereby leading to a spatial spillover effect that impedes carbon emission reduction. 3. Urban technology inflow does not exhibit a significant spillover effect on carbon emission reduction. This lack of effect may arise from the strong competitiveness and exclusivity of resources among enterprises, resulting in ineffective technology diffusion to surrounding cities and, consequently, no significant indirect carbon emission reduction spillover effect. Additionally, the surrounding areas may lack sufficient technical stock, technical absorption and assimilation capacity, or may be hindered by inadequate economic conditions and policy environments, which collectively prevent the realization of a tangible carbon emission reduction spillover effect. In conclusion, this study provides support for Hypotheses 3, 5, and 6, while Hypothesis 4 is not substantiated.

4.4. Robustness Analysis

To further evaluate the effectiveness of urban technology transfer on carbon emissions, this study conducts a robustness test by modifying the spatial weight matrix. The spatial weight matrix is known to significantly influence the result of spatial econometric regression analyses. In this investigation, the geographical adjacency matrix utilized in the initial regression is replaced with both a composite matrix and a border length matrix. The composite matrix is generated by multiplying the geographical adjacency matrix with the geographical inverse distance matrix, thereby accounting for both geographical proximity and spatial distance relationships among regions. The weight assigned in the geographical inverse distance matrix is determined by the reciprocal of the distance between two cities, expressed as W_ij = 1/d_ij. Conversely, the weight in the border length matrix is based on the length of adjacent borders between cities, defined as W_ij = g_ij (where cities i and j are adjacent), and W_ij = 0 otherwise. This border length matrix facilitates a more nuanced understanding of regional adjacency relationships. The economic weight matrix was not selected for this analysis due to its higher potential for endogeneity, which could compromise the scientific rigor of the regression model. The results of the robustness regression are presented in

Table 4. Robustness test regression results and decomposition of independent variable effects under the border length matrix.

	Composite Matrix			Border Length Matrix
	Direct Effect	Indirect Effect	Total Effect	Direct Effect	Indirect Effect	Total Effect
Inzc	−0.0666 ** (0.0610)	0.0521 * (0.1098)	−0.0146 (0.1330)	−0.0368 ** (0.0609)	0.1485 ** (0.0989)	0.1116 (0.1189)
Inzr	−0.0638 *** (0.0611)	0.1005 (0.0949)	0.0367 (0.1202)	−0.0671 ** (0.0534)	−0.1144 (0.0813)	−0.1819 (0.1190)
ρ	0.1041 *** (0.0263)			0.0486 ** (0.0220)
R2	0.11661			0.1654
N	2690			2690

The numbers in parentheses are standard errors; ***, **, * respectively indicate that the variable is significant at the 1%, 5%, and 10% significance levels.

. It is evident that, regardless of whether urban technology is flowing in or out, the regression outcomes remain largely consistent with those of the initial regression after altering the spatial weight matrix. This finding suggests that the spatial Durbin model employed in this study effectively captures the impact of the variables on carbon emissions.

To enhance the verification of the reliability of the stability test,

Table 5. Results of suitability test for changing spatial weight matrix.

Test Method	Test Purpose	Composite Matrix		Border Length Matrix
		Statistical Measure	p-Value	Statistical Measure	p-Value
Wald-lag	Whether to Degenerate to SAR Model	43.69	0.0000	65.70	0.0000
Wald-err	Whether to Degenerate to SEM Model	44.44	0.0000	67.49	0.0000
LR-lag	Whether to Degenerate to SAR Model	43.24	0.0000	64.94	0.0000
LR-err	Whether to Degenerate to SEM Model	44.92	0.0000	66.77	0.0000

presents the results of the Wald test and the Likelihood Ratio (LR) test following modifications to the spatial weight matrix. The findings indicate that the Wald-lag, Wald-error, LR-lag, and LR-error tests all significantly reject the null hypothesis of degeneration to the spatial autoregressive (SAR) model and the spatial error model (SEM) at the 1% significance level. This suggests that the spatial Durbin model (SDM) is more reliable and robust in comparison to both the SAR and SEM models.

4.5. Regional Heterogeneity Analysis

Given the vast geographical expanse of China, significant disparities are evident in regional resource endowments, economic development, technological capabilities, and levels of environmental regulation. This study aims to evaluate whether variations in the impact of technology transfer on carbon emissions manifest across different regions. To achieve this, the cities within the country are categorized into four distinct groups, eastern, central, western, and northeastern regions, thereby enabling an analysis of regional heterogeneity. The results of the conducted test are presented in

Table 6. Regional regression results.

	Urban Technology Outflow			Urban Technology Inflow
	Direct Effect	Indirect Effect	Total Effect	Direct Effect	Indirect Effect	Total Effect
East	−0.0387 ** (0.0589)	0.3049 (0.1040)	0.2661 ** (0.1241)	−0.0889 ** (0.0550)	−0.3876 *** (0.0912)	−0.4766 *** (0.1133)
Central	−0.1204 ** (0.1096)	−0.4661 (0.1728)	−0.3457 * (0.1902)	−0.3209 *** (0.0939)	0.3893 ** (0.1243)	0.0684 (0.1455)
West	0.1017 (0.1245)	0.8824 *** (0.1850)	0.9841 *** (0.2324)	−0.0495 (0.1077)	0.4318 ** (0.1619)	0.4813 ** (0.2137)
Northeast	−0.1247 (0.1427)	−1.0947 ** (0.2578)	−1.2193 *** (0.3141)	0.2510 ** (0.1159)	−0.0639 *** (0.1824)	0.3148 *** (0.2218)

The numbers in parentheses are standard errors; ***, **, * respectively indicate that the variable is significant at the 1%, 5%, and 10% significance levels.

.

The analysis of urban technology outflow indicates that the estimated coefficients for the direct effects in the eastern and central regions are both significantly negative at the 5% significance level, with impact coefficients of −0.0387 and −0.1204, respectively. This finding suggests that urban technology outflow in these regions plays a significant role in reducing local carbon emissions. In contrast, the indirect effects are not statistically significant, implying that urban technology outflow does not produce a notable reduction in carbon emission in neighboring cities within the eastern and central regions. In the western and northeastern regions, the direct effects are not statistically significant; however, the indirect effects are significant at the 1% and 5% levels, respectively. Notably, the impact coefficient for the western region is 0.8824, while that for the northeastern region is −1.0947. This suggests that urban technology outflow in the western region does not significantly contribute to local carbon emission reduction and is, in fact, associated with an increase in carbon emissions in surrounding cities. Conversely, the technology outflow from the northeastern region significantly enhances carbon emission reduction in neighboring cities. Several factors may explain these findings: 1. The eastern and central regions exhibit relatively high levels of economic development, characterized by stronger environmental regulations and increased awareness of green low-carbon practices. Consequently, the transfer of technologies that are not conducive to green development can significantly facilitate local carbon emission reductions. Such technologies are typically transferred to the western and northeastern regions, which possess relatively underdeveloped industrial structures, thereby failing to exert a significant impact on the carbon emissions of neighboring cities in the eastern and central regions. 2. The technology market in the western and northeastern regions is relatively underdeveloped, resulting in a scarcity of scientific and technological innovation resources. This limitation hampers the effective flow of technology resources through outflow activities. Additionally, the concentration of labor-intensive manufacturing industries in these regions diminishes the potential for technology outflow to significantly promote local carbon emission reductions. Compared to the northeastern region, the western region has a weaker foundation in advanced manufacturing industries; thus, even outdated technologies can enhance labor productivity. Consequently, the technologies introduced by western cities are predominantly transferred to other cities within the western region, leading to a significant increase in carbon emissions in neighboring cities as a result of urban technology outflow. In contrast, the northeastern region experiences a pronounced homogenization of industrial structures, characterized by high industry similarity and a predominance of resource-intensive industries. As a result, recipients of technology outflow are often neighboring regions, allowing technology outflow to partially address the green technology challenges faced by these areas, thereby significantly mitigating carbon emissions in neighboring regions.

From the perspective of urban technology inflow, the direct effects on cities in the eastern and central regions are significantly negative at the 5% and 1% levels, respectively, with impact coefficients of −0.0889 and −0.3209. The indirect effects for both regions are significant at the 10% level, with impact coefficients of −0.3876 and 0.3893. This suggests that technology inflow in eastern cities positively influences carbon emission reduction in both local and surrounding areas, whereas technology inflow in central cities has a significant local carbon-reducing effect but results in increased carbon emissions in adjacent regions. Furthermore, the direct effect in the western region is not statistically significant; however, the indirect effect is significant at the 5% level, with an impact coefficient of 0.4318. This indicates that urban technology transfer in the western region does not substantially contribute to local carbon emission reduction and may even exacerbate carbon emissions in surrounding areas. In the northeastern region, the direct effect is significantly positive at the 5% level, with an impact coefficient of 0.2510, while the indirect effect is significantly negative at the 1% level, with an impact coefficient of −0.0639. This implies that technology inflow in northeastern cities leads to an increase in local carbon emissions while effectively promoting carbon emission reduction in surrounding areas. Several factors may account for these findings: (1) The eastern and central regions possess relatively advanced levels of technological innovation and high-end manufacturing, making them more receptive to external low-carbon technologies. Consequently, technology inflow can enhance local technological innovation and provide a foundation for carbon emission reduction. Compared to central cities, the more developed eastern cities host numerous universities, research institutes, and enterprises engaged in green technology innovation, facilitating closer interactions among various innovative entities and resulting in pronounced knowledge spillover effects. Central cities, facing developmental pressures, exhibit significant disparities in regional economic and technological development. technology inflow may lead to the relocation of energy-intensive industries, such as manufacturing, to less developed surrounding areas, thereby transferring carbon emissions along with the industries and significantly increasing emissions in adjacent cities. (2) Although the industrial structure in the western region is progressively rationalized, with high-energy-consuming and high-polluting industries being gradually phased out, carbon emissions theoretically should continue to decline. However, the western region suffers from inadequate infrastructure and a limited low-carbon technology foundation, still relying on traditional energy sources during the industrial structure upgrade process, which may contribute to increased carbon emissions. The inability to effectively balance economic interests with environmental sustainability may result in the direct effect of urban technology inflow in the western region being insignificant. Additionally, the importation of non-green, low-carbon innovative technologies may also extend to surrounding areas with lower technological capabilities, thereby generating a significant positive spillover effect. (3) The industrial structure of older industrial cities primarily follows a “secondary, tertiary, primary” model, with insufficient industrial upgrading. Consequently, technology inflow in northeastern cities tends to enhance energy efficiency and energy consumption, leading to increased local carbon emissions. Simultaneously, technology inflow may exert a siphoning effect on the population and financial resources of surrounding areas, thereby significantly reducing the carbon emission intensity of those regions.

5. Conclusions and Recommendations

This paper investigates the impact of urban technology transfer on carbon emissions and its spatial spillover effects, utilizing a spatial Durbin model with a geographically weighted matrix. This analysis is grounded in technology transfer and carbon emission data from 269 prefecture-level cities in China, spanning the period from 2010 to 2019. This study yields several key findings: Firstly, there exists a significant positive spatial correlation among technology inflow, technology outflow, and carbon emissions across Chinese cities. Additionally, a notable spatial spillover effect in carbon emissions is observed among these cities. Secondly, the basic regression analysis reveals that urban technology inflow and outflow have a significant inhibitory effect on local carbon emissions. However, concerning the carbon emissions of neighboring cities, technology inflow does not contribute to a reduction spillover effect, while technology outflow is associated with an increase in carbon emissions spillover effects. Lastly, the regional heterogeneity analysis reveals that technology inflow and outflow in the eastern and central regions demonstrate significant direct effects in reducing carbon emissions. In contrast, the direct carbon reduction effects of technology inflow and outflow in the western and northeastern regions are not statistically significant; notably, urban technology inflow in the northeastern region is linked to an increase in carbon emissions. Regarding indirect effects, only technology inflow from the eastern region and both technology inflow and outflow from the northeastern region demonstrate significant spillover effects in reducing carbon emissions. Conversely, the spillover effects of technology outflow from the eastern region and central regions, as well as technology inflow and outflow from the western region, are associated with a significant increase in carbon emissions.

In consideration of the conclusions presented above, this study proposes the following policy recommendations.

Firstly, local governments should, in alignment with the prevailing circumstances related to technology transfer and carbon emissions, as well as the overarching directives of the central government, transcend administrative boundaries. They should promote the development of regional technology transfer initiatives that are tailored to local conditions and collaboratively address regional challenges associated with carbon emissions.

Furthermore, it is imperative to further develop and refine the technology transfer system while optimizing the market allocation of resources and environmental components. By enhancing the cyclical processes associated with technology, we can improve the efficiency of the flow and allocation of technological resources and leverage the innovative collaboration and coordination capabilities inherent within society. The acceleration of technology aggregation, research and development, as well as the promotion and application of advanced energy-saving and decarbonization technologies, are critical for optimizing and adjusting the energy consumption structure. These initiatives will act as primary catalysts for carbon reduction and emissions mitigation, thereby facilitating the growth and advancement of the new energy sector.

In conclusion, addressing the disparities in the impact of technology transfer on carbon emissions across different regions requires a more strategic approach to the radiating and demonstration effects associated with the inflow and outflow of technology in the eastern and central regions. It is crucial to highlight the “latecomer advantage” of technology inflow in the western region, while simultaneously enhancing the synergistic carbon-reducing spillover effects of technology outflow from the eastern, central, and western regions on neighboring cities. Additionally, it is imperative to mitigate the existing development paradigms that foster technology and “beggar-thy-neighbor” policies.