Technological diffusion and cooperation in emission reduction alliances: Reshaping regional paths to low-carbon development

Abstract

Amid escalating climate pressures and uneven regional development, cross-regional technology cooperation has become essential for coordinated low-carbon transitions. This study proposes a novel “knowledge-equipment-institution” synergy framework to address gaps in understanding technology diffusion mechanisms and their emission reduction impacts within China’s regional diversity. Four regional cooperation alliances are identified, distinguishing between technology frontier and backward provinces. A multi-regional dynamic computable general equilibrium (CGE) model is constructed to assess cooperation outcomes. Two technology diffusion channels are integrated in the model: knowledge spillovers and machine and equipment (M&E) trade. Our findings reveal a fundamental trade-off between short-term efficiency and long-term resilience. Cooperation through M&E trade yields immediate energy saving and GDP growth benefits, but its long-term effectiveness is constrained. In contrast, cooperation via knowledge spillovers, despite modest short-term economic costs, fosters greater long-term adaptive capacity. The effectiveness of cooperation is highly heterogeneous across regions. For instance, the Beijing–Tianjin–Hebei–North-east alliance shows fluctuating carbon productivity, while the Yangtze River Economic Belt experiences productivity losses due to heavy industry lock-in. Ultimately, we find that a win-win dynamic is achievable, where technology frontier provinces accelerate industrial upgrading (tertiary sector share + 0.15%) and backward provinces enhance production efficiency. However, a sustained partnership hinges on the backward provinces’ commitment to enhancing their own absorptive capacity to avoid “free-riding” and low-level equilibrium traps. The findings provide a quantitative basis for designing differentiated and phased regional cooperation strategies to support adaptive climate governance.

Graphical abstract

Keywords

technology diffusion / carbon emission reduction / technical cooperation alliances / knowledge capital / machinery-equipment trade

1 Introduction

Technological change and diffusion are fundamental drivers of long-term economic growth and development (Barbieri et al., 2023). In an era of global economic integration, technology diffusion, facilitated by trade, investment, and factor mobility, has profoundly reshaped regional development patterns (Bradlow and Kentikelenis, 2024; Mercure et al., 2014). Meanwhile, latecomer regions are increasingly constrained by the technological path dependencies of more advanced areas (Baldwin, 2016). The cross-regional diffusion of low-carbon technologies has emerged as a crucial mechanism for bridging technological disparities and accelerating global decarbonization (Barbieri et al., 2023; Nicholas et al., 2013). Established in 2015, the Paris Agreement speaks of the vision of fully realizing technology development and transfer for improving resilience to climate change and reducing GHG emissions. As the world’s largest developing country, China’s uneven resource endowments and industrial structures have widened China’s provincial technological disparities (Jiang, 2021; Rohe and Mattes, 2022). China’s achievement of its “dual carbon” goals depends on systematically addressing regional disparities in low-carbon technological capabilities. Therefore, we consider the possibility of establishing an institutional framework for cross-regional technology cooperation (Sun et al., 2025) based on the characteristics of technology diffusion. It aims to transform discrete technology transfer into systemic capacity building by contractual mechanisms, essential for breaking regional low-carbon lock-in and rebalancing development rights with emission reduction responsibilities.

However, both theory and practice show that the pathway to effective technology cooperation is fraught with persistent challenges. These challenges can be broadly categorized into two types (Jiang et al., 2023). The first category involves endogenous barriers inherent in the technology diffusion process. For instance, the “black box” nature of the machinery and equipment (M&E) trade can trap recipients in an inefficient import-imitation cycle. This occurs when they lack sufficient capabilities to decode the technology (Garsous and Worack, 2022). Additionally, non-physical knowledge diffusion is often hampered by spatial decay and institutional barriers. Furthermore, technological-resource mismatches in latecomer regions may trigger ‘rebound effects’ that offset efficiency gains (Shi et al., 2025; Stern, 2022). The second category concerns institutional frictions at multiple levels that impede collaborative efforts. At the international level, progress is often stalled despite the vision set by frameworks like the Paris Agreement and the UNFCCC’s Technology Mechanism (Bortoletto et al., 2023). Key obstacles include conflicts over interest distribution (Yang and Sattar, 2024), intellectual property barriers (Carattini et al., 2024), and the persistent North–south technology divide (Bradlow and Kentikelenis, 2024). Similarly, within national borders, cooperation is also constrained. For example, in China’s pioneering regional integration zones (Liu et al., 2022; Shao et al., 2025), collaborative initiatives often progress slowly due to administrative barriers, differing path dependencies, and the lack of robust benefit-sharing mechanisms (Jin and Xu, 2024; Wang et al., 2024).

In summary, the existing literature provides a multifaceted understanding of technology cooperation. However, research on domestic cross-regional low-carbon technology cooperation, in particular, reveals three key limitations: (1) Most studies focus on single technology diffusion channels (e.g., patent transfers, equipment trade, foreign direct investment, or knowledge spillovers) rather than integrated mechanisms (Touboul et al., 2023; Witajewski-Baltvilks, 2025). (2) The research either evaluates historical cooperation performance or focuses on static policy design (Hattori et al., 2022), lacking dynamic simulations to characterize the evolutionary patterns of technology diffusion and assess its potential impacts. (3) Existing models often assume spatial homogeneity and a single-regional scale (Cui et al., 2022; De Paulo and Porto, 2023), neglecting China’s regional heterogeneity in resource endowments, institutional capacities, and technological capabilities. This simplification prevents models from capturing the nonlinear characteristics of technology diffusion and the heterogeneous effects of cooperation across regions. These shortcomings reveal theoretical gaps related to multi-dimensional cooperative path design, dynamic policy tools, and heterogeneous response mechanisms.

To address the identified gaps in the literature, this study operates within the policy framework of China’s Emissions Trading Scheme (ETS). We delineate multiple regional emission reduction alliances and adopt two core technology cooperation strategies within them: strengthening knowledge spillovers and adjusting cross-regional M&E trade flows (in terms of both volume and direction). We have developed a multi-regional dynamic computable general equilibrium (CGE) model. Using this model, we employ counterfactual scenario analysis to quantify the net benefits of cooperation compared to a baseline scenario. This paper then evaluates cooperation outcomes across three analytical levels—national, alliance, and provincial—to reveal the heterogeneous effects of technology cooperation. Our primary objective is to build an exploratory cooperative framework. This framework integrates M&E trade, knowledge spillovers, and alliance mechanisms in order to assess the potential benefits and constraints of low-carbon technology cooperation. Ultimately, our findings provide policymakers with actionable insights for developing differentiated low-carbon partnerships.

This paper makes three primary contributions. First, we establish regional low-carbon alliances focused on technology cooperation and identify each province’s role within them. We then construct a cross-regional ‘knowledge-equipment-institution’ framework to reveal the complementarities and conflicts among different forms of technology diffusion (Cameron, 1975; Knez, 2023). Second, we integrate endogenous technological change and productivity improvements arising from knowledge capital investment and equipment trade into a multi-regional dynamic CGE model. This approach captures the role and synergistic mechanisms of both tangible and intangible technology spillovers. Third, our study comprehensively analyzes the impacts of cooperation at the national, alliance, and provincial levels. This analysis explores multiple interests and differentiated cooperation responses. The findings provide new empirical evidence for the ‘core-periphery’ theory in economic geography (Krugman, 1991). More importantly, they offer a scientific basis for establishing differentiated and allied regional climate governance systems.

The remainder of the paper is structured as follows: Section 2 outlines the methodological framework for alliance technology cooperation, incorporating knowledge spillovers and M&E trade adjustment into the multi-regional dynamic CGE model. Section 3 illustrates the details of the counterfactual scenario setting and data sources. Section 4 presents the simulation results and analyzes the impact of technology cooperation. Section 5 offers concluding remarks.

2 Methodology

2.1 Research framework and theoretical foundations

To explore inter-regional technology cooperation on carbon mitigation, we develop a four-step research framework: (1) the research boundaries of technology cooperation are defined by the spatial spillover characteristics of technology and inter-regional economic interaction patterns within a national context. (2) The Emission Reduction Alliances are constructed through provincial spatial agglomeration, and the technological frontier is delineated based on technological capability. (3) extending a multi-regional dynamic CGE model with endogenous knowledge capital accumulation and decomposed trade modules to capture knowledge spillover and M&E trade adjustment. (4) The framework evaluates 2020–2060 policy scenarios using dynamic counterfactual analysis, systematically investigating the multi-dimensional impacts of cooperation at national, alliance, and provincial levels while assessing the potential of collaborative decarbonization pathways. The research framework is illustrated in Fig. 1.

Grounded in the theoretical frameworks established in earlier studies, technology generation and spread occur in two primary approaches: “standing on the shoulders of giants” and “standing on the shoulders of neighbors” (Bosetti et al., 2008; Sagar and Van der Zwaan, 2006). That means an open region’s knowledge creation involves autonomous technological innovation (knowledge accumulation from past innovations) and technology diffusion from other areas (Jin, 2016). Low-carbon technology diffusion is further categorized into two types: hardware-embedded (e.g., clean M&E trade) (Garsous and Worack, 2022; Popp, 2011) and knowledge-spillover (e.g., cross-regional R&D or FDI) (Leibowicz et al., 2016).

Technology diffusion follows a directional spatiotemporal evolutionary process. The “Schmidt’s Law” (Grubler et al., 2018) indicates that when technology spreads across regions, it creates core, rim, and periphery groups. The core, or technology frontier, with sufficient innovation capacity and economic motivation, leads to technology development. Technology diffuses from the core to the rim and then to the periphery, a pattern applicable to both physical and non-physical technologies diffusion (Leibowicz et al., 2016; Verdolini and Galeotti, 2011).

2.2 Emission reduction alliance formation and technology frontier identification

To promote cross-regional technology cooperation, we establish contractual emission reduction alliances between different regions in China. The formation of emission reduction alliances serves two key purposes. First, it narrows the scope of cooperation and clarifies partners, ensuring the establishment and maintenance of cooperative relationships. Second, it enables the evaluation of alliance performance, offering insights for designing differentiated cooperation strategies. The cooperation goal is to improve the efficiency of carbon emission reduction of the subjects within the alliance and the alliance as a whole to promote the process of carbon neutrality. Drawing on the main channels of technology diffusion (Pan et al., 2021), we categorize provinces within alliances into Technology Frontier (hereinafter TF) and Technology Backward (hereinafter TB) entities. The cooperation framework prioritizes two core modalities: (1) intensifying inter-provincial knowledge (non-physical technology) spillovers and (2) adjustment of inter-provincial M&E (physical technology) trade flows. Specifically, TF provinces (characterized by advanced R&D capabilities and manufacturing outputs) serve as knowledge dissemination hubs, while TB provinces (with lower technological readiness) focus on absorption and adaptation. This structure aims to optimize the spatial allocation of technological resources while mitigating path dependency in regional low-carbon transitions.

2.2.1 Constructing emission reduction alliances

A systematic approach to alliance member selection and role clarification forms the foundational step, ensuring compatibility between provincial resource endowments and assigned cooperative functions. Following the existing classification (Li and Xing, 2024), China’s 30 provincial-level administrative units are grouped into four alliances: Beijing–Tianjin–Hebei and North-East (BTHNE), Yellow River Basin (YRB), Yangtze River Economic Belt (YREB), and Pan-Pearl River Delta (PPRD). The constituent provinces of each alliance are listed in Table 1.

While this regional classification differs from traditional administrative divisions, its rationality is grounded in a set of integrated principles crucial for analyzing technology diffusion and collaborative governance:

• Alignment with national strategic regions. The YREB and YRB alliances directly correspond to the “Yangtze River Economic Belt Development” and the “Ecological Protection and High-Quality Development of the Yellow River Basin” strategies, respectively. Similarly, the BTHNE alliance integrates the “Beijing–Tianjin–Hebei Coordinated Development” and “Northeast Revitalization” strategies. The PPRD alliance encompasses the “Guangdong–Hong Kong–Macao Greater Bay Area” and its economic hinterland. This ensures the policy relevance of our framework and reflects the institutional reality within which cooperation occurs.

• Geographic proximity and economic interconnectedness. The provinces within the same alliance are geographically adjacent and share strong, pre-existing economic ties. This proximity minimizes transaction costs and facilitates the flow of goods, capital, and knowledge (Weidenfeld et al., 2021). The natural corridors of the Yangtze and Yellow Rivers (Jin and Xu, 2024; Ma et al., 2025) have historically fostered deep infrastructural and industrial integration within the YREB and YRB alliances. Likewise, the BTHNE region shares deep linkages in heavy industry and energy (Tian et al., 2019), while the PPRD is a highly integrated hub for manufacturing and trade (Li et al., 2018).

• Significant internal technology gradient. This division intentionally creates a “core-periphery” structure (Jia et al., 2025) within each alliance. By grouping technologically advanced provinces (potential TF provinces) with less developed ones (potential TB provinces), we ensure that a significant technology gap exists in each cooperative unit. This internal heterogeneity is a necessary precondition for meaningfully modeling and analyzing the dynamics of technology diffusion, knowledge spillovers, and the potential gains from cooperation (Xie et al., 2024).

2.2.2 Identifying the technology frontier within the alliance

This study employs a dual-criterion approach to identify technological frontiers within alliances according to physical and non-physical technology spillover characteristics. Based on the base year’s data, provinces within each alliance are ranked according to their knowledge capital stock and M&E output. The R&D capital stock reflects innovation intensity (Fujimori et al., 2023), while the production of M&E products reflects the R&D investment and the technology or knowledge involved in using these products (Weidenfeld et al., 2021). Thus, those ranking high in both dimensions are designated as TF (Technologically Frontier) provinces, while others are classified as TB (Technologically Backward). Provinces.

2.3 The CGE model

2.3.1 Core model architecture

This paper employs a widely utilized top-down framework—the multi-regional, multi-sectoral, and dynamic computable general equilibrium (CGE) model—to examine the economy’s response to shifts in climate policy, shocks, or exogenous factors. By simulating various policy scenarios, the CGE model offers insights into their potential economic ramifications and facilitates the comparison of alternative options. We aggregate the 42 sectors into 11 sectoral groupings. Following previous research, the model integrates modules for production, income-expenditure, trade, factors, closure, energy and emissions, and dynamics (Hu et al., 2023; Wang et al., 2009) (Fig. 2). We detail the production and trade modules in the methodology section. Descriptions of other modules are in the Electronic Supplementary Material.

2.3.2 Production block: Knowledge capital integration

In this study, a new factor, “knowledge capital,” is introduced to incorporate “knowledge” as an additional input in the production function (Garau and Lecca, 2015; Křístková, 2012). Knowledge creation is the source of endogenous technological change in the model. Additionally, modifications are made to the model’s nested structure, treating knowledge as a component of value-added and allowing for substitution between knowledge and other production factors (Sue Wing, 2008; Wang et al., 2009). The production structure of the model is illustrated in Fig. 3. As shown in Eq. (1), the intermediate input QINTA_(R,A) and value-added QVA_(R,A) are nested into departmental output QA_(R,A) through the Leo production function. Knowledge is assumed to be non-excludable and transferable across sectors (Sue Wing, 2008). Labor input QLD_(R,A), energy and capital bundle input QKE_(R,A), and knowledge input QHD_(R,A) are combined within a CES production function (Eq. (2)). QEE_(R,A) is the aggregate of fossil fuel QFOSI_(R,A) and electric power inputs ELE_(R,A). The subscripts R and A represent region and sector, respectively. The setting of the substitution elasticities in the production function are based on existing research (Hu et al., 2023; Parrado and De Cian, 2014; Witajewski-Baltvilks, 2025). These settings are presented in Table A1. The method of knowledge capital accounting is achieved in the Electronic Supplementary Material.

(1)\begin{array}{c}Q{A}_{R,A}=LE{O}_1\{QINT{A}_{\text{(R,A)}},CE{S}_2[QL{D}_{\text{(R,A)}},QH{D}_{\text{(R,A)}},\\ CE{S}_3(QK{D}_{\text{(R,A)}},CE{S}_4⟨QEL{E}_{\text{(R,A)}},QFOS{I}_{\text{(R,A)}}⟩)]\},\end{array}

(2)q=f\left(K,H,L\right)=A{[{\delta }^k{K}^{\rho }+{\delta }^h{H}^{\rho }+{\delta }^l{L}^{\rho }]}^{\frac{1}{\rho }},\hspace{0.33em}\hspace{0.33em}\hspace{0.33em}\hspace{0.33em}{\delta }^k+{\delta }^h+{\delta }^l=1.

The R&D capital stock determines the knowledge factor of sector A in region R in year T (H_{(\text{R,A,T})}). It is influenced by the R&D investment in time T (R{D}_{\text{H}}{}_{\text{(R,A,T)}}), the previous year’s R&D capital stock (H_{(\text{R,A,T-1})}), and the depreciation rate ({\delta }_{\text{H}}), as shown in Eq. (3). Based on existing studies, we set the depreciation rate for knowledge capital at 15% (Buonanno et al., 2003; Bye and Jacobsen, 2011; Wang et al., 2009), and physical capital at 3% (Jin, 2012; Křístková, 2012).

(3)H_{\text{(R,A,T)}}=H_{\text{(R,A,T-1)}}\times (1-{\delta }_{\text{H}})+R{D}_{\text{H}}{}_{\text{(R,A,T)}}.

2.3.3 Trade block: Multi-tiered market decomposition

Fig. 4 shows the commodity flows in the local market. The Armington assumption is used to describe the substitute between local (Q{D}_{\text{(R,C)}}), imports from other regions of the country (QM{R}_{\text{(R,C)}}), and imported goods (Q{M}_{\text{(R,C)}}) (Eq. (4)). Locally produced goods (Q{X}_{\text{(R,C)}}) are divided into local markets (Q{D}_{\text{(R,C)}}), exported to other regions of the country (QD{R}_{\text{(R,C)}}), and exported abroad (Q{E}_{\text{(R,C)}}) through the CET function (see Eq. (5)).

(4)\begin{array}{c}Q{Q}_{\text{(R,C)}}=\{Q{D}_{\text{(R,C)}},Ar{m}_1[Q{M}_{\text{(R,C)}},\\ Ar{m}_2\left(QMR{S}_{\text{(R,C,R1)}},\dots ,QMR{S}_{\text{(R,C,R30)}}\right)]\},\end{array}

(5)\begin{array}{c}Q{X}_{\text{(R,C)}}=\{Q{D}_{\text{(R,C)}},CE{T}_1[Q{E}_{\text{(R,C)}},\\ CE{T}_2\left(QDR{S}_{\text{(R,C,R1)}},\dots ,QDR{S}_{\text{(R,C,R30)}}\right)]\}.\end{array}

2.3.4 Carbon emission trading block

To achieve carbon neutrality, China’s future ETS should gradually expand its scope, introduce auction mechanisms, and reduce the proportion of free allowances (Chen et al., 2023; Jia and Lin, 2020). We assume that all sectors participating in the ETS follow the grandfathering allocation scheme. The description for the ETS module in the CGE model is shown as Eqs. (6)–(9):

(6)Quot{a}_{(\mathrm{A},t)}=T{A}_t\times (Q{A}_{(\mathrm{A},t\text{-}1)}/{\sum }_{\text{A}}Q{A}_{(\mathrm{A},t\text{-}1)}),

(7)C{C}_{(\mathrm{A},t)}=p\times (E{M}_{(\mathrm{A},t)}-{\beta }_{t}Quot{a}_{(\mathrm{A},t)}),

(8)C{R}_{(\mathrm{A},t)}=p\times (Quot{a}_{(\mathrm{A},t)}-E{M}_{(\mathrm{A},t)}),

(9){\sum }_{\text{A}}C{C}_{(\mathrm{A},t)}+{{\sum }_{\text{A}}CR}_{(\mathrm{A},t)}=p\times (1-{\beta }_t)T{A}_t,

where Quota_(A,t) is the quota of sector A in year t. TA_t is the total quotas of all sectors in year t, exogenously determined by government planning, the total emission cap is set as TA_t according to existing research (Hahn et al., 2024; Hu et al., 2023; Zickfeld et al., 2023). It is also shown in Fig. B3. QA_(A,t-1) is the last year’s domestic output in sector A. CC_(A,t) and CR_(A,t) are the carbon cost or carbon revenue in sector A. If the sector is not included in ETS, then CC_(A,t) and CR_(A,t) equal 0. p is the endogenous carbon price. {\beta }_{\text{t}} is the proportion of free quota. In this model, the carbon market is assumed to be perfectly competitive. Equilibrium is achieved when carbon supply meets demand, resulting in a uniform carbon quota price across all participating industries (Wen and Jia, 2023; Zhang et al., 2022b).

Drawing on the 2016 planning guidelines set by the National Development and Reform Commission (NDRC) and the four-phase experience of the EU-ETS, we propose a three-phase carbon ETS for China. In each phase, new sectors are integrated into the trading market, and the proportion of carbon quota auctions is progressively increased, as shown in Table 2.

2.4 Technology diffusion mechanisms in CGE modeling

2.4.1 Non-physical technology (knowledge) spillover

Under the framework of induced technological change, cross-regional knowledge spillovers occur when R&D investments made by TF provinces benefit TB provinces. The frontier knowledge pool is considered a public good, though only a fraction is absorbed. TB provinces need some absorption capacity to benefit from R&D in TF provinces (Romer, 1990; Weyant and Olavson, 1999). This paper defines knowledge spillovers using three key components: the frontier knowledge pool, absorptive capacity, and knowledge utilization (Bosetti et al., 2008; Jin and Zhang, 2016).

Frontier knowledge pool. The R&D capital stock is an essential indicator for measuring the advancement of non-physical technology (knowledge) (Hübler et al., 2012). Therefore, the sum of the historical cumulative R&D capital stock (H_{\text{(R,T)}}) of all TF provinces in the alliance constitutes a knowledge pool (KP). For TB provinces, the knowledge pool available for absorption in time T is K{P}_{\text{(R,T)}} (Bosetti et al., 2008).

(10)K{P}_{\text{(R,T)}}={\sum }_{\text{R}\in \text{TF}}{H}_{\text{(R,T)}}-H_{\text{(R,T)}},(R\in TB).

Absorbing ability. The knowledge-absorbing ability (Bosetti et al., 2008) of a TB province depends on its relative distance from the frontier knowledge pool ({\gamma }_{\text{(R,T)}}) (Hübler et al., 2012; Jin, 2015).

(11){\gamma }_{\text{(R,T)}}=\frac{H_{\text{(R,T)}}}{\sum _{\text{R}\in \text{TF}}^{}{H}_{\text{(R,T)}}}.

Knowledge utilization. Absorbed knowledge is incorporated into the R&D investments of TB provinces and depreciates over time. The updated knowledge capital stock equation accounts for locally generated and absorbed knowledge from TF provinces. Noting that, for any TB province, its knowledge absorption of all frontier provinces is non-discriminatory, regardless of whether the alliance exists. Based on Eq. (3), the knowledge capital stock of province R in year T after knowledge absorbing ({H^{\mathrm{\prime }}}_{(\text{R,A,T})}) can be expressed as follows:

(12)\begin{array}{rl}{H^{\mathrm{\prime }}}_{(\text{R,A,T})}=& {H^{\mathrm{\prime }}}_{(\text{R,A,T-1})}\times (1-{\delta }_{\text{H}})+R{D}_{\text{H}}{}_{\text{(R,A,T)}}+{\gamma }_{\text{a}}{}_{\text{(R,T)}}\\ & \times K{P}_{\text{a}}{}_{\text{(R,T)}}+{\gamma }_{\text{na}}{}_{\text{(R,T)}}\times K{P}_{\text{na}}{}_{\text{(R,T)}},(\mathrm{R}\in \mathrm{T}B),\end{array}

where K{P}_{\text{a}}{}_{\text{(R,T)}} and K{P}_{\text{na}}{}_{\text{(R,T)}} denote the alliance and non-alliance knowledge pools relative to region R (TB province), {\gamma }_{\text{a}}{}_{\text{(R,T)}} and {\gamma }_{\text{na}}{}_{\text{(R,T)}} correspond to their absorbing abilities, respectively.

2.4.2 Physical technology (machine-equipment) diffusion

The acquisition and use of new M&E are recognized as critical channels for regional technological advancement and economic growth (Long and Summers, 1991; Jaffe et al., 2005). Production technology is spatially transferred when M&E is traded across provinces from TF to TB provinces.

From M&E trade to productivity growth. Existing studies have shown that cross-regional M&E trade significantly increases the factor productivity (including physical capital (K), knowledge capital (H), and energy (E)) (Long and Summers, 1991; Jaffe et al., 2005), both in international and domestic trade contexts. This finding suggests that endogenous technological change induced by cross-regional trade can be captured in a CGE model by incorporating a module that links M&E trade to growth in capital and energy productivity (Carraro and De Cian, 2013). In this framework, the technology diffusion resulting from increased M&E imports is reflected through factor productivity gains. Furthermore, empirical evidence indicates a linear relationship (Parrado and De Cian, 2014) between changes in M&E trade () from TF provinces and changes in sectoral energy and capital productivity (gTF{P}_{(\text{R,A,F})}).

(13)

(14){\zeta }_{（\text{R,A,F}）}={\zeta }_{\text{0F}}M{S}_{\text{R}}C{R}_{(\text{R,TF})}C{S}_{(\text{R,A})},

where gTF{P}_{\text{(R,A,F)}} represents the growth rate of capital or energy productivity, denotes the growth rate of M&E trade from TF provinces, and {\zeta }_{（\text{R,A,F}）} is the elasticity coefficient. MS_R refers to the proportion of region R’s M&E output relative to the national total, reflecting the region’s share of total M&E production. A higher MS_R indicates a more significant knowledge base, suggesting that the region is better equipped to comprehend and utilize the knowledge and technologies embedded in imported machines and equipment. C{R}_{\text{(R,TF)}} measures the proportion of M&E imported by region R from the TF provinces relative to its total commodity import from TF provinces; C{S}_{\text{(R,A)}} captures the share of M&E purchased by sector A in region R relative to the region’s total M&E imports. C{R}_{\text{(R,TF)}} and C{S}_{\text{(R,A)}} serve as indicators of regional and sectoral preferences for sourcing M&E from the TF provinces, reflecting their willingness to engage in technology cooperation. Finally, {\zeta }_{\text{0}} represents the adjustment coefficient.

3 Scenarios and data

3.1 Carbon reduction mechanism of technology cooperation

Mechanism of carbon emission reduction through knowledge spillover enhancement. Based on the existing knowledge capital stock within alliance members, TF provinces decide whether to join the cooperation by increasing knowledge capital investment in eight carbon-intensive sectors, thereby expanding the knowledge pool; TB provinces decide whether to join the cooperation by increasing knowledge capital investment in eight carbon-intensive sectors, thereby enhancing their absorption ability. For TF provinces, increased knowledge capital in eight carbon-intensive sectors drives carbon emission reduction directly through two main channels. First, it stimulates a shift in the structure of production factor inputs, accelerating the replacement of carbon-intensive technologies in high-emission sectors, optimizing the energy mix, and reducing emissions (Gillingham et al., 2008). Second, it fosters the growth of knowledge-intensive industries, promoting a low-carbon industrial transition. The exact mechanisms apply to TB provinces. Moreover, the improved knowledge absorption capacity further amplifies the TB provinces’ emission reduction effects, accelerating the overall decarbonization process (An et al., 2023; Goulder and Mathai, 2000). The emission reduction mechanism is illustrated in Fig. 5.

Mechanism of carbon emission reduction of M&E trade adjustment. According to Eq. (13), increasing the procurement of M&E from TF provinces could directly enhance TB provinces’ capital and energy factor productivity. Specifically, improvements in capital productivity drive higher sectoral output, while enhancements in energy productivity lead to higher energy efficiency (Parrado and De Cian, 2014). The combined effect of these two improvements accelerates the achievement of emission reduction targets. Compared to the carbon reduction pathway driven by knowledge spillovers, the impact of trade adjustment on emissions reduction is more direct. The specific mechanism is illustrated in Fig. 5.

3.2 Scenario setting

The CGE model, by its nature, is designed to conduct “what-if” analyses to quantify the potential economic and environmental outcomes under specific, predefined policy shocks. Therefore, our cooperation scenarios assume that the institutional arrangements for cooperation have been successfully established, and we do not endogenously model the complex political bargaining or specific contractual incentives that would lead to this outcome. The primary goal is to assess the potential benefits and costs of cooperation, which in turn provides a quantitative basis for designing the necessary real-world incentive mechanisms, as discussed in the implications Subsection 5.2.

3.2.1 Baseline (Non-cooperative) scenario

Baseline Scenario (Base). The baseline scenario in this study is not a regular business-as-usual (BAU) scenario but rather a counterfactual benchmark designed to quantify the net effects of the cooperation mechanisms. This baseline scenario accounts for the spatial spillover effects of knowledge and the productivity gains in capital and energy factors driven by M&E trade, yet it does not incorporate regional cooperation. Specifically, TB provinces maintain their existing M&E procurement patterns, while TF provinces do not take the initiative to increase knowledge capital investment. The following provides a detailed explanation of how to determine the key variables in the baseline scenario and their reference basis.

• GDP projection and validation. Provincial GDP growth is endogenously projected within the model, driven by fundamental assumptions about demographic trends and labor productivity that are consistent with China’s long-term development goals. To ensure the realism of our baseline, this paper validates the simulated GDP trajectories against actual historical data for each province up to 2023, with the comparison charts provided in Figs. B1 and B2.

• Carbon emissions pathway. The national carbon emissions trajectory is exogenously imposed on the model. This pathway is carefully designed to align with China’s official targets of peaking emissions before 2030 and achieving carbon neutrality by 2060. Crucially, this national carbon cap is identical across all scenarios (both baseline and cooperation scenarios). This clarifies that our study does not assess if the climate goal is met, but rather evaluates the differing economic impacts and structural changes of achieving this pre-defined goal through different cooperation pathways.

• Energy and power structure. The evolution of the energy and power mix is endogenously determined by the model. It is a result of the complex interplay between economic growth, technological progress (such as TFP growth), and the national carbon cap, rather than being an external assumption.

3.2.2 Single cooperation scenarios

Knowledge spillover with unilateral investment (KUS). In this scenario, only the TF provinces within the alliance increase R&D capital investment in eight carbon-intensive sectors. The additional R&D investment, GR_H, is set at 10% of the previous year’s knowledge capital investment. The increased funding for R&D is sourced from the TF provinces’ government revenue. This setting simulates the impact of unidirectional knowledge spillovers, where TF provinces spill over knowledge to TB provinces. In contrast, TB provinces passively receive the spillover without making additional investments in local knowledge development. This scenario’s assumption is grounded in China’s top-down regional development strategies. It reflects the role of designated “growth poles” (Zheng et al., 2024) or “pilot demonstration zones” (Zhang et al., 2023). For instance, core cities like Shanghai (in the YREB) and Shenzhen (in the PPRD) are mandated by national strategy to lead in technological innovation. Their governments make substantial fiscal investments to enhance their own R&D capabilities.

Knowledge spillover with bilateral investment (KBS). The alliance’s TF and TB provinces enhance R&D capital investment in eight carbon-intensive sectors. The additional R&D investment is a fixed proportion (GR_H) of the previous year’s knowledge capital investment (GR_H = 10%). Under this scenario, TB provinces absorb more advanced knowledge from TF provinces and actively strengthen their R&D investment. This bilateral investment model is also highly realistic and is increasingly practiced in China’s major regional integration initiatives. It mirrors the establishment of “joint innovation funds” and “cross-regional collaborative R&D platforms” (Li et al., 2024; Liu et al., 2023), such as those within the Yangtze River Delta Science and Technology Innovation Community.

Trade adjustment of machine-equipment (T). TB provinces within the alliance increase the proportion of M&E purchased from TF provinces and decrease the proportion of M&E purchased from other provinces. The additional purchase is set as AR_T (AR_T = 10%). The TB and TF provinces do not adjust the R&D investment. The assumption of a targeted trade shift is supported by existing Chinese policies. These policies are designed to guide industrial procurement. Examples include government green procurement rules and the “First Set” equipment policy. Green procurement mandates the use of high-efficiency equipment. The “First Set” policy provides subsidies for firms to adopt advanced domestic machinery. Therefore, the T scenario models a plausible outcome of targeted industrial policies, not just a free-market result.

3.2.3 Combined cooperation scenarios

Integrated unilateral investment and trade (KUS-T). This scenario combines unilateral knowledge spillover with M&E trade adjustment. Based on the KUS scenario, TB provinces within the alliance increase the share of M&E imports from TF provinces. This setting explores the synergy between unidirectional knowledge absorption and hardware upgrading. TB provinces benefit from a dual strategy that enhances knowledge absorption while upgrading their technological base through imported equipment, leading to emission reduction gains.

Integrated bilateral investment and trade (KBS-T). This scenario combines bilateral knowledge investment with M&E trade adjustment. Based on the KBS scenario, TB provinces increase the proportion of M&E imports from TF provinces. This setting assesses the combined effect of bidirectional knowledge investment and hardware upgrading, where TB provinces enhance their R&D capacity of eight carbon-intensive sectors, strengthen knowledge absorption, and upgrade their technological base simultaneously, thereby achieving emission reduction benefits. Table 3 provides a structured overview of all the scenarios designed in this paper to facilitate readability and indexing.

3.3 Data sources and processing

This paper’s Social Accounting Matrix (SAM) includes data from China’s 26 provinces and four municipalities (excluding Hong Kong, Macao, Taiwan, and Xizang). The compilation of the Social Accounting Matrix (SAM) in the base year (2017) refers to the CEADs Mainland China Provincial MRIO Table for 30 Provinces (42 sectors) (Zheng et al., 2021). The R&D investment data are sourced from the China Science and Technology Yearbook and provincial statistical yearbooks. Calculations related to energy consumption, carbon emissions, total factor productivity, and parameter adjustments are consistent with previous research (Hu et al., 2023, 2020b). From 2017 to 2023, the regional GDP, industrial structure, and GDP estimates based on different accounting methods align closely with the statistics published by the National Bureau of Statistics. The General Algebraic Modeling System (GAMS) and the Mixed Complementarity Problem (MCP solver) solver are used to solve the CGE model. Parameter calibration is conducted based on statistical data from 2017 to 2023, with the calibrated results and parameter settings provided in the Electronic Supplementary Material.

4 Results analysis and discussion

4.1 GDP, welfare, and energy impacts

4.1.1 GDP growth dynamics

Simulation results in Fig. 6 show that different modes of technology cooperation have significantly differentiated impacts on the national GDP growth trajectory. Compared to the Base scenario, pure knowledge spillover cooperation (KUS, KBS) exhibits a unique “convergent recovery” pattern: GDP initially drops by 0.01%–0.02%, then gradually narrows until it greater than zero by 2060 in the KBS scenario. This pattern highlights the cyclic process of “learning-absorption-innovation” required for knowledge accumulation in eight carbon-intensive sectors. In the early stages, investments in knowledge capital crowd out other productive investments and lead to a decline in GDP. However, as technological absorption capacity improves, growth dividends are gradually realized (Aghion et al., 2009). Compared to the unidirectional knowledge flow in the KUS scenario, the KBS scenario enhances knowledge absorption and conversion efficiency through bidirectional investment, breaking the “core-periphery” dependency structure.

However, cooperation models dominated by M&E trade (T, KUS-T, KBS-T) consistently yield positive economic effects, with GDP growth following a “U-shaped” trajectory: the growth rate declines from 0.08% in 2020 to 0.04%–0.05% by 2035, then recovers to about 0.13% by 2060. This trend suggests that M&E trade cooperation facilitates the cross-regional capital and technology flows, enabling TB provinces to break through technological barriers and integrate into advanced production systems. During the early stages of cooperation, capital reallocation and industrial restructuring create short-term adjustment costs. However, economic growth momentum strengthens as technology spillover benefits materialize and industrial chains reorganize (Rodrik, 2018). The “J-curve Effect” during technology transfer explains the initial decline and subsequent rise of GDP growth rates.

Notably, when knowledge spillover is integrated into M&E trade adjustment (KUS-T, KBS-T), the crowding-out effect of productive investment results in relatively lower GDP growth rates than the pure M&E trade cooperative scenario (T). By 2060, the differences in economic effects among the three trade-dominated cooperation scenarios gradually converge, confirming the “path dependency saturation effect” in technology diffusion—when the intensity of M&E trade surpasses a critical threshold, the marginal benefits of cooperation diminish.

4.1.2 Household welfare effects

The simulation results in Fig. 6 reveal that all cooperation scenarios lead to varying welfare loss (The calculation of household welfare is shown in Appendixes). Pure knowledge spillover cooperation (KUS, KBS) results in relatively moderate welfare reductions, as increased knowledge capital investment in eight carbon-intensive sectors partially crowds out investments in public welfare sectors. The more significant welfare loss in the KBS scenario stems from TB provinces suffering from insufficient absorptive capacity and bearing additional sunk costs of knowledge capital inputs (Ronayne et al., 2021).

In contrast, the trade-dominated cooperation (T, KUS-T, KBS-T) shows an increasing trend in welfare losses, rising from about −0.35% in 2020 to −0.8% by 2060. This reflects the “dual cost shock” of M&E trade: in the short-term, the relocation of production M&E influences local employment opportunities, while in the long-term, technological lock-in exacerbates regional development imbalances (Rodrik, 2016). The intensity of welfare loss follows a consistent ranking: T < KUS-T < KBS-T, indicating that the synergy between knowledge spillovers and M&E trade amplifies welfare losses.

The observed trends in welfare loss across different cooperation scenarios reflect the necessity to further systematically review and optimize the institutional design of the cooperation framework. The GDP growth driven by M&E trade cooperation essentially shifts the cost of technology upgrading, such as labor skill transformation and industrial restructuring, to the household (−0.8% welfare loss). In knowledge spillover scenarios, although bidirectional knowledge investment (KBS scenario) helps mitigate technological disparities (as evidenced by continuous electrification improvements), the higher welfare losses (0.15% more than in the KUS scenario) suggest the implicit costs of knowledge spillover. This paper assumes that all additional R&D investment comes from government revenues. Thus, the extra social investments made by TB provinces in education and R&D to absorb external knowledge are converted to reductions in household consumption under the current accounting framework. This “cost-shifting” mechanism prevents the economic gains (GDP growth) and environmental benefits (reduced energy consumption) of technology cooperation from translating into welfare improvements, revealing the limitations of the “Kaldor-Hicks’s efficiency” principle (Idisi et al., 2018). Even if cooperation generates net overall benefits, without appropriate compensation mechanisms, welfare losses for specific groups may hinder the sustainability of cooperation. An appropriate subsidy mechanism or a requirement that part of the R&D investment originates from enterprises may alleviate this dilemma to a certain extent.

4.1.3 End-use energy consumption patterns and electrification trajectories

Figures 7(a) and 7(b) show that all technology cooperation modes can effectively reduce the national end-use energy consumption and optimize the energy consumption structure. However, compared with pure knowledge spillover cooperation (KUS and KBS), trade-dominated cooperation (T, KUS-T, KBS-T) demonstrates more substantial energy saving and upgrading potential through the “technology substitution effect” and “industry restructuring effect.” The former is the immediate energy-saving effect of advanced M&E replacing high-energy-consuming ones. At the same time, the latter optimizes the national industrial layout by transferring high-energy-consuming industries to low-technology provinces. The energy consumption reduction in the three trade-dominated cooperation scenarios shows a monotonically increasing trend: from −0.015% in 2020 to −0.06% by 2060. This reveals the “cumulative energy saving effect” of equipment updates: large-scale replacement of advanced M&E immediately improves energy efficiency and continuously optimizes energy consumption through technological lock-in effects (Arthur, 1989; Goldstein et al., 2023). The continuous increase in the proportion of non-fossil energy consumption also provides further evidence for this.

The impact of different technology cooperation modes on the national electrification rate and the share of fossil fuel power generation (Figs. 7(c) and 7(d)) shows significant differentiation. Unidirectional knowledge capital investment (KUS) fails to boost the electrification rate, remaining below the Base scenario until 2060. This partly confirms the absorption capacity threshold theory (Cohen and Levinthal, 1990): TB provinces’ knowledge capital accumulation must reach a critical level to convert external knowledge effectively flows into technological change; otherwise, knowledge spillovers face transmission blockages due to excessive “technological gradient.” Bilateral knowledge capital investment scenarios (KBS) show a sustained increase in electrification rate, proving that enhancing TB provinces’ absorption capacity through simultaneous knowledge capital input further optimizes the knowledge spillover transmission efficiency. It forms a positive cycle of “knowledge-capability-innovation.” Trade-dominated cooperation scenarios (T, KUS-T, KBS-T) significantly improve the electrification rate. Still, this effect diminishes over time: from 2020 to 2040, the growth rate continuously narrows, and the T scenario turns to negative growth compared to the Base scenario after 2041. This indicates that the short-term efficiency improvements from mere factor mobility (M&E trade) cannot meet long-term low-carbon transition needs. Initially, advanced equipment replaces traditional energy equipment, rapidly increasing the electrification rate. However, when the TB provinces lack accompanying knowledge-capital accumulation, the embedded technology’s digestion and absorption efficiency in M&E will decrease with technological generational leaps. This will eventually exhaust the equipment update potential and stabilize the energy demand structure (Leibowicz et al., 2016).

Notably, the ranking of energy consumption and electrification rate improvement effects in trade-dominated cooperation scenarios (T < KUS-T < KBS-T) further confirms that knowledge accumulation improves efficiency in the use of M&E, and M&E updates provide a carrier for knowledge application, thus creating a “1+1>2” system synergy. It also reaffirms the importance of “bilateral knowledge flow”: the TB provinces’ absorption capacity building is the key to cooperation effectiveness, adding a receptor constraint condition to traditional technology diffusion theories (Bosetti et al., 2008; Leibowicz et al., 2016; Verdolini and Galeotti, 2011).

Combining the impacts of cooperation on GDP, welfare, and energy consumption reveals that M&E trade cooperation and knowledge spillover cooperation represent two distinct paths of technology diffusion: “hardware efficiency” and “software capability.” They show significant decoupling in time and space. M&E trade can boost production efficiency and energy saving in the short term through the physical embedding of M&E. Still, the electrification rate’s decay trajectory (turning negative in 2041) exposes the fatal flaw of “rootless technology.” Without continuous local knowledge capital investment, hardware technology cannot adaptively iterate, causing technological generational gaps. In contrast, despite initially suppressing GDP growth and worsening welfare losses, the “knowledge-capability-innovation” chain built by the knowledge spillover path demonstrates long-term growth potential, confirming the “efficiency and resilience” trade-off in technology diffusion. It means that immediate gains from physical technology diffusion (M&E trade) come at the expense of system adaptability, while nonphysical technology (knowledge) building exchanges short-term costs for long-term transformation resilience.

4.2 Alliance-level carbon reduction efficiency

Given the heterogeneity in natural resource endowment, industrial structure, and economic development stage, it is essential to examine the impact of cooperation from each alliance’s perspective. Focusing on the core goal of inter-provincial technology cooperation, i.e., to enhance regional carbon reduction efficiency through collaborative innovation, this paper takes the dynamic evolution of alliance carbon productivity (CP) and the response of eight carbon-intensive sectors’ energy intensity (EI) as core indicators (Zhang et al., 2022a), building a contrastive analysis framework between non-cooperation (Base) and cooperation scenarios. The changes in the alliance’s carbon productivity (\mathrm{\Delta }C{P}_{\text{alliance}}) and energy intensity (\mathrm{\Delta }E{I}_{\text{alliance}}) are as follows.

(15)\mathrm{\Delta }C{P}_{\text{alliance}}=\frac{C{P}_{\text{alliance,CS}}-C{P}_{\text{alliance,Base}}}{C{P}_{\text{alliance,Base}}}\times 100\mathrm{\%},

(16)\mathrm{\Delta }E{I}_{\text{alliance}}=\frac{E{I}_{\text{alliance,CS}}-E{I}_{\text{alliance,Base}}}{E{I}_{\text{alliance,Base}}}\times 100\mathrm{\%},

where C{P}_{\text{alliance,CS}} and E{I}_{\text{alliance,CS}} are the alliance’s carbon productivity and energy intensity of cooperation scenarios (CS), respectively; C{P}_{\text{alliance,Base}} and E{I}_{\text{alliance,Base}} are the alliance’s carbon productivity and energy intensity of the non-cooperation scenario (Base). The cooperation mechanism established in this study may alter the emission reduction efficiency gaps between and within alliances through factor recombination. Moreover, the changes in these gaps offer crucial insights into the regional economic system’s collaborative response mechanism. Therefore, they are also what we focus on in the result analysis.

4.2.1 Changes in alliances’ carbon reduction efficiency

Under trade-dominated cooperation scenarios, cooperation significantly affects the immediate control of energy intensity in eight carbon-intensive sectors (except for the YRB alliance). However, the long-term effects of cooperation on carbon productivity vary across alliances (Fig. 8).

Beijing-Tianjin-Hebei and North-east (BTHNE) alliance. Carbon productivity (\mathrm{\Delta }C{P}_{\text{BTHNE}}) exhibits a double-peak fluctuation, reflecting the dynamic interplay between policy sensitivity and technology diffusion. Initially, high-efficiency M&E replaces outdated production capacity, creating a peak. Later, technological adaptation recedes. In 2035, the reduced free carbon allowances triggered a secondary wave of technological upgrades, though the policy shock gradually weakened.

Yellow River Basin (YRB) alliance. After cooperation, the growth rate of carbon productivity (\mathrm{\Delta }C{P}_{\text{YRB}}) outpaces other alliances, attributed to the synergistic effect of carbon allowance constraints and technological upgrades. The carbon market mechanism accelerates the elimination of coal-dependent industries and redirects factor inputs to low-carbon sectors. However, the energy intensity of eight carbon-intensive sectors in this alliance (\mathrm{\Delta }E{I}_{\text{YRB}}) decreases and rebounds. As a coal-rich region, cooperation initially lowers energy intensity but fails to alter the energy consumption structure (coal consumption still exceeds 60%). With deepening cooperation, producers may expand capacity, or consumers may increase energy consumption due to the cost advantage from energy efficiency improvements, leading to the “energy rebound effect” (Rodrik, 2018). Consequently, the \mathrm{\Delta }E{I}_{\text{YRB}} reduction effect stagnates or even rebounds after 2035. It validates the new manifestation of the “resource curse” theory (Sachs and Warner, 2001) in low-carbon transitions: traditional energy endowments hinder technology absorption through path dependence.

Pan-Pearl River Delta (PPRD) alliance. The carbon productivity improvement rate (\mathrm{\Delta }C{P}_{\text{PPRD}}) slows in later years due to insufficient knowledge capital accumulation in TB provinces. When the technological gap between imported M&E and local capabilities surpasses the absorption threshold, marginal benefits decline more rapidly (as evidenced by the near-overlapping carbon productivity curves of KUS and KBS scenarios).

Yangtze River Economic Belt (YREB) alliance. Contrary to expectations, technology cooperation leads to a decline in carbon productivity (\mathrm{\Delta }C{P}_{\text{YREB}}). As a hub for heavy chemical production capacity (accounting for 45% of national capacity), the YREB alliance heavily relies on traditional energy (e.g., coal) and high-carbon technologies (e.g., steel and petrochemicals). Instead of facilitating low-carbon substitution, technological improvements encourage production expansion, reinforcing structural rigidity. This suggests that the proposed cooperation framework fails to resolve and may even exacerbate the deep-seated technological lock-in within the carbon-intensive industry chain and traditional energy in the YREB alliance.

The alliance-level analysis reveals a critical insight. The success of technology cooperation depends less on the technology itself. Instead, it relies on the complex interplay between the diffusion method and a region’s unique endowments and industrial structure. Our findings offer a new perspective on classic development theories like the “resource curse” and “path dependency.” For the YRB, its deep reliance on coal creates an energy rebound effect, negating the benefits of imported hardware. This shows that technology diffusion alone is insufficient without a forced decoupling from traditional energy sources. The YREB’s productivity decline illustrates a similar challenge. In this heavy industry hub, new technology can paradoxically reinforce high-carbon structures by improving their efficiency, rather than replacing them. Our disparity analysis reinforces this point. Cooperation may reduce efficiency gaps in the short-term, but it can amplify them in the long run if these underlying structural issues are ignored. Ultimately, a “one-size-fits-all” cooperation framework is destined to fail. Policy must evolve beyond simply promoting technology flow. It must orchestrate profound, region-specific structural transformations.

4.2.2 Between-alliance and within-alliance disparity analysis

To further analyze changes in emission reduction efficiency disparities both between and within alliances, we computed the Theil index for carbon productivity (CP) and energy intensity (EI) in eight carbon-intensive sectors (Fig. 9). It is important to note that the core value of cooperation is not to eliminate heterogeneity between or within alliances but to promote the complementary flow of production factors through institutional interventions (trade adjustment and knowledge spillover), breaking the dual path dependencies of resource lock-in and technological lock-in. This process aims to achieve a systematic leap in low-carbon technology levels across all regions, ultimately restructuring the stability of regional economic systems. The results indicate that technology cooperation based on pure knowledge spillover minimally impacts whole and between-group disparities in emission reduction efficiency. In contrast, trade-dominated technology cooperation can reduce the whole Theil index in the short-term. However, its long-term effect is constrained by regional resource endowments and industrial structure rigidities. As the technological gap continues to narrow, a β-convergence effect is triggered (Barro, 1991), which aligns with the diminishing marginal utility of technology diffusion (Romer, 1990).

The within-group Theil index exhibits a similar trend of an initial decline followed by a rebound. TB provinces narrow their emission reduction efficiency gap in the early stage by acquiring advanced machines and equipment. However, as cooperation progresses, TF provinces gradually establish “innovation barriers,” causing TB provinces to become passively dependent on machines and equipment, which reverses within-group disparities after 2035.

Decomposing the contribution degree of within-group differences reveals that the Yellow River Basin (YRB) alliance’s internal differences are the core driver of overall divergence, with its within-group Theil Index contribution being the largest among the four alliances (75% of within-group Theil Index contribution) and continuously rising after cooperation (the decomposing results and analysis are shown in Appendixes).

4.3 Impacts of cooperation on provinces with different roles

The formation and sustainability of cooperative relationships depend on mutual benefits for participating entities. Therefore, it is necessary to quantify the gains from cooperation and validate the feasibility as well as the sustainability of the mechanism. The key performance indicators of both TF and TB provinces are systematically compared under cooperation and non-cooperation scenarios. The analysis focuses on the industrial structure of TF provinces, the production elasticity of M&E in TB provinces, and the evolution of carbon intensity in both groups. The dual-dimensional analytical approach goes beyond traditional one-sided benefit assessments, providing a quantitative basis for designing incentive mechanisms in regional coordinated emission reduction efforts.

4.3.1 Industrial upgrading in technology frontier (TF) provinces

The response of the tertiary sector share in TF provinces (Fig. 10) exhibits significant path dependence: pure knowledge spillover cooperation leads to an average decline of 0.025% in the tertiary sector share across the four alliances. In contrast, trade-dominated cooperation systematically increases this indicator. This divergence stems from differences in factor allocation mechanisms between the two technology diffusion models. Knowledge spillovers reinforce technological path dependence, strengthening factor aggregation in the secondary sector while suppressing service sector development (Acemoglu et al., 2012; Safarzyńska et al., 2012). In contrast, M&E trade stimulates service sector demand through industrial linkages, such as technology maintenance and financial services, increasing the service sector’s value-added elasticity. This finding aligns with the “snakes and ladders” theory of value chains (Baldwin, 2016), which posits that hardware-embedded cooperation facilitates the upgrading of technology frontiers toward R&D and services. In contrast, due to diminishing marginal returns and learning curve effects, pure knowledge spillover cooperation exacerbates the industrial lock-in of manufacturing in TF provinces.

It is worth noting that TF provinces’ industrial structure response to knowledge investment decisions in TB provinces also varies according to cooperation modes. Under the knowledge spillover framework, TB provinces’ knowledge investment significantly impacts TF provinces’ structural upgrade (the tertiary industry ratio drops by 0.025%). In contrast, under the M&E trade framework, such investment has a negligible effect. We attempt to explain this phenomenon. In knowledge cooperation, increased knowledge investment by TB provinces enhances absorption capacity, shortens the technological generational gap within the alliance, and forces TF provinces to maintain manufacturing scale and technological edge. The space for capital accumulation in the TF provinces’ service sector transformation is thus narrowed, causing significant divergence in tertiary industry ratios between KUS and KBS scenarios. In trade cooperation, TF provinces have locked into the stable division of labor through industrial chain position, with service sector development mainly driven by M&E output scale. In combined cooperation scenarios (KUS-T and KBS-T), knowledge investment by TB provinces, as downstream adaptation input, lacks a direct transmission channel to the industrial structure changes of TF provinces, resulting in convergence of the impacts across KUS-T and KBS-T scenarios.

4.3.2 Production elasticity of M&E in technology backward (TB) provinces

Fig. 11 illustrates the differentiated evolution of M&E production elasticity in TB provinces across the four alliances. Production elasticity continues to rise in the BTHNE, YREB, and PPRD alliances, indicating a more substantial capacity for absorbing and utilizing physical technologies after cooperation (Cohen and Levinthal, 1990). Scenario comparisons reveal significant additional elasticity gains in combined cooperation compared to pure trade (KUS-T > KBS-T > T). Local knowledge capital investment catalyzes shorter technical digestion cycles and better output performance. Absorbing new external knowledge further boosts the marginal output of manufacturing equipment in TB provinces. This validates the Cohen-Levinthal absorption capacity hypothesis (Cohen and Levinthal, 1990): the simultaneous promotion of hardware embedding and knowledge capacity building can accelerate the breakthrough of the “inefficient equilibrium trap.”

However, the TB provinces in the Yellow River Basin (YRB) alliance present an inverse U-shaped elasticity curve (peaking at 0.87), with negative values in both the early (−3.06 on average) and late (−15.55 on average) stages. It may be due to significant differences in coal dependence among provinces in the YRB alliance and a pronounced knowledge capital gradient. This reflects two key issues: (1) the misalignment between introduced technology and local factor endowments in TB provinces, resulting in negative elasticity at the initial stage, and (2) lagging investment in knowledge capital in TB provinces hampers technology absorption, leading to technological generational gaps, and ultimately causing a decline in elasticity in later stages. This finding validates the original appropriate technology hypothesis (Basu and Weil, 1998) and extends Los and Timmer’s discussion on technology adaptation dynamics (Los and Timmer, 2005). It reveals the existence of dual adaptation thresholds in technology transfer: only when the complexity of imported technology aligns with local factor endowments vertically, and its evolutionary trajectory synergizes with local knowledge infrastructure horizontally can the full economic potential of transferred technology be realized.

4.3.3 Carbon intensity changes in TB and TF provinces

In TF provinces, the change rate of carbon intensity (CI) initially rises and then falls (Fig. 12). It increases compared to the Base scenario during 2020–2035 but decreases from 2037, reaching a 0.14%–0.15% reduction by 2060. Initially, expanded M&E production causes a carbon intensity upsurge due to factor recombination costs. As cooperation deepens, scale effects trigger technological generational leaps and increase marginal returns on knowledge capital investment, reducing carbon intensity.

In TB provinces, the carbon intensity reduction narrows from about −0.18% (2020–2030) to about −0.8% (2050–2060). This mainly stems from two constraints: the pace of equipment replacement lags behind the rate of technological frontier diffusion, and the limitations of knowledge capital density thresholds and diminishing marginal returns of the TB provinces’ learning curve, leading to a gradual decline in technology absorption. This confirms the core argument of technology diffusion theory: as the technological potential gap narrows, the driving force of spillover effects on carbon intensity improvement progressively weakens.

The provincial-level impacts offer a nuanced picture, revealing both the “win-win” potential and the inherent risks of the cooperation framework. The divergent outcomes for TF and TB provinces confirm that cooperation is not a simple, one-way transfer. It is a dynamic restructuring of the entire regional economy. For TF provinces, the choice of cooperation strategy is a pivotal one. Promoting M&E trade can help them upgrade toward a service-based economy, while focusing on knowledge spillovers may instead reinforce their manufacturing leadership. For TB provinces, the results are also revealing. The dramatic increase in M&E output elasticity validates the benefits of cooperation. However, the stark failure of the YRB serves as a critical cautionary tale about “technology adaptation.” Simply acquiring advanced hardware is not enough; the technology must match local factor endowments and be supported by sufficient absorptive capacity. Finally, the carbon intensity trajectories highlight a temporal mismatch in costs and benefits. TF provinces bear initial carbon costs for long-term gains, while TB provinces see immediate benefits that diminish over time. This underscores the need for dynamic incentive mechanisms, such as cross-period compensation, to ensure the long-term sustainability of the alliance.

4.4 Sensitivity analysis

Given the inherent limitations of the CGE model, we conduct a sensitivity analysis from the perspective of both model parameters and scenario parameters to assess the robustness of the model results and the effectiveness of the policy instruments. In the previous studies that we refer to (Hu et al., 2020a; Wang et al., 2009), most of the key model parameters have been verified. This paper selects the Total Factor Productivity (TFP), the depreciation rate of knowledge capital ({\delta }_{\text{H}}), the substitution elasticities between knowledge factor inputs and non-knowledge inputs ({\sigma }_{\text{VA}}), as the model parameter to be examined. The increase ratio of knowledge capital input (GR_H) and the M&E trade adjustment ratio (AR_T) are selected as the scenario parameters to be discussed.

As shown in Table 4, this paper introduces a ±1% variation in TFP, a ± 3% variation in the depreciation rate ({\delta }_{\text{H}}) and a ± 5% variation in other parameters ({\sigma }_{\text{VA}}, GR_H, AR_T), subsequently examining the changes in GDP and knowledge capital stock (QHD) across various scenarios by the year 2060. As the parameters are adjusted, each scenario’s comparative advantage remains consistent. This indicates that the model results are robust to changes in the exogenous TFP and other parameters. Table 4 also reveals that GDP is highly sensitive to exogenous changes in TFP. Compared to the baseline, a mere 1% variation in TFP across different policy scenarios can lead to more than 10% of GDP changes.

5 Conclusions and implications

5.1 Main conclusions

Considering the necessity of technology cooperation and the significance of establishing cooperative relationships in the form of alliances within China’s administrative context, this paper constructs a dual-channel technology cooperation module for M&E and knowledge capital based on the characteristics of technology spatial diffusion. This framework explores the potential impacts of cross-regional technology cooperation on emission reduction. The cooperation effects are analyzed from two aspects: the national economy and energy system, and the alliance’s carbon reduction efficiency and equity among provincial roles. We hope to provide a reference framework for specifying differentiated regional collaborative emission reduction solutions. The following conclusions are drawn.

Cross-regional technology cooperation in M&E trade and knowledge spillover shows distinct impacts on the national GDP, welfare, and energy consumption. This paper reveals a fundamental trade-off between short-term efficiency and long-term resilience in choosing cross-regional technology cooperation pathways. M&E trade, through hardware integration, rapidly reduces energy consumption in the short-term (up to 0.06%) and drives GDP growth (following a U-shaped trajectory). However, its long-term effectiveness is constrained by technological generational gaps, leading to a stagnation in electrification rates after 2041. In contrast, cooperation focusing on knowledge spillover embodies a “short-term pain for long-term gain” dilemma. It initially restrains economic growth (with GDP declining by 0.02%) due to the crowding-out effect of R&D investment, but is crucial for building endogenous innovation capacity and enhancing long-term economic resilience. This highlights a critical choice for policymakers between immediate, tangible benefits and foundational, long-term transformation. The combined cooperation scenarios (KUS-T, KBS-T) demonstrate synergistic energy-saving effects (with reductions ranked as T < KUS-T < KBS-T). However, they also intensify welfare losses, reaching −0.8% in the KBS-T scenario. This highlights the fundamental trade-off between efficiency and equity in regional collaborative emission reduction.

The differentiated impacts of cooperation on each alliance’s carbon reduction efficiency indicate the adaptability limitations of single technology diffusion modes, which struggle to accommodate diverse regional characteristics. The Yellow River Basin (YRB) alliance, with a coal-dominated energy structure (over 60%), triggers the “Jevons Paradox,” where the energy rebound effect offsets initial energy-efficiency improvements. In the Beijing-Tianjin-Hebei-North-east (BTHNE) alliance, the interplay between policy sensitivity and dynamic technology diffusion results in a dual-peak fluctuation in carbon productivity. The Yangtze River Economic Belt (YREB) alliance, constrained by heavy industry lock-in (45% capacity), sees a 0.3% carbon productivity drop after cooperation (average of five scenarios). The cooperation mechanism drives short-term β-convergence of within-alliance technological gradients (significant Theil index drop). Still, long-term regional endowment rigidity constraints lead to system evolution into a higher equilibrium (overall carbon productivity rise), with intensified within-group disparities (YRB contributes about 75%). These findings underscore the necessity of a customized low-carbon transition strategy that integrates institutional frameworks, technology diffusion, and resource endowment dynamics.

An appropriate technology cooperation scheme can generate mutual benefits for both participants, although the benefits of each party may manifest themselves differently. Technology frontier provinces enhance their tertiary industry ratio through industrial chain restructuring (e.g., average growth of 0.15% in BTHNE, YREB, and YRB alliances) and gain long-term competitiveness. Technology-backward provinces significantly boost their manufacturing equipment output elasticity (reaching 9.27 and 2.84 in BTHNE and YREB alliances by 2060), alongside continued improvement in carbon intensity control. However, the technology-backward provinces in the YRB alliance show an inverted U-shaped output elasticity trajectory (peaking at 0.87 before declining sharply to −15.55 in later years), underscoring the importance of technology adaptation conditions. Practical cooperation requires a multidimensional alignment between the complexity of introduced technologies, local factor endowments, knowledge capital density, and the absorption cycle. Without this alignment, cooperation risks systemic efficiency losses or regional development imbalances. This finding confirms the long-term sustainability potential of regional technology cooperation. It offers a quantitative decision-making framework for differentiated cooperation strategies in heterogeneous regions (e.g., the cooperation model proposed in this paper may not be optimal for the Yellow River Basin alliance).

Finally, technology-backward provinces’ knowledge absorption capacity building is key to sustaining cooperation relationships. Simulation results show the gradual convergence pattern of technology diffusion across key indicators, including growth rates of national GDP (annual convergence rate 0.13%), electrification rate (reduction narrowing to 0.1%), and carbon productivity (between-group disparities reduced by 37.2%), driven by the diminishing marginal returns of TB provinces’ learning curve and the dynamic decline in the elasticity of knowledge absorption. Thus, without sufficient local knowledge capital accumulation, M&E trade cooperation will fall into the “low-level equilibrium trap.” This finding confirms the unsustainability of the TB provinces’ “free-riding” behavior, and technology-backward regions must match R&D investment to break through the path lock-in of technology transfer.

5.2 Implications

Policy design should be grounded in equipment trade’s short-term benefits and knowledge spillovers’ long-term resilience. To mitigate the short-term economic suppression caused by knowledge spillover cooperation, policymakers should adopt hybrid and phased strategies. This paper suggests that M&E trade provides an initial GDP boost. Therefore, a “hybrid” approach that simultaneously promotes M&E trade to secure short-term economic stability while initiating long-term knowledge capital investment can effectively smooth the transition. Furthermore, a “phased” implementation, starting with pilot projects in key sectors before scaling up, can manage the initial fiscal burden. To balance equity and efficiency in technology cooperation, it is essential to optimize the cost-sharing mechanism. On the one hand, a collaborative financing model combining government subsidies and market financing is recommended, such as covering part of equipment procurement costs through carbon allowance pledging. On the other hand, enterprises should be provided with increased R&D tax deductions to diversify R&D funding sources, thereby mitigating fiscal pressure and potential welfare losses.

A differentiated pathway should be designed to accommodate the heterogeneous endowments across alliances. Based on our findings of severe path dependency, specific and targeted intervention programs are necessary. The technology-backward provinces within the YRB alliance should establish a “Technology-Resource Decoupling Program.” This program offers recipient provinces access to advanced M&E from TF provinces. To receive it, they must meet annual targets for renewable energy expansion. The transition will be subsidized using revenues from carbon allowance auctions. The BTHNE alliance could leverage its policy sensitivity to develop a carbon productivity fluctuation early-warning and response system. This system would trigger cross-provincial capacity reallocation (e.g., Liaoning accommodating part of Hebei’s steel production capacity) when the detected fluctuation valley exceeds a predefined threshold. The YREB alliance is encouraged to establish a capacity phase-out mechanism to break its industrial lock-in. This can be operationalized by linking technology cooperation to legally binding “Structural Transformation Commitments.” For example, TB provinces must agree to a scheduled reduction in the production capacity of specific heavy industries (e.g., steel, petrochemicals) to qualify for cooperative technology transfer programs.

A bilateral benefit-balancing mechanism between technology-frontier and technology-backward provinces should be further developed to ensure the sustainability of technology partnerships and a positive cycle. First, cross-period financial and carbon quota compensation mechanisms should be introduced to offset the short-term costs borne by technology frontier provinces. Second, mandatory government or corporate co-financing in R&D should be implemented in backward provinces to enhance the technological absorptive capacity, preventing free-riding risks. For example, a “Cooperative R&D Fund” could be established at the alliance level, with mandatory contributions from both the benefiting enterprises in TB provinces and the M&E-exporting enterprises in TF provinces. It will shift the fiscal burden away from being solely a government expenditure that crowds out public welfare.

5.3 Limitations and future research

This study presents an exploratory framework for inter-provincial technology cooperation, but several limitations should be acknowledged, providing directions for future research.

First of all, it is important to recognize the inherent limitations of the CGE modeling approach employed. The model operates under the standard assumption of perfect competition and market clearing, which may not fully capture real-world market imperfections such as monopolistic power or barriers to entry. Second, our simulation is calibrated to a 2017 Social Accounting Matrix. While the model is dynamic, the base-year economic structure is static, which introduces uncertainty as long-term structural changes may deviate from historical patterns. Third, although we conducted a sensitivity analysis on key parameters like substitution elasticities, these values remain a source of uncertainty in any CGE analysis. Future research could incorporate features of imperfect competition or employ probabilistic simulations to address these uncertainties.

Beyond the model’s structural limitations, our analytical framework also has scope for refinement. First, the current framework lacks a dynamic optimization mechanism for alliance formation and provincial role specialization. Subsequent investigations should develop adaptive algorithms to optimize alliance composition and establish differentiated provincial strategies through multi-agent evolutionary game simulations. Second, this paper focuses on M&E trade and knowledge spillover pathways, leaving other critical technology cooperation modalities underexplored (such as patent licensing agreements and cross-regional joint R&D ventures), and should be considered in future research. Finally, it should be emphasized that this paper’s schemes and implications apply only to China at present. Further verification and generalization of our modeling framework and findings are needed if they are used in other fields.

6 Appendixes

6.1 Calculation of Household Welfare

In this CGE model, household welfare is measured by the level of household consumption, which is modeled using the Stone-Geary utility function and implemented through Lluch’s Extended Linear Expenditure System (ELES). The core logic of the ELES framework is to disaggregate household consumption into two components: subsistence consumption (a fixed basket of goods required to meet basic survival needs) and discretionary consumption (after subsistence needs are met, the remaining disposable income is allocated to various goods based on the household’s marginal propensity to consume). The consumption expenditure for each commodity is determined by the following equation:

(C1)\begin{array}{c}\begin{array}{rl}& P{Q}_{(\text{R,C})}\times Q{H}_{\text{(R,C,H)}}=P{Q}_{(\text{R,C})}\times LESsu{b}_{(\text{R,C,H})}\\ & +LESbet{a}_{\mathit{\text{(R,C,H)}}}\times (E{\mathrm{H}}_{\text{(R,H)}}\\ & -{{\sum }_{\text{C}}LESsub}_{\text{(R,C,H)}}\times P{Q}_{\text{(R,C)}}),\end{array}\end{array}

where QH_(R,C,H) is the consumption quantity of a commodity, PQ_(R,C) is the price of the commodity, LESsub_(R,C,H) represents the subsistence quantity of the commodity, LESbeta_(R,C,H) is the marginal propensity to consume the commodity, EH_(R,H) is the household’s total disposable income. Therefore, any policy shock that affects household disposable income (EH_(R,H)) or market prices (PQ_(R,C)) will lead to a change in the overall consumption level (QH), which is interpreted as a change in household welfare.

6.2 Calculation and decomposition of alliance carbon reduction efficiency disparities

Theil Index and its Decomposition. In this study, the Theil Index analyzes the differences in carbon emission reduction efficiency within and between alliances. The whole national differences, differences within alliances, differences between regions, and the associated contribution rates have been calculated.

(D1)\begin{array}{c}T=\frac{1}{\text{k}}\sum _{\text{q}=1}^{\text{k}}\left(\frac{S_{\text{q}}}{\overline{S}}\times \ln \frac{S_{\text{q}}}{\overline{S}}\right),\end{array}

(D2)\begin{array}{c}{T}_{\text{p}}=\frac{1}{k_{\text{p}}}\sum _{\text{q}=1}^{k_{\text{p}}}\left(\frac{S_{\text{pq}}}{\overline{S{}_{\text{p}}}}\times \ln \frac{S_{\text{pq}}}{\overline{S_{\text{p}}}}\right),\end{array}

(D3)\begin{array}{c}T=T_{\text{w}}+T_{\text{b}}=\sum _{\text{p}=1}^4\left(\frac{k_{\text{p}}}{\text{k}}\times \frac{\overline{S_{\text{p}}}}{S}\times T_{\text{p}}\right)+\sum _{\text{p}=1}^4\left(\frac{k_{\text{p}}}{\text{k}}\times \frac{\overline{S_{\text{p}}}}{S}\times \ln \frac{\overline{S_{\text{p}}}}{S}\right).\end{array}

In Eq. (D1), T represents the overall Theil Index for carbon emission reduction efficiency differences, with a value range of [0, 1]. A smaller Theil Index indicates lower overall differences, while a more extensive index suggests more significant differences. q denotes a province, k is the total number of provinces in the country, S_q is the carbon emission reduction efficiency of province q, and S is the national average carbon emission reduction efficiency. In Eq. (D2), T_p represents the overall Theil Index for carbon emission reduction efficiency differences within alliance p, k_p is the number of provinces in alliance p, S_pq is the carbon emission reduction efficiency of province q within alliance p, and S_p is the average carbon emission reduction efficiency of alliance p. In Eq. (D3), the overall Theil Index for carbon emission reduction efficiency differences is further decomposed into the within-region difference Theil Index (T_w) and the between-region difference Theil Index (T_b).

Additionally, Eqs. (D4) and (D5) define the contribution rates of within-region differences (D_w) and between-region differences (D_b) to the overall differences, respectively. Equation (C6) defines the contribution rate of each alliance to the overall within-alliance differences (D_p), where S_p represents the sum of carbon emission reduction efficiencies of all provinces within alliance p, and S is the total national carbon emission reduction efficiency.

(D4)\begin{array}{c}{D}_{\text{w}}=\frac{T_{\text{w}}}{T},\end{array}

(D5)\begin{array}{c}{D}_{\text{b}}=\frac{T_{\text{b}}}{T},\end{array}

(D6)\begin{array}{c}{D}_{\text{p}}=\frac{S_{\text{p}}}{S}\times \frac{T_{\text{p}}}{T}.\end{array}

The calculation results are shown in Figs. D1 and D2.

The Yellow River Basin (YRB) alliance’s internal differences are the core driver of overall divergence. This may be due to significant differences in coal dependence among provinces in the YRB alliance and a pronounced knowledge capital gradient. The cooperation mechanism further amplifies these dual heterogeneities: high-resource-dependent provinces experience slower technological substitution, widening the carbon productivity and energy efficiency gap with low-resource-dependent provinces. Meanwhile, cooperation enables TF provinces to accumulate technological advantages, while TB provinces experience a gradual decline in local R&D capabilities due to resource crowding-out effects. Thus, when implementing technology cooperation, the YRB alliance’s low-carbon transition must overcome dual thresholds of resource lock-in (reducing coal dependency) and technology lock-in (enhancing R&D intensity); otherwise, cooperation may exacerbate internal disparities. This finding offers a key theoretical basis for balancing efficiency and fairness in regional emission reduction cooperation.

References

[1] Acemoglu D, Aghion P, Bursztyn L, Hemous D, (2012). The environment and directed technical change. American Economic Review, 102( 1): 131–166

[2] Aghion P, Blundell R, Griffith R, Howitt P, Prantl S, (2009). The effects of entry on incumbent innovation and productivity. Review of Economics and Statistics, 91( 1): 20–32

[3] An K, Wang C, Cai W, (2023). Low-carbon technology diffusion and economic growth of China: An evolutionary general equilibrium framework. Structural Change and Economic Dynamics, 65: 253–263

[4] Arthur W B, (1989). Competing technologies, increasing returns, and lock-in by historical events. Economic Journal (London), 99( 394): 116–131

[5] Baldwin R E, (2016). The world trade organization and the future of multilateralism. Journal of Economic Perspectives,, 30( 1): 95–116

[6] Barbieri N, Marzucchi A, Rizzo U, (2023). Green technologies, interdependencies, and policy. Journal of Environmental Economics and Management, 118: 102791

[7] Barro R J, (1991). Economic growth in a cross section of countries. Quarterly Journal of Economics, 106( 2): 407–443

[8] Basu S, Weil D N, (1998). Appropriate technology and growth. Quarterly Journal of Economics, 113( 4): 1025–1054

[9] Bortoletto W W, Pacagnella Junior A C, Cabello O G, (2023). Exploring the scientific literature on clean development mechanisms: A bibliometric analysis. Energy Policy, 183: 113806

[10] Bosetti V, Carraro C, Massetti E, Tavoni M, (2008). International energy R&D spillovers and the economics of greenhouse gas atmospheric stabilization. Energy Economics, 30( 6): 2912–2929

[11] Bradlow B H, Kentikelenis A, (2024). Globalizing green industrial policy through technology transfers. Nature Sustainability, 7( 6): 685–687

[12] Buonanno P, Carraro C, Galeotti M, (2003). Endogenous induced technical change and the costs of Kyoto. Resource and Energy Economics, 25( 1): 11–34

[13] Bye B, Jacobsen K, (2011). Restricted carbon emissions and directed R&D support; an applied general equilibrium analysis. Energy Economics, 33( 3): 543–555

[14] Cameron R, (1975). The diffusion of technology as a problem in economic history. Economic Geography, 51( 3): 217–230

[15] Carattini S, Kim G, Melkadze G, Pommeret A, (2024). Carbon taxes and tariffs, financial frictions, and international spillovers. European Economic Review, 170: 104883

[16] Carraro C, De Cian E, (2013). Factor-augmenting technical change: An empirical assessment. Environmental Modeling and Assessment, 18( 1): 13–26

[17] Chen X Q, Ma C Q, Ren Y S, Lei Y T, (2023). Carbon allowance auction design of China’s ETS: A comprehensive hierarchical system based on blockchain. International Review of Economics & Finance, 88: 1003–1019

[18] Cohen W M, Levinthal D A, (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35( 1): 128–152

[19] Cui W, Li L, Chen G, (2022). Market-value oriented or technology-value oriented? —Location impacts of industry-university-research (IUR) cooperation bases on innovation performance. Technology in Society, 70: 102025

[20] Long J B D, Summers L H, (1991). Equipment investment and economic growth. Quarterly Journal of Economics, 106( 2): 445–502

[21] De Paulo A F, Porto G S, (2023). Unveiling the cooperation dynamics in the photovoltaic technologies’ development. Renewable & Sustainable Energy Reviews, 187: 113694

[22] Fujimori S, Oshiro K, Hasegawa T, Takakura J, Ueda K, (2023). Climate change mitigation costs reduction caused by socioeconomic-technological transitions. npj Climate Action, 2( 1): 9

[23] Garau G, Lecca P, (2015). The impact of regional R&D subsidy in a computable general equilibrium model. International Regional Science Review, 38( 4): 319–357

[24] Garsous G, Worack S, (2022). Technological expertise as a driver of environmental technology diffusion through trade: Evidence from the wind turbine manufacturing industry. Energy Policy, 162: 112799

[25] Gillingham K, Newell R G, Pizer W A, (2008). Modeling endogenous technological change for climate policy analysis. Energy Economics, 30( 6): 2734–2753

[26] Goldstein J E, Neimark B, Garvey B, Phelps J, (2023). Unlocking “lock-in” and path dependency: A review across disciplines and socio-environmental contexts. World Development, 161: 106116

[27] Goulder L H, Mathai K, (2000). Optimal CO₂ abatement in the presence of induced technological change. Journal of Environmental Economics and Management, 39( 1): 1–38

[28] Grubler A, Wilson C, Bento N, Boza-Kiss B, Krey V, McCollum D L, Rao N D, Riahi K, Rogelj J, De Stercke S, Cullen J, Frank S, Fricko O, Guo F, Gidden M, Havlík P, Huppmann D, Kiesewetter G, Rafaj P, Schoepp W, Valin H, (2018). A low energy demand scenario for meeting the 1.5 °C target and sustainable development goals without negative emission technologies. Nature Energy, 3( 6): 515–527

[29] Hahn T, Morfeldt J, Höglund R, Karlsson M, Fetzer I, (2024). Estimating countries’ additional carbon accountability for closing the mitigation gap based on past and future emissions. Nature Communications, 15( 9707): 1–10

[30] Hattori T, Nam H, Chapman A, (2022). Multilateral energy technology cooperation: Improving collaboration effectiveness through evidence from International Energy Agency Technology Collaboration Programmes. Energy Strategy Reviews, 43: 100920

[31] Hu X, Liu J, Yang H, Meng J, Wang X, Ma J, Tao S, (2020a). Impacts of Potential China’s Environmental Protection Tax Reforms on Provincial Air Pollution Emissions and Economy. Earth’s Future, 8( 4): e2019EF001467

[32] Hu X, Pollitt H, Pirie J, Mercure J F, Liu J, Meng J, Tao S, (2020b). The impacts of the trade liberalization of environmental goods on power system and CO2 emissions. Energy Policy, 140: 111173

[33] Hu X, Wu H, Ni W, Wang Q, Zhou D, Liu J, (2023). Quantifying the dynamical interactions between carbon pricing and environmental protection tax in China. Energy Economics, 126: 106912

[34] Hübler M, Baumstark L, Leimbach M, Edenhofer O, Bauer N, (2012). An integrated assessment model with endogenous growth. Ecological Economics, 83: 118–131

[35] Idisi P O, Ogwu I J, Zorto M N, (2018). Kaldor-hicks compensation criterion: A measure of social welfare. International Journal of Management and Commerce Innovations, 6( 2): 623–628

[36] Jaffe A B, Newell R G, Stavins R N, (2005). A tale of two market failures: Technology and environmental policy. Ecological Economics, 54( 2-3): 164–174

[37] Jia L, Zhang J, Li R, Wang L, Wu H, Wang P, (2025). Spatial correlation investigation of carbon emission efficiency in the yangtze river delta of China: The role of low-carbon pilot cities. Ecological Indicators, 172: 113282

[38] Jia Z, Lin B, (2020). Rethinking the choice of carbon tax and carbon trading in China. Technological Forecasting and Social Change, 159: 120187

[39] Jiang H D, Dong K, Qing J, Teng Q, (2023). The role of technical change in low-carbon transformation and crises in the electricity market: A CGE analysis with R&D investment. Energy Economics, 125: 106897

[40] Jiang Y (2021). The dynamic evolution analysis of regional development differences of high-tech industries: Based on China’s data from 1997 to 2019. In: 2021 5th Annual International Conference on Data Science and Business Analytics , 343–348

[41] Jin W, (2012). Can technological innovation help China take on its climate responsibility? An intertemporal general equilibrium analysis. Energy Policy, 49: 629–641

[42] Jin W, (2015). Can China harness globalization to reap domestic carbon savings? Modeling international technology diffusion in a multi-region framework. China Economic Review, 34: 64–82

[43] Jin W, (2016). International technology diffusion, multilateral R&D coordination, and global climate mitigation. Technological Forecasting and Social Change, 102: 357–372

[44] Jin W, Zhang Z, (2016). On the mechanism of international technology diffusion for energy technological progress. Resource and Energy Economics, 46: 39–61

[45] Jin Y, Xu Y, (2024). Carbon reduction of urban form strategies: Regional heterogeneity in yangtze river delta, China. Land Use Policy, 141: 107154

[46] Knez K, (2023). Technology diffusion and uneven development. Journal of Evolutionary Economics, 33( 4): 1171–1195

[47] Křístková Z, (2012). Impact of R&D investment on economic growth of the Czech Republic: A recursively dynamic CGE approach. Prague Economic Papers, 21( 4): 412–433

[48] Krugman P, (1991). Increasing returns and economic geography. Journal of Political Economy, 99( 3): 483–499

[49] Leibowicz B D, Krey V, Grubler A, (2016). Representing spatial technology diffusion in an energy system optimization model. Technological Forecasting and Social Change, 103: 350–363

[50] Li L, Ye F, Li Y, Tan K H, (2018). A bi-objective programming model for carbon emission quota allocation: Evidence from the Pearl River Delta region. Journal of Cleaner Production, 205: 163–178

[51] Li X, Xing H, (2024). Regional carbon emission pathways under central-local coordination: Objective constraints and collaborative incentives. Sustainable Production and Consumption, 47: 156–165

[52] Li Y, Chen J, Li L, Huang X, (2024). Government innovation awards, innovation funds acquisition and enterprise innovation. Economic Analysis and Policy, 82: 846–864

[53] Liu H, Cui C, Chen X, Xiu P, (2023). How can regional integration promote corporate innovation? A peer effect study of R&D expenditure. Journal of Innovation & Knowledge, 8( 4): 100444

[54] Liu X, Wang W, Wu W, Zhang L, Wang L, (2022). Using cooperative game model of air pollution governance to study the cost sharing in yangtze river delta region. Journal of Environmental Management, 301: 113896

[55] Los B, Timmer M P, (2005). The ‘appropriate technology’ explanation of productivity growth differentials: An empirical approach. Journal of Development Economics, 77( 2): 517–531

[56] Ma Z, Mao T, Ma W, Niu X, Yang J, (2025). The green trade benefits and horizontal ecological compensation from the perspective of embodied carbon emissions: A case study of the yellow river basin, China. Research in Cold and Arid Regions, 17( 1): 55–64

[57] Mercure J F, Pollitt H, Chewpreecha U, Salas P, Foley A M, Holden P B, Edwards N R, (2014). The dynamics of technology diffusion and the impacts of climate policy instruments in the decarbonisation of the global electricity sector. Energy Policy, 73: 686–700

[58] Nicholas B, Mark S, John Van R, (2013). Identifying technology spillovers and product market rivalry. Econometrica, 81( 4): 1347–1393

[59] Pan X, Wei Z, Han B, Shahbaz M, (2021). The heterogeneous impacts of interregional green technology spillover on energy intensity in China. Energy Economics, 96: 105133

[60] Parrado R, De Cian E, (2014). Technology spillovers embodied in international trade: Intertemporal, regional and sectoral effects in a global CGE framework. Energy Economics, 41: 76–89

[61] Popp D, (2011). International technology transfer, climate change, and the clean development mechanism. Review of Environmental Economics and Policy, 5( 1): 131–152

[62] Rodrik D, (2016). Premature deindustrialization. Journal of Economic Growth, 21( 1): 1–33

[63] Rodrik D, (2018). Populism and the economics of globalization. Journal of International Business Policy, 1( 1-2): 12–33

[64] Rohe S, Mattes J, (2022). What about the regional level? Regional configurations of technological innovation systems. Geoforum, 129: 60–73

[65] Romer P M, (1990). Endogenous technological change. Journal of Political Economy, 98( 5, Part 2): S71–S102

[66] Ronayne D, Sgroi D, Tuckwell A, (2021). Evaluating the sunk cost effect. Journal of Economic Behavior & Organization, 186: 318–327

[67] Sachs J D, Warner A M, (2001). The curse of natural resources. European Economic Review, 45( 4-6): 827–838

[68] Safarzyńska K, Frenken K, Van den Bergh J C J M, (2012). Evolutionary theorizing and modeling of sustainability transitions. Research Policy, 41( 6): 1011–1024

[69] Sagar A D, Van der Zwaan B, (2006). Technological innovation in the energy sector: R&D, deployment, and learning-by-doing. Energy Policy, 34( 17): 2601–2608

[70] Shao Z, Zheng X, Zhao J, Liu Y, (2025). Evaluating the health impact of air pollution control strategies and synergies among PM_2.5 and O₃ pollution in Beijing-Tianjin-Hebei region, China. Environmental Research, 274: 121348

[71] Shi B, Lin S, Shi M, (2025). When differentiated carbon tax policy meets LBD of renewable energy and electrification of energy end-use: Policy implications of sectoral differentiation of carbon productivity and carbon emission. Environmental Impact Assessment Review, 110: 107724

[72] Stern N, (2022). Towards a carbon neutral economy: How government should respond to market failures and market absence. Journal of Government and Economics, 6: 100036

[73] Sue Wing I, (2008). The synthesis of bottom-up and top-down approaches to climate policy modeling: Electric power technology detail in a social accounting framework. Energy Economics, 30( 2): 547–573

[74] Sun Y, Jiang L, Cao C, (2025). Multiple proximities and inter-agency collaboration within a policy network: The case of innovation policymaking in China. Technovation, 141: 103169

[75] Tian Y, Jiang G, Zhou D, Ding K, Su S, Zhou T, Chen D, (2019). Regional industrial transfer in the jingjinji urban agglomeration, China: An analysis based on a new “transferring area-undertaking area-dynamic process” model. Journal of Cleaner Production, 235: 751–766

[76] Touboul S, Glachant M, Dechezleprêtre A, Fankhauser S, Stoever J, (2023). Invention and global diffusion of technologies for climate change adaptation: A patent analysis. Review of Environmental Economics and Policy, 17( 2): 316–335

[77] Verdolini E, Galeotti M, (2011). At home and abroad: An empirical analysis of innovation and diffusion in energy technologies. Journal of Environmental Economics and Management, 61( 2): 119–134

[78] Wang K, Wang C, Chen J, (2009). Analysis of the economic impact of different Chinese climate policy options based on a CGE model incorporating endogenous technological change. Energy Policy, 37( 8): 2930–2940

[79] Wang Y, Li W, Li G, Tan S, (2024). Assessing the impact of regional coordinated development on a low-carbon future: Evidence from the beijing-tianjin-hebei coordinated development strategy. Journal of Cleaner Production, 481: 144140

[80] Weidenfeld A, Makkonen T, Clifton N, (2021). From interregional knowledge networks to systems. Technological Forecasting and Social Change, 171: 120904

[81] Wen S, Jia Z, (2023). A penny saved is a penny earned: Exploring the synergistic effect between carbon trading and energy carbon-content tax. Journal of Cleaner Production, 430: 139618

[82] Weyant J P, Olavson T, (1999). Issues in modeling induced technological change in energy, environmental, and climate policy. Environmental Modeling and Assessment, 4( 2-3): 67–85

[83] Witajewski-Baltvilks J, (2025). Endogenous technological change adapted to the CGE framework. Journal of Global Economic Analysis, 10( 1): 106–174

[84] Xie Q, Wang D, Bai Q, (2024). “Cooperation” or “Competition”: Digital finance enables green technology innovation—a new assessment from dynamic spatial spillover perspectives. International Review of Economics & Finance, 93: 587–601

[85] Yang F, Sattar U, (2024). COP29: Technology development and transfer framework. Frontiers in Environmental Science, 12: 1349843

[86] Zhang H, Feng C, Zhou X, (2022a). Going carbon-neutral in China: Does the low-carbon city pilot policy improve carbon emission efficiency. Sustainable Production and Consumption, 33: 312–329

[87] Zhang Y, Qi L, Lin X, Pan H, Sharp B, (2022b). Synergistic effect of carbon ETS and carbon tax under China’s peak emission target: A dynamic CGE analysis. Science of the Total Environment, 825: 154076

[88] Zhang Z, Wang J, Feng C, Chen X, (2023). Do pilot zones for green finance reform and innovation promote energy savings? Evidence from China. Energy Economics, 124: 106763

[89] Zheng H, Bai Y, Wei W, Meng J, Zhang Z, Song M, Guan D, (2021). Chinese provincial multi-regional input-output database for 2012, 2015, and 2017. Scientific Data, 8( 244): 1–13

[90] Zheng Y, Collins A, Yao S, (2024). Promoting sustainable and high‐quality economic development in China via regional innovation poles. Journal of Regional Science, 64( 3): 671–699

[91] Zickfeld K, MacIsaac A J, Canadell J G, Fuss S, Jackson R B, Jones C D, Lohila A, Matthews H D, Peters G P, Rogelj J, Zaehle S, (2023). Net-zero approaches must consider earth system impacts to achieve climate goals. Nature Climate Change, 13( 12): 1298–1305

RIGHTS & PERMISSIONS

Higher Education Press