How Does the Carbon Emission Trading Policy Enhance Corporate Green Technology Innovation? Evidence from Advanced Manufacturing Enterprises

Abstract

As global climate change progresses and the “dual carbon” strategy advances, market-based carbon emission trading systems are of great theoretical and practical importance for green technology innovation. In this paper, A-share listed advanced manufacturing enterprises in pilot regions from 2010 to 2023 are taken as samples, and a multi-period difference-in-differences (DID) method is employed to probe into the mechanism by which this policy influences green technology innovation in China’s advanced manufacturing enterprises. Empirical analysis reveals that carbon emission trading exerts a remarkable promoting impact on green technology innovation in China’s advanced manufacturing enterprises. The study indicates that the policy’s influence on enterprises is indirect; specifically, government policies encourage enterprises to raise their R&D investment, thus facilitating green technology innovation to some degree. Moreover, carbon emission rights prices play a positive moderating role, which is vital for maintaining the policy’s incentive function in long-term green transition—within a reasonable range, carbon prices can enhance the policy’s promoting effect. In addition, enterprise-specific features like company size and asset-liability ratio have certain effects on enterprises’ green technology innovation behaviors. The research findings will offer a theoretical foundation and practical reference for optimizing China’s carbon market mechanism in advanced manufacturing and advancing the green transformation of China’s advanced manufacturing industry.

1. Introduction

Since the reform and opening-up, China’s overall economic development level has ranked among the world’s top. However, due to the lack of national awareness of environmental protection and the long-term “high pollution, high energy consumption, high emission” development model in manufacturing, ecological environment damage and resource waste have been unprecedentedly severe. According to the 2022 Global Environmental Performance Index (EPI) Report, China ranks 160th among all participating countries and regions, highlighting the urgency of green governance. The report of the 20th National Congress of the Communist Party of China comprehensively deployed “promoting green development and promoting harmony between human and nature”, emphasizing that “we must firmly establish and practice the concept that lucid waters and lush mountains are invaluable assets, and plan development from the perspective of harmony between human and nature”. At the 75th session of the United Nations General Assembly, General Secretary Xi Jinping announced that “China strives to peak carbon dioxide emissions before 2030 and achieve carbon neutrality before 2060”. Implementing environmental protection planning policies has become an important part of the development of the times.

Enterprises are the world’s largest carbon emitters. Against the backdrop of tackling climate change, enterprises need to adjust their production processes to meet stricter carbon emission standards. Emission reduction by enterprises can not only cut down environmental pollution but also enhance their brand image and competitiveness. Green technology innovation is not only a crucial means for enterprises to boost their competitiveness in the low-carbon era but also an important pillar for achieving regional sustainable development [1]. Green technology innovation serves as a vital pathway for China to realize low-carbon emission reduction and develop a green economy. Making the most of the positive interaction and complementary effects between the carbon trading system and green technology innovation can not only advance the progress of China’s emission reduction technologies but also propel the green upgrading of China’s industries. This provides significant theoretical support for attaining the “dual carbon” goals and promoting high-quality economic and social development. Market-oriented environmental regulations like carbon trading policies can effectively stimulate enterprises’ inherent drive for green technology innovation by adjusting the cost–benefit structure [2].

Advanced manufacturing encompasses industries such as high-end equipment manufacturing, intelligent manufacturing, new energy, and new materials. It leverages high technology, innovative technologies, and modern management approaches to enhance manufacturing efficiency, quality, and product added value, and facilitate industrial transformation and upgrading [3]. Owing to its features of high technology and high energy consumption, advanced manufacturing has become the main force in green technological innovation and a key node in carbon emission control. Made in China 2025 clearly stipulates that technological innovation should steer advanced manufacturing toward low-carbonization and intelligence. The development quality of advanced manufacturing is closely tied to the level of green technology innovation, and its green transformation can notably improve regional green total factor productivity [4]. Whether the carbon trading policy can effectively spur the motivation for green technology innovation in this sector will directly influence the coordinated development of the “dual carbon” goals and industrial upgrading.

While extant studies have explored the nexus between carbon trading and green innovation, three critical gaps remain. First, most studies focus on the broader industrial sector or high-energy-consuming industries, neglecting advanced manufacturing—a sector that combines high-tech intensity (critical for innovation scalability) and substantial emission contributions (critical for sustainability impact). Second, the integrated analysis of R&D investment as a mediating pathway and carbon price as a moderating factor remains fragmented, failing to capture the joint mechanisms that shape policy effectiveness in driving innovation. Third, few studies explicitly link carbon trading policy effects to SDG targets, limiting the policy’s implications for global sustainability agendas [5,6]. To address these gaps, this paper takes China’s A-share listed advanced manufacturing enterprises from 2010 to 2023 as the research object, uses the multi-period difference-in-differences (DID) method to study the impact of the carbon emission trading policy on enterprises’ green technology innovation behaviors, and further explores the mediating role of R&D investment and the moderating role of carbon emission quota prices—with a specific focus on implications for SDG implementation.

2. Literature Review

Currently, there are still disagreements in existing research concerning the influence of carbon emission trading policies on enterprises’ green technology innovation. A portion of scholars maintains that these policies exert a positive promotional impact on enterprises’ green technology innovation. Feng, through empirical research, discovered that even though the implementation of carbon emission trading policies generally has a restraining effect on enterprises’ innovation, this restraint mainly targets non-green innovation, while still boosting green innovation [7]. Chen S utilized a multiple mediation model and found that China’s carbon emission trading policy has decreased the total carbon emissions by 13.39%, and this decreasing trend is on the rise year after year. Nevertheless, only high-quality innovation has played an effective mediating role, and there is still considerable potential for emission reduction [8]. Zhang carried out an analysis by employing a multi-period difference-in-differences model, taking Chinese A-share listed companies as the research sample. Their core viewpoint is that carbon emission trading policies can indirectly foster green technology innovation by motivating enterprises to increase R&D investment, and fluctuations in carbon prices play a crucial regulatory role in this process. Different situations of carbon price fluctuations have a notable influence on the policy’s promotional effect [3]. Moreover, Liu and Sun indicated that the carbon trading system’s facilitation of low-carbon innovation not only enhances enterprises’ competitiveness but also lowers regional carbon intensity, which is of great significance for attaining environmental sustainability [9]. Lyu investigated the impact of carbon emissions trading systems on low-carbon technological innovation and found that carbon emissions trading policies can advance low-carbon technological innovation and offer certain policy incentives for enterprises’ green technology innovation [10]. Lv and Bai assessed China’s carbon emissions trading policy from the perspective of corporate innovation and concluded that the policy has a positive influence on corporate innovation, facilitating green technology innovation activities and enhancing enterprises’ green innovation capabilities [11]. Pan pointed out that corporate environmental governance can reinforce green technology innovation, and carbon emissions trading policies, as a form of environmental governance measure, can encourage enterprises to enhance environmental management, thereby propelling green technology innovation [12]. Peng explored the cultivation mechanism of green technology innovation in the manufacturing industry from an ecological niche perspective and argued that carbon emissions trading policies help create an environment favorable for green technology innovation and promote green technology innovation in manufacturing enterprises [13].

However, the effectiveness of the policy remains a subject of debate. Some scholars contend that carbon trading policies might not be able to effectively boost green innovation; in fact, they could even have adverse effects, which would be detrimental to sustainable development endeavors. A number of scholars both domestically and internationally hold the view that carbon trading-related environmental policies will not only fail to exert a positive influence on enterprises’ green innovation but may actually have a negative one. Anderson pointed out that under the EU carbon trading system, numerous companies are more inclined to utilize new equipment or acquire other technologies to deal with the rise in carbon pricing instead of enhancing their own green technology innovation [14]. Du, by using a spatial econometric model, found that although China’s carbon emission trading pilot policy has spurred green innovation in pilot areas and generated positive spillover effects in adjacent regions, at the individual enterprise level, the policy’s direct incentive effect on green technology innovation is not remarkable [15]. Yu highlighted the significant impact of enterprise heterogeneity, and their research demonstrated that the influence of carbon emission trading policies on the green technology innovation of enterprises with different characteristics differs. Among these enterprises, the impact on state-owned enterprises and those with relatively high financial constraints is more pronounced, while the incentive effect on other types of enterprises is relatively weak. This also reflects that the policy is limited by the enterprises’ own characteristics in the process of functioning [16]. Song examined how China’s current carbon trading policy affects carbon prices and found that there are certain problems in the formation of carbon prices under this policy. These problems may diminish enterprises’ motivation for green technology innovation and restrict the policy’s effectiveness in promoting such innovation [17]. Liu gave an overview of China’s carbon emissions trading system and identified several challenges that might lessen the policy’s incentive effects on enterprises’ green technology innovation, making it hard to effectively drive the development of green technology [18]. Zhao analyzed the efficiency of China’s carbon trading market and found that market inefficiencies hinder the policy’s ability to stimulate green technology innovation, preventing it from fully achieving its intended function [19].

Table 1. Comparison of Core Dimensions Between Existing Literature and This Study.

Research Perspective	Representative References and Core Arguments	Key Focus Areas	Distinctions in Current Study
Supportive Perspectives	[7]: Policy inhibits non-green innovation but promotes green innovation.	Economic effects (e.g., cost efficiency) Technological effects (e.g., innovation output)	Provides holistic analysis of joint effects (R&D and carbon price) on long-term sustainability. Addresses gaps by incorporating environmental and social sustainability outcomes aligned with UN SDGs, not just economic/technological factors.
	[8]: Policy reduces emissions by 13.39% annually, mediated by high-quality innovation.
	[9]: Carbon trading enhances low-carbon innovation, competitiveness, and emission reduction.
	[10]: Policy provides direct incentives for low-carbon technological innovation in firms.
	[11]: Policy stimulates green innovation activities and boosts innovation capabilities.
	[12]: Carbon trading strengthens environmental management to drive lifecycle green innovation.
	[13]: Policy cultivates green innovation niches in manufacturing sectors.
Critical Perspectives	[14]: Firms prefer buying external tech over self-innovation under EU ETS.	Regulatory barriers (e.g., carbon price volatility) Firm-level constraints (e.g., heterogeneity in ownership/finances)	Focuses on advanced manufacturing heterogeneity, not treating manufacturing as monolithic.
	[15]: Weak firm-level incentives due to implementation barriers limit direct policy effects.
	[16]: Policy effectiveness varies by ownership/financial heterogeneity.
	[17]: Carbon price volatility undermines long-term green investment and emission goals.
	[18]: Market design flaws (e.g., quota allocation) reduce innovation incentives.
	[19]: Low market efficiency weakens price signals and innovation mechanisms.
This Study	Constructs a unified “policy-R&D investment-carbon price-green innovation” framework, analyzing advanced manufacturing enterprises.	Integrated mediator-moderator analysis (R&D investment + carbon price) Sustainability linkage (environmental outcomes)

presents the comparison of core dimensions between existing literature and this study (for a detailed version, please refer to the Supplementary Materials). To summarize, existing studies have made valuable contributions to understanding the relationship between carbon trading policies and corporate green innovation, but three critical gaps persist from a sustainability perspective. First, the literature primarily focuses on the economic and technological effects while neglecting the environmental and social sustainability outcomes of green innovation—factors that are core to the UN SDGs [20]. Second, most studies treat the manufacturing sector as a whole, failing to address the heterogeneity of advanced manufacturing enterprises, which are key drivers of sustainable industrial upgrading [3]. Third, the integrated analysis of R&D investment (mediator) and carbon prices (moderator) remains fragmented, with limited discussion on how these two factors jointly enhance the policy’s contribution to long-term sustainable development. This study fills these gaps by focusing on advanced manufacturing enterprises, constructing a unified framework of ‘policy-R&D investment-carbon price-green innovation’, and explicitly linking the research findings to sustainable development goals.

3. Theoretical Analysis and Research Hypotheses

3.1. Impact of Carbon Emission Trading Policy on Corporate Green Technology Innovation

Amid global efforts to address climate change and advance sustainable development, carbon emission trading, as a market-oriented environmental regulatory instrument, has garnered extensive attention worldwide [10]. As a typical market-incentive environmental regulation tool, the carbon emission trading policy can effectively internalize environmental externalities and steer enterprises in adjusting their production and operation decisions [21]. To leverage market mechanisms in lowering enterprises’ emission reduction costs and encourage them to achieve energy conservation and emission reduction through technological innovation, pilot carbon emission trading was launched in eight provinces and cities across the country starting from 2013.

Integrating the Porter Hypothesis and Induced Innovation Theory, we analyze the policy’s impact: The Porter Hypothesis posits that appropriate environmental regulations can spur enterprises to innovate, thereby offsetting environmental protection costs and enhancing their competitiveness. Induced Innovation Theory emphasizes that changes in factor prices drive enterprises to reallocate resources toward innovative activities alleviating constraints from scarce/costly factors. In the carbon emission trading context, carbon pricing’s cost pressures and economic incentives (Porter Hypothesis logic) intersect with the quota price as a factor price signal (Induced Innovation Theory logic), jointly inducing green technology innovation.

From a cost-driven viewpoint, exceeding carbon limits forces quota purchases, raising operating costs and reducing competitiveness. To cut costs, enterprises transform production modes (Porter Hypothesis: innovation to offset costs) [22]. From an economic incentive perspective, surplus quotas from green innovation can be resold, stimulating internal motivation (Induced Innovation Theory: price-driven resource reallocation to innovation) [23].

Studies show the policy directly promotes emission reduction and indirectly drives green innovation by adjusting cost–benefit structures [24]. Combining both theories, enterprises are encouraged to innovate from cost and benefit angles to cope with carbon constraints. Long-term policy implementation forms stable expectations, facilitating increased green R&D investment [25]. Thus:

Hypothesis 1.

Carbon emission trading policies can significantly promote green technology innovation in advanced manufacturing enterprises.

3.2. Mediating Effect of R&D Investment

The incentive impact of carbon emission trading policies is typically conveyed via R&D investment. Research indicates that R&D investment is a crucial factor for enterprises to realize technological innovation and a vital connection in advancing the conversion of enterprises’ innovation achievements [26,27]. Theoretically, carbon emission trading policies can urge enterprises to optimize resource allocation through cost-based incentives, thus supplying more funds for R&D investment in green technologies. Under the constraints of carbon emissions, the cost pressure solely from purchasing quotas will encourage enterprises to proactively conduct research and development, and overcome technical bottlenecks by increasing R&D personnel, equipment, and technology introduction. In addition, the regulatory influence of economic incentives on R&D investment will also be strengthened. The quota income that enterprises gain from green technology innovation can be fed back to R&D, thereby forming a favorable cycle. That is, the extra income generated from emission reduction can offer financing for subsequent R&D, and the increased R&D expenses can further enhance innovation efficiency and enlarge the emission reduction scope. Green technology innovation cannot be separated from continuous R&D investment, and the carbon emission trading policy provides enterprises with a sustainable funding source for R&D by creating economic incentives [28]. For instance, to meet the objectives of the carbon market, high-end manufacturing enterprises can turn part of their income into R&D funds, which are specifically used for improving environmental protection processes and researching and developing new energy technologies, and achieve technological breakthroughs in the form of patents. In summary, carbon emission trading policies indirectly boost green technology innovation by stimulating enterprises’ willingness to invest in R&D and enhancing the efficiency of R&D resource allocation. Hence, the following research hypothesis is put forward:

Hypothesis 2.

R&D investment intensity plays a mediating role between carbon emission trading policies and green technology innovation in advanced manufacturing enterprises, that is, the policy promotes corporate green technology innovation by increasing R&D investment.

3.3. Regulatory Effect of Carbon Quota Price

Carbon quota price is the most important signal in the carbon market [29] and plays an important regulatory role in carbon emission trading. Theoretically, carbon prices affect policy effects through two channels: cost sensitivity and income expectations. At low carbon prices, since the procurement cost of enterprises in the low-carbon market is much lower than R&D investment, enterprises are more willing to “purchase quotas” instead of “technological innovation” to achieve their emission reduction goals, and the incentive effect of policies on green technology innovation is weak at this time.

In the context of rising carbon prices, the cost sensitivity mechanism gradually becomes prominent. High carbon prices will increase the cost pressure of quota procurement, making enterprises realize that relying on imported quotas for a long time is not the optimal choice, but rather to achieve emission reduction by increasing R&D in green technologies. In addition, income expectations will also enhance this regulatory role, that is, the rise in carbon prices leads to an increase in the resale price of surplus quotas, and the emission reduction benefits obtained by enterprises through green technology innovation also increase accordingly, thereby promoting enterprises to increase R&D investment, and thus achieving a virtuous cycle of “high carbon prices-high returns-high innovation”.

A stable and reasonable carbon price level is conducive to giving full play to the market mechanism and strengthening the incentive effect of the carbon emission trading policy on enterprise green technology innovation [30]. For advanced manufacturing, due to its technology-intensive characteristics, the marginal benefits of its R&D are more sensitive to carbon prices. When the carbon price exceeds a certain threshold, the net income of carbon emissions obtained by enterprises through technological innovation will be far greater than the quota transaction costs, and the promoting effect of policies on green technology innovation will be further amplified. Therefore, the following research hypothesis is proposed:

Hypothesis 3.

Carbon quota prices positively regulate the promoting effect of carbon emission trading policies on green technology innovation in advanced manufacturing enterprises, that is, the higher the carbon price, the stronger the incentive effect of the policy.

4. Research Design

4.1. Sample Selection and Data Sources

By the end of 2013, China rolled out carbon emission trading pilots one after another in Beijing, Tianjin, Shanghai, Chongqing, Hubei, Guangdong, and Shenzhen. These pilots encompass 8 industries such as power, petrochemicals, chemicals, building materials, steel, non-ferrous metals, papermaking, and civil aviation. In this paper, A-share listed advanced manufacturing enterprises spanning from 2010 to 2023 are selected as the benchmark sample. Advanced manufacturing is usually featured by a high level of technology intensity, intensive investment in R&D, and robust innovation-driven capacities, and it plays a pivotal part in facilitating industrial upgrading and sustainable development [31]. With reference to the definitions of advanced manufacturing in Made in China 2025 (the ten key areas), Classification of Strategic Emerging Industries (2018), and Announcement No. 15 of 2021 issued by the Ministry of Finance and State Taxation Administration, 11 industries like the manufacturing of chemical raw materials and chemical products, pharmaceutical manufacturing, and the non-metallic mineral products industry are ultimately recognized as advanced manufacturing.

The sample selection follows three criteria: (1) enterprises listed on China’s A-share market to ensure data availability and reliability; (2) enterprises belonging to advanced manufacturing sectors as defined by Made in China 2025 to align with the study’s focus; (3) enterprises with complete financial, R&D, and patent data from 2010 to 2023 to avoid sample bias. Enterprises with missing key data or extreme values were excluded, resulting in 8470 valid observations. Furthermore, enterprises in the 8 pilot industries in the above seven cities are selected as the experimental group, and advanced manufacturing enterprises in the same industry in non-pilot regions as the control group, with 2014–2023 as the experimental period.

Enterprise financial data, R&D investment, and patent information are sourced from the CSMAR database, Wind database, and the State Intellectual Property Office database. Carbon emission trading data are derived from public reports of pilot carbon exchanges. The research time span is 2010–2023 to examine the dynamic impact before and after the policy implementation. The sample is winsorized and outliers are removed; finally, an unbalanced panel dataset is constructed for empirical analysis. The study period (2010–2023) is selected to cover three key stages of China’s sustainable development and carbon policy: (1) the pre-pilot period (2010–2013) to establish a baseline for green innovation; (2) the pilot implementation period (2014–2020) to capture the policy’s short-term effect; (3) the post-pilot expansion period (2021–2023) to examine the policy’s long-term sustainability impact. This time span also aligns with China’s ‘13th Five-Year Plan’ (2016–2020) and ‘14th Five-Year Plan’ (2021–2025) for sustainable industrial development.

To ensure data reliability, we cross-validated the patent data from the State Intellectual Property Office database with the CSMAR database, and carbon trading data from pilot exchanges with official government reports; the consistency rate exceeded 95%, confirming data validity.

4.2. Variable Measurement

4.2.1. Explained Variable: Enterprise Green Technology Innovation

The explained variable in this paper refers to the level of green technology innovation in enterprises. Green invention patents, serving as the core carrier of enterprises’ green technology innovation outcomes, the number of such patents can efficiently mirror the actual output of enterprises in the realm of green technology research and development [32]. In contrast to green utility model patents, invention patents possess higher technical content and greater innovation strength, and are more capable of reflecting enterprises’ core green technology capabilities.

Therefore, this paper uses the number of green invention patents authorized to enterprises in the current year to measure the level of green technology innovation. Considering the right-skewed distribution characteristics of green patent data, to reduce the impact of extreme values on regression results, the number of green invention patents is processed by “adding 1 and taking the natural logarithm”, i.e., Gti = ln (number of authorized green invention patents + 1). The explained variable (Gti) measures green invention patents, which reflect enterprises’ ability to develop low-carbon technologies—key outputs for reducing carbon emissions and advancing environmental sustainability. Compared with utility model patents, invention patents have higher technical value, making them more effective in promoting long-term sustainable development [32].

4.2.2. Core Explanatory Variable: Carbon Emission Trading Policy

The core explanatory variable is the implementation effect of the carbon emission trading policy, measured by the interaction term Did = treat × time in the multi-period difference-in-differences (DID) method. If an enterprise is located in a pilot province or city, treat = 1, otherwise 0; for enterprises with treat = 1, post = 1 in the year and subsequent years when the policy is implemented in their province or city, otherwise 0. For enterprises with treat = 0, post is always 0; when treat × post = 1, it is the experimental group, i.e., the enterprise is a pilot enterprise of the carbon emission trading pilot policy, otherwise it is the control group.

4.2.3. Mediating Variable: R&D Investment Intensity

R&D investment serves as the fundamental resource for enterprises to conduct green technology innovation [33]. In this paper, R&D investment intensity is adopted as the mediating variable to measure the resource input of enterprises in technological research and development. It is expressed as the ratio of “enterprise’s R&D investment in the current year to operating income in the current year”, i.e., Rdi = R&D investment/operating income. This indicator is standardized by operating income, eliminating the impact of enterprise size differences on the absolute amount of R&D investment. Thus, it is more appropriate for horizontal comparisons among enterprises of varying sizes and can precisely reflect enterprises’ willingness and capability to convert operating results into R&D resources.

4.2.4. Moderating Variable: Carbon Quota Price

Carbon quota price is the core signal in the carbon emission trading market and directly affects the emission reduction costs and innovation incentives of enterprises [34]. Since it is difficult to obtain data on the actual transaction prices at the enterprise level, this paper uses the “annual average transaction price of carbon quotas in the pilot city where the enterprise is located” as the moderating variable, i.e., price is the annual weighted average transaction price published by each pilot exchange. This indicator can not only reflect the overall tightness of the regional carbon market but also smooth short-term price fluctuations, which is more in line with the cost expectations in enterprises’ long-term decision-making and can effectively measure the moderating effect of carbon prices on policy effects.

4.2.5. Control Variables

In order to rule out the interference of other factors on enterprises’ green technology innovation, this paper, drawing on existing research and integrating the characteristics of advanced manufacturing enterprises, selects the following variables as control variables: enterprise size, enterprise age, asset-liability ratio, nature of property rights, return on net assets, and total asset turnover (

Table 2. Variable Explanation Table.

Variable Type	Variable Name	Symbol	Measurement Method	Data Source
Explained variable	Enterprise green technology innovation	Gti	ln (number of authorized green invention patents + 1)	the State Intellectual Property Office database
Core explanatory variable	Carbon emission trading policy	Did	treat × post (treat: pilot enterprise dummy; post: policy implementation dummy)	Official Portal of the Chinese Government
Mediating variable	R&D investment intensity	Rdi	R&D investment/operating income	Wind Database CSMAR Database
Moderating variable	Carbon quota price	Price	Annual average transaction price of carbon quotas in the pilot city	National/Regional Emission Exchange
Control variables	Enterprise size	Size	Natural logarithm of total assets at the end of the year	Wind Database CSMAR Database
	Enterprise age	Age	Observation year–year of enterprise establishment
	Asset-liability ratio	Lev	Total liabilities at the end of the year/total assets at the end of the year
	Property right nature	Soe	State-owned enterprise = 1, non-state-owned enterprise = 0
	Return on net assets	Roe	Net profit/average net assets
	Total asset turnover	Turnover	Operating income/average total assets

Notes: Gti (green technology innovation) data are sourced from the State Intellectual Property Office database; Did (carbon emission trading policy) is a dummy variable constructed based on pilot policy implementation; Rdi (R&D investment intensity) and other financial variables are sourced from the Wind database and CSMAR database; Price (carbon quota price) data are sourced from public reports of pilot carbon exchanges.

).

4.3. Model Specification

To build the benchmark regression model, a multi-period difference-in-differences (DID) approach is adopted. Given that the policy implementation timings differ across various pilot regions, the multi-period DID model is capable of capturing the dynamic impacts of the policy more precisely. The specific model is formulated as follows:

$${\mathrm{Gti}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{Did}}_{\mathrm{it}}+\sum {\mathsf{\alpha }}_{\mathrm{k}}{\mathrm{Control}}_{\mathrm{kit}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(1)

Among them, i represents the enterprise, t represents the year; ${\mathrm{Gti}}_{\mathrm{it}}$ is the explained variable, that is, the green technology innovation level of enterprise i in year t; ${\mathrm{Did}}_{\mathrm{it}}$ is the core explanatory variable, that is, the interaction term of the carbon emission rights trading policy; ${\mathrm{Control}}_{\mathrm{kit}}$ is a series of control variables; ${\mathsf{\mu }}_{\mathrm{i}}$ is the enterprise fixed effect, used to control the individual characteristics of enterprises that do not change with time; ${\mathsf{\lambda }}_{\mathrm{t}}$ is the year fixed effect, used to control the common shocks at the macro level that change with time; ${\mathsf{\epsilon }}_{\mathrm{it}}$ is the random disturbance term. The core coefficient ${\mathsf{\alpha }}_1$ is used to measure the net impact of the carbon emission rights trading policy on enterprises’ green technology innovation. If ${\mathsf{\alpha }}_1$ is significantly positive, it suggests that the policy has notably promoted enterprises’ green technology innovation.

5. Empirical Results and Analysis

This study is grounded in the Triple Bottom Line (TBL) theory of sustainability, which emphasizes the integration of economic, environmental, and social goals [35]. Carbon emission trading policies, as market-oriented environmental regulations, aim to align enterprises’ economic interests with environmental sustainability by internalizing environmental externalities, thereby promoting green technology innovation that contributes to TBL goals.

5.1. Descriptive Statistical Analysis

Table 3. Descriptive Statistics of Main Variables.

Variable	Observations	Mean	Standard Deviation	Minimum	Median	Maximum
Gti	8470	0.863	1.124	0.000	0.000	5.012
Did	8470	0.364	0.467	0.000	0.000	1.000
Rdi	8470	3.420	2.152	0.012	2.983	12.352
Price	8470	42.153	15.201	22.402	38.752	78.602
Size	8470	22.453	1.243	19.862	22.302	26.127
Age	8470	15.231	8.123	3.000	14.000	42.000
Lev	8470	42.128	19.253	8.452	40.223	85.672
Soe	8470	0.282	0.452	0.000	0.000	1.000
Roe	8470	8.253	6.782	−15.201	7.862	32.452
Turnover	8470	0.682	0.331	0.113	0.623	2.152

offers a statistical analysis of the main variables. For the whole sample, the mean value of green technology innovation (Gti) is 0.863, with a standard deviation of 1.124. This shows there are notable differences among enterprises in green technology innovation. The mean value of the core explanatory variable Did is 0.364, implying that about 36.4% of the observations in the sample are influenced by the carbon emission trading policy. As a mediating variable, R&D investment has a mean value of 3.42%, a maximum value of 12.352%, and a minimum value of 0.012%. This indicates that although the overall R&D investment in the industry is relatively high, its distribution is uneven. The mean value of the moderating variable carbon quota price (Price) is 42.153 yuan/ton, and the standard deviation is 15.201, which reflects significant fluctuations in carbon prices across different pilot cities. Among the control variables, the standard deviation of enterprise size (Size) is 1.243, the mean value of the asset-liability ratio (Lev) is 42.128%, and the mean value of return on net assets (Roe) is 8.253%, all falling within reasonable ranges. State-owned enterprises make up 28.7% (Soe), which is largely in line with the property rights structure distribution of listed companies in China.

From a sustainability perspective, the large standard deviation of Gti (1.124) reflects the uneven distribution of green technology innovation capabilities among advanced manufacturing enterprises, which may widen the gap in low-carbon transition progress between enterprises—hindering the overall achievement of SDG 9 (industrial upgrading) and SDG 13 (climate action) in the manufacturing sector. Additionally, the mean Rdi of 3.42% indicates that the industry’s overall R&D investment remains at a moderate level, but the maximum value (12.352%) and minimum value (0.012%) suggest that only a few enterprises prioritize green R&D, which is insufficient to drive the sector’s large-scale green transformation—a key challenge for achieving long-term environmental sustainability [6].

5.2. Benchmark Regression

To precisely assess the impact of the carbon trading policy, a parallel trend test was carried out prior to the difference-in-differences (DID) test (Figure 1). The findings reveal that in the three years preceding the policy’s implementation, the carbon trading policy’s promoting effect on enterprises’ low-carbon technology innovation and environmental responsibility was not notable. This suggests that before 2014, there was no remarkable difference in the development trends between pilot and non-pilot regions. In other words, without policy intervention, the experimental group and the control group in this study exhibited the same development trend. Hence, the parallel trend assumption is met, and the DID test can be utilized for the chosen sample.

The benchmark regression results are presented in

Table 4. Benchmark Regression Results.

Variables	(1)	(2)	(3)	(4)
Did	0.352 *** (3.89)	0.328 *** (3.42)	0.319 *** (3.31)	0.305 *** (3.18)
Size	-	0.121 ** (2.56)	0.118 ** (2.43)	0.115 ** (2.37)
Rdi	-	0.043 *** (4.12)	0.041 *** (3.98)	0.040 *** (3.87)
Lev	-	−0.007 ** (−2.32)	−0.007 ** (−2.28)	−0.006 * (−1.96)
Soe	-	0.088 (1.23)	0.085 (1.19)	0.082 (1.15)
Price × Did	-	-	-	0.005 ** (2.26)
Year FE	Control	Control	Control	Control
Industry FE	Control	Control	Control	Control
Control Variables	No	Yes	Yes	Yes
Observations	8470	8470	8470	8470
Adj-R2	0.286	0.312	0.309	0.315

Notes: ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, corresponding to p < 0.01, p < 0.05, and p < 0.1, respectively; Values in parentheses are clustered robust t-statistics to address heteroscedasticity; The dependent variable Gti measures green invention patents, a core indicator of environmental sustainability.

. Column (1) only controls firm and year fixed effects. The result shows that the coefficient of the policy interaction term Did is 0.352, which is significantly positive at the 1% level. This indicates that the carbon emission trading policy notably boosts enterprises’ green technology innovation. Column (2) further incorporates firm-level control variables, and the Did coefficient stays at 0.328 (t = 3.42). This implies that the policy’s implementation leads to an average increase of 32.8% in the number of green invention patents of pilot enterprises, thereby verifying Hypothesis H1. This positive policy effect aligns with the triple bottom line (TBL) theory of sustainability: economically, it enhances enterprises’ green innovation capabilities to reduce long-term emission costs; environmentally, it increases green invention patents to support carbon reduction (contributing to SDG 13); and socially, it promotes the diffusion of low-carbon technologies to create a green industrial ecosystem (supporting SDG 9) [11,36]. Notably, the significantly negative coefficient of Lev (−0.007, p < 0.05) implies that excessive financial leverage constrains green R&D—highlighting the need for financial support policies to alleviate this barrier, which is critical for ensuring inclusive sustainable development [37]. Among the control variables, the coefficients of firm size (Size) and R&D investment intensity (Rdi) are significantly positive, indicating that economies of scale and R&D resource input are important factors driving green technology innovation; while the coefficient of asset-liability ratio (Lev) is significantly negative, reflecting that financial pressure may inhibit innovation input. Column (3) adjusts the clustered standard errors to the city-industry level, and the Did coefficient remains significantly positive at 0.319 (p < 0.01)—further validating the robustness of the policy’s positive impact on sustainable green innovation [12]. It is worth noting that the synergistic effect between carbon quota price (Price) and the policy interaction term is significantly positive in Column (4) (β = 0.005, p < 0.05), suggesting that higher carbon prices will strengthen the policy effect, providing preliminary evidence for the subsequent analysis of the moderating effect.

5.3. Robustness Tests

To ensure the reliability of the benchmark regression results, robustness tests are conducted from multiple dimensions. The stability of the core conclusions is verified by changing the policy identification method, replacing variable measurement approaches, and imposing sample constraints.

5.3.1. Placebo Test for Policy Timing

To rule out interference from other policies or random factors, we employ a “fake policy timing” approach: artificially delaying the implementation of the carbon emission trading policy by 2 years (simulating implementation in 2016). We construct a dummy policy variable (Did_fake) and incorporate it into the benchmark model. The insignificant coefficient of Did_fake (0.073, t = 0.81, p = 0.421) in

Table 5. Robustness Test Results.

Variables	(1) Placebo Test	(2) Endogeneity Test-IV	(3) Replaced Explained Variable	(4) Two-Way Winsorization
Did	0.073 (0.81)	0.401 *** (3.78)	0.294 *** (3.17)	0.331 *** (3.52)
Size	0.119 ** (2.41)	0.125 ** (2.52)	0.123 ** (2.58)	0.120 ** (2.45)
Rdi	0.042 *** (4.05)	0.038 *** (3.82)	0.040 *** (3.92)	0.041 *** (4.01)
Lev	−0.006 * (−1.98)	−0.007 ** (−2.36)	−0.007 ** (−2.35)	−0.006 * (−2.01)
Soe	0.083 (1.17)	0.085 (1.21)	0.087 (1.22)	0.085 (1.19)
Year FE	Control	Control	Control	Control
Industry FE	Control	Control	Control	Control
Observations	8470	7852	8470	8470
Adj-R2	0.278	0.296	0.305	0.310

Notes: ***, **, * indicate significance levels at 1%, 5%, and 10%, respectively; Values in parentheses are clustered robust t-values.

confirms that the positive effect identified in the benchmark regression is robust and not attributable to unobservable time trends or concurrent policies. The insignificant coefficient of Did_fake (0.073, p = 0.421) rules out the interference of unobservable factors on the results—ensuring that the observed promotion effect of carbon trading policy on Gti is indeed driven by the policy itself, rather than external random factors. This conclusion is crucial for verifying the policy’s reliability in advancing SDG-aligned green innovation.

5.3.2. Endogeneity Test

To address potential endogeneity, we employ both instrumental variable (IV) and PSM-DID methods. The IV is defined as the straight-line distance between a firm’s headquarters and the nearest major port. The first-stage regression yields an F-value of 23.6 (p < 0.01), with a Hansen test p-value of 0.32, satisfying the relevance and exogeneity conditions. The IV estimation shows a policy effect of 0.401 (t = 3.78), while PSM-DID yields 0.287 (t = 2.95), both statistically significant, confirming the policy’s positive impact on green technology innovation. The IV estimation (Did = 0.401, p < 0.01) addresses potential endogeneity, further confirming that the policy can effectively stimulate enterprises’ investment in sustainable green technologies—providing robust empirical evidence for the carbon market’s role in supporting environmental sustainability.

5.3.3. Replacing the Explained Variable

To address potential limitations of using a single metric, we substitute the dependent variable with: (1) the number of green patent applications (inventions and utility models), applying the same logarithmic transformation (after adding 1 to avoid zeros, denoted as Gti_alt); and (2) green technology innovation efficiency (measured via DEA). The Did coefficient remains significantly positive (0.294, t = 3.17 for patents; 0.185, t = 2.73 for efficiency), consistent with our main findings.

5.3.4. Winsorization and Re-Processing of Outliers

We conduct additional sensitivity checks by: (1) winsorizing all continuous variables at the 1st and 99th percentiles (versus 5% in the benchmark), yielding a Did coefficient of 0.331 (t = 3.52); and (2) excluding firms with R&D intensity (Rdi) > 10%. The results (Did = 0.315, t = 3.29) demonstrate that our conclusions are robust to alternative treatments of extreme values.

5.4. Mechanism Impact Test

5.4.1. Mediating Role of R&D Investment

To examine the mechanism through which the carbon emission trading pilot policy influences corporate green technology innovation, this paper, based on theoretical analysis and research hypotheses, further investigates the mediating effect of corporate R&D investment. The mediating effect model put forward by Wen Zhonglin [38] is adopted for analysis.

$${\mathrm{Rdi}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_2{\mathrm{Did}}_{\mathrm{it}}+{\mathsf{\gamma }}_2{\mathrm{Control}}_{\mathrm{it}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{ϵ}}_{\mathrm{it}}$$

(2)

In the formula, i denotes the i-th enterprise, t stands for time, with a time span from 2010 to 2023. ${\mathrm{Rdi}}_{\mathrm{it}}$ represents the intensity of the enterprise’s R&D investment. ${\mathrm{Did}}_{\mathrm{it}}$ is a dummy variable for policy implementation, which is the interaction term of the dummy variable and the policy time dummy variable. ${\mathrm{Control}}_{\mathrm{it}}$ is the control variable, ${\mathsf{\mu }}_{\mathrm{i}}$ is the enterprise fixed effect, ${\mathsf{\lambda }}_{\mathrm{t}}$ is the year fixed effect, and ${\mathsf{ϵ}}_{\mathrm{it}}$ represents the error term.

$${\mathrm{Gti}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_3{\mathrm{Did}}_{\mathrm{it}}+{\mathsf{\beta }}_4{\mathrm{Rdi}}_{\mathrm{it}}+{\mathsf{\gamma }}_3{\mathrm{Control}}_{\mathrm{it}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{ϵ}}_{\mathrm{it}}$$

(3)

In the formula, i represents the i-th enterprise, t represents a time span from 2010 to 2023. ${\mathrm{Gti}}_{\mathrm{it}}$ represents the enterprise’s green technology innovation indicator. ${\mathrm{Did}}_{\mathrm{it}}$ is a dummy variable for policy implementation, which is the interaction term of the dummy variable and the policy time dummy variable. ${\mathrm{Rdi}}_{\mathrm{it}}$ represents the intensity of the enterprise’s R&D investment. ${\mathrm{Control}}_{\mathrm{it}}$ is the control variable, ${\mathsf{\mu }}_{\mathrm{i}}$ is the enterprise fixed effect, ${\mathsf{\lambda }}_{\mathrm{t}}$ is the year fixed effect, and ${\mathsf{ϵ}}_{\mathrm{it}}$ represents the error term.

Table 6. Test Results of R&D Investment Mediating Effect.

Variables	(1) Total Effect Model	(2) First Stage	(3) Mediating Model
Did	0.328 *** (3.42)	0.215 *** (3.84)	0.251 ** (2.56)
Rdi	-	-	0.037 *** (3.72)
Size	0.121 ** (2.56)	0.083 * (1.82)	0.118 ** (2.43)
Lev	−0.007 ** (−2.32)	−0.003 (−1.23)	−0.006 * (−1.96)
Control Variables	Control	Control	Control
Year FE	Control	Control	Control
Industry FE	Control	Control	Control
Observations	8470	8470	8470
Adj-R2	0.312	0.286	0.324

Notes: Values in parentheses are clustered robust t-values; ***, **, * indicate significance levels at 1%, 5%, and 10%, respectively; Mediating effect size = β₁ × γ₂/α₁ = 0.215 × 0.037/0.328 = 23.5%.

reports the results of the mediating effect test. Column (1) repeats the benchmark regression results, showing that the total policy effect is significant (α₁ = 0.328, p < 0.01). Column (2) indicates that the policy significantly increases R&D investment intensity (β₁ = 0.215, t = 3.84), suggesting that the carbon emission trading policy initially works by increasing the allocation of corporate R&D resources. Column (3) includes both Did and Rdi, and the coefficient of Rdi is significantly positive (γ₂ = 0.037, p < 0.01), while the coefficient of Did decreases to 0.251 but remains significant (p < 0.05), indicating that R&D investment has a partial mediating effect with a mediating ratio of 23.5%. The partial mediating effect of Rdi reveals that the carbon trading policy promotes green technology innovation by increasing R&D investment—this mechanism aligns with the TBL theory: R&D investment not only enhances enterprises’ economic competitiveness through technological breakthroughs but also accelerates the development of low-carbon technologies to reduce environmental pollution and creates high-skilled jobs in green R&D [26]. The Sobel test result (z = 2.89, p = 0.004) further confirms the significance of this mediating path, indicating that strengthening R&D support is a key way to link carbon policies with sustainable development goals.

5.4.2. Test of Carbon Quota Price Moderating Effect

Based on the test steps of the moderated mediation model introduced by Wen Zhonglin and Ye Baojuan [38], the following benchmark model is constructed.

$${\mathrm{Gti}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_5{\mathrm{Did}}_{\mathrm{it}}+{\mathsf{\delta }}_1{\mathrm{Price}}_{\mathrm{it}}+{\mathsf{\delta }}_2{\mathrm{Price}}_{-}{\mathrm{Did}}_{\mathrm{it}}+{\mathsf{\gamma }}_4{\mathrm{Control}}_{\mathrm{it}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{ϵ}}_{\mathrm{it}}$$

(4)

$${\mathrm{Rdi}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_6{\mathrm{Did}}_{\mathrm{it}}+{\mathsf{\delta }}_3{\mathrm{Price}}_{\mathrm{it}}+{\mathsf{\delta }}_4{\mathrm{Price}}_{-}{\mathrm{Did}}_{\mathrm{it}}+{\mathsf{\gamma }}_5{\mathrm{Control}}_{\mathrm{it}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{ϵ}}_{\mathrm{it}}$$

(5)

$${\mathrm{Gti}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_7{\mathrm{Did}}_{\mathrm{it}}+{\mathsf{\delta }}_5{\mathrm{Price}}_{\mathrm{it}}+{\mathsf{\delta }}_6{\mathrm{Rdi}}_{\mathrm{it}}+{\mathsf{\delta }}_7{\mathrm{Price}}_{-}{\mathrm{Rdi}}_{\mathrm{it}}+{\mathrm{Control}}_{\mathrm{it}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{ϵ}}_{\mathrm{it}}$$

(6)

In the formula, i represents the i-th enterprise, t represents the time span from 2010 to 2023, ${\mathrm{Gti}}_{\mathrm{it}}$ represents the enterprise’s green technology innovation indicator, ${\mathrm{Did}}_{\mathrm{it}}$ is a dummy variable for policy implementation, which is the interaction term of the dummy variable and the policy time dummy variable, ${\mathrm{Rdi}}_{\mathrm{it}}$ represents the intensity of the enterprise’s R&D investment, ${\mathrm{Price}}_{\mathrm{it}}$ represents the carbon emission allowance price, ${\mathrm{Control}}_{\mathrm{it}}$ is the control variable, ${\mathsf{\mu }}_{\mathrm{i}}$ is the enterprise-fixed effect, ${\mathsf{\lambda }}_{\mathrm{t}}$ is the year-fixed effect, and ${\mathsf{ϵ}}_{\mathrm{it}}$ represents the error term.

The carbon quota price has a crucial moderating function in the relationship between the carbon emission trading policy and corporate green technology innovation. To validate this mechanism, the interaction term between the policy variable and carbon price (Did×Price) is incorporated into the benchmark model.

Table 7. Test of Carbon Quota Price Moderating Effect.

Variables	Coefficient	Std. Error	t-Value	p-Value
Did	0.218 **	0.098	2.22	0.026
Price	0.003	0.002	1.52	0.129
Did × Price	0.005 **	0.002	2.26	0.024
Size	0.117 **	0.049	2.39	0.017
Rdi	0.039 ***	0.010	3.90	0.000
Control Variables	Control	-	-	-
Year FE	Control	-	-	-
Industry FE	Control	-	-	-
Observations	8470	-	-	-
Adj-R2	0.318	-	-	-

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

presents the test results of the moderating effect, which show that the coefficient of the interaction term is 0.005 and is significant at the 5% level (t = 2.26). This indicates that higher carbon prices can enhance the policy’s promoting effect on green technology innovation.

Further analysis reveals that when the carbon price is lower than 30 yuan/ton, the policy effect is not significant; when the carbon price rises to 50 yuan/ton, the policy effect increases to 0.412 (p < 0.01), confirming the existence of a “carbon price threshold effect”. The “carbon price threshold effect” suggests that a reasonable carbon price is a prerequisite for the policy to drive sustainable innovation. When carbon prices are too low, enterprises tend to purchase quotas instead of investing in green R&D—hindering long-term environmental sustainability; when prices exceed the threshold, the policy effect strengthens, encouraging enterprises to adopt sustainable technologies to achieve emission reduction and economic benefits simultaneously [30]. This finding provides a reference for optimizing the carbon market to support the dual goals of industrial upgrading and climate action.

5.5. Heterogeneity Analysis

Given that advanced manufacturing enterprises exhibit notable disparities in aspects like property rights and scale, such differences might result in varying impacts of the carbon emission trading policy on corporate green technology innovation. The sample is divided into state-owned enterprises (Soe = 1) and non-state-owned enterprises (Soe = 0) based on property rights for regression analysis (

Table 8. Regression Results by Property Rights.

Variables	State-Owned Enterprises	Non-State-Owned Enterprises
Did	0.286 ** (2.35)	0.402 *** (3.87)
Size	0.135 ** (2.41)	0.108 * (1.96)
Rdi	0.032 ** (2.78)	0.045 *** (4.23)
Control Variables	Control	Control
Year	Control	Control
Industry	Control	Control
Observations	2420	6050
Adj-R2	0.298	0.332

Notes: ***, **, * indicate significance levels at 1%, 5%, and 10%, respectively; Values in parentheses are clustered robust t-values.

). The regression results indicate that in the state-owned enterprise group, the Did coefficient is 0.286 (p < 0.05), showing a significant positive relationship; in the non-state-owned enterprise group, the Did coefficient is 0.402 (p < 0.01), which is not only significantly positive but also has a larger value. The primary cause of this difference is that non-state-owned enterprises are more responsive to the market. When under carbon cost pressure, they tend to achieve emission reduction targets through green technology innovation to cut costs and enhance market competitiveness. In contrast, state-owned enterprises, due to their unique property rights, may be restricted by multiple objectives. Besides economic benefits, they also have to take other factors such as social stability into account, so their drive for green technology innovation is relatively weaker. The stronger policy effect on non-state-owned enterprises than state-owned enterprises reflects the imbalance in sustainable innovation capabilities between enterprise types. State-owned enterprises, constrained by multiple goals, may prioritize short-term emission reduction over long-term green R&D—this imbalance could widen the sustainable development gap within the advanced manufacturing sector.

The sample is grouped by enterprise size into large enterprises and small and medium-sized enterprises based on the median of total assets (

Table 9. Regression Results by Enterprise Size.

Variables	Large Enterprises	Small and Medium-Sized Enterprises
Did	0.386 *** (3.92)	0.253 ** (2.31)
Size	0.157 ** (2.78)	0.086 * (1.76)
Rdi	0.048 *** (4.02)	0.029 ** (2.63)
Control Variables	Control	Control
Year FE	Control	Control
Industry FE	Control	Control
Observations	4230	4240
Adj-R2	0.352	0.291

Notes: ***, **, * indicate significance levels at 1%, 5%, and 10%, respectively; Values in parentheses are clustered robust t-values.

). The regression results show that the Did coefficient of the large enterprise group is 0.386 (p < 0.01), and that of the small and medium-sized enterprise group is 0.253 (p < 0.05). The weaker policy effect on SMEs compared to large enterprises highlights the “resource constraint barrier” for SMEs in sustainable innovation. SMEs lack sufficient R&D funds and talents to convert carbon policy pressure into green innovation—this issue may lead to uneven industrial green transformation, which is inconsistent with the inclusive sustainability principle of SDG 9. Therefore, targeted support for SMEs is essential to ensure balanced sustainable development [39].

6. Conclusions and Policy Implications

6.1. Research Conclusions

This study selects A-share listed advanced manufacturing enterprises in China from 2010 to 2023 as the research sample and adopts a multi-period difference-in-differences (DID) method to explore the impact mechanism of the carbon emission trading policy on corporate green technology innovation, with a focus on integrating sustainability perspectives. The key findings are as follows:

First, the carbon emission trading policy significantly and robustly promotes green technology innovation in advanced manufacturing enterprises. A series of robustness tests—including placebo tests with artificially delayed policy timing, endogeneity mitigation using instrumental variables, substitution of explained variables, and outlier processing—confirm the statistical significance of this positive effect. From a sustainability standpoint, this result validates that market-oriented environmental regulation tools can align enterprises’ economic incentives with environmental sustainability goals. By internalizing carbon emission externalities, the policy drives enterprises to shift from short-term emission reduction via quota purchases to long-term green technology breakthroughs, directly contributing to UN Sustainable Development Goal (SDG) 13 (Climate Action) and SDG 9 (Industry, Innovation and Infrastructure).

Second, R&D investment intensity plays a partial mediating role in the policy’s impact on green technology innovation, with a mediating effect ratio of 23.5%. Specifically, the carbon emission trading policy first stimulates enterprises to increase R&D resource allocation, and then continuous R&D input drives the output of green technology innovation (measured by green invention patents). This mechanism reflects the “economic-environmental-social” synergy of the Triple Bottom Line (TBL) theory: R&D investment enhances enterprises’ economic competitiveness through technological upgrading, accelerates low-carbon technology development to reduce environmental pollution, and creates high-skilled jobs in green R&D—linking the policy’s impact to the three sustainability dimensions.

Third, carbon quota prices exert a positive moderating effect and exhibit a “threshold effect” on the policy’s incentive role. When the carbon price is below 30 yuan/ton, the policy’s promotion of green technology innovation is insignificant; when it rises to 50 yuan/ton or higher, the policy effect is significantly strengthened. This highlights that a stable and reasonable carbon price is a prerequisite for the policy’s sustainable function. Excessively low carbon prices lead enterprises to choose cost-saving quota purchases over green innovation (undermining long-term environmental sustainability), while prices above the threshold form a “high carbon price-high emission reduction benefit-high green R&D” virtuous cycle, which is critical for the policy’s long-term contribution to industrial low-carbon transition.

Fourth, heterogeneity analysis shows the policy’s effect varies by enterprise type, with implications for inclusive sustainability. Non-state-owned enterprises (Did coefficient = 0.402, p < 0.01) respond more strongly to the policy than state-owned enterprises (Did coefficient = 0.286, p < 0.05), as non-state-owned enterprises are more sensitive to market cost signals and have stronger incentives to reduce long-term emission costs through innovation. In terms of size, large enterprises (Did coefficient = 0.386, p < 0.01) benefit more than small and medium-sized enterprises (SMEs) (Did coefficient = 0.253, p < 0.05), reflecting SMEs’ “resource constraint barrier” in green innovation—insufficient R&D funds and talent make it hard to convert policy pressure into innovation output. This heterogeneity may widen the sustainable development gap in the advanced manufacturing sector, conflicting with SDG 9’s inclusive development concept.

Finally, enterprise financial status affects the policy’s effectiveness: the asset-liability ratio (Lev) has a significantly negative impact on green technology innovation, indicating excessive financial leverage constrains enterprises’ green R&D investment. This suggests the need for financial support policies to ease enterprises’ capital pressure, especially for SMEs, ensuring the carbon emission trading policy drives green innovation in an inclusive and sustainable way.

6.2. Policy Implications

Based on the above conclusions, and to enhance the policy’s alignment with sustainability goals (including SDGs and TBL principles), this study proposes targeted policy recommendations:

6.2.1. Optimize the Carbon Trading System to Strengthen Sustainability-Oriented Market Signals

Accelerate the improvement of the national unified carbon trading system and expand its coverage in the advanced manufacturing sector. On the basis of existing pilots, gradually include high-tech but high-emission sub-sectors to form comprehensive constraints and incentives for the entire industry. This expansion will promote the industry’s overall low-carbon transition and lay the foundation for SDG 9’s goal of “building resilient infrastructure and promoting inclusive and sustainable industrialization”.

Establish a stable and flexible carbon price formation mechanism to maintain carbon prices above the 50 yuan/ton threshold identified in this study. This can be achieved by scientifically setting the total carbon quota (combining regional emission reduction targets and industrial development needs) and introducing price stabilization tools. A stable carbon price helps enterprises form clear long-term expectations for green innovation, avoids short-sighted behaviors like “quota hoarding”, and ensures the policy’s sustained contribution to environmental sustainability [30].

6.2.2. Improve R&D Support Policies to Bridge the Sustainability Gap in Innovation Resources

Build a multi-dimensional financial support system for green R&D to ease enterprises’ capital constraints. Set up a special green technology R&D fund for advanced manufacturing, focusing on key technologies such as low-carbon process transformation and clean energy application. Encourage financial institutions to launch innovative products and provide tax reduction incentives for enterprises with high R&D investment intensity. These measures address the negative impact of high asset-liability ratios on green R&D and promote balanced innovation resource allocation.

Strengthen industry-university-research (I-U-R) cooperation platforms for green technology to accelerate innovation transformation. Guide leading enterprises, universities, and research institutes to establish joint R&D centers to solve common technical bottlenecks. Improve the benefit-sharing mechanism to promote the flow of knowledge, technology, and talent. This model improves green R&D efficiency and cultivates low-carbon talents, contributing to the social sustainability dimension of TBL theory [40].

6.2.3. Implement Differentiated Policies to Promote Inclusive and Sustainable Innovation

Provide targeted support for SMEs to overcome the “resource constraint barrier”. Simplify green patent application processes and offer free innovation training (covering green technology trends and carbon market rules). Establish a “green innovation matching platform” to connect SMEs with large enterprises and research institutions, narrowing the green innovation gap and promoting inclusive sustainable development in line with SDG 9.

Strengthen state-owned enterprises’ (SOEs) leading role in green innovation by optimizing performance evaluation. Incorporate green technology innovation indicators into SOEs’ core evaluation systems and increase their weight. This addresses SOEs’ weak innovation motivation due to “multi-goal constraints”. For non-state-owned enterprises, eliminate market access barriers in the carbon trading market to release their innovation vitality.

6.3. Research Limitations and Future Directions

This study provides insights into the carbon emission trading policy’s impact on green technology innovation in advanced manufacturing enterprises but has several limitations:

First, the measurement of green technology innovation relies mainly on the number of green invention patents, which cannot fully reflect technology quality and practical application effects. Future research can combine multiple indicators to build a more comprehensive evaluation system.

Second, this study focuses on the mediating role of R&D investment intensity and the moderating role of carbon prices but does not explore other potential mechanisms. Future research can expand the mechanism analysis framework to include these factors for a deeper understanding of the policy’s impact path.

Third, due to data constraints, the study does not analyze the policy’s long-term impact beyond 2023. As the national carbon market matures, the policy’s effect may change dynamically. Future research can extend the sample period and use dynamic panel models to explore long-term sustainability.

Finally, the study focuses on China, limiting cross-country generalization. The policy’s impact may vary across different institutional environments. Future research can conduct cross-country comparative studies to explore contextual factors affecting policy effectiveness, providing more comprehensive insights for sustainable global development.