Green Innovation, Export Synergy, and Total Factor Productivity: Evidence from China’s Marine Enterprises

Abstract

In the context of China’s “dual carbon” goals and rising green trade barriers, green transformation is key to improving total factor productivity (TFP) and competitiveness in marine industries. This study uses panel data of Chinese listed marine enterprises (2014–2023) and a multidimensional fixed-effects model to examine how green innovation, export, and R&D investment jointly affect TFP. Results show that (1) green innovation has an inverted “S”-shaped nonlinear effect on TFP, with marginal returns rising, then accelerating, and finally declining; (2) positive synergies exist between green innovation and both exports and R&D, while the export–R&D interaction negatively affects TFP, indicating coordination challenges.; and (3) ownership heterogeneity matters, as state-owned enterprises benefit from stronger institutional support, mitigating negative effects, while private firms are more vulnerable due to weaker green technology mechanisms. This study emphasizes green innovation as a driver for sustainable productivity growth in marine enterprises and suggests policies that improve institutional frameworks, incentives, and resource allocation to support high-quality green innovation.

1. Introduction

The 20th National Congress of the Communist Party of China explicitly emphasized the strategic objectives of “strengthening the nation’s strategic scientific and technological capabilities” and “developing the marine economy, protecting the marine ecological environment, and accelerating the construction of a maritime power [1]”. Against the backdrop of growing global demand for the development and utilization of marine resources, marine technology has increasingly emerged as a core driver of marine economic development. Currently, China’s marine gross product is making a significant and growing contribution to national economic growth. According to the China Marine Economic Statistical Bulletin, in 2024, China’s marine output exceeded CNY 10 trillion for the first time, reaching CNY 10.5438 trillion, a year-on-year increase of 5%, raising its share in GDP to 7.8% [2]. Recognizing the importance of high-quality marine economic development, the Chinese government has issued a series of policy documents focusing on marine energy exploitation, high-end equipment manufacturing, and eco-friendly marine utilization, aiming to build a systematic support framework. Cultivating a new quality productive force in the marine sector has become a critical strategy for China to gain a competitive edge in the global marine arena [3]. This transformation relies on the upgrading of marine human resources, the optimization of marine production objects, and the innovation of marine production tools. Achieving a balance between economic efficiency and ecological sustainability is fundamental to the development of the marine economy. It is also essential for building an intelligent, green, and integrated modern marine manufacturing system [4]. In this context, green innovation within the marine industry has become a key pathway to improving firms’ total factor productivity (TFP), while a higher level of opening up to global markets further accelerates the development of new productive forces in the marine economy. Enhancing the TFP of marine enterprises is not only an important means of facilitating industrial transformation and upgrading but also a critical engine for the high-quality development of the marine economy.

Against the backdrop of simultaneous economic growth and environmental challenges, the relationship between green innovation and total factor productivity (TFP) has garnered increasing attention. Green innovation is recognized as a key engine driving sustainable development. It refers to technological innovation activities that aim to reduce environmental impacts, improve resource efficiency, and promote green transformation. A defining characteristic of green innovation is its capacity to balance economic performance with environmental outcomes [5]. The European Union defines green innovation as technological innovation activities that comply with ecological economic principles, conserve energy and resources, and prevent, eliminate, or mitigate environmental pollution and degradation [6]. Prior studies have emphasized that green innovation includes new products, processes, or technological solutions that are intended to protect the environment and support sustainable development [7,8]. Green innovation has been shown to deliver ecological dividends through resource savings and pollution reduction. Furthermore, it contributes to improvements in green total factor productivity (GTFP) through industrial upgrading and the optimization of energy structures [9,10,11,12].

In addition, growing attention has been paid to the dual externalities of green innovation. On the one hand, green innovation creates positive externalities through knowledge and technology spillovers. These benefits often cannot be fully internalized by the innovating firms, even though they bear the costs of innovation. On the other hand, the negative externalities associated with environmental pollution weaken incentives for green innovation when emission costs are not reflected in production costs [13]. Although some studies have reported that technological innovation may increase energy use and carbon emissions [14], many firms view green technology adoption as a strategic response to long-term development goals and reputational concerns. These firms often aim to enhance production efficiency and product value through green innovation. Empirical findings indicate that green innovation has a significant positive effect on GTFP. Moreover, this relationship is often nonlinear. Invention-type green patents have been found to produce stronger effects on carbon emission reduction and factor agglomeration, and their impacts often exhibit threshold effects and spatial spillovers [15,16,17,18].

Total factor productivity, which is closely linked to green innovation, serves as a comprehensive measure of a firm’s ability to convert inputs such as capital and labor into outputs efficiently. It reflects technological progress, institutional quality, and the efficiency of resource allocation. TFP is widely acknowledged as a fundamental source of long-term economic growth and industrial advancement [19]. It also plays a central role in explaining differences in income across countries. According to the standard real business cycle model [20], TFP shocks affect economic dynamics through changes in investment and labor supply. Theories that incorporate innovation as an endogenous driver of TFP growth emphasize the challenge of financing the fixed costs associated with innovation. This issue is addressed by granting temporary monopoly rights to innovators, allowing them to recover their initial investments through profits earned from patented technologies [21,22]. In addition, policies such as market expansion, greater access to skilled labor, and research and development subsidies can reduce the marginal cost of innovation, thereby accelerating technological progress and raising TFP growth rates [23].

At the industrial level, green TFP has become a critical indicator for assessing sustainable development capacity [24]. In the case of China, the green manufacturing system has been gradually strengthened. By 2024, a total of 6430 national-level green factories had been established, accounting for approximately 20% of the total industrial output value. Green innovation has been found to significantly improve TFP and enhance carbon emission performance [25]. In particular, within the marine economy, TFP reflects not only the governance capability, input efficiency, and resource allocation of enterprises, but also their comprehensive ability to address ecological constraints and respond to environmental policy requirements [26,27]. However, the impact of green innovation on TFP is not uniform and may follow nonlinear patterns. Therefore, further investigation is required to identify the underlying mechanisms and dynamic marginal effects of green innovation on productivity.

With respect to the relationship between exports and total factor productivity (TFP), trade liberalization is commonly associated with a selection effect, wherein increased openness leads to the survival of the most productive firms [28]. Due to firm-level productivity heterogeneity, trade induces a process of “survival of the fittest”, where only more efficient firms can successfully enter export markets [29]. According to the self-selection hypothesis, high-productivity firms are more likely to choose to export, while less productive firms tend to focus on domestic markets or are eventually eliminated from competition [30,31,32,33]. Empirical evidence from German and Japanese manufacturing sectors provides strong support for the self-selection mechanism [34,35]. In general, trade openness facilitates technology diffusion and knowledge spillovers from foreign trade partners, thereby contributing to income and productivity growth. Exporting firms tend to exhibit a productivity advantage over non-exporters, particularly in the early stages of entering international markets [36]. However, the link between exports and productivity remains controversial. Some studies argue that there is no significant causal relationship between exporting and productivity gains. In fact, research on Chinese firms has identified an “export-productivity paradox”, whereby high-productivity firms are more inclined to sell domestically, while relatively less productive firms disproportionately engage in export activities [37,38]. This counterintuitive finding suggests that the relationship between export status and firm productivity may be shaped by institutional factors, market distortions, or industry-specific dynamics in developing economies.

Considering green innovation in conjunction with export behavior, the rapid expansion of China’s economy has led an increasing number of firms to engage in international trade. A growing body of research highlights the close linkage between export activity and breakthrough green innovation, with export-oriented firms generally exhibiting a higher share of patent applications compared to their non-exporting counterparts in most years [39]. Green innovation has been shown to exert a significantly positive impact on firm-level total factor productivity (TFP). Although higher per capita export values may attenuate the marginal effect of green innovation on TFP, export intensity itself remains positively associated with productivity improvements [6]. Moreover, there exists notable heterogeneity in firms’ green management awareness. Firms with a stronger environmental consciousness are more inclined to pursue green technological innovation as a strategic pathway for green transformation and sustainable development. These firms also place greater emphasis on environmental, social, and governance (ESG) performance and the cultivation of a green reputation. Institutional mechanisms such as the official designation of “green factories” in China can effectively enhance green innovation capability, thereby improving firms’ reputational capital and facilitating higher levels of outward foreign direct investment [40]. The mechanism through which exports promote firm performance lies in the positive spillover effects of green technology innovation, which enhances TFP and the technological sophistication of export products. As a result, firms are better positioned to increase the export volume of green products [41]. This suggests a virtuous cycle wherein green innovation not only drives internal productivity gains but also reinforces firms’ external competitiveness in international markets through environmentally sustainable exports.

While extensive research has been conducted on the total factor productivity (TFP) of marine enterprises and their export performance individually, studies that simultaneously examine the combined effects of green innovation and export behavior on TFP in the marine sector remain scarce. Compared with general manufacturing or service firms, marine enterprises exhibit distinct characteristics, such as strong resource dependency, a highly complex operational environment, stringent regulatory constraints, and high technological intensity. Marine enterprises rely heavily on marine resources, and their development must strike a delicate balance between resource utilization and environmental sustainability. Moreover, industries such as fisheries, maritime transport, and shipbuilding are subject to rigorous governance under international maritime law, high seas resource management agreements, and national-level environmental regulations. As a result, marine firms must navigate complex international trade barriers and rising environmental compliance standards. Additionally, firms operating in sectors such as deep-sea exploration, intelligent aquaculture, and marine energy development face high dependence on advanced technologies and equipment. These activities demand significant R&D investment and are often hindered by elevated technological thresholds. Consequently, marine enterprises frequently encounter financing constraints, inefficiencies in resource allocation, and weak innovation capabilities, all of which inhibit improvements in total factor productivity [42]. To enhance their core competitiveness, marine enterprises must respond not only to the unique demands of natural marine environments and regulatory frameworks but also embrace technological innovation and ecologically sustainable strategies. Integrating into global value chains (GVCs) offers a pathway for marine firms to transcend geographical limitations, leverage international resources, restructure industrial linkages, and elevate their technological and market competitiveness [43]. However, the integration process into GVCs is not without challenges: marine enterprises often face technological bottlenecks, market access restrictions, and geopolitical power imbalances in maritime domains, which can significantly constrain their participation and upgrading in the global economy [44].

Given the unique characteristics of marine enterprises, such as high dependence on natural resources, complex operating environments, stringent regulatory constraints, and technology-intensive production, it remains uncertain whether green innovation investments can effectively enhance firm productivity and how export behavior influences productivity dynamics. This study aims to investigate the impact of green innovation investment and export activity on total factor productivity (TFP) at the firm level in the marine sector. Specifically, it seeks to compare the effects of green innovation on firm productivity and to examine the productivity differences associated with firms’ export decisions. Employing production function estimation methods, we analyze the influence of green innovation and export behavior on the operating revenue of Chinese marine enterprises and further estimate firm-level TFP. Additionally, we explore how TFP responds under different scenarios of green innovation investment and export participation to uncover the underlying mechanisms. Given the absence of a standardized classification of marine enterprises in the literature, this study adopts the “Classification of Marine and Related Industries” [45], issued by the China Marine Standardization Technical Committee in 2022, to ensure scientific and consistent identification of marine firms. Based on this taxonomy, we identify 139 marine-related A-share listed companies whose core business activities correspond to the standard’s major, intermediate, or minor industry categories, including offshore oil, gas, and mineral extraction and equipment R&D, marine communication, information, and environmental product manufacturing and services, marine aquatic product processing, shipbuilding and marine engineering equipment manufacturing, marine-related equipment production, marine chemical and petrochemical manufacturing, marine transportation, and marine research and development.

This study contributes to the existing literature in the following aspects. First, based on firm-level panel data from China’s marine industries, it constructs a fixed-effects model to systematically evaluate the impact mechanisms of green innovation and export behavior on total factor productivity (TFP). It focuses in particular on the marginal effects and interactions among green patents, export performance, and R&D investment, thereby enhancing the understanding of how green transformation and internationalization interact at the enterprise level. Second, the study identifies a significant inverted-S-shaped nonlinear relationship between green innovation and TFP, revealing the heterogeneous productivity effects of green patents across different stages of development. This finding emphasizes that green transformation should not rely solely on the growth in the quantity of green patents, but must also prioritize their industrial applicability and economic value. Third, the results demonstrate that while green innovation exhibits synergistic effects with both exports and R&D, the interaction between exports and R&D is significantly negative. This suggests that, in the absence of a solid foundation, simultaneous expansion in both areas may lead to resource misallocation and efficiency losses. Finally, the study conducts a heterogeneity analysis based on ownership structure. It finds that state-owned enterprises (SOEs) demonstrate more stable synergy between green innovation and R&D, while private enterprises, due to limited green technology accumulation and inefficient transformation pathways, face stronger nonlinear impacts. These results underscore the need for targeted and differentiated policy support. In summary, this study enriches the theoretical foundation and empirical evidence on the relationship between green innovation and productivity by addressing micro-level mechanisms, nonlinear effects, synergy pathways, and ownership heterogeneity.

2. Theoretical Analysis and Hypothesis

The impact mechanism of green innovation on total factor productivity (TFP) is multidimensional and complex. Theoretically, innovation is regarded as a fundamental driver of long-term productivity growth [46,47], with innovation outputs significantly enhancing firms’ technological efficiency and resource allocation capacity [48]. As an environmentally-oriented technological advancement, green innovation not only improves resource utilization efficiency and production processes to achieve the dual goal of increasing desirable outputs while reducing undesirable ones, but also plays a pivotal role in improving energy efficiency and optimizing industrial structures [12,49,50]. Empirical studies at the firm level indicate that green technology investments and R&D activities substantially boost productivity [51], with particularly evident effects observed in state-owned enterprises under supportive institutional environments [52]. Furthermore, green innovation facilitates the accumulation and diffusion of technical knowledge, strengthens firms’ absorptive capacity for external green technologies, and promotes reinvention, contributing to sustained technological progress [53,54]. Simultaneously, green innovation supports the transition from resource-intensive growth models to knowledge-intensive, innovation-driven development, thereby improving the overall innovation environment and accelerating the growth of green total factor productivity (GTFP) [55]. As a key indicator of high-quality green economic development, GTFP reflects the coordination between firm-level efficiency and environmental performance, and serves as an important metric for assessing the effectiveness of green policies and the vitality of enterprises [42].

However, some scholars have raised concerns regarding the positive impact of green innovation. They argue that due to the positive externalities of knowledge spillovers, firms may lack sufficient incentives to innovate, while negative externalities in the form of pollution abatement efforts may come at the cost of reduced TFP [13,56]. In addition, existing studies often overlook the spatial, nonlinear, and heterogeneous characteristics of green innovation impacts [15]. Therefore, although green innovation has the potential to improve TFP, it may also face efficiency bottlenecks due to underdeveloped incentive mechanisms, warranting a comprehensive investigation in both empirical identification and policy design. Based on the above discussion, this study posits the following hypothesis:

Innovation is the core driver of long-term productivity growth [46,47]. Empirical studies have demonstrated that innovation output has a positive impact on total factor productivity (TFP) [48]. Green innovation investment can effectively improve firms’ production structure and output efficiency by enhancing resource utilization efficiency and promoting technological progress. Empirical analyses at the firm level also find that R&D activities significantly increase firm productivity [51]. Moreover, green innovation is considered a key link in coordinating environmental regulation and firm performance growth [49], promoting industrial upgrading and improving energy utilization efficiency [12]. Under China’s institutional environment, investment in green technologies has a significant and heterogeneous effect on the productivity improvement of state-owned enterprises [52]. However, although green innovation’s contribution to green total factor productivity continues to increase, its heterogeneity, spatial characteristics, and nonlinear effects are often neglected in current research [15]. At the same time, some scholars question whether green innovation truly has a positive effect on firm-level TFP. The double externalities associated with green innovation may lead to insufficient innovation incentives and low innovation efficiency due to a lack of external motivation [13]. Firms tend to engage in green innovation primarily to meet government and market expectations rather than to achieve substantive breakthroughs in pollution reduction. Some studies indicate that the pollution reduction effects of green patents may come at the expense of firm-level total factor productivity to some extent [56]. Based on the above analysis, green innovation, as environmentally driven technological innovation, plays a crucial role in the sustainable development of marine-related enterprises. Therefore, this study proposes the following hypotheses:

H1. 

Green innovation has a significant positive effect on the total factor productivity of marine-related enterprises.

The relationship between international trade and firm productivity mainly involves two aspects: first, the positive impact of trade liberalization on firms’ total factor productivity (TFP); second, the role of foreign direct investment in enhancing TFP. The theory of “learning-by-exporting” suggests that firms entering international markets must continuously learn and improve technologies to adapt to foreign customer demands, meet higher quality standards, and face intense competition, thereby promoting productivity growth [46,57,58]. Studies have shown that exporting helps firms improve resource allocation efficiency and achieve economies of scale, thus enhancing TFP. Some scholars found that exporting firms generally outperform non-exporting firms in terms of technical efficiency and productivity, further confirming that the “export-induced TFP improvement” effect is also significant in China’s regional industrial clusters [59]. Moreover, a firm’s productivity is influenced by its previous export activities [60]. Differences in trade patterns can explain variations in total factor productivity [61]. Based on the above theories, this study posits that firm export behavior still entails learning effects and innovation incentives, and therefore proposes the following hypothesis:

H2. 

Exporting has a significant positive effect on the total factor productivity of marine-related enterprises.

In the absence of green innovation, economic openness and export diversification may suppress improvements in ecological efficiency [62]. Green innovation and exports not only individually enhance firm total factor productivity (TFP), but their synergistic effect can realize a complementarity between “endogenous technological capabilities” and “external market experience”, further amplifying productivity gains [63]. Exporting firms that simultaneously invest in innovation experience significantly greater productivity improvements compared to those investing in only one dimension. Using data from Japanese manufacturing firms, scholars have also confirmed the reinforcing effect of the interaction between green innovation and exports on TFP [64]. Trade has a significant positive impact on TFP, benefiting from improved resource allocation and access to new markets [65]. In summary, this study argues that green innovation helps improve firms’ overseas export performance, while export behavior also incentivizes green innovation, thereby enhancing total factor productivity. Therefore, the following hypothesis is proposed:

H3. 

There is an interactive and reinforcing effect between green innovation and export behavior in enhancing the total factor productivity of marine-related enterprises.

Firm ownership structure influences strategic decision-making logic and resource allocation efficiency. Some scholars argue that Chinese state-owned enterprises (SOEs) exhibit weaker motivation for strategic green innovation, while non-state-owned enterprises (NSOEs) are more inclined to pursue profits through such innovation [66]. However, others suggest that firms adopting market-based strategies are less likely to establish government-business ties, making them more likely to pay pollution fees without receiving environmental subsidies, yet more willing to engage in green innovation [67]. SOEs generally possess superior resource endowments but face weaker innovation incentives. In contrast, private firms are more market-driven and often respond more swiftly to green innovation and export opportunities. Significant differences in innovation efficiency exist across firms with different ownership structures [68]. Resource misallocation in SOEs limits their TFP growth potential [69]. State ownership of enterprises has a positive impact on green innovation [70], further confirming that “ownership heterogeneity” plays a crucial role in policy responsiveness and technology pathway selection:

H4. 

The impact of green innovation and export on the total factor productivity of marine-related enterprises varies across different ownership types.

3. Production Function Estimation

3.1. Production Function Model

In the basic production function model, labor and capital are typically considered as the primary input factors, with value added used as the dependent variable. The portion of output that cannot be explained by these input factors is defined as total factor productivity (TFP).

$$\mathit{ln}\left(Y_{it}\right)=\alpha \mathit{ln}\left(K_{it}\right)+\beta \mathit{ln}\left(L_{it}\right)+\gamma \mathit{ln}\left({R\&D}_{it}\right)+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

In the production function, Y_it denotes the output of firm i in year t (typically measured by operating revenue or gross output). K_it, L_it, and RD_it represent capital input, labor input, and R&D investment, respectively. The parameters α, β, and γ correspond to the output elasticities of capital, labor, and R&D input. μ_i captures firm fixed effects to control for time-invariant unobservable heterogeneity across firms, while λ_t denotes year fixed effects to account for systematic shocks over time. ε_it is the error term capturing random disturbances not explained by the model.

This study begins by estimating the production function to construct firm-level total factor productivity (TFP). We apply both the Olley-Pakes (OP) and Ackerberg–Caves–Frazer (ACF) estimation methods. Compared to the OP method, the ACF approach demonstrates superior performance in several respects when estimating firm production functions.

Table 1. Comparison of OP and ACF estimation results.

Variable	(1) Olley-Pakes Method	(2) ACF Method
$\mathrm{l}\mathrm{n} \left(Capital\right)$	0.273 ***	0.288 ***
	(0.016)	(0.012)
$\mathrm{l}\mathrm{n} \left(Employees\right)$	0.408 ***	0.385 ***
	(0.027)	(0.023)
$\mathrm{l}\mathrm{n} (R\&D)$	0.363 ***	0.402 ***
	(0.019)	(0.017)
N	1081	1390
Adj. R2	0.836	0.892
RMSE	0.56	0.56

Note: p < 0.1, *** p < 0.001. Source: Own elaboration.

shows that capital, labor, and R&D inputs all have statistically significant and positive effects on output at the 1% level under both methods. Specifically, the coefficient for capital input is 0.288 under the ACF method, higher than the 0.273 obtained with the OP method. The coefficient for R&D input increases from 0.363 (OP) to 0.402 (ACF), suggesting that the ACF method better captures the marginal contribution of R&D to output. Although the labor input coefficient slightly decreases, the overall change is negligible. In terms of model fit, the R-squared of the ACF estimation reaches 0.892, substantially higher than the 0.836 from the OP method, indicating that the ACF method has stronger explanatory power regarding output variation. Moreover, the ACF method retains more observations (1390 vs. 1081), enhancing the robustness of the estimation. In conclusion, by addressing the identification issues associated with the investment function in the OP method, the ACF method more accurately captures firms’ dynamic production decisions, especially in complex production environments involving intermediate inputs such as R&D. Therefore, this study ultimately adopts the ACF method to estimate firm-level TFP, ensuring the accuracy and interpretability of the results.

$\mathrm{l}\mathrm{n}\left(Capital\right)$$\mathrm{l}\mathrm{n}\left(Employees\right)$$\mathrm{l}\mathrm{n}(R\&D)$

3.2. Trends in TFP Changes Among Marine Enterprises

The Figure 1 illustrates that the total factor productivity (TFP) of Chinese marine enterprises exhibited a generally slow upward trend from 2014 to 2023. Although there were significant differences among individual firms, with noticeable fluctuations in TFP levels, the overall trajectory remained relatively stable without sharp volatility. The solid black line represents the average TFP of all sample firms, indicating a slight increase in overall productivity after 2016 and reaching a relatively stable peak around 2020, followed by a steady trend. The red dashed line, which represents the fitted trend line, further confirms the gradual improvement in TFP. This pattern may reflect the growing effectiveness of China’s supportive policies for the marine economy, continued improvements in firms’ technological investment and management efficiency, and the positive impact of emerging factors such as digital technologies on productivity. Nevertheless, the multicolored trajectories at the lower level of the figure reveal substantial heterogeneity in both the level and trend of TFP among firms, suggesting that there remains considerable room to improve resource allocation efficiency within the industry.

4. Data and Methodology

4.1. Model Setting

$$\begin{gathered} \mathbf{l}\mathbf{n}{\left(\mathit{T}\mathit{F}\mathit{P}\right)}_{\mathit{i}\mathit{t}}={\mathit{\beta }}_0+{\mathit{\beta }}_1·\mathbf{ln}\left({\mathit{G}\mathit{r}\mathit{e}\mathit{e}\mathit{n}\mathit{I}\mathit{n}\mathit{n}\mathit{o}\mathit{v}\mathit{a}\mathit{t}\mathit{i}\mathit{o}\mathit{n}}_{\mathit{i}\mathit{t}}\right)+{\mathit{\beta }}_2·\mathbf{ln}\left({\mathit{E}\mathit{x}\mathit{p}\mathit{o}\mathit{r}\mathit{t}\mathit{s}}_{\mathit{i}\mathit{t}}\right)+{\mathit{\beta }}_{\mathit{n}} \\ ·{\mathit{I}\mathit{n}\mathit{t}\mathit{e}\mathit{r}\mathit{a}\mathit{c}\mathit{t}\mathit{i}\mathit{o}\mathit{n}}_{\mathit{i}\mathit{t}}+{\mathit{\beta }}_{\mathit{n}}·{\mathit{C}\mathit{o}\mathit{n}\mathit{t}\mathit{r}\mathit{o}\mathit{l}\mathit{s}}_{\mathit{i}\mathit{t}}+{\mathit{\mu }}_{\mathit{i}}+{\mathit{\lambda }}_{\mathit{i}}+{\mathit{\epsilon }}_{\mathit{i}\mathit{t}} \end{gathered}$$

4.2. Variable Selection and Data Description

Independent variables: The core independent variables in this study are green innovation and exports. Data on patent applications and authorizations of listed firms in heavily polluting industries were manually collected from the search platform of the State Intellectual Property Office (SIPO) of China, along with their corresponding IPC classification codes. Green patents were identified by matching these IPC codes with the “IPC Green Inventory” released by the World Intellectual Property Organization (WIPO) in 2010. To ensure the precision of green patent identification, the selected samples primarily focus on key technological domains such as energy efficiency enhancement, pollution control, marine ecological restoration, and green ship equipment, aligning closely with the development objectives of the marine green industry. Furthermore, to enhance the representativeness of technological quality, the dataset distinguishes between invention patents and utility model patents during the processing stage. The number of authorized green invention patents is adopted as the primary indicator of green innovation, as these patents undergo substantive examination and better reflect the depth and originality of technological advancements. Based on this, the number of authorized green patents for each firm from 2014 to 2023 was calculated annually. Export data were obtained from the annual reports of listed companies in China.

Control variables: As shown in

Table 2. Variable selection and data interpretation.

Variable Type	Variable Name	Variable Description	Meaning
Dependent Variable	$ln TFP$	Total Factor Productivity (TFP)	The natural logarithm of firm i’s total factor productivity in year t, estimated using the ACF method.
Independent Variables	$lnGreen Innovation$	Green Innovation	The natural logarithm of the number of green invention patents (+1 to avoid ln(0)).
	$lnExports$	Export	The natural logarithm of total export value (+1 to handle zero values).
Control Variables	$Firm Age$	Firm Age	Observation year minus listing year plus one.
	$Firm Size$	Firm Size	The natural logarithm of firm revenue.
	$R\&D$	R&D Intensity	R&D expenditure as a share of revenue, reflecting R&D resource allocation efficiency.
	$Capital Intensity$	Capital Intensity	The natural logarithm of capital per employee.
	$Ownership Type$	Ownership Type	Type of firm ownership.

Source: Own elaboration.

, the control variables include (1) Firm age, calculated as the observation year minus the listing year plus one. (2) Firm size, measured by total annual revenue as disclosed in the firms’ annual reports. (3) R&D intensity, defined as the ratio of annual R&D expenditure to revenue, indicating the degree and efficiency of resource allocation to innovation activities. This variable reflects a firm’s commitment to technological innovation and strategic emphasis on R&D. (4) Capital intensity, measured as the natural logarithm of capital per employee. (5) Ownership type, a dummy variable where non-state-owned firms are coded as 1 and state-owned firms as 0. The descriptive statistics of all variables are shown in

Table 3. Data description.

Variable	N	Mean	Sd	Min	Max
$\ln TFP$	933	1.224	0.168	0.000	1.779
$\ln Green Innovation$	933	0.449	0.777	0.000	4.500
$\ln Exports$	933	5.564	2.295	0.000	10.935
$\ln Firm Age$	933	2.896	0.369	0.000	3.611
$\ln Firm Size$	933	7.656	1.695	2.715	12.914
$R\&D$	933	0.067	0.076	0.000	0.791
$\ln Capital Intensity$	933	0.355	0.313	0.000	2.286

Source: Own elaboration.

.

4.3. Regression Results

The empirical results based on the two-way fixed-effects model indicate that the total factor productivity (TFP) of Chinese marine-related enterprises is influenced by multiple factors, including green innovation, export performance, and firm characteristics, exhibiting a relatively clear mechanism of action. The baseline regression results are presented in

Table 4. Baseline regression results.

Variables	Two-Way FE Model
$\ln Green Innovation$	−0.004
	(0.004)
$\ln Exports$	0.006 *
	(0.003)
$\ln Firm Age$	0.111 **
	(0.037)
$\ln Firm Size$	0.090 ***
	(0.005)
$R\&D$	−2.126 ***
	(0.087)
$\ln Capital Intensity$	−0.120 ***
	(0.016)
Num.Obs.	933
R2	0.607
R2Adj.	0.549
RMSE	0.05

Note: p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Source: Own elaboration.

. Among these, export performance (ln Exports) has a significantly positive impact on TFP (coefficient = 0.006, p < 0.05), supporting the “learning-by-exporting” hypothesis. This suggests that participation in international markets compels marine enterprises to meet higher standards and face stronger competitive pressure, which in turn drives technological progress and managerial improvements, ultimately enhancing production efficiency.

In contrast, green innovation shows a negative but statistically insignificant impact on TFP (coefficient = −0.004), indicating that the current increase in the number of green patents has not yet translated into productivity gains. Possible reasons include low patent quality, limited commercialization, or the fact that many green technologies are still in the early R&D phase. Regarding the control variables, capital intensity has a significantly negative effect on TFP (−0.120, p < 0.001), suggesting potential issues such as redundant capital investment or inefficient resource allocation within some enterprises. Meanwhile, firm size and firm age both have significantly positive impacts on TFP (0.090 and 0.111, respectively), implying that larger and more mature firms benefit from stronger capabilities in resource integration and accumulated experience, helping them maintain high productivity levels during green transitions. Notably, R&D intensity exerts a significantly negative effect on TFP (−2.126, p < 0.001), which contradicts theoretical expectations. This may reflect inefficiencies in R&D resource allocation, weak commercialization capabilities, and the lack of an effective mechanism for transforming innovation inputs into productivity gains among some marine firms. Overall, the model demonstrates good explanatory power (R² = 0.607, adjusted R² = 0.549), highlighting exports as a key driver of productivity growth in marine-related enterprises. However, green innovation and R&D investments still need to shift from a focus on quantity to one on quality. Future policies should emphasize the improvement of green technology quality and application efficiency, and enhance the synergy between green innovation, exports, and R&D to facilitate the dual goals of green transformation and high-quality development in the marine industry.

In the baseline regression, although export performance was found to have a significant positive effect on the total factor productivity (TFP) of marine-related enterprises, the direct impact of green innovation was not statistically significant, and the model has certain limitations. Specifically, it fails to fully capture the complex mechanisms through which green innovation may influence firm productivity. Therefore, this study further constructs an extended model by incorporating interaction terms and nonlinear components to explore the marginal effects of green innovation and its synergies with other key variables.

According to the results of the panel fixed-effects model, as shown in

Table 5. Panel fixed-effects model results.

Variables	Two-Way FE Model
$\ln Green Innovation$	−0.064 ***
	(0.017)
$\ln Exports$	0.037 ***
	(0.004)
$\ln Firm Age$	0.036
	(0.035)
$\ln Firm Size$	0.142 ***
	(0.006)
$R\&D$	−1.331 ***
	(0.075)
$R\&D\times GREEN$	0.007 *
	(0.003)
$R\&D\times EXP$	−0.009 ***
	(0.001)
$GREEN\times EXP$	0.002
	(0.002)
$\ln {Green Innovation}^{^2}$	0.020
	(0.012)
$ln{Green Innovation}^{^3}$	−0.007 *
	(0.003)
$\ln Capital Intensity$	−0.109 ***
	(0.015)
Num.Obs.	933
R2	0.6664
RMSE	0.05
Fixed Effects	Firm FE + Year FE

Note: p < 0.1, * p < 0.05, *** p < 0.001. Source: Own elaboration.

. First, from the perspective of interaction effects, the extended model includes interaction terms between green innovation and R&D investment (R&D × GREEN), green innovation and exports (GREEN × EXP), and exports and R&D (R&D × EXP). The coefficient of the interaction between green innovation and R&D is positive and significant (0.007, p < 0.05), indicating that firms with a certain level of R&D capacity are more likely to translate green innovation into productivity gains. This supports the mechanism that R&D enhances the efficiency of green innovation transformation, suggesting that green technologies do not automatically lead to productivity improvement but require complementary internal capabilities. Although the interaction between green innovation and exports is not statistically significant, its positive coefficient (0.002) implies that green innovation may strengthen export competitiveness through product differentiation and compliance with environmental standards. In contrast, the interaction between exports and R&D shows a significant negative effect (−0.009, p < 0.001), indicating that marine-related firms may face resource misallocation and managerial inefficiencies when simultaneously pursuing export expansion and technological innovation without sufficient R&D capacity, thus hampering productivity growth. To further verify this mechanism, this study constructs a panel fixed-effects model by introducing a one-period lagged interaction term. The results show that the lagged interaction between export and R&D (Interact_RD_EXP_lag1) has a coefficient of −0.0187 (p < 0.001), and its squared term is 0.0000699 (p < 0.001), indicating a significant U-shaped nonlinear lagged effect. Specifically, the negative linear term suggests that pursuing export expansion and R&D simultaneously may cause short-term efficiency “growing pains”. In contrast, the positive quadratic term implies that as firms enhance their collaborative capacity and technological maturity, the negative impact will gradually diminish and may even transform into a positive driver of productivity. This mechanism echoes prior findings on the threshold effect of green innovation on GTFP [71,72]. Therefore, while the initial phase of export–R&D synergy may hinder productivity, sustained investment and effective integration are likely to turn it into a crucial driver for long-term TFP improvement. According to Figure 2, the scatter plot and the smoothed trend line suggest a nonlinear relationship between the number of green patents and total factor productivity (TFP).

In exploring the impact mechanism of green innovation on the total factor productivity (TFP) of Chinese marine-related enterprises, this study extends the baseline regression by constructing a nonlinear model that incorporates cubic polynomial terms of the green patent variable (linear, quadratic, and cubic terms) to capture the potentially complex relationship between green innovation and TFP. As TFP is a key component of both long-term and short-term economic growth dynamics [73], identifying the marginal effects of green innovation from a nonlinear perspective is crucial to revealing its intrinsic mechanism. The regression results show that the coefficient of the linear term is −0.064 (p < 0.001), the quadratic term is 0.020 (not significant), and the cubic term is −0.007 (p < 0.05), indicating a typical inverted “S”-shaped curve. This suggests that the marginal effect of green innovation on TFP does not increase linearly but follows a dynamic pattern of “limited effect in the early stage—accelerated growth in the middle stage—diminishing marginal returns in the later stage”. The marginal effects of green innovation on total factor productivity are shown in Figure 3.

In the first stage (initial phase of green innovation), marine-related enterprises often face challenges such as low technological maturity, limited R&D resources, and insufficient institutional incentives. As a result, green patents at this stage have not yet translated into productivity gains. The marginal benefits of green innovation remain low and may even exert short-term pressure on firm performance due to excessive R&D investment. In the second stage (green patent accumulation and technological synergy), as green technology accumulates and firms gradually establish a more integrated green technology system, green innovation becomes increasingly embedded in production processes. This leads to energy savings, pollution reduction, and process optimization, resulting in a significant rise in marginal returns. This phase reflects the agglomeration, synergy, and knowledge spillover effects of green technologies, with green patents exerting their peak impact on TFP. In the third stage (green innovation “overload” phase), when green patent volumes become excessive and decoupled from market demand, their marginal returns begin to decline. Issues such as “patent stockpiling” and redundant R&D may emerge, increasing maintenance and compliance costs, misallocating resources, and ultimately suppressing productivity growth. This suggests that, in the absence of strategic guidance and effective commercialization mechanisms, green innovation may yield negative externalities.

The above findings are generally consistent with existing literature in terms of overall trends, while also presenting certain theoretical distinctions. On one hand, the marginal effect of renewable energy development on green total factor productivity (GTFP) has been shown to follow a “strong-then-weak” nonlinear pattern, which aligns with this study’s conclusion regarding the diminishing marginal returns of green innovation at later stages of technological accumulation [74]. Additionally, threshold regression results demonstrate that only when green patent levels surpass a critical point can they significantly promote GTFP growth, supporting this paper’s finding that green innovation plays a limited role in the early stages [71]. From a regional perspective, the coexistence of positive and nonlinear effects of green innovation on GTFP further confirms the spatial-temporal heterogeneity in the externalities and transformation efficiency of green technologies [72].

On the other hand, this study expands upon prior research by investigating the heterogeneity of green innovation effects across different development stages at the firm level. It emphasizes that under varying conditions of foundational capabilities, commercialization mechanisms, and policy alignment, the actual impact of green technologies on productivity exhibits significant nonlinear fluctuations.

In summary, the relationship between green innovation and firm TFP follows a complex nonlinear mechanism. Its positive effects depend on the combined influence of technological accumulation, conversion capacity, and institutional environment. Therefore, policymakers should shift from a “quantity-driven” to a “quality-oriented” approach in promoting green transformation, focusing on improving the efficiency of green technology commercialization and its industrial value, while mitigating the risk of diminishing marginal returns. This will help achieve truly high-quality green development.

4.4. Robustness Analysis

4.4.1. Random Effects Model with Clustered Standard Errors

To further validate the robustness of the baseline regression results, this study employs a random effects (RE) model with firm-level clustered robust standard errors. As shown in

Table 6. Random effects model with clustered SE.

Variables	(1)
(Intercept)	0.403 ***
	(0.085)
$\ln Green Innovation$	−0.058 **
	(0.019)
$\ln Exports$	0.028 ***
	(0.006)
$\ln Firm Age$	0.044 *
	(0.018)
$\ln Firm Size$	0.122 ***
	(0.011)
$R\&D$	−1.677 ***
	(0.214)
$R\&D\times GREEN$	0.006
	(0.005)
$R\&D\times EXP$	−0.008 ***
	(0.001)
$GREEN\times EXP$	0.003
	(0.003)
$\ln {Green Innovation}^{^2}$	0.015
	(0.016)
$ln{Green Innovation}^{^3}$	−0.006 *
	(0.004)
$\ln Capital Intensity$	−0.115 ***
	(0.032)
$Ownership Type$	−0.112 ***
	(0.024)
Num.Obs.	933
R2	0.715
R2Adj.	0.711
AIC	−2924.9
BIC	−2857.1
RMSE	0.05

Note: p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Source: Own elaboration.

, the coefficient of the green innovation variable is −0.058 and statistically significant at the 5% level, indicating that merely increasing the number of green patents may not continuously enhance firm productivity. In the absence of effective commercialization pathways, such efforts may even result in resource misallocation. The coefficient for exports is 0.028 and highly significant at the 1% level, reaffirming the “learning-by-exporting” effect—marine-related enterprises improve productivity by entering international markets and adapting to competitive pressures and stricter technical standards. Among the control variables, capital intensity is negatively significant (−0.115), which may reflect inefficiencies in capital allocation. In contrast, firm size and firm age show positive and significant effects, suggesting that enterprises with greater resource integration capabilities and institutional experience are more likely to achieve productivity gains.

Moreover, the interaction term between R&D and exports (R&D × EXP) is negative (−0.008) and significant at the 1% level, suggesting that excessive simultaneous efforts in export expansion and R&D may lead to efficiency losses due to resource strain or managerial overload. Although the square and cubic terms of green innovation are statistically insignificant, their consistent directional signs indicate a downward-sloping S-shaped trend, further supporting the notion of diminishing marginal returns to green innovation. Overall, the adjusted R² of the model is 0.711. The coefficient signs and levels of significance remain consistent with those in the baseline regression, thereby reinforcing the study’s core conclusions regarding the impacts of green innovation, export behavior, and their interactions on the productivity of marine enterprises.

4.4.2. GMM and Instrumental Variable (2SLS) Approaches

To address potential endogeneity issues related to green innovation and export variables in the model, this study further adopts the system generalized method of moments (GMM) for robustness testing. In the GMM model, the first and second lags of green innovation and exports are used as instrumental variables for the endogenous regressors. In addition, firm revenue, capital intensity, and firm age are included as control variables to better identify the causal effects on total factor productivity (TFP). The empirical results show that the lagged terms of green innovation and exports are not statistically significant (p-values well above 0.1), suggesting that these variables do not exhibit strong lagged effects on TFP, thus alleviating concerns about endogeneity. Furthermore, the Sargan test for over-identifying restrictions yields a p-value of 0.05088, indicating that the chosen instruments are generally valid, though the result lies close to the margin of statistical significance, implying that the validity of instruments should be interpreted with caution. The autocorrelation tests confirm the absence of first- and second-order serial correlation, enhancing the reliability of the GMM estimates.

To mitigate estimation bias caused by omitted variables, reverse causality, or measurement errors, this study not only employs the system GMM method but also conducts a robustness test addressing endogeneity in green innovation through the use of instrumental variable (2SLS) estimation, following the approaches of [75,76,77]. Specifically, the natural logarithm of firms’ ESG ratings (lnESG) is used as an instrumental variable for green innovation, measured by the number of authorized green patents. This choice is grounded in the argument that ESG performance significantly affects firms’ green strategic implementation and resource allocation decisions, while it is not directly driven by the immediate output of green patents, thus meeting the theoretical requirement of exogeneity. In the first-stage regression, the coefficient of ln ESG on green patents is 0.113 (t = 3.95, p < 0.001), with an F-statistic of 51.6, well above the conventional threshold of 10, confirming both instrument strength and validity. In the second stage, the estimated coefficient for green innovation on total factor productivity (TFP) is significantly negative (−0.041, p < 0.05), lending support to the “green burden hypothesis”. Meanwhile, export behavior (ln Exports) shows a significant positive effect on TFP (coefficient = 0.023, p < 0.001), suggesting that export expansion contributes to improved production efficiency in the current stage.

To ensure the robustness of the estimation, the model controls for key firm characteristics such as capital intensity, revenue size, R&D intensity, firm age, and ownership type. Firm and year fixed effects are also included, and standard errors are clustered at the firm level, thereby enhancing the rigor of causal identification and the credibility of the findings.

Given the research context, where China’s marine enterprises are simultaneously advancing green innovation and internationalization strategies, this study incorporates the system GMM estimation method to effectively address the potential endogeneity of green innovation and export variables. This provides methodological support for analyzing the dynamic impact mechanisms of these factors on total factor productivity (TFP). The GMM results further confirm the robustness of the fixed-effects model after accounting for endogeneity, thereby enhancing the reliability and explanatory power of the empirical analysis. In addition, the study employs an instrumental variable (2SLS) approach, using firms’ ESG ratings as an exogenous instrument to identify and correct the potential endogeneity of green innovation from an alternative perspective. This enhances the strength of causal inference between green innovation and TFP and provides additional evidence supporting the robustness and validity of the research findings.

4.5. Heterogeneity Analysis

Green innovation contributes to the improvement of green total factor productivity (GTFP), but its effects vary depending on the type of patent, firm characteristics, and technological gap [78]. While green innovation facilitates environmental protection, it is also a key driver of economic efficiency [15]. Compared with green utility models and green design innovations, green invention patents are more effective in enhancing GTFP. Heterogeneity analysis further indicates that marine enterprises receiving less government subsidy tend to experience greater improvements in TFP [42]. Moreover, high carbon intensity is significantly associated with weak corruption control and insufficient R&D investment [79]. In state-owned enterprises (SOEs), the potential for rent-seeking due to privileged access to resources and policy enforcement may facilitate the persistence of high-carbon traditional industrial chains, thereby hindering green innovation. In contrast, private firms are often constrained by limited access to finance and weaker policy incentives, which restrict their investment in R&D and hamper their technological upgrading and green product development during the low-carbon transition.

To better capture the differentiated mechanisms through which green innovation and exports affect TFP, this study further categorizes the sample into SOEs and non-SOEs. By separately examining the impact of green innovation and exports on TFP across ownership types, the analysis aims to uncover the heterogeneous effects of ownership structure on policy responsiveness and resource allocation efficiency.

As shown in

Table 7. Comparison of regression results: state-owned vs. private firms.

Variables	(1) State-Owned Firms	(2) Private Firms
$\ln Green Innovation$	−0.084 **	0.015
	(0.026)	(0.031)
$\ln Exports$	0.045 ***	0.041 ***
	(0.008)	(0.005)
$\ln Firm Age$	0.073 **	0.023
	(0.023)	(0.017)
$\ln Firm Size$	0.128 ***	0.146 ***
	(0.011)	(0.007)
$R\&D$	−1.414 ***	−1.324 ***
	(0.125)	(0.090)
$R\&D\times GREEN$	0.009 *	−0.004
	(0.004)	(0.005)
$R\&D\times EXP$	−0.010 ***	−0.011 ***
	(0.001)	(0.001)
$GREEN\times EXP$	0.002	0.003
	(0.003)	(0.002)
$\ln {Green Innovation}^{^2}$	0.027	−0.022
	(0.017)	(0.028)
$ln{Green Innovation}^{^3}$	−0.009 *	0.006
	(0.003)	(0.008)
$\ln Capital Intensity$	−0.024	−0.214 ***
	(0.022)	(0.021)
N	420	513
R2	0.581	0.771
RMSE	0.05	0.04
Fixed Effects	Firm Fixed Effects	Firm Fixed Effects

Note: p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Source: Own elaboration.

, there are significant differences between state-owned enterprises (SOEs) and private firms in terms of how green innovation and export activities affect total factor productivity (TFP). Regarding green innovation, the coefficient of green patent output for SOEs is −0.084 (standard error = 0.026), significant at the 5% level, indicating a negative impact of green technologies on their TFP. Furthermore, the cubic term is also significantly negative (coefficient = −0.009, standard error = 0.003, 10% level), further confirming the diminishing marginal returns of green innovation. In contrast, the corresponding coefficient for private firms is 0.015 (standard error = 0.031) and statistically insignificant, suggesting that green innovation has not yet played a substantial role in promoting their TFP. In terms of exports, both SOEs and private firms show significantly positive coefficients—0.045 and 0.041, respectively (both significant at the 1% level)—which supports the “learning-by-exporting” hypothesis. This suggests that engagement in international markets enables firms to improve efficiency by gaining experience and adapting to higher technical standards. Capital intensity exhibits a significantly negative effect on the TFP of private firms (coefficient = −0.214, standard error = 0.021, significant at the 1% level), indicating potential issues of capital misallocation or inefficient investment. In contrast, the impact on SOEs is not significant (coefficient = −0.024, standard error = 0.022), possibly reflecting better access to or utilization of capital. Firm age has a significantly positive effect on TFP among SOEs (coefficient = 0.073, standard error = 0.023, 5% level), suggesting advantages from institutional stability and accumulated experience. However, the effect is insignificant for private firms (coefficient = 0.023).

For the interaction terms, SOEs exhibit a significant and positive synergy between R&D and green innovation (coefficient = 0.009, standard error = 0.004, 10% level), indicating a complementary relationship. In contrast, the interaction term for private firms is negative and insignificant (−0.004), suggesting the absence of effective synergy mechanisms. Moreover, the interaction between R&D and exports is significantly negative for both ownership types (−0.010 for SOEs and −0.011 for private firms, both at the 1% significance level), indicating that excessive cross-investment or short-term resource dispersion may hinder productivity gains. Lastly, R&D intensity (R&D/Revenue) shows a significantly negative effect on TFP in both groups (−1.414 for SOEs and −1.324 for private firms), implying that current R&D investments have yet to be effectively translated into productivity improvements.

According to the comparison of regression coefficients in Figure 4, state-owned enterprises (SOEs) and privately owned enterprises (POEs) exhibit significant differences in the impacts of green innovation, exports, and R&D investment on total factor productivity (TFP). Overall, SOEs possess an institutional foundation for green innovation–R&D synergy. However, green technologies have yet to become a productivity-enhancing advantage. POEs, by contrast, display greater market sensitivity but still face substantial challenges in improving the efficiency of green innovation and capital utilization. Both types of enterprises need to optimize R&D resource allocation and strengthen mechanisms for green technology commercialization to achieve sustained improvements in TFP. These findings provide empirical support for Hypothesis 4 (H4), which posits that the impacts of green innovation and exports on the TFP of marine-related enterprises are heterogeneous across different ownership types.

4.6. Mechanism Analysis

Table 8. Impact of green innovation and exports on TFP (fixed-effects models).

Variables	(1) Model 1	(2) Model 2	(3) Model 3	(4) Model 4	(5) Model 5
$\ln Green Innovation$		−0.004	−0.005	−0.039 ***	−0.063 ***
		(0.004)	(0.004)	(0.011)	(0.017)
$\ln Exports$			0.006 *	0.036 ***	0.037 ***
			(0.003)	(0.004)	(0.004)
$\ln Firm Age$	−0.043 ***	−0.041 ***	0.009	0.024 *	0.020
	(0.011)	(0.011)	(0.015)	(0.014)	(0.014)
$\ln Firm Size$	0.107 ***	0.108 ***	0.091 ***	0.138 ***	0.139 ***
	(0.004)	(0.004)	(0.005)	(0.006)	(0.006)
$R\&D$	−1.355 ***	−1.352 ***	−1.751 ***	−1.353 ***	−1.345 ***
	(0.049)	(0.049)	(0.075)	(0.076)	(0.076)
$R\&D\times GREEN$				0.002	0.007 *
				(0.003)	(0.003)
$R\&D\times EXP$				−0.009 ***	−0.009 ***
				(0.001)	(0.001)
$GREEN\times EXP$				0.004 *	0.002
				(0.002)	(0.002)
$\ln {Green Innovation}^{^2}$					0.019
					(0.012)
$ln{Green Innovation}^{^3}$					−0.007 *
					(0.003)
$\ln Capital Intensity$	−0.130 ***	−0.129 ***	−0.122 ***	−0.115 ***	−0.114 ***
	(0.017)	(0.017)	(0.017)	(0.015)	(0.015)
N	1390	1390	933	933	933
R2	0.603	0.603	0.591	0.659	0.665

Note: p < 0.1, * p < 0.05, *** p < 0.001. Source: Own elaboration.

presents the fixed-effects panel regression results examining the impact of green innovation and exports on firms’ total factor productivity (TFP). To more comprehensively identify the mechanisms of action between variables and their marginal dynamics, five model specifications are constructed, with green innovation, exports, and their interaction and nonlinear terms introduced step by step to enhance explanatory power and robustness. In Model 1, only baseline control variables are included. The results show that firm size has a consistently significant positive impact on TFP, while capital intensity exhibits a stable negative effect, highlighting the critical role of resource allocation efficiency. Model 2 introduces the green innovation variable. The coefficient is negative and statistically insignificant, providing preliminary evidence against Hypothesis 1 (H1) in the short term. This suggests that green innovation may initially hinder productivity, possibly due to high implementation costs or inefficiencies in the technology conversion process. In Model 3, the export variable is added and shows a significant positive effect on TFP, supporting Hypothesis 2 (H2). This result indicates that exports promote firm efficiency through mechanisms such as economies of scale, international market competition, and technology spillovers. Model 4 incorporates interaction terms. The interaction between green innovation and R&D investment is significantly positive, suggesting a synergistic effect, while the interaction between green innovation and exports is also positive but statistically insignificant. These findings provide initial support for Hypothesis 3 (H3), which posits that green innovation and exports can jointly enhance productivity under certain conditions. However, the R&D–export interaction is significantly negative, indicating that export-oriented R&D may suppress productivity if not supported by adequate institutional or market conditions. In Model 5, a nonlinear structure is constructed by introducing the quadratic and cubic terms of green innovation. The results reveal an inverted S-shaped marginal effect of green innovation on TFP—limited effect in the early stage, accelerated improvement in the mid-stage, followed by diminishing and even negative returns. This suggests that simply increasing the number of green patents without ensuring their quality or commercialization efficiency is insufficient for sustained productivity gains.

To ensure the robustness of the model specification, this study further employs a natural spline regression to examine the nonlinear relationship between green innovation and firms’ total factor productivity (TFP). The estimation results confirm the existence of a stable inverted S-shaped relationship. Specifically, the three basis function terms of ns (ln Green Innovation) reflect the marginal effect of green innovation at low, medium, and high intensity levels. The coefficient of ns (ln Green Innovation) 1 is −0.0136 and not statistically significant, suggesting that when the accumulation of green patents is low, its contribution to TFP is weak or even slightly negative. The coefficient of ns (ln Green Innovation) 2 is 0.1212 and significantly positive at the 5% level, indicating that at a moderate level of green innovation, productivity gains are more pronounced. In contrast, ns (ln Green Innovation) 3 yields a coefficient of −0.2416, also statistically significant at the 5% level, implying that excessive green innovation input results in diminishing or even negative marginal returns. This may be attributed to resource misallocation, technological redundancy, or declining conversion efficiency, which in turn hinders firm performance. These segmented results are highly consistent with the previously identified inverted S-shaped structure based on the cubic polynomial of green patents, thereby reinforcing the robustness of the findings from both the model form and empirical estimation perspectives.

Across all models, R&D intensity consistently exhibits a highly significant negative effect on TFP, indicating that the efficiency of resource allocation remains a key determinant of productivity. Overall, the effects of green innovation and exports on firm productivity are neither linear nor isolated. Instead, they reflect a dynamic mechanism characterized by synergy and structural constraints. The empirical findings verify the conditions under which H1, H2, and H3 hold, providing important insights for promoting high-quality and green development of China’s marine-related enterprises under the dual pressures of carbon neutrality goals and international competition.

5. Conclusions

This study employed micro-level panel data from China’s marine-related industries and constructed a fixed-effects model to systematically analyze the impact mechanisms of green innovation and exports on firms’ total factor productivity (TFP). Special attention was given to the interaction effects and marginal impacts among green patents, export performance, and R&D investment, as well as to the heterogeneity of these mechanisms across state-owned enterprises (SOEs) and privately owned enterprises (POEs). The results reveal a significant inverted “S-shaped” nonlinear relationship between green patents and TFP. In the early stage, green innovation may suppress productivity due to immature technologies and underdeveloped transformation mechanisms. In the middle stage, as green technology accumulates and diffuses, it positively promotes TFP. However, in the later stage, excessive investment or declining efficiency in resource allocation can lead to diminishing marginal returns. This suggests that enterprises’ green transformation should not rely solely on the quantity of green patents but also focus on the industrialization efficiency and actual economic contribution of technologies. Further analysis identifies synergistic effects among green patents, exports, and R&D, indicating that effective integration of green technologies enhances firms’ international competitiveness and endogenous growth momentum. Nevertheless, the interaction between exports and R&D exhibits a significant negative effect, implying that when marine enterprises pursue export expansion and intensive R&D simultaneously without solid foundational capabilities, resource misallocation and efficiency losses may occur. Regarding ownership heterogeneity, SOEs demonstrate stronger institutional foundations and resource allocation capacity for the coordination of green innovation and R&D, with the negative impact of green patents on TFP being relatively weaker. In contrast, POEs, despite possessing greater market responsiveness, suffer from lagging green technology accumulation and inefficient transformation pathways, resulting in a more pronounced inverted “S-shaped” pattern in the green innovation–TFP relationship. This highlights the urgent need for policy mechanisms to guide and support private enterprises in improving green innovation effectiveness.

6. Policy Recommendations

Based on the empirical findings and in light of China’s current policy framework emphasizing green transformation and the “going global” strategy in the marine-related industries, the following policy recommendations are proposed:

6.1. Prioritize Green Innovation to Build Sustainable International Competitiveness

Enterprises should comprehensively integrate ecological principles into all aspects of product design, production processes, and resource utilization, promote the adoption of efficient and clean technologies, and enhance resource recycling efficiency. By actively shaping a green corporate image, firms can improve their adaptability and competitiveness under the evolving global green trade rules. Meanwhile, the government should fully consider the internal and external development environment faced by marine enterprises, clarify industrial planning, and strengthen market-oriented mechanisms, particularly in light of the unique characteristics of marine-related industries. Moreover, it is essential to strengthen full-process supervision over policy implementation, enabling real-time monitoring and timely feedback to ensure policy effectiveness.

6.2. Upgrade Green Innovation from Quantitative to Qualitative and Efficiency-Oriented Development

The government should establish a quality evaluation system for green technologies and support the cultivation of high-value green patents, particularly in key sectors such as marine energy, ecological aquaculture, and equipment manufacturing. This aims to enhance the practical effectiveness of green technologies and avoid resource waste caused by excessive patent accumulation. In addition, R&D funding should be actively directed toward green innovation projects. Taking into account differences in marine resource endowments, industrial development trends, and the actual operational conditions of enterprises, efforts should be made to promote the establishment of green development management systems within firms and to strengthen their green innovation capacity.

6.3. Establish Synergistic Mechanisms Linking Green Innovation, Export, and R&D

It is recommended to encourage marine enterprises to integrate green technologies into their export product systems. By promoting government-led green export demonstration projects and advancing international green technology certification, a coordinated development model of “green innovation–R&D–export” can be fostered, helping to address efficiency losses caused by the misalignment between export expansion and R&D directions. In addition, enterprises should be guided to strengthen independent innovation in green technologies and patents, driving green transformation through external demand and promoting green technology spillovers. Furthermore, the construction of green financing platforms and channels, along with the promotion of carbon trading and eco-environmental products, should be advanced to facilitate the green development of the marine economy.

6.4. Implement Ownership-Specific Support Policies

Despite the growing body of research on green innovation, several gaps remain in the existing literature. Most studies remain at the conceptual level or rely on qualitative comparisons, lacking systematic quantitative analyses that examine differences in green innovation intensity, return cycles, and policy responsiveness across firms with different ownership structures. Whether the “political connectedness” of state-owned enterprises (SOEs) can be effectively translated into green innovation performance has yet to be rigorously tested at the micro-firm level. Additionally, research on the efficiency of green innovation in SOEs remains limited, and whether they face a structural dilemma of “high input–low output” has not been thoroughly explored. In light of these limitations, differentiated policy support paths should be formulated based on ownership type. For SOEs, a performance evaluation system centered on the practicality of green technologies and the conversion rate of innovation outcomes should be established to strengthen the mechanisms for marketizing green technologies. Leveraging their institutional advantages and responsiveness to policy, efforts should be made to promote synergy between R&D and green innovation by providing dedicated funding, performance-based incentives, and platforms for technology commercialization. These measures will help guide SOEs to shift from a “policy-driven” to a “performance-driven” model, thereby enhancing their leadership in green transformation. For private enterprises, whose green innovation efforts have not yet significantly improved TFP, early-stage support is urgently needed. Specifically, policy tools such as fiscal subsidies, tax incentives, and green incubator platforms should be utilized to lower the threshold for green innovation. Given issues such as inefficient capital allocation and insufficient digital transformation, these firms should be guided to increase investment in green equipment and intelligent technologies to improve capital efficiency. To address the observed lack of synergy between exports and R&D, export-oriented green R&D subsidy mechanisms should be established to help these firms align with international green supply chains and enhance the added value and global competitiveness of their green products. Furthermore, green finance support should be expanded, regional shared innovation platforms should be developed, and green talent recruitment mechanisms should be strengthened. Considering that private enterprises in certain regions and industries still face unequal access to resources and difficulties in policy implementation, the government should, under the premise of ensuring fair competition, introduce more targeted and actionable support policies. This will enhance the participation and benefit-sharing capacity of private enterprises in green innovation and promote coordinated and inclusive development across ownership types in the process of green transformation.

6.5. Build a Unified Green Technology Evaluation and Policy Feedback System

A dynamic monitoring system for green technology indicators, export shares, and transformation efficiency should be developed. Strengthening interdepartmental coordination platforms and enhancing fiscal policy support will improve the government’s ability to allocate resources proactively, accurately, and effectively, thereby creating a conducive institutional environment for green innovation and high-quality development in marine enterprises. In view of the distinctive characteristics of the marine industry, it is necessary to establish a dedicated dynamic monitoring system for green technology indicators, export shares, and transformation efficiency. Efforts should be made to strengthen interdepartmental coordination mechanisms, improve the fiscal policy support framework, and build a data-driven policy implementation and feedback loop to ensure the real-time adjustment and precise delivery of green policies. In addition, the incentive mechanisms for marine green innovation should be further refined. Enterprises that achieve technological breakthroughs in key green sectors—such as marine energy, seawater desalination, and environmentally friendly aquaculture—should be granted targeted policy support, including tax incentives and special subsidies. At the same time, government procurement policies should give priority to certified green technology products and services, thereby enhancing market-based incentives.