How heterogeneous environmental regulations shape marine sustainable development: evidence from coastal China

1 Introduction

The United Nations Sustainable Development Goals (SDGs) highlight the conservation and sustainable use of oceans, seas, and marine resources as a key global development objective. This emphasizes the importance of the transition of marine economy from an extensive model to green, low-carbon and efficient pathways (Lee et al., 2020). As the global marine economy rapidly expands, issues such as marine environmental degradation, excessive resource exploitation, and ecosystem deterioration have become increasingly evident (Veríssimo et al., 2021). These emerging pressures call for a fundamental transformation of marine development strategies, shifting away from growth models centered on scale and speed toward approaches that prioritize quality, efficiency, and sustainability (Stebbings et al., 2020). As one of the world’s major maritime countries, China administers nearly three million square kilometers of sea area and possesses abundant marine resources. According to the 2024 China Marine Economic Statistical Bulletin, China’s gross ocean product exceeded RMB 10 trillion in 2024, accounting for approximately 8% of national GDP, and has become an important component of the national economy. However, the ecological costs associated with marine economic development have become increasingly evident. According to the Bulletin of Marine Ecology and Environment Status of China in 2024, approximately 83% of China’s coastal waters met the good-quality seawater standards, while the eutrophicated sea area reached 30,160 km² in summer, and only six typical marine ecosystems were classified as healthy. These indicators reflect continued ecological pressure in China’s coastal and marine regions.

The sustainable development of the marine economy fundamentally relies on efficient resource use and effective environmental protection, which highlights the importance of productivity-oriented performance assessment. However, conventional measures of total factor productivity have predominantly concentrated on desirable outputs, while neglecting undesirable outputs such as environmental pollution and resource consumption (Xie et al., 2021). In response, researchers have integrated environmental factors into efficiency evaluation frameworks, leading to the conceptualization of green total factor productivity (GTFP) (Chung et al., 1997). This productivity framework has subsequently been applied to the marine economy by incorporating resource inputs, desirable economic outputs, and undesirable environmental outputs into the assessment of marine green productivity (Liu et al., 2021; Ren and Ji, 2021). With intensifying ecological pressures and continued economic expansion, enhancing MGTFP has emerged as a crucial strategy for achieving a green transformation. In this study, marine sustainable development is operationalized as the efficiency of transforming marine economic inputs into desirable economic outputs while minimizing undesirable environmental outputs, and is accordingly measured by MGTFP.

Environmental regulation has long been recognized as an essential policy instrument for mitigating marine pollution and enhancing environmental quality, and its role in reconciling economic growth with ecological conservation has become increasingly prominent (Ren and Ji, 2021). From a theoretical perspective, however, its productivity consequences remain ambiguous. Traditional environmental economics emphasizes the compliance costs imposed by regulation, arguing that stricter environmental requirements may raise production costs, crowd out productive investment, and consequently reduce productivity. In contrast, the Porter hypothesis suggests that properly designed environmental regulation can stimulate innovation and improve efficiency (Porter and van der Linde, 1995). Subsequent studies further distinguish between the weak, narrow, and strong versions of the Porter hypothesis, emphasizing innovation stimulation, flexible regulatory design, and productivity enhancement (Lanoie et al., 2008; Ambec et al., 2013). In this theoretical context, existing empirical studies have reported mixed evidence on the relationship between GTFP, which can generally be grouped into three perspectives. Consistent with the Porter hypothesis, the first perspective emphasizes the productivity-enhancing role of environmental regulation, arguing that well-designed regulatory policies can stimulate improvements in production processes, promote industrial restructuring, and induce technological progress, thereby enhancing GTFP. For instance, Cheng and Kong (2022) argued that rational adjustments to industrial structure can facilitate reallocation of resources from underperforming to high-performing sectors, thereby enhancing overall resource utilization and environmental performance. Related studies further suggest that regulatory pressure can enforce firms to adopt green technologies, which ultimately improves production efficiency and environmental outcomes (Song et al., 2022; Wen et al., 2022). Building on this line of inquiry, Lian et al. (2024) empirically confirmed that effectively formulated environmental regulations can significantly boost GTFP and promote more sustainable economic development. In line with the conventional compliance-cost argument, the second perspective highlights the potential inhibitory impacts. Specifically, Jin et al. (2019) have demonstrated environmental regulations hinder GTFP improvement in industrial water resource utilization. This inhibitory effect is further verified by Tang et al. (2020), who found that environmental regulation significantly suppresses firm’s GTFP, with persistent and lagged negative effects. Meanwhile, Yang et al. (2022) revealed that while environmental regulations generally restrain China’s GTFP, this negative impact can be alleviated by mediating mechanisms. A third perspective suggests that these two theoretical mechanisms may coexist rather than be mutually exclusive. Environmental regulation can initially generate compliance costs but produce productivity gains once innovation compensation becomes sufficiently strong, resulting in nonlinear effects. Empirical studies have identified U-shaped or inverted U-shaped relationships, suggesting that the effects of regulation depend critically on its strength and implementation stage (Wu et al., 2020; Qiu et al., 2021). Moreover, Tian and Feng (2022) demonstrated that different regulatory instruments may exhibit distinct nonlinear trajectories, further emphasizing the complexity of the regulation-GTFP relationship.

Within the marine context, closely related studies have further examined environmental regulation in relation to marine economic efficiency, marine GTFP, industrial green transformation, and high-quality marine development (Zheng et al., 2020; Ren and Ji, 2021; Qin et al., 2021; Liu and Chen, 2022; Sun et al., 2023; Wu et al., 2023). Taken together, this literature shows that marine environmental regulation is neither homogeneous in design nor uniform in its productivity consequences. However, the existing evidence remains relatively fragmented in explaining why different regulatory instruments generate divergent marine green productivity outcomes. In particular, it remains less clear whether CER and MER are associated with MGTFP through systematically different transmission patterns when the same candidate channels are evaluated within a common framework. This question is also relevant to the Porter hypothesis, because subsequent theoretical developments emphasize that the balance between compliance costs and compensatory gains depends partly on regulatory design. In addition, existing nonlinear marine studies have also identified important conditioning roles for regulatory intensity, technological innovation, industrial structure, economic development, and FDI. However, less attention has been paid to whether external conditions surrounding policy implementation affect CER and MER differently. Given the distinct administrative and incentive logics of the two instruments, their relationships with MGTFP may respond differently to local fiscal capacity and broader economic openness. These unresolved issues shift the research focus from whether heterogeneous environmental regulation matters to how and under what conditions the differences between CER and MER emerge in marine green productivity. Against this background, the central research question of this study is: how and under what conditions do CER and MER differ in their associations with MGTFP?

In summary, existing studies have provided substantial theoretical and empirical evidence on the relationship between environmental regulation and GTFP, establishing a valuable foundation for further research. Building on this literature, this study makes three incremental contributions. First, it provides an instrument-specific extension of the Porter-hypothesis discussion in the marine productivity context. Rather than treating environmental regulation as a homogeneous policy input, the study compares CER and MER against the same MGTFP outcome and examines whether compliance-oriented administrative regulation and incentive-oriented market regulation exhibit different productivity associations. Second, the study moves from documenting heterogeneous regulatory outcomes to comparing their transmission structure. MIS and MTI are evaluated as common candidate channels for both CER and MER, allowing the analysis to examine whether the two regulatory instruments are associated with MGTFP through different potential pathways. This comparative design helps distinguish instrument-specific transmission patterns from the more general roles of industrial structure and technological innovation identified in previous marine studies. Third, the study extends the analysis of nonlinear regulatory effects to the external conditions surrounding policy implementation. Fiscal capacity and economic openness are introduced as threshold variables for both CER and MER to examine whether the contextual dependence of regulation itself differs across instruments. This framework provides a more conditional comparison of whether the differences between CER and MER vary across fiscal conditions, and external economic environments.

The remainder of this paper is organized as follows. Section 2 examines the mechanisms through which different environmental regulations influence MGTFP and develops the corresponding research hypotheses. Section 3 introduces the research framework, including variable selection, model specification, and data sources. Section 4 presents the empirical findings. Finally, Section 5 presents the conclusion, policy implications, discussion, and limitations and future research.

2 Research hypothesis

2.1 Direct impact of environmental regulations on MGTFP

Marine-specific evidence suggests that environmental regulation may influence green productivity through two competing forces. One is the compliance-cost effect, whereby regulatory requirements increase pollution-control expenditure and constrain productive investment. The other is the compensatory effect emphasized by the Porter hypothesis, whereby regulation induces technological adjustment, resource reallocation, and efficiency improvement. Existing marine studies further indicate that the relative strength of these forces may differ across regulatory instruments (Ren and Ji, 2021; Liu and Chen, 2022; Sun et al., 2023).

2.1.1 Direct impact of CER on MGTFP

CER is defined as the government’s implementation of mandatory constraints on pollutant emissions through laws, regulations, and administrative measures. This form of regulation is characterized by its explicit directive nature and enforcement authority (Peng et al., 2021). In the marine context, CER is mainly reflected in mandatory pollution-control standards, emission limits, technical requirements, environmental inspections, and other administrative compliance obligations. These instruments directly constrain the production and pollution-control decisions of regulated entities and typically require corresponding pollution-control effort and expenditure.

From the compensatory-gain perspective, CER may improve MGTFP by forcing regulated entities to adjust inefficient production practices. Mandatory environmental requirements can accelerate the elimination of outdated and highly polluting capacity, encourage cleaner production processes, and improve the efficiency of resource use (Qin et al., 2021). When these adjustments generate sufficient efficiency gains, the costs imposed by regulation may be partly or fully offset. This prediction is consistent with the broader Porter-hypothesis argument that environmental constraints can induce productivity-enhancing responses under appropriate conditions (Porter and van der Linde, 1995; Ambec et al., 2013).

A competing prediction follows from the conventional compliance-cost view. Mandatory standards and administrative enforcement require regulated entities to devote additional resources to pollution treatment and regulatory compliance. These expenditures may crowd out productive investment and increase operating costs, thereby reducing productivity when the resulting burden exceeds the compensatory gains generated by regulatory adjustment (Barbera and McConnell, 1990). These competing effects imply that the productivity consequences of CER depend on the relative strength of regulatory benefits and compliance costs. If administrative intervention accelerates the exit of inefficient capacity and stimulates green transformation, CER may improve MGTFP. However, when rigid compliance costs dominate these compensatory gains, CER may instead reduce MGTFP. This possibility may be particularly relevant to command-and-control instruments because regulated entities have relatively limited flexibility in choosing how to satisfy mandatory requirements. Consequently, the following hypotheses are proposed:

H1a CER can promote MGTFP.

H1b CER can inhibit MGTFP.

2.1.2 Direct impact of MER on MGTFP

MER refers to regulatory arrangements that influence environmental and resource-use decisions through economic incentives and price signals rather than mandatory administrative requirements (Blackman et al., 2018). In the marine context examined in this study, the relevant mechanism is the paid use and pricing of marine resources. By attaching an explicit economic cost to the occupation and utilization of marine space, this mechanism incorporates resource scarcity and environmental costs into the production and investment decisions of marine entities.

Rather than imposing uniform mandatory standards, MER primarily affects MGTFP through flexible guidance and incentive-based channels (Guo and Yuan, 2020). This characteristic is consistent with the narrow version of the Porter hypothesis, which argues that flexible instruments are more likely to stimulate innovation and efficiency improvements. First, MER incorporates marine resource scarcity and environmental costs into production and investment decisions, thereby encouraging marine production entities to improve resource-use efficiency and adjust extensive production practices. Second, price-based incentives provide regulated entities with greater flexibility in choosing how to respond to environmental and resource-use costs. This flexibility can facilitate resource reallocation toward more efficient production activities and strengthen incentives for technological and operational adjustment, thereby enhancing marine resource utilization efficiency and boosting MGTFP (Li and Ramanathan, 2018). These efficiency and adjustment effects provide a theoretical basis for expecting a positive relationship between the price-based MER mechanism examined in this study and MGTFP. Based on the above analysis, the following hypothesis is proposed:

H2 MER can promote MGTFP.

2.2 Indirect impact of environmental regulations on MGTFP

2.2.1 Marine industrial structure (MIS)

Environmental regulations primarily facilitate industrial restructuring and upgrading by guiding the entry and exit mechanisms of marine production entities and facilitating resources flow between industries. Rigorous environmental regulations can stimulate the relocation of polluting entities, thereby promoting local industrial upgrading (Zhu et al., 2014; Shen et al., 2017). When coastal regions implement strict environmental policies, marine production entities face increased operational costs due to regulatory pressure. To maintain profitability, highly polluting entities will choose to adjust or relocate their production bases. Additionally, environmental regulations raise entry barriers within the marine industry, which not only restricts the access of high-pollution, high-emission marine production entities but also accelerates the phase-out of incumbent inefficient facilities. Consequently, this process reduces the proportion of highly polluting enterprises while advancing the expansion of high-technology and environment friendly sectors, collectively facilitating the upgrading and transformation of the MIS. However, environmental regulations may also increase production cost, leading to reduced profitability, which could potentially inhibit industrial optimization (Gray, 1987; Cohen and Tubb, 2018).

The structural responses to regulation may nevertheless differ across regulatory instruments. CER relies primarily on mandatory standards and administrative enforcement. By increasing compliance requirements and accelerating the adjustment or exit of inefficient and pollution-intensive activities, CER may alter the relative composition of marine industries (Zhu et al., 2014; Shen et al., 2017). MER, in contrast, operates through economic incentives and price signals. By changing the relative costs of marine resource use, MER may influence the allocation of production factors across marine activities and encourage resources to move toward more efficient uses (Guo and Yuan, 2020; Zhang et al., 2019). Thus, although both regulatory instruments may affect marine industrial structure, the underlying adjustment mechanisms differ.

Numerous studies identify industrial structure upgrading as a key driving factor for GTFP (Zheng et al., 2020; Ren and Ji, 2021; Wei et al., 2021; Wang W. et al., 2025). From a structural perspective, the reallocation of resources toward higher value-added and less pollution-intensive sectors can reduce environmental burdens while improving production efficiency. Zhao et al. (2022) demonstrate that structural transformation can enhance GTFP through pollution mitigation. These findings are reinforced by Yu and Wang (2021), who confirmed that coordinated implementation of environmental regulation and industrial structure upgrading contributes to lower emissions, more efficient resource utilization, and improved allocation efficiency, thereby accelerating green economic development. Building upon this theoretical foundation, this study argues that environmental regulation may influence MGTFP indirectly by reshaping MIS. Accordingly, the following hypothesis is proposed:

H3a MIS serves as a transmission channel through which CER is associated with MGTFP.

H3b MIS serves as a transmission channel through which MER is associated with MGTFP.

2.2.2 Marine technological innovation (MTI)

The Porter hypothesis, one of the most influential theoretical frameworks regarding the effect of environmental regulation, emphasizes that appropriately designed regulatory policies can generate innovation incentives and ultimately enhance firms’ productivity performance and competitive advantage (Porter and van der Linde, 1995). Specifically, the weak version argues that environmental regulation can stimulate technological innovation. In the marine economy, such innovation responses can arise not only within firms but also through the broader regional marine innovation system, in which research institutions play an important role in generating and diffusing marine technological knowledge (Ren and Ji, 2021; Liu et al., 2023). Environmental regulation can increase the demand for cleaner technologies, pollution-control solutions, and more efficient production processes, thereby creating incentives for technological development and application.

The innovation response may nevertheless differ across regulatory instruments. CER relies on mandatory standards and administrative requirements. Stricter pollution-control obligations can increase demand for technologies that help regulated entities satisfy environmental standards, thereby generating a technology-forcing effect. At the same time, compliance expenditure may crowd out resources that could otherwise be allocated to innovation, particularly when regulatory costs become dominant (Ren and Ji, 2021; Wu et al., 2020). The relationship between CER and MTI therefore depends on the balance between technology-forcing effects and compliance-cost pressures.

MER operates through economic incentives and price signals. By increasing the economic cost of inefficient resource use, market-incentive regulation can strengthen incentives to adopt and develop technologies that improve resource-use efficiency and reduce environmental costs. Its greater flexibility also allows regulated entities to choose technological responses that better match their production conditions, which is consistent with the Porter hypothesis regarding the innovation-inducing role of flexible regulatory design (Porter and van der Linde, 1995; Ambec et al., 2013).

Technological innovation is widely recognized as a key driver of sustainable economic development. At the regional level, advances in marine technological knowledge can improve resource-use efficiency, facilitate cleaner production, and strengthen pollution-control capacity (Jiang and Sun, 2025). Technologies generated by marine research institutions may contribute to marine production through technology transfer, collaborative R&D, licensing, and commercialization, thereby linking regional innovation supply with productivity improvement (Liu et al., 2021). With continuous advancement of technology, product quality steadily improves, leading to corresponding increases in labor productivity (Sun, 2022). In this process, inefficient production capacity is gradually eliminated. This process not only enables marine entities to mitigate environmental compliance expenditures but also strengthens their market competitiveness (Liu et al., 2021), further driving industrial green transformation and improving resource allocation efficiency, thereby achieving sustained growth in MGTFP. Building upon this theoretical foundation, the following hypothesis is proposed:

H4a MTI serves as a transmission channel through which CER is associated with MGTFP.

H4b MTI serves as a transmission channel through which MER is associated with MGTFP.

3 Materials and methods

3.1 Variable selection

3.1.1 Dependent variable

• Slacks-Based Measure (SBM) model

The dependent variable is marine sustainable development, proxied by MGTFP and constructed by combining the super-efficiency SBM model with the Global Malmquist-Luenberger (GML) productivity index. Traditional BCC and CCR models are radial and may fail to capture input or output slacks. To address this limitation, Tone (2001) developed the non-radial Slacks-Based Measure (SBM) model, which directly incorporates slack variables and provides a more accurate efficiency assessment. Nevertheless, the traditional SBM model fails to differentiate performance among decision-making units that all achieve an efficiency score of 1. Tone (2002) subsequently introduced the super-efficiency SBM model by integrating super-efficiency principles with the SBM foundation, enabling comparative assessment of multiple efficient decision-making units within the same timeframe. In this study, the super-efficiency SBM model provides the efficiency measures required for constructing the GML productivity index under desirable and undesirable outputs. The SBM efficiency score itself and the final MGTFP index therefore have different interpretations. The model is specified in Equation 1:

y * = min 1 + 1 θ ∑ i = 1 m s i − / x i g t 1 − 1 q + r ∑ j = 1 q s j + ω j g t + ∑ k = 1 r s k − b k g t s . t ∑ t = 1 T ∑ g = 1 , g ≠ 0 N x i g t λ g t X λ − s i − ≤ x i g t ∑ t = 1 T ∑ g = 1 , g ≠ 0 N ω j g t λ g t + s j + ≤ ω j g t ∑ t = 1 T ∑ g = 1 , g ≠ 0 N b k g t λ g t − s k − ≤ b k g t λ , s i − , s j + , s k − ≥ 0 (1)

Wherein, y * represents the optimal efficiency value of the model measure; x denotes input variables; ω represents desired output; b denotes undesired output; i indicates the number of input variables; s i − is the slack variable for inputs; x i g represents the g th DMU’s input variable i; s k − denotes the slack variable for undesired output; b k g represents the g th DMU’s undesired output variable k.

• 2. The Global Malmquist-Luenberger (GML) index is calculated according to Equation 2:

G M L t t + 1 = x t , ω t , b t ; x t + 1 , ω t + 1 , b t + 1 = 1 + D g t x t , ω t , b t 1 + D g t x t + 1 , ω t + 1 , b t + 1 (2)

Wherein, D g t x t , ω t , b t and D g t x t + 1 , ω t + 1 , b t + 1 represent the efficiency scores for t and t + 1. When G M L t t + 1 > 1, it indicates an improvement in the MGTFP of the decision-making unit between time periods t and t + 1; conversely, it indicates a decrease.

Following Oh (2010), the GML index is constructed using a global reference technology and measures the change in green productivity between two adjacent periods. A value greater than one indicates productivity improvement, whereas a value below one indicates productivity decline. Because the annual GML index represents productivity change rather than a directly comparable productivity level, this study sets the MGTFP value in the first sample year 2007, to one and obtains the MGTFP values for subsequent years through cumulative multiplication of the adjacent-period GML indices. This treatment converts the annual productivity-change indices into an intertemporally comparable chained MGTFP level series.

• 3. Input indicators and output indicators

Based on the research of Zheng et al. (2022) and Ren and Ji (2021), this study establishes three input indicators: marine capital stock, marine labor input, and marine energy consumption. The gross ocean product is specified as the desired output, while discharges of industrial wastewater, sulfur dioxide, and solid waste from marine industries are designated as undesired output. This comprehensive set of indicators effectively captures both the resource utilization efficiency and environmental consequences of marine economic activities. Detailed procedures for constructing marine capital stock, labor and energy inputs, desirable output, and undesirable outputs are reported in Supplementary Appendix A.

3.1.2 Independent variables

The core explanatory variables represent two distinct types of environmental regulation, CER and MER. CER is proxied by the intensity of marine pollution control investment, which is measured as the ratio of total marine pollution control expenditure to regional gross ocean product. This proxy captures the observable pollution-control effort associated with compliance with mandatory administrative requirements and is therefore interpreted as an implementation-based measure of CER (Xie et al., 2017; Wang X. et al., 2025). MER is proxied by the sea-use fee per unit area, calculated as the amount of collected sea-use fees divided by the area of marine zones with confirmed rights. Although sea-use fees are collected by the government, they are not merely administrative charges. Instead, they represent a market-oriented governance instrument based on marine resource-right allocation and pricing mechanisms. By increasing the cost of marine resource utilization, sea-use fees encourage users to consider resource scarcity and environmental costs, thereby generating incentive effects. Following Chen and Qian (2020), this indicator reflects the intensity of economic regulation and environmental supervision over marine-use right holders, and captures the application of market-based instruments in marine environmental governance.

3.1.3 Mechanism variables

This study employs MIS and MTI as mechanism variables. Referring to Shao (2020) and Sun et al. (2023), MIS is measured as the ratio of value added in the marine tertiary industry to that in the marine secondary industry, which captures the relative shift of marine economic activities toward tertiary sectors. MTI is represented by the natural logarithm of the total number of invention patents held by regional marine research institutions, which captures regional marine technological knowledge and innovation supply.

3.1.4 Threshold variables

This study selects fiscal capacity (FC) and economic openness (EO) as threshold variables. According to Liu and Chen (2022) and Ren and Ji (2021), FC is measured as the ratio of local fiscal revenue to regional GDP and reflects the fiscal revenue capacity available to local governments, and economic openness is quantified as the proportion of total import-export value to GDP.

3.1.5 Control variables

To better control potential factors influencing MGTFP, this study introduces a set of control variables following established researches (Hong et al., 2023; Ren and Ji, 2021; Du and Li, 2022; Wang et al., 2023), including social consumption level (SC), technology market transaction volume (TMT), marine tertiary industry agglomeration level (TA), and regional informatization level (RIL) (Table 1). To ensure comparability across the 2007–2021 sample period, all monetary variables used in the empirical analysis are expressed in real terms at 2006 constant prices.

TABLE 1. Definition of the variables.

3.2 Empirical model

3.2.1 Benchmark regression model

In the benchmark regression model, MGTFP is employed as the dependent variable, and CER and MER are employed as core independent variables. In the benchmark analysis, CER and MER are examined both separately and jointly across different specifications. The models are first estimated without control variables and are subsequently re-estimated with the unified set of control variables defined in Section 3.1.5.

The standard errors are clustered at the provincial level to account for potential within-province correlation and heteroskedasticity. The model also includes province and year fixed effects to control for unobserved provincial characteristics and common time shocks. To improve the stability of the estimation results, the natural logarithmic transformations are applied to marine technology innovation (MTI) and technology market transaction volume (TMT), which exhibit relatively large numerical scales. The benchmark regression model is specified in Equation 3:

M G T F P i t = β 0 + β 1 C E R i t + β 2 M E R i t + α X + μ i + μ t + ε i t (3)

Wherein, i and t denote province and year, respectively; μ i and μ t represent provincial and time fixed effect; and X represents the set of control variables. The coefficients β 1 and β 2 measure the respective effects of CER and MER on MGTFP.

3.2.2 Mechanism analysis model

To examine the potential transmission mechanisms through which CER and MER are associated with MGTFP, this study considers MIS and MTI as mechanism variables. The mechanism analysis models are specified in Equations 4, 5:

M i t = β 0 + β 1 C E R i t + β 2 M E R i t + α X + μ i + μ t + ε i t (4)

M G T F P i t = β 0 + β 1 C E R i t + β 2 M E R i t + β 3 M i t + α X + μ i + μ t + ε i t (5)

Wherein, M i t denotes the mechanism variables through which CER and MER influence MGTFP, including MIS and MTI. For each mechanism variable, the indirect effects associated with CER and MER are calculated as α 1 β 3 and α 2 β 3, respectively.

Because an indirect effect is the product of two estimated coefficients and its sampling distribution may deviate from normality, this study employs a bootstrap procedure with 5,000 replications to estimate the indirect effects and their confidence intervals. The 95% bootstrap confidence interval is reported. A transmission pathway is considered statistically supported when the corresponding bootstrap confidence interval excludes zero.

3.2.3 Threshold effect model

To examine whether the associations of CER and MER with MGTFP vary across different levels of fiscal capacity and economic openness, this study applies the fixed-effects panel threshold model proposed by Hansen (1999). FC and EO are separately introduced as threshold variables, while CER and MER are examined as threshold-dependent explanatory variables. The panel threshold models are specified in Equations 6, 7.

M G T F P i t = β 0 + M E R + β 1 C E R i t × I W i t ≤ τ 1 + β 2 C E R i t × I τ 1 < W i t ≤ τ 2 + · · · + β k C E R i t × I W i t > τ k + α X + μ i + μ t + ε i t (6)

M G T F P i t = λ 0 + C E R + λ 1 M E R i t × I W i t ≤ τ 1 + λ 2 M E R i t × I τ 1 < W i t ≤ τ 2 + · · · + λ k M E R i t × I W i t > τ k + α X + μ i + μ t + ε i t (7)

Wherein, I · denotes the indicator function; τ 1, τ 2, …, τ k are threshold values; W i t is the threshold variable. The threshold values are estimated by minimizing the residual sum of squares over the admissible range of the threshold variable. Following Hansen (1999), threshold effects are tested sequentially. The single-threshold specification is examined first. If a statistically significant threshold is identified, an additional threshold is subsequently tested to determine whether a double-threshold specification is supported (Hansen, 1999; Wang, 2015). Province fixed effects are controlled in the threshold estimation, and year fixed effects are included to account for common temporal shocks. Because the threshold parameter is not identified under the null hypothesis of no threshold effect, the conventional asymptotic distribution of the test statistic is not applicable. Following Hansen (1999), this study therefore employs bootstrap inference with 1,000 replications to obtain the p-values for the threshold-effect tests. The 95% confidence intervals of the estimated threshold values are constructed based on the likelihood-ratio statistic, following the threshold inference procedure of Hansen (1999), Hansen (2000). The estimated threshold values and their corresponding 95% confidence intervals are reported.

3.3 Data sources

Considering the availability and comparability of data, this study selects 11 coastal provinces and municipalities in China and covers the period from 2007 to 2021. The data are mainly collected from official statistical publications, including the China Marine Economic Statistics Yearbook, China Marine Economic Statistics Bulletin, and China Statistical Yearbook. It should be noted that after 2018, the China Marine Economic Statistics Yearbook discontinued reporting marine-related employment. Since marine labor input is directly incorporated into the construction of MGTFP, missing observations may influence the measurement of the dependent variable. Therefore, this study adopts a linear interpolation method to estimate missing employment values based on the observed trends before and after the missing period. Linear interpolation is suitable for short-term missing observations in annual panel data because it preserves the continuity of temporal changes while avoiding excessive assumptions about future trends. To further examine whether the interpolation method affects the empirical results, this study conducts additional robustness checks using alternative approaches to reconstruct marine employment, including ARIMA forecasting and an estimation method based on the ratio of marine output to regional GDP combined with regional employment.

Before conducting the regression analysis, we report the descriptive statistics and correlation diagnostics of the main variables. The results are presented in Supplementary Appendix B. The descriptive statistics show the basic distribution characteristics of the variables, while the correlation matrix and variance inflation factor (VIF) tests indicate that there is no severe multicollinearity problem among the explanatory variables.

4 Results

4.1 Spatiotemporal evolution of MGTFP

Figure 1A illustrates the temporal evolution of China’s MGTFP over the period 2007–2021. As explained in Section 3.1.1, the reported MGTFP is a chained productivity level index with 2007 set to 1, rather than a contemporaneous cross-sectional efficiency score. Values above or below one therefore indicate cumulative productivity performance relative to the 2007 reference level. Overall, MGTFP remained at a relatively low and fluctuating level between 2007 and 2015. After entering a rapid growth phase in 2016, MGTFP reached a peak value of 1.19 in 2018, experienced a slight decline in subsequent years, and rebounded to 1.18 in 2020. These temporal patterns coincided with broader changes in China’s marine economic development and the external economic environment. During 2007–2015, China’s marine economy remained relatively dependent on traditional marine industries and extensive development patterns, while the 2008 global financial crisis represented an important external economic shock during this period (Zhao et al., 2014; Zhu et al., 2021). These factors provide relevant contextual background for the relatively slow improvement in MGTFP. The increase in MGTFP after 2016 also coincided with the implementation of the 13th Five-Year Plan and a growing policy emphasis on high-quality marine development and ecological governance (Cao et al., 2017; Zheng et al., 2020). The subsequent fluctuations after 2018 occurred alongside heightened external economic uncertainty and escalating trade frictions (Zhang et al., 2019; Benguria et al., 2022). The COVID-19 pandemic emerged only toward the end of the sample period and may therefore be more relevant to the fluctuations observed around 2020–2021 than to the decline immediately following 2018.

To conduct a more in-depth comparative analysis of MGTFP across China’s coastal regions, this study illustrates the evolutionary trends of regional MGTFP over the research period, as shown in Figures 1B, 2. Horizontal comparison shows that Guangdong, Shanghai, and Shandong maintained an average MGTFP above 1.1, indicating relatively stronger cumulative marine green productivity performance over the sample period. In contrast, the MGTFP in Hainan, Hebei, and Guangxi was below 1, indicating relatively weaker cumulative productivity performance relative to the reference level. This study further categorizes the results into three levels. In 2008, regions with high average MGTFP were concentrated mainly in the Bohai Rim area, while Hainan and Guangxi exhibited relatively low levels of development. By 2012, the southeastern coastal regions (Zhejiang, Fujian, and Guangdong) showed rapid MGTFP growth, whereas within the Bohai Rim, only Shandong maintained a relatively high level. From 2016 to 2021, the high-level zone covered Guangdong and Shanghai, the medium-level zone expanded to eight provinces and municipalities, and only Hainan remained in the low-level zone. Overall, the spatial distribution of MGTFP shows persistent differences across coastal regions. These patterns may be associated with regional variation in industrial structure, technological capacity, economic development, and environmental governance conditions (Guo et al., 2022; Ren and Ji, 2021). The results therefore provide descriptive evidence of uneven spatial evolution in marine green productivity across China’s coastal regions.

4.2 Benchmark regression results

Table 2 summarizes the benchmark regression estimates for the associations between heterogeneous environmental regulations and MGTFP.

TABLE 2. Benchmark regression results.

*p < 0.01, p < 0.05, *p < 0.1.

After accounting for both province and year fixed effects, the empirical evidence reveals distinct relationships between different regulatory instruments and MGTFP. Specifically, CER is associated with a statistically significant negative relationship with MGTFP, while MER exhibits a statistically significant positive relationship. The negative association of CER with MGTFP may reflect the compliance costs and rigid constraints associated with administrative regulation, which may place additional pressure on marine production activities. In contrast, the positive association between MER and MGTFP may be related to the economic incentives generated by market-oriented regulatory instruments, which can influence production decisions and resource allocation. Importantly, these patterns remain unchanged after the inclusion of the unified set of control variables defined in Section 3.1.5. Both CER and MER retain their respective signs and remain statistically significant, indicating that the opposite associations of the two regulatory instruments with MGTFP are stable across the benchmark specifications. Given that CER and MER are constructed using different indicators and measured on different scales, their raw coefficient magnitudes are not directly comparable. Accordingly, we focus on the direction and statistical significance of the estimated associations rather than comparing their relative magnitude. Overall, the benchmark results provide evidence of heterogeneous associations between different types of environmental regulation and MGTFP, supporting Hypotheses H1b and H2.

4.3 Robustness and endogeneity tests

• To address the potential concern that the baseline results may depend on the specific measurement of MER, this study replaces the original proxy with provincial pollution emission fees, following previous research. Pollution emission fees reflect the use of market-based instruments to reduce pollutant emissions by imposing economic costs on pollution discharge. As shown in column (1) of Table 3, the coefficient of MER remains positive and significant at the 1% level, while CER continues to show a significantly negative effect on MGTFP. This confirms that the main conclusions are not sensitive to alternative measurements of MER.
• The robustness of the results is further examined by adopting an alternative method to measure MGTFP. Given that marine employment is directly included in the MGTFP measurement, this study further examines whether the treatment of missing observations affects the main findings. Specifically, two alternative approaches are adopted to reconstruct marine employment. First, the ARIMA model is employed to forecast missing employment values after 2018. Second, marine employment is estimated using the ratio of marine output to regional GDP combined with regional employment levels. Based on these alternative labor input measures, MGTFP is recalculated and the baseline regression models are re-estimated. The results reported in column (2) and column (3) of Table 3, and remain consistent with the benchmark regression.
• To address the potential endogeneity problem, this study adopts an instrumental-variable approach. Following Hering and Poncet (2014), Shi and Xu (2018), and Yu and Zhang (2022), the instruments for CER and MER are constructed by interacting the local ventilation coefficient with the contemporaneous leave-one-out average of the corresponding regulatory variable in other provinces within the same marine economic region. This construction combines meteorologically driven variation in local pollution-dispersion conditions with changes in the external regional regulatory environment. The ventilation coefficient is determined by natural atmospheric conditions, while the leave-one-out regional average captures regulatory variation external to the focal province, providing a plausible source of exogenous variation for CER and MER.

TABLE 3. Robustness and endogeneity tests.

This study re-estimates the baseline model using two-stage least squares (2SLS), and the results are presented in column (4) of Table 3. The estimated coefficient of CER is −0.028 and is statistically significant at the 1% level, indicating that the negative effect of command-and-control environmental regulation on MGTFP remains after accounting for potential endogeneity. The estimated coefficient of MER is 0.035 and is also significant at the 1% level, suggesting that market-incentive environmental regulation continues to promote MGTFP. The signs and significance of both regulatory variables are consistent with the benchmark results, providing additional support for the stability of the main findings. Regarding instrument strength, the Kleibergen-Paap rk LM statistic is 36.338 and is significant at the 1% level, rejecting the null hypothesis of underidentification. The Kleibergen-Paap rk Wald F statistic is 1002.512, indicating that weak identification is unlikely to be a concern. In addition, the Cragg-Donald Wald F statistic of 885.615 substantially exceeds the corresponding Stock-Yogo critical values. Overall, the IV estimates provide additional evidence that the main coefficient patterns are robust to this endogeneity treatment.

4.4 Mechanism analysis

Table 4 reports the bootstrap estimates of the indirect effects through MIS. The indirect effect of CER on MGTFP through MIS is 0.0023, with a percentile 95% bootstrap confidence interval that excludes zero. This result provides statistical evidence consistent with MIS serving as a potential transmission channel between CER and MGTFP. In contrast, the indirect effect of MER through MIS is 0.0005, and its percentile 95% confidence interval includes zero. Therefore, the proposed MIS pathway is not statistically supported for MER. Therefore, H3a is supported by the sample data.

TABLE 4. Bootstrap tests of indirect effects.

Bootstrap confidence intervals are based on 5,000 replications. An indirect effect is considered statistically significant when its 95% bootstrap confidence interval does not include zero.

Table 4 further reports the indirect effects through MTI. Technological innovation may improve resource-use efficiency, facilitate cleaner production, and enhance pollution-control capacity, thereby providing a potential channel linking environmental regulation with MGTFP. The indirect effect of MER on MGTFP through MTI is 0.0072, with a percentile 95% bootstrap confidence interval that excludes zero. This finding provides evidence consistent with MTI serving as a potential transmission channel for MER. By contrast, the indirect effect of CER through MTI is −0.0012, and the corresponding confidence interval includes zero, providing no statistical support for this pathway. Therefore, H4b is supported by the sample data.

Taken together, the bootstrap results indicate that the transmission mechanisms differ across regulatory instruments. The evidence is consistent with an MIS-related pathway for CER and an MTI-related pathway for MER, whereas the corresponding alternative pathways are not statistically supported. These findings suggest instrument-specific transmission patterns rather than a uniform mechanism across different forms of environmental regulation.

4.5 Threshold effect results

This section further employs a panel threshold regression model to examine whether the effects of environmental regulation on MGTFP exhibit nonlinear characteristics under different institutional and economic conditions. Fiscal capacity and economic openness are introduced as threshold variables, and the significance of the threshold effects is tested through a bootstrap procedure based on 1000 replications.

Table 5 reports the threshold-effect tests for fiscal capacity and economic openness. For fiscal capacity, MER exhibits a double-threshold effect at 0.1492 and 0.1641, while no significant threshold is identified for CER. One possible explanation lies in the relatively rigid institutional characteristics of CER. This type of regulation relies mainly on mandatory standards, administrative supervision, and compliance requirements. Once these obligations are established, their core requirements are less likely to adjust systematically with local fiscal capacity. Local governments also have relatively limited discretion to alter mandatory regulatory requirements solely in response to fiscal conditions (Kostka, 2016; Zhu et al., 2022). For economic openness, MER exhibits a single threshold at 1.0775, whereas CER shows two thresholds at 0.6070 and 1.0225. Overall, the results indicate that the relationship between environmental regulation and MGTFP varies with fiscal capacity and economic openness, with distinct threshold patterns across regulatory instruments.

TABLE 5. Threshold effect test results.

For statistically insignificant threshold tests, the reported threshold estimates and confidence intervals are provided for completeness and are not interpreted as economically meaningful regime boundaries.

Table 6 reports the regime-specific estimates for MER across different levels of fiscal capacity. When FC is at or below 0.1492, the coefficient of MER is 0.020 and statistically insignificant. When FC lies between 0.1492 and 0.1641, the coefficient increases to 0.084 and is significant at the 1% level. When FC exceeds 0.1641, the coefficient remains positive at 0.022 but declines in magnitude and is significant only at the 10% level. These results indicate that the positive association between MER and MGTFP is strongest within the intermediate fiscal-capacity regime. A possible explanation is that sufficient fiscal capacity provides stronger public governance conditions for market-based environmental regulation, including environmental administration, monitoring, information provision, and related public services. However, further increases in fiscal capacity do not necessarily translate proportionately into stronger regulatory outcomes.

TABLE 6. Threshold effect regression results of FC.

Table 7 reports the regime-specific estimates across different levels of economic openness. For MER, the coefficient is positive when EO is at or below 1.0775 but turns negative once EO exceeds this threshold, indicating a change in the direction of the MER–MGTFP relationship. For CER, the coefficient is significantly negative at low levels of openness, significantly positive within the intermediate regime, and statistically insignificant at high levels of openness. These results suggest that CER is most favorable to MGTFP under moderate economic openness, possibly because such conditions provide better competitive and technological support for firms to respond to mandatory environmental requirements (Qiu et al., 2021; Luo et al., 2021). When openness becomes very high, this positive association is no longer statistically detectable.

TABLE 7. Threshold effect regression results of EO.

Taken together, the threshold results reveal clear instrument-specific heterogeneity. Fiscal capacity significantly conditions the relationship between MER and MGTFP but does not generate a statistically detectable threshold for CER. Economic openness, by contrast, exhibits significant threshold effects for both regulatory instruments. These findings suggest that the effectiveness of environmental regulation varies not only across regulatory instruments but also across the fiscal and external economic conditions in which they operate.

4.6 Further analysis

4.6.1 Spatial heterogeneity analysis

To conduct a systematic examination of potential spatial heterogeneity, this study classifies China’s 11 coastal provinces and municipalities into three major marine economic zones. Following the National Marine Economic Development Plan for the 12th Five-Year Plan Period issued by the State Council, this study divides the coastal provinces into the northern, eastern, and southern marine economic circles. The northern marine economic circle comprises Liaoning, Hebei, Tianjin, and Shandong; the eastern marine economic circle comprises Jiangsu, Shanghai, and Zhejiang; and the southern marine economic circle comprises Fujian, Guangdong, Guangxi, and Hainan. This classification follows the regional division of China’s marine economic development planning and has also been adopted in related studies on regional marine economic development (Guo et al., 2022; Su et al., 2021).

Table 8 presents the regression results for different marine economic zones. For CER, the estimated coefficient is negative but statistically insignificant in the southern marine economic circle, while it is significantly negative in the northern and eastern marine economic circles. For MER, the estimated coefficient is positive and statistically significant in the southern and eastern marine economic circles, whereas the coefficient is positive but statistically insignificant in the northern marine economic circle.

TABLE 8. Spatial heterogeneity regression results.

Pairwise Chow tests examine whether the coefficients of CER, and MER, differ significantly across marine economic circles. The null hypothesis is that the coefficients are equal between two groups.

To further examine whether the observed differences in coefficients across regions are statistically significant, this study conducts pairwise Chow tests for coefficient equality. The results are reported at the bottom of Table 8. The tests examine whether the coefficients of CER and MER differ significantly between different marine economic circles. The results show that the coefficient differences of CER between the southern and northern marine economic circles and between the northern and eastern marine economic circles are statistically significant at the 10% level, while the coefficient difference of CER between the southern and eastern marine economic circles is not significant. In contrast, the coefficient differences of MER across all three regional comparisons are statistically insignificant. Therefore, the regional differences in CER effects receive stronger statistical support, whereas the heterogeneity of MER effects should be interpreted more cautiously.

The heterogeneous effects of CER may be related to differences in regional industrial structures, marine economic development patterns, and environmental governance conditions. In regions with greater dependence on resource- and pollution-intensive industries, administrative regulatory measures may exert a more pronounced influence on pollution-intensive activities and green productivity transformation (Zhou et al., 2022). However, in regions where economic activities are more diversified, the effectiveness of command-based regulation may depend on the compatibility between regulatory requirements and local industrial characteristics. Given the relatively small regional subsamples, the regional heterogeneity results should be interpreted as suggestive evidence of spatial variation in regulatory effects rather than as definitive structural differences across regions.

4.6.2 Temporal heterogeneity analysis

In 2012, China incorporated ecological civilization construction into the national development strategy, marking an important transition in China’s development orientation and environmental governance framework. The selection of 2012 as the breakpoint is based on several considerations. In 2012, the 18th National Congress of the Communist Party of China incorporated ecological civilization construction into the overall framework of the “Five-Sphere Integrated Plan” for national development and proposed the strategic goal of building China into a maritime power (Hu, 2012). Following this transition, marine ecological governance mechanisms were gradually strengthened and institutionalized (Jiang et al., 2020; Lin et al., 2022). However, the year 2012 is not treated as an exogenous policy shock in this study. Instead, it is regarded as a stage boundary reflecting broader changes in China’s institutional environment, marine governance system, and economic development model. To examine whether the effectiveness of environmental regulation varies across policy phases, this study divides the research period into two stages: a pre-strategy phase (2007–2012) and a post-strategy phase (2013–2021).

The regression results reported in Table 9 reveal differences in the effects of environmental regulation across the two periods. During the pre-2012 period, neither CER nor MER showed statistically significant effects on MGTFP, suggesting that the role of environmental regulation in promoting green productivity was relatively limited at this stage. During the post-2012 period, CER and MER exhibit coefficient patterns consistent with the benchmark results, with CER showing a negative effect and MER showing a positive effect on MGTFP. To further examine whether these differences are statistically significant, this study conducts a Chow-type coefficient stability test. The Chow-type coefficient stability test yields a p-value of 0.536 for CER and 0.032 for MER. Therefore, no significant temporal change is identified for CER, whereas MER differs significantly across the two periods.

TABLE 9. Temporal heterogeneity regression results.

These findings indicate that temporal heterogeneity mainly exists in the effect of market-incentive environmental regulation. The strengthened role of MER after 2012 may be associated with the gradual improvement of environmental governance institutions and the increasing integration of ecological objectives into economic development policies. Nevertheless, given that the post-2012 period also coincided with China’s transition toward a new economic development stage, the observed changes may reflect the changing role of environmental regulation under a broader process of institutional and economic transformation.

4.6.3 Lagged effect analysis

Given the time lag inherent in policy transmission, a temporal gap often exists between the implementation of environmental regulations and subsequent behavioral adjustments and technological upgrading by marine production entities. To capture these delayed effects, this study employs one-period and two-period lag treatments for CER and MER.

The regression results reported in Table 10 reveal notable differences in the dynamic impacts of CER and MER. For CER, the coefficients of both the one-period and two-period lags are statistically insignificant, indicating that its delayed effect on MGTFP is not robustly supported. In contrast, MER demonstrates both more immediate and more persistent effects. The coefficients of M E R t − 1 and M E R t − 2 are both positive and statistically significant, suggesting that the positive effect of MER on MGTFP persists over time. This may reflect the relatively continuous incentive signals generated by market-based regulatory instruments. Overall, the lagged results provide stronger evidence for the temporal persistence of MER, whereas no significant delayed effect is observed for CER.

TABLE 10. Lagged effect regression results.

5 Conclusion, policy implications and discussion

5.1 Conclusion

Using panel data for 11 coastal provinces and municipalities in China from 2007 to 2021, this study examines the heterogeneous relationships between environmental regulation and marine sustainable development, as measured by MGTFP. The key conclusions are summarized below. (1) Overall, CER is significantly negatively associated with MGTFP, while MER significantly enhances MGTFP, indicating clear differences in the effectiveness of regulatory instruments. This contrast suggests that the productivity consequences of environmental regulation depend critically on regulatory design, which is consistent with the central preposition of the Porter hypothesis. (2) The transmission patterns are instrument-specific. Bootstrap evidence supports an MIS-related pathway for CER and an MTI-related pathway for MER. The MER–MIS and CER–MTI pathways are not statistically supported. The positive innovation response to MER provides further support for the innovation promotion mechanism emphasized by the weak version of the Porter hypothesis. (3) Fiscal capacity exhibits a double-threshold effect for MER but no significant threshold for CER, while economic openness significantly conditions both regulatory instruments. MER is more favorable at lower openness, whereas CER performs best at an intermediate level of openness. (4) Spatial differences are more strongly supported for CER than for MER. Temporally, the MER–MGTFP relationship strengthens significantly after 2012 and shows persistent positive lagged associations, while no significant temporal change or delayed association is identified for CER.

Overall, CER and MER differ in their productivity associations, transmission pathways, contextual dependence, and temporal persistence. These findings highlight the importance of considering regulatory design and implementation conditions when evaluating heterogeneous marine environmental regulation.

5.2 Policy implications

Building on the above findings, the following recommendations are proposed to strengthen the contribution of environmental regulation to marine sustainable development.

• Improving the coordination of different environmental regulatory instruments. The empirical results show that CER and MER exhibit different associations with MGTFP, suggesting that marine environmental governance may benefit from greater attention to the characteristics of different regulatory instruments. The use of CER and MER could therefore be coordinated according to their respective regulatory mechanisms and observed effects, while avoiding reliance on a uniform regulatory approach. In particular, the design and implementation of CER may place greater emphasis on regulatory consistency and appropriate implementation flexibility, whereas the operation of MER may focus on improving the functioning of market-based environmental governance mechanisms.
• Taking account of the different transmission pathways of environmental regulation. The mechanism analysis provides evidence consistent with an MIS-related pathway for CER and an MTI-related pathway for MER. This suggests that the design of environmental regulation may consider its potential implications for marine industrial structure and technological innovation. For CER, attention may be given to its relationship with marine industrial restructuring when regulatory measures are formulated. For MER, policy design may place greater emphasis on maintaining incentive conditions that are conducive to marine technological innovation.
• Considering regional and temporal differences in regulatory effects. The spatial analysis indicates that the estimated effects of environmental regulation vary across the southern, northern, and eastern marine economic circles, although the statistical evidence for cross-regional differences is stronger for CER than for MER. Accordingly, regional differences may be considered when environmental regulations are implemented, while avoiding a uniform interpretation of subgroup estimates. The temporal analysis further indicates that the association between MER and MGTFP strengthened after 2012, whereas the corresponding change for CER is not statistically significant. These findings suggest that changes in the broader institutional and economic environment may also be relevant when evaluating regulatory effectiveness.
• Incorporating fiscal capacity and economic openness into the implementation of environmental regulation. The threshold results indicate that the association between MER and MGTFP varies across different levels of fiscal capacity, while no statistically significant FC threshold is identified for CER. This suggests that local fiscal conditions may be considered when designing the supporting governance arrangements for MER. For CER, the absence of a significant FC threshold implies that attention may be directed more toward regulatory design and implementation arrangements than toward fiscal conditions alone. Economic openness also exhibits threshold effects for both CER and MER, suggesting that the external economic environment may be considered when coordinating different environmental regulatory instruments.

5.3 Discussion

The findings highlight the importance of regulatory design in explaining the relationship between environmental regulation and marine green productivity. The negative association of CER with MGTFP is consistent with the compliance-cost argument, whereas the positive association of MER is more consistent with the Porter hypothesis, particularly its emphasis on flexible and incentive-based regulation. These results complement existing marine studies by showing that different regulatory instruments may generate distinct productivity associations (Ren and Ji, 2021; Liu and Chen, 2022; Sun et al., 2023). Second, previous studies have linked marine environmental regulation to industrial restructuring (Chen and Qian, 2020) and marine technological innovation to high-quality marine development (Liu et al., 2021). Our results further distinguish these channels across regulatory instruments, with MIS supported for CER and MTI supported for MER. Third, existing studies show that regulatory outcomes may vary with external conditions such as openness and regional characteristics (Qiu et al., 2021; Luo et al., 2021). Our results extend this evidence by showing that fiscal capacity mainly conditions MER, while economic openness conditions both instruments, with MER also exhibiting stronger temporal persistence.

5.4 Limitations and future research

First, the availability of marine economic data remains limited, particularly at the prefecture-city level, and marine employment data after 2018 require reconstruction. Future research could use more complete and disaggregated datasets to improve measurement accuracy. Second, the sample contains only 11 coastal provincial-level regions, and the regional subsamples are relatively small, which limits the number of clusters and the precision of regional inference. The China-specific institutional setting may also limit the generalizability of the findings. Future studies could use finer-level data and comparative evidence from other marine economies. Third, the proxies for CER and MER capture only specific dimensions of environmental regulation. Although alternative measures and IV estimation are employed, potential reverse causality and omitted-variable concerns cannot be completely eliminated. Future research could construct more comprehensive regulatory indicators and employ stronger quasi-experimental designs. Fourth, MIS is measured as the ratio of marine tertiary to secondary value added, which is a relatively appropriate proxy under current data availability but mainly captures relative changes in industrial composition. MTI, measured by invention patents held by marine research institutions, primarily reflects regional innovation supply and may not fully capture enterprise-level innovation or technology transfer. Future research could employ weighted or composite industrial-structure indicators incorporating advancement and rationalization, together with firm-level marine patent and technology-transfer data, to improve the measurement of these mechanisms.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenvs.2026.1953310/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Juan Antonio Morente-Molinera, University of Granada, Spain

Reviewed by: Wei Wang, Guangdong Ocean University, China

Abdelhak Lefilef, Abdelhafid Boussouf University, Algeria