Heterogeneous environmental regulation tools and green economy development: evidence from China

The implementation of environmental regulation policy by the government is usually an effective tool to reduce local pollution emissions. The super-efficient SBM model of unexpected output was used in this study to estimate green economy development levels in 30 Chinese provinces from 2010 to 2019 and constructed a panel econometric model. It empirically tested the theoretical hypothesis and mechanism of heterogeneous environmental regulation affecting green economy development by using the least squares estimation, the instrumental variable method, spatial panel regression, the mediating effect model, and further models. The relationship between command-controlled environmental regulation and green economy development was U-shaped, whereas green economy development was not significantly influenced by market-incentive environmental regulations. Command-controlled environmental regulation affected green economy development via the progress of pollution control technology and production technology. The strengthening of command-controlled environmental regulation progressed pollution control technology as regional enterprises continually improved, while production technology was initially suppressed, then promoted. Market-incentive environmental regulation mainly affected the green economy development level through pollution control technology progress, while the mediating effect of production technology progress was insignificant. This study provides some empirical support for the verification of Porter’s win-win hypothesis and the realization of green economic transformation in emerging countries such as China.


Introduction
China has experienced unprecedented rapid economic growth in recent decades through extensive development modes featuring high energy consumption; this has caused high environmental pollution, high emissions, resource waste, and low production efficiency. Resources and the environment in China are reaching a critical point in terms of their carrying capacity. Ecological-and environmental problems frequentl y occur which threaten people's livelihoods and cause social instability (Cui et al 2019. Therefore, China's green development transformation should increase total factor productivity and decrease the environmental damage caused by economic activities. Its essence lies in optimizing environmentally adjusted multifactor productivity (Rodríguez et al 2018). Tolliver et al (2021) argued that green innovation and green finance are two key components of sustainable development. And the extent to which they can promote environmentally adjusted multifactor productivity growth, green patent registrations, and other indicators are important factors for sustainable growth. Rodríguez (2018) also extended the analytical framework for measuring multifactor productivity to account for environmental services. Existing literature shows that environmental regulation is the most targeted and powerful tool to control environmental pollution. It mainly regulates economic activities directly or indirectly through formulating corresponding pollutant emission standards by the government (Hancevic 2016. Green economy development has become a hot academic topic by maximizing environmental regulation tools to achieve a 'win-win' outcome. Whether environmental regulation can ensure a win-win outcome of controlling environmental pollution and achieving economic development, as well as the specific implementation mechanisms of different environmental regulations are the issues we need to discuss in this paper. Environmental resources are non-excludable, competitive public resources. The private sector has no incentive to improve the ecological environment without regulation. Therefore, implementing government control is necessary to achieve Pareto's optimal solution. The classical government intervention theory holds that the government has complete market information and can internalize the pollution externality cost and control the pollution emission level of enterprises through environmental regulation (Hye-Young Joo, Min (2018)). Cameron Hepburn (2010) argued that leaving environmental protection to the free market, relying on notions of corporate social responsibility and altruistic consumer and shareholder preferences, would not deliver optimal results. However, according to some scholars, environmental regulations raise the costs of production for businesses and discourage investments in technological innovation, thereby limiting the development of the economy.Zhang et al (2021) adopted China's 2000-2017 construction industry as a case study, pointing out that different types of environmental regulations have differing effects on green technology innovation efficiency, including hindering investment in technology innovation. 'Porter's win-win hypothesis' suggests that strict environmental regulation tools implemented by the governments can force companies to make technological innovations, thus offsetting the environmental regulation costs incurred for pollution (Porter and der Linde 1995). Various literature has confirmed the existence of 'Porter's win-win hypothesis' (Guo et al 2017). For example, Mohr (2002) modeled external economies of scale and discrete technological changes using a general equilibrium model, while strict environmental regulations improve industrial enterprise productivity (Boyd andPang 2000, Manello 2017). In research by Johnstone et al (2010), the intensity of environmental regulation was positively correlated with the number of enterprise patent applications. Environmental regulation promotes industrial innovation in China and attracts more direct foreign investment and drives economic growth (Fahad et al 2022). Policy measures for preventing and controlling air pollution promulgated by the Chinese government in 2012 greatly improved the innovation performance of enterprises (Mbanyele and Wang 2022).
Meanwhile, the 'cost of compliance theory' of neoclassical economic theory believes that the government cannot know exactly what each manufacturer's external costs are, and unified environmental regulation tools often violate the marginal production principle of manufacturers which reduces production efficiency (Palmer et al 1995). An empirical study of American industrial enterprises found that environmental regulations reduce total factor productivity by 4.6% for every 1% increase in intensity (Gray and Shadbegian 2003). Ambec et al (2013) believe that pollution control can force enterprises to innovate and adopt green technology for production; however, inefficient production resources and environmental protection equipment lead to a decline in productivity, which will bring an adverse impact on economic development.
Can environmental regulation tools ensure 'win-win' outcomes in terms of economic development and environmental protection? Ultimately, the realization of the 'win-win' is based on the extent of the 'innovation compensation' benefit, that is, what role environmental regulations can play in promoting technological progress and offsetting compliance costs (Wang 2020). A comprehensive comparison of the positive and negative effects of environmental regulation and technological innovation reveals a U-shaped relationship (Ouyang and Du 2020). Some scholars believe that there is a time difference between the positive-and negative effects of environmental regulation tools on technological progress. The negative effects tend to occur in a short time, while technological innovation requires a long accumulation period which leads to a long lag period for the positive effects of technological innovation, resulting in a U-shaped relationship (Lanoie et al 2008, Du et al 2021, Li et al 2022. There is a rich amount of existing literature involving environmental regulation to facilitate the 'win-win' pattern of green transformation and economic development. However, we believe that the existing literature has at least two deficiencies: first, environmental regulation is roughly classified into command-controlled environmental regulation and market-incentive environmental regulation. The former mainly order enterprises to meet environmental quality objectives through mandatory administrative measures and emission standards of the government. The latter brings externalities into the production costs of enterprises, and uses a market regulation mechanism and the environment to promote protection and utilization of resources, mainly including emissions trading, sewage charges (taxes), and other tools. The two environmental regulations have different policy means and implementation priorities, thus forming different impacts on enterprise technology production , Shen et al 2019, Sun et al 2021. The economic principles behind the two environmental regulation tools are different, so the mechanism of action for green economy development may be heterogeneous. However, most literature only considers the effects of command-controlled environmental regulation on green economy development, or simply confuses the two regulatory tools. Second, most of the existing literature regards technological progress as a mechanism for regulating the environment affecting green economy development based on Porter's hypothesis. However, technological progress also includes production-and pollution control. Confusing the two kinds of technological progress will produce defects in its theoretical basis which is not conducive to clarifying the internal mechanism of environmental regulation affecting green economy development.
The marginal contribution of this study can be summarized as follows: first, the mathematical model of the technical progress under the enterprise behavior decomposition for pollution treatment technology and production technology progress was taken in the same framework to interpret the influence of environmental regulation 'win-win' theory to realize economic growth and environmental protection, respectively. Secondly, as a way to uncover the mechanism of influence of rules on green economic development, an analysis of the impact of different environmental regulations was conducted. This study provided support for the verification of Porter's hypothesis in China and provided an empirical reference for China and other emerging green economies' transformation and sustainable development.
The following is the structure of the study: the second section is the theoretical analysis and research hypothesis; the third section is the methods and data, including the construction of empirical models and selection of variables; the fourth section is the empirical analysis used to evaluate environmental regulation's impact on green economy development, and the fifth section is the conclusion and policy recommendations.

Theoretical model
This study drew on the benchmark model ideas of Copeland and Taylor (2004) and , and integrated pollution control technology and production technology into the analysis framework. To clarify the theoretical mechanisms of how environmental regulation affects the development of green economies, this study research two types of technological progress and their impacts on environmental regulation. First, it was assumed that the production function of a single enterprise with a constant return to scale is expressed as Y A K L F K L , , .
r r p p = ( ) ( ) The total factor productivity A is assumed to be Hicks neutral and determined by technological capital input K r and labor input L . r The output F at a given technology level is determined by capital inputs K p and L . p Y is the total output per unit amount, and output will be treated as a by-product. The government corrects the externalities of pollutants through environmental regulations; that is, the government sets the maximum pollution discharge of enterprises as G. Therefore, the enterprise must use part of the total output for pollution control: E eY , = e 0 1. < < It was assumed that the enterprise pollution emissions for D Y E , , ( ) and D Y E , ( )partially meet: This means that the stricter the regulation of the environment, the lower the G and D. D E / ¶ ¶ denotes the pollution control input coefficient of pollutant discharge per unit while D Y / ¶ ¶ denotes the pollutant Faced with environmental regulation G, enterprises generally adopt two ways to decrease pollution emissions: improve the level of pollution control technology and increase investment in pollution control technology, or improve production technology, and promote production efficiency to make up for pollution control costs. Therefore, it can be assumed that the technical level T A E T T , , where T A stands for production technology progress, T E represents the progress of pollution control technology, and T A E , We also assumed that T E is determined by the pollution control expenditure e, and T E E ( ) is a diminishing marginal return: If a typical enterprise faces a completely competitive market environment and the final product price is P, the prices of K , r K , p L r and L p are r , r r , p w r and w , p respectively. The optimal production decision of an enterprise is described by the following equation: The Lagrange multiplier method was used to solve the optimal solution of equation (1) to generate the following equation: where l is the Lagrange operator. According to equation (3), According to equation (4), equation (2) can be rewritten as: can be expressed as: . Therefore, the pollutant emission of enterprises D is negatively correlated with the proportion of pollution control investment e. Consequently higher environmental regulation G increases the proportion of pollution control investment e.
It can be deduced from equation ( This means that the pollutant emissions of enterprises continue to decrease with improved environmental regulation, while the pollution control technology of enterprises improves, the production technology declines, and the overall technical level declines.
cannot be determined at this time.
However, when e 1, With improved regulation, enterprises will be able to cut their pollutant emissions even further, pollution control technology will be improved, production technology will show an upward trend, and the overall technical level will improve. In summary, increasing the intensity of regulations results in environmental protection a gradual decrease in enterprise pollutant emissions, and a gradual improvement in pollution control technology. Production technology and total technology levels show a U-shaped trend: first decreasing, then increasing. Green economic development needs to meet economic growth and pollution reduction, and economic growth is usually determined by productivity factors or production technology levels. Therefore, a U-shaped relationship is also believed to exist between environmental regulation intensity and green economic development.

Effects of different environmental regulations on green economy development
At present, China's environmental regulation tools are mainly command-controlled and supplemented by market-incentives (Yu and Li 2020). The government plays a decisive role in environmental governance.
(1) Command-controlled environmental regulation This requires limited emission of waste gas, wastewater, solid waste, and other ancillary products produced by enterprises. It mainly requires pollutant discharge enterprises to take actions to meet environmental objectives according to the regulations or orders issued by government departments with the management department issuing rewards or sanctions according to whether the pollutant discharge department meets the standards. Governments and societies can easily supervise and control this policy since it is simple and direct (Blackman et al 2018, Tang et al 2020, Zhang et al 2018. The environmental regulation variable G set in the above theoretical model is very similar to the imperative environmental regulation. The mandatory pollutant emission standards set by command-controlled environmental regulation do not consider the differences in pollution control technologies of heterogeneous enterprises. Therefore, some enterprises may bear higher compliance costs in the short term, resulting in lower production efficiency . Despite this, the widespread application of green production technology makes enterprises improve production quality and promote green economy development in the region in the long term. This information, together with the theoretical model analysis allows this study to propose the following hypotheses: Hypothesis 1. there is a U-shaped curve between command-controlled environmental regulation and green economy development.
Hypothesis 2. command-controlled environmental regulation affects green economy development through pollution control technology and production technology progress.
(2) Market-incentive environmental regulation This differs from command-controlled environmental regulation which is mainly achieved through emissions trading, collecting emissions fees, and so on. Emission trading is a policy tool with the connotation of 'establishing a market' which uses the idea of the Coase Theorem to correct the externality of pollutant emissions. Collecting sewage charges is a policy tool with the connotation of 'using the market' based on Pigouvian tax theory so that enterprises can consider the constraints of sewage costs in production decisions and correct the externality of pollutant emissions. However, emissions trading is based on the total amount of pollutant control which is related to the regional environmental capacity. Therefore, the total pollutants determined at this stage are a target of the total optimal amount of emissions. This final total amount of pollutants is likely to exceed the environmental capacity and cause environmental damage (Shin 2013). China's emissions trading market is usually traded in pattern unit price bidding (the British auction), with the bidder's bid considered without regard to the order of the required quantity of sewage's discharge right. This leads to the buyers often not caring about the price of small demands, which results in abnormal emissions prices and reduces enterprise production efficiency (Wang and Zhang 2022).
At present, market-incentive environmental regulation tools in China are still in the trial stage and face challenges such as the small scope of recipients, weak awareness of the environmental responsibility by enterprises, and inconsistent performance goals of local governments. Currently, China's market-oriented environmental regulation is unsatisfactory (Zhao et al 2016). Thus, the following hypothesis is proposed in this study: Hypothesis 3: The development of China's current market-incentive environmental regulation tools is imperfect and has no significant impact on green economy development.

Methods and data
3.1. Model setting We constructed a panel data model to test the relation between heterogeneous environmental regulation and green economic efficiency. First, the primary-and secondary terms of environmental regulation were incorporated into the regression model to determine whether a U-shaped relationship existed between them. The model was set as follows: where, GEE it is the green economic efficiency of region i in t year, ER it represents the environmental regulation of region i in t year, X it is the control variable, i m and t l are fixed effects of region i and fixed effect of time t, respectively, and it e is the residual term.
Secondly, a mediating effect model was constructed to verify whether a heterogeneous environment impacts green economy efficiency through technological progress. This was verified based on three stepwise panel regressions: where T it is the mediating variable, and the testing steps of the mediating effect are as follows: step 1, if the regression coefficient 1 b in equation (8) is significant and positive, environmental regulation positively impacts green economic efficiency; step 2, if the regression coefficient 1 a of equation (9) is significant, environmental regulation affects the mediating variable; step 3, if 1 c and 2 c are significant, and the absolute value of 1 c is greater than that of , 1 b a partial mediation effect exists. If 1 c is insignificant, but 2 c is significant, there is a complete mediation effect.

Variable selection (1) Explained variable
Green economic efficiency (GEE) refers to the economic efficiency generated under green development constraints. The traditional economic efficiency evaluation system is modified to incorporate the consumption of energy, and the environmental pollution index is taken as the undesired output. The greater the efficiency value of the green economy, the higher the green economy efficiency. In order to solve the problem that the input slack variable is not considered and the output value is reduced, the super-efficiency SBM model of unexpected output was used to measure the green economic efficiency of each province (Tone 2001) as follows:  Where l is the weight vector; the larger the value of an objective function , * d the more efficient the decisionmaking unit. Each province is equivalent to a production decision-making unit, with a total of j (1, 2KKN) decision-making units. Each decision-making unit needs to input m (1, 2KKN) elements x m (including goodand bad input), and obtain S 1 (1, 2KKN) expected outputs y g and s 2 (1, 2KKN) unexpected outputs y . b Table 1 shows the green economic efficiency index system constructed in this paper. The regional green economy development level of 30 provinces in China was calculated from 2010 to 2019, and their temporal and spatial distribution characteristics are depicted in figure 1. The overall level of green economy development in China is not high, but the spatial distribution of green economy development presents obvious characteristics of the Northeast-Southwest boundary (the geographical order is the Hu Huanyong line). The area east of the demarcation line is densely populated, with high economic growth and relatively strong green economic development. The area west of the demarcation line is sparsely populated with low economic development. Furthermore, it occupies China's main consumable energy resources including coal, natural gas, and oil; therefore, the pollution emission problem is relatively serious, and green economy development is relatively low.
Note: the spatial distribution of GEE is shown in 2011, 2015, and 2019, and the classification criteria are the global quartile of the panel sequence of GEE.
(2) Core explanatory variable: environmental regulation Environmental regulation policy (ER) includes command-controlled environmental regulation and market-incentive environmental regulation. Among them, command-controlled environmental regulation (ERC) is usually expressed by indicators such as the standard rate of industrial waste gas emission, the removal rate of industrial sulfur dioxide, the harmless treatment rate of domestic waste, and the removal rate of industrial smoke and dust under the emission standard. Market-incentive environmental regulation (ERM) is expressed Table 1. Green economy efficiency index system.

Element index input
Unexpected output Expected output Labor input Wastewater discharge Actual gross regional product Capital investment Sulfur dioxide emissions Total energy consumption Solid waste discharge Note: labor input is the total number of employed people in each region. Capital input is the total amount of physical capital stock calculated by the perpetual inventory method using 2010 as the base period (Li and Hu 2012). The actual gross regional domestic product is obtained following adjustment with the base period.
by the emission fee and the amount actively invested in pollution control projects under the emission trading system (Guo and Yuan, 2020). The above indicators were standardized, followed by employing the entropy method to measure and calculate the index weight. The comprehensive score of the two environmental regulations was calculated as the proxy variable of heterogeneous environmental regulations. Table 2 shows the measurement indicators of environmental regulation in this paper.
(3) Intermediary variable: technological progress Technological progress can be measured in two ways: production technological progress (PTE) and pollution control technological progress (WTE). The production technology level uses the Malmquist index in data envelopment analysis This method expresses total factor productivity and divides it into three parts: the change rate of technological progress, pure technical efficiency, and scale efficiency. The change rate of technological progress is the index constructed by this paper to calculate the degree of production technological progress. The extent of pollution control technology is generally measured by the number of patent applications related to environmental pollution control; however, there is a current lack of relevant statistical data. Therefore, the amount of pollution control must be greater when the amount of pollution production remains unchanged during high levels of pollution control technology; therefore, it is wise to use the amount of pollution control corresponding to the unit of pollution production to express the level of pollution control technology.
(4) Control variable This paper combines the existing literature (Wang 2020, Jia et al 2022 to set the following control variables: 1. Actual utilization of foreign direct investment (FDI) expressed by the actual utilization of FDI flow data of each region each year; 2. Level of economic development (AGDP). GDP per capita was used to reflect the economic development level for each province. The square term of GDP per capita was set as a control variable according to the classical environmental Kuznets hypothesis (Roberts and Grimes 1997). 3. Education level (EDU): the average number of years of education of the population in each province was adopted. 4. Industrial

Overall indicators Classification index Basic indicators Unit
The intensity of environmental regulation

Command-controlled environmental regulation
The standard rate of industrial wastewater discharge

Empirical analysis 4.1. Benchmark regression results
The air circulation index and regional greening rate were further introduced as instrumental variables of environmental regulation (Hering and Poncet 2014) based on the data from the ERA-interim database (ECMWF 2019) to account for the endogenous problems caused by possible measurement errors and missing variable errors in the model. First, the Davidson MacKinnon endogenous test was performed based on the premise that the instrumental variables are valid; this produced a p-value of 0.004. Therefore, the original assumption that 'all explanatory variables are exogenous variables' can be rejected at a significance level of 1% and the environmental regulation variable coefficient has endogenous bias. Second, the validity of instrumental variables using the over-identification test showed a p-value of 0.702. Accordingly, 'instrumental variables are exogenous' can be accepted as the original assumption. At last, the two-stage least square regression (2SLS) was used to obtain the coefficient estimates of regulatory variables.
The lag period of green economic efficiency was simultaneously included in the explanatory variables to form a dynamic panel regression equation by eliminating the autocorrelation between variables and using the generalized method of moments (GMM) to evaluate parameters (Bond 2002). The p-value of Hansen's J statistic of the GMM estimation of the system was above 10% which accepts the original hypothesis indicating that the selection of instrumental variables of dynamic panel regression was reasonable. Moreover, the AR (1) statistic testing whether the residuals in the difference equation exist in first-order autocorrelation in the estimation was at the 5% level indicating that it is significant. Meanwhile, the AR (2) statistic testing whether the residuals in the difference equation exist in second-order autocorrelation in the System GMM was at the 10% significant level attributing that it was insignificant. Therefore, the model setting of the System GMM was relatively reasonable (table 4).
A significance level of 1% and 5% was found for the first and second terms of command-controlled environmental regulation, respectively, demonstrating that green economy effectiveness was linked to it in a nonlinear U-shaped curve (table 4). Both the two-stage least square method and the generalized moment estimation method showed that command-controlled environmental regulation restrains green economic efficiency before the inflection point value, then promotes it. Enterprise R&D is used for pollution control at the beginning of environmental regulation policy implementation. This squeezes out product R&D investment which hinders improvements to enterprise technological innovation levels and leads to declining production efficiency. However, strengthening mandatory environmental regulation increases investment in pollution control which promotes the progress of pollution control technology, upgrades the regional green industrial structure, and promotes regional green economic efficiency improvement.
Only at a significance level of 10% are the results of the 2SLS-and GMM estimation of the quadratic coefficient of market-motivated environmental regulation significant, suggesting that the U-shaped curve relationship between green economic efficiency and market-incentive environmental regulation is not very strong. The reason for this may be that environmental regulation affects negatively technological innovation due to the low level of regulation.
There is also a positive 'U' curve link between per capita GDP and green economic efficiency, as shown by the primary term coefficient of per capita GDP being negative at the significance level of 1% and the secondary term coefficient being positive at the significance level of 5%. This point verifies the theory of the 'environmental Kuznets curve.' A significance level of 1% indicated positive and significant industrial structure coefficients; therefore, the green economic efficiency showed an upward trend following improvements in industrial structure. There is evidence that industrial improvement contributes to regional green economic efficiency and reduces industrial pollution. In the presence of foreign direct investment, there is a positive and significant coefficient at the significance level of 1%, suggesting that foreign direct investment contributes to the development of green economies. The technology spillover effect brought by foreign investment results in the Note: * , ** and *** are significant at the level of 1%, 5%, and 10%, respectively. t statistics are in parentheses and P values are in square brackets. Models (1) and (4) are OLS regression results, models (2) and (5) are 2SLS estimation results, and models (3) and (6) are GMM estimation results. advancement of enterprise innovation and the overall level of the green economy in the region. It is possible that the insignificant impact of education level on green economy efficiency is due to the long period of regional human capital investment that cannot be converted into a new driver for green economy growth on a short-term basis. Factor endowments have a negative impact on green economy development. This indicated that the rise in the capital-labor ratio in various regions of China was not obvious. Most provinces in China are still dominated by labor-intensive industries and cannot quickly transform into capital-intensive industries.

Robustness test based on the spatial model
The first law of geography holds that economic variables generally have spatial relevance (Kondo 2015). A full consideration of the spatial dependence between cross-sectional units can include more information about regional heterogeneity in the model and enhance the validity of the conclusion. Therefore, this study set the spatial Doberman model as the robustness test method for the above conclusions.
where ER it represents the environmental regulation of region i in year t, X it is the control variable, it w is the spacing coefficient; W it stands for the spatial weight matrix; i m and t l are the regional fixed effect and the time fixed effect, respectively, and it e is the residual term.
This paper as two kinds of spatial weight matrices: the geographical distance spatial weight matrix W , D and an economy-geography nested spatial weight matrix W . C The elements of geographical distance spatial weight matrix W D can be expressed as: where d is the spherical geographical distance between region i and region j. As a general rule, nested spatial weight matrices W C are formed by the combination of two or more single-weight matrices, especially when economic spatial weight matrices are combined with geospatial weight matrices (Drukker et al 2013). W C was calculated as follows: , ..., . 15 Environmental regulation under command had a significant U-shaped relationship with green economic efficiency; however, it was unclear how green economic efficiency and market-incentive environmental regulation were related. The spatial Dobbin model (SDM) was used to quantify the effect of heterogeneous environmental regulation on green economic performance in 30 provinces of China and its spatial effects.
Prior to using spatial panel analysis, the core variable must be autocorrelated on a spatial basis. If there is a spatial correlation of core variables, the spatial panel model can be used for empirical analysis. This study used Moran's I index to measure spatial correlation. The three core variables passed the spatial autocorrelation test at the 10% significance level, and their Moran's I index was positive, indicating that these variables have a significant positive spatial correlation (table 5).
We initially determined whether the samples were generated by the spatial error model (SEM) or spatial lag model (SAR) through the Lagrange multiplier test (LM) and robust Lagrange multiplier test (LM robust) before using the spatial econometric model for analysis. Both LM spatial-and LM-spatial error tests passed the significance level test of 5% (table 6). Secondly, to find out whether the spatial Durbin model (SDM) can be simplified to the spatial error model and spatial lag model, the Wald test and likelihood ratio (LR) test were used.  I is approximately subject to a normal distribution, and a hypothesis test can be carried out. I I a n dI 0, 0, 0, > < = represent spatial positive correlation, spatial negative correlation, and that variable x is spatially randomly distributed, respectively. The spatial weight matrix used here is the geographic distance weight matrix W .

D
Wald-Spatiallag, Wald-Spatiallerror, LR-Spatiallerror, and LR-Spatiallerror all passed the 1% significance test under two different spatial weights which rejected the original assumption that SDM can degenerate into SAR or SEM. At the same time, the Hausman test result of fixed effect and random effect selection was 103.11 (P value < 1%). Therefore, this study chose the SDM model with a fixed effect to fit the samples. At a 5% level of significance, the command-controlled environmental regulation's primary term was considerably negative, and its quadratic coefficient was significantly positive. The empirical findings in table 6 further demonstrated that there was a U-shaped curve link between command-control environmental regulation and green economic efficiency when weighing geographic and economic distance. The connection between market-incentive environmental regulation ERM and economic efficiency is unclear which is comparable with previous research results. There was a positive spatial spillover effect of green economic efficiency indicating that the green economic development of the region promotes the same type of development in the neighboring region according to the estimated spatial lag coefficient of the variable. The regression results of Model (2) show that command-controlled environmental regulation also has positive spillover effects. Regulation of regional environmental pollution and emissions can positively impact green economy development in surrounding areas.
A robustness test showed that the results were in line with the estimation outcome of core variables in the prior empirical results, which indicated that environmental regulation had a reliable and robust impact on green economic efficiency.

Mechanism analysis
According to the above methods, green economic efficiency and environmental regulation are related in a U-shaped curve. The theoretical model indicated that environmental regulation affects green economy efficiency through technological progress. The theoretical mechanism was tested using the mediating effect model of stepwise regression. First, 30 provinces were divided into four types of samples according to the Note: * , ** , and *** are significant at the level of 10%, 5%, and 1%, respectively; t statistics are in round brackets; p-values are in square brackets.
inflection point value of command-controlled environmental regulation (0.042) and the inflection point value of market-incentive environmental regulation (0.059), while the production technology level and pollution control technology level were used as intermediary variables for regression estimation (tables 7 and 8).
A command-controlled environmental regulation below the inflection point value of 0.042 impacts green economic efficiency and production technology progress at the significance level of 5% (−0.098 and −0.094, respectively) (table 7). The inclusion of both technological progress and environmental regulation in the model regression results in coefficients of 0.011 and −0.027, respectively at a significance level of 1%. A commandcontrolled environmental regulation above the inflection point value results in a positive influence coefficient on green economic efficiency and production technology progress which passes the significance level test of 10% and 5%, respectively. This demonstrates how the U-shaped relationship between command-controlled environmental regulation and green economic efficiency with growing environmental regulation first suppresses production technology levels before promoting them. This is in line with the conclusions drawn from the study's theoretical model.
According to model (4) and model (6), when the intensity of market-incentive environmental regulation below the inflection point value, the impact of environmental regulation on green economic efficiency is negative and significant at the level of 1%; However, the improvement of environmental regulation on production technology was insignificant. Green economy efficiency was significantly improved by a marketincentive regulation above the inflection point value; the progress of production technology was still insignificant. Therefore, production technology progress was not taken into account when analyzing the effects of market-incentive regulation on the development of the green economy. This might be the case because, based on sewage charge (tax), the effect of environmental control on supporting enterprise production technological improvement was minimal. Investigating the mediating impact of manufacturing technological advancement from the perspective of market-incentive environmental control was therefore unachievable.
Model (2) showed a significantly positive impact of command-controlled environmental regulation on the progress of pollution control technology whether the intensity of it was before-or after the inflection point, (table 8). This empirica l conclusion was consistent with the derivation result of the theoretical model: the enhancement of command-controlled environmental regulation improved pollution control technology and reduced enterprise pollution emissions. At the same time, a combination of model (1) and model (3) showed that pollution control technology progress played an intermediary effect in the process of command-controlled environmental regulation affecting green economy development. Model (5) showed market-incentive environmental regulation always positively affected pollution control technology progress, yet showed an intermediary effect on pollution control technology progress according to. A combination of model (4) and model (6) showed that market-incentive environmental regulation affected green economy development through pollution control technology, not production technology.

Research conclusion
In this study, environmental regulation was empirically examined in relation to green economy development using the theoretical context of 'heterogeneous environmental regulation-technological innovation-green economy development.' A theoretical model was initially established to deduce the theoretical effects of environmental regulation on the progress of pollution abatement technology and production technology, discuss the mechanism of government regulation on green economy development, and put forward theoretical assumptions. Secondly, the green economy development level was measured by the super-efficiency SBM of unexpected output and other methods at the provincial level. This was combined with sample data and the theoretical analysis to create a panel regression model to quantitatively test the connection between heterogeneous environmental regulation and green economy development. The mechanism of heterogeneous environmental regulation on the growth of the green economy was finally tested using the intermediary effect model. The main conclusions are stated below: First, OLS, 2SLS, and GMM estimation results of the benchmark model all showed that green economy development is impacted differently by heterogeneous environmental regulations, and the relationship between environmental regulations controlled by command and the efficiency of the green economy followed a U-shaped curve. Command-controlled environmental regulation below the inflection point resulted in environmental regulation inhibiting the growth of green economic efficiency. The development of green economic efficiency was facilitated by command-controlled environmental regulation above the inflection point. Market-incentive environmental regulation did not significantly affect green economy development. Second, empirical results using spatial panel regression as the robustness test supported the benchmark regression results, and command-controlled environmental regulation had a positive spillover effect. Regulation of regional environmental pollution and emissions can positively impact the green economic development of surrounding areas. Third, the intermediary effect model showed that command-controlled-and marketincentive environmental regulations affect green economy efficiency by improving pollution control technology. Increasing command-controlled levels of regulation before the inflection point inhibited the progress of production technology which affected green economy development. Meanwhile, commandcontrolled environmental regulation intensity after the inflection point enhanced green economy development through the progress of production technology. However, through the advancement of production technology, market-incentive environmental regulations did not significantly impact green economy efficiency. In short, the empirical conclusions of this study were consistent with the assumptions put forward in the theoretical model.

Policy recommendations
The conclusions of this study can be used as a reference to promote China's green economic transformation and environmental regulation policy transformation.
First, the government should establish an awareness that environmental protection and economic development complement each other and avoid perpetuating the wrong perception that environmental regulation is incompatible with economic development. For example, local governments have incentives to maintain short-term regional economic growth and cover up enterprises' emission behavior under the development mode of GDP-only local government competition and performance concept . Therefore, local governments should remove the original thinking and firmly establish the concept of green development from a long-term perspective.
Second, in order to accomplish technological advancement, green economic transformation, and sustainable development, stringent application of environmental regulating rules is necessary. Currently, China's environmental regulation tools are largely based on command and control which supports Porter's hypothesis that strict environmental regulation can force enterprises to achieve technological progress and improve economic efficiency. However, market-incentive environmental regulation is a more flexible marketoriented policy tool that has not significantly promoted green economy development. It follows that the construction of China's emission trading market is still in the pilot stage, with problems such as low binding force, small market capacity, and imperfect rules. The emission fee system has problems such as insufficient rigidity of law enforcement and administrative intervention. To be specific, active reform of the sewage charge system should be done to encourage enterprises to discharge less, pay fewer taxes, and form a positive incentive to achieve technological progress. In addition, a unified emission trading system should be established to compete with the quota issuance mechanism and make full use of the market mechanism to stimulate the compensation effect of technological innovation in Porter's hypothesis.

Limitations of the study
This study examines how heterogeneous environmental regulation affects green economy development on both the theoretical and empirical levels. However, there was no theoretical discussion of the mathematical model of the impact of market-incentive environmental regulation on green economy development; more work is needed to verify the relationship between the two from an empirical point of view. Market-incentive Environmental regulations typically provide businesses more freedom to decide how to behave and give them a chance to take the lead in balancing the development of their economic performance and pollution reduction. This is the main direction of China's environmental regulation reform in the future (Tang et al 2020, Yuan 2020, Song andHan 2022). A focus for future research on related topics should be the effects of market-incentive environmental regulation on green economy development.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).