How green industrial policy affects the constancy of green technology innovation: a fresh proof from the innovation motivation perspective

Green industrial policy is an important means to achieve coordinated development of the environment and industry. Enterprises are the primary bodies involved in the green transformation of the market. Thus, it is important for governments and policymakers to recognize the micro-effects of policy implementation on enterprise innovation. From the perspective of the green-tech innovation motivation of enterprises, this study distinguishes green enterprises using text mining analysis methods. Based on propensity score matching (PSM) samples and using panel data from 1,391 listed enterprises in China for the period of 2008–2019, a log log survival analysis model was constructed, and the effectiveness of green industrial policy on the green-tech innovation constancy of enterprises was dynamically analyzed. The results demonstrate that enterprises supported by the green industrial policy can significantly reduce the risk rate of stopping green-tech innovation constancy. Green industrial policy can significantly improve the constancy of the green technological innovation capabilities of enterprises. The innovation constancy of state-owned enterprises is higher than that of private enterprises. Furthermore, intermediary effect tests indicate that different types of green industrial policies stimulate enterprises to generate different innovation motivations. Command-control policies can promote substantive innovation constancy through the industrial competition effect and exert a significantly positive effect. Market-oriented policies can induce enterprises to perform strategic innovation constancy through fiscal incentives and financial investment effects and can produce significant negative effects. Based on these findings, this study proposes improvements in the policy construction of a long-term mechanism to strengthen the substantive innovation of enterprises. This study provides a theoretical basis and policy reference for improving the incentive efficiency of green industrial policies and promoting constant enterprise innovation under the Green and High-Quality Development Goals of China.


Introduction
With the increasingly prominent conflict between economic growth and environmental pollution, green development has gradually become an important strategy for governments to achieve their development goals.Since the financial crisis in 2008, as a response international organizations such as the United Nations have put forward the Green New Deal initiative calling on all economies to vigorously develop the green economy and achieve the transformation of economic growth mode to meet the various challenges in the process of sustainable development.Therefore, many countries and regions have begun to implement green development measures as a national development strategy (Liu et al 2021).In practice, new green deals, green industries, and low-carbon societies are being vigorously developed in the United States and Europe (Schlosser 2021, Xiao et al 2021, Pan et al 2022).In emerging economies, green industrial policies are widely used to solve environmental governance problems in industrial structure upgrades (Pegels 2014).For China, to achieve coordination between economic development and the ecological environment, the Central Committee of the Communist Party of China (CPC) has proposed a new approach to promote the formation of green, low-carbon, and circular development from the perspective of ecological civilization construction (Chen andLi 2021, Shen et al 2023).The 19th National Congress of the Communist Party of China proposed a legal system and policy guidelines to accelerate the establishment of a green economic system (Li et al 2019).In this context, industrial policy, as an important driver in promoting industrial structure transformation and economic efficiency, has played an important role in effectively guiding governments to promote market mechanisms and economic structure reform (Guilhot 2022).Green industrial policy is an industrial policy with environmental objectives that aims to promote the coordinated development of the environment and industry (Li and Ding 2022).Compared to traditional industrial policy, green industrial policy places more emphasis on environmental externalities and inclusiveness to make the economic structure consistent with the needs of sustainable development (Altenburg and Pegels 2012, Fischer 2017).In 2019, the National Development and Reform Commission of China issued the 'Green Industry Guidance Catalogue' ('Catalogue ', hereinafter).From the policies and measures on investment, price, finance, taxation, and others, the catalogue focuses on strengthening the six major aspects to accelerate the green transformation of the industry (Zhang et al 2022).Green industrial policies have become an important policy choice for governments to promote green development ideas (Altenburg et al 2017).If it can promote the level, quality, and efficiency of a green economy and promote industrial transformation and upgrading, it is of great importance to the sustainable development of an economy.
In the process of green development, enterprises are not only the subject of market transformation but also the main consumers of resources and energy.Green technological innovation is at the core of effectively improving the sustainable development ability of enterprises for green-tech innovation (Luo and Liu 2022).Green-tech innovation can not only improve the production efficiency and change the production mode of enterprises, but it can also promote the continuous improvement of green industry and achieve a 'win-win' in regard to economic development and environmental improvement.The existing literature primarily analyzes the relationship between industrial policy and economic efficiency, investment efficiency, and production efficiency at the macro level.However, in China, industrial policy is typically selective, and this is manifested as 'government choice' rather than 'market choice' in the context of regulation.Through development planning of the industry, the government provides financial support such as tax reduction, government subsidies, and land allocation to relevant enterprises and then guides the flow of resources to industries in the supporting areas (Dai et al 2021).Regarding the micro aspect, studies indicate that industrial policy may induce enterprises to carry out policy arbitrage of 'production for subsidies' (Tong et al 2014), thus leading to the distortion of enterprise innovation behavior ( Li and Wang 2019).Enterprises supported by industrial policies face two types of innovation that include strategic and substantive innovation (Chen et al 2023).
Based on this analysis, it is of interest to determine the effect of government-led green industrial policy on the motivation and result of green-tech innovation of enterprises under a green development background.It is important to assess if the introduction of green industrial policy would stimulate the constancy of enterprise green-tech innovations.Additionally, it should be determined if implementing green industrial policy would render arbitrage.Enterprises may also conduct strategic green innovation to obtain support from the government that will result in irrational investment in the capital market and policy inefficiency.Based on this, it is important to identify which 'market oriented' or 'policy oriented' approaches enterprises should follow to conduct their green innovations.To answer these questions, this study focuses on the micro effect of green policy implementation from the perspective of motivation for green technological innovation.The contributions of this research are as follows.(1) Existing studies do not provide a clear definition or standard for the classification of green enterprises.In this study, we classify green enterprises using a text mining method based on a green industrial policy that provides a novel idea for the classification of green enterprises.(2) The underlying path of the green industrial policy effect on the green innovation of enterprises is discovered, and this not only provides a potential theory for the improvement of green industrial policy but also enriches the research focused on green policy effects on enterprise behavior.
(3) This study explored the essence of policy-driven green innovation from the perspective of motivation, thus providing a new perspective in enterprise innovation strategy studies.(4) Existing studies have predominantly examined the static effects of industrial policy without considering the constancy of green innovation.This study adds to the existing research archives.
The remainder of this paper is organized as follows.In section 2, we summarize related studies examining industrial policy and green technological innovation.Section 3 presents the theoretical framework of the study.Section 4 explains the methodology and the basic model, and section 5 reports the empirical results.Section 6 presents the robustness test, and section 7 concludes the manuscript and proposes suggestions.

Related works
Green industrial policies have gradually become an important concern in recent years.Extensive research has been conducted examining the theoretical basis and effects of green industrial policies.From a theoretical perspective, green industrial, traditional industrial, and environmental policies intersect.Essentially, all three types of policies aim to correct market failures caused by externalities (Altenburg et al 2017).Aghion et al (2012) believe that through encouragement, restriction, or elimination, industrial policies can effectively guide enterprises to make investment and financing decisions and improve industrial productivity.However, traditional industrial policies focus on the government implementation of various preferential policies such as taxes, credits, and financial subsidies for specific industries to improve productivity and obtain returns on capital and labor.Green industrial policies seek sustainable development within ecological boundary systems.Government-supported enterprises are subject to choices of the government.Through this mechanism, the government can guide the flow of resources and factors to designated lowcarbon industries and promote their green upgrading (Lei and Wang 2023).
For the policy effects, it was observed that green industrial policies can effectively promote the transformation and upgrading of enterprise structure, improve production efficiency, and play a central role in the transformation of enterprises to a low-carbon structure (Harrison et al 2017, Liu et al 2020, Anzolin and Lebdioui 2021).In terms of energy efficiency, green industrial policy acts as a policy coordination mechanism for high-pollution industries, technology-intensive industries, and renewable energy sources (Chen et al 2021, Shen et al 2021).Additionally, as a market subject and an important component of the industrial chain, the green technological innovation ability of enterprises is an important focus of green industrial policies (Peng et al 2021).Improving the green technology innovation level is the primary means to improve carbon emission efficiency and the green economy (Zhao et al 2021, Dong et al 2022).However, certain studies have reported that although the government can influence the development of specific industries through industrial policies, its ability to intervene in the market is limited (Karp and Stevenson 2012).Interventions with industrial policies may reduce the production efficiency of relevant industries (Lee 1996).Compared to traditional industrial policy, green industrial policy is not only accompanied by the risk and high uncertainty of enterprise R&D activities but must also solve more environmental externalities (Li et al 2019).The construction of a green industrial policy system must effectively guide the corresponding green production technology reform and drive solutions to social distribution and equity issues (Altenburg et al 2017).Therefore, effectively improving the level of green technological innovation has become key to achieving the green transformation of industrial structures.
Certain scholars have analyzed the impact of different types of industrial policies on green-tech innovation by constructing an indicator system, and they observed that different types of policy tools presented a threshold effect.Green technological innovation can be promoted only when policy strength is maintained within a reasonable range.Therefore, it is necessary to formulate policy guidelines for their combination and differentiation (Yi et al 2019, Wu et al 2022).Additionally, industrial policies such as enterprise subsidies and consumer subsidies will affect the green innovation decision-making of enterprises and the public and will exert a significant direct promotion effect on green technological innovation (Gong and Dai 2022).However, it may also lead to the crowding-out effect and prompt enterprises to undertake strategic innovations such as subsidy seeking (Dai et al 2021, Nguyen 2023).Moreover, from a micro perspective, the existing literature primarily focuses on industrial policy and enterprise innovation constancy, financial behavior, and decision-making (Meckling and Nahm 2019).The research examining the relationship between specific green industrial policies and enterprise behavior has primarily been conducted from the perspective of investment behavior and restructuring (Huang and Yuan 2021), and few studies have explored the impact mechanism of green industrial policies on green technological innovation constancy from the perspective of motivation.Consequently, this study attempts to fill this gap.

Mechanism of green industrial policy affecting enterprise green-tech innovation
The essence of industrial policy is a type of policy that allows the government to support specific industries to promote transformation and upgrading through selecting support objectives in a targeted manner (Cai and Tian 2019).Industrial policies can not only directly affect the market but also send signals to the market to guide enterprises to make decisions in line with government guidance and achieve the flow of resource elements.In China, green industrial policy, as a selective industrial policy, is the concentrated embodiment of the national green development strategy (Yan et al 2023).Presently, the green industrial policy tools of China are primarily divided into command-control tools that participate in mandatory administrative orders and market-oriented tools that stimulate market-oriented behavior (Chang and Andreoni 2020).Under the influence of policy tools, the government has issued numerous support measures to coordinate green industrial policy and has played a direct and indirect role in guiding the market.Specifically, the government will directly guide changes in industrial competition patterns through command-control policies, including market access mechanisms, project approval restrictions, energy conservation, and emission reduction requirements.This will facilitate the transformation of enterprises into green low-carbon enterprises to meet the requirements of policies and regulations.Market-oriented policies primarily include financial subsidies, tax preferences, green financing, and other direct economic benefits.The government can improve the financial situation and financing constraints of enterprises through indirect guidance to stimulate them to strengthen green technological innovation (Dai et al 2021).Therefore, enterprises supported by green industrial policy obtain more resources, ultimately promoting the constancy of enterprise innovation levels.Thus, we propose Hypothesis 1.
Hypothesis 1: Green industrial policy promotes consistency in the green tech innovation of enterprises.
Furthermore, due to the information asymmetry between enterprises and government policymakers, enterprises may possess different innovation motives in the actual implementation process (Arqué-Castells 2013).Combined with the research purpose and from the perspective of enterprise green tech innovation motivation, we referenced Li and Zheng (2016) and divided the green tech innovation behavior of enterprises into two types.The first is the green-tech innovation behavior of enterprises generated under the pressure of the market effect to gain market competitive advantages, and this is defined as substantive innovation constancy.The green tech innovation behavior of enterprises driven by the attraction of the market effect to obtain benefits, and this is defined as strategic innovation constancy.In practice, the direct guidance of command-control policies forms a pattern of industrial competition effects.Under the pressure of this effect, to maintain and improve market competitiveness enterprises will prioritize green technology innovation that meets the requirements of the specification, increase research and development efforts, and increase capital investment.Under the influence of a competitive mechanism, enterprises focus on substantive innovation constancy.Under the indirect guidance of market-oriented policies, the government typically provides financial subsidies such as energy-saving and environmental protection subsidies, green technological subsidies, clean energy subsidies, and environmental protection tax incentives.Previous research has demonstrated that financial subsidies and tax preferences can improve the initiative for independent innovation and the ability to allocate resources to enterprises (Wang et al 2021).However, direct economic benefits attract enterprises with short-term wealth and financial incentives.Enterprises obtain more government subsidies through simple innovation or the pursuit of innovation.Meanwhile, green financial measures in market-oriented policies such as green credit and green project guarantees ease the financing constraints of enterprises and reduce financing costs through credit mechanisms (Tian et al 2022, Zhang et al 2022).Therefore, these green financial measures will stimulate enterprises to increase the number of green-tech innovations with the help of financing dividends and make green industrial policies a way to pool funds.Under the comprehensive effect of short-term wealth, financial incentives, and financial investment effects, this will be more attractive to enterprises.Enterprises conduct strategic green technological innovations in pursuit of policy support.Based on the above analysis, we propose Hypothesis 2.
Hypothesis 2: Command-control green industrial policy stimulates enterprises to carry out market-oriented substantial green-tech innovation constancy (a), and this stimulates enterprises to carry out policy arbitrage green-tech innovation constancy (b).
The theoretical mechanism of this paper is shown in figure 1.

Research methods and basic models
Enterprise innovation is a dynamic process.We divided innovation into substantive and strategic innovation from the perspective of the green-tech innovation motivation of enterprises.As it is difficult to identify an appropriate index to directly measure innovation motivation, survival analysis is a dynamic research method.We use the survival analysis method to describe the two types of innovation behavior and explore the constancy of enterprise green-tech innovation.
As the green industry involves a wide range of industries, the China Securities Regulatory Commission has not classified it.We organized the national 'Eleventh Five-Year Plan' to 'Thirteenth Five-Year Plan' planning documents and referred the 'Green Industry Guidance Catalogue' in 2019 (Chen et al 2010).This study uses a text-mining analysis method to identify 38 keywords such as pumped storage, green packaging, sand control, and ecological gardens 3 .Furthermore, we compared the stock concept section of Sina Finance and classified the basic information of the companies, thus identifying them as green enterprises.In practice, it is not random if an enterprise can obtain support from a green industrial policy.To avoid endogeneity problems, overcome sample selection bias, and avoid adverse effects on the estimation results, we adopted the propensity score match (PSM) method to screen the samples.Therefore, we set green enterprises that were supported by the green industrial policy for the first time as the treatment group and enterprises that had never been supported by the green industrial policy as the control group.Furthermore, we estimate the probability of enterprises receiving support from green industrial policies.To more accurately measure the pairing situation between the treatment and control groups, we conduct a matching balance test that is a t-test of the variables that affect the probability of green industry policy support for the treatment and control groups before and after matching.By matching the enterprises, the paired enterprises in the two matched sample groups differed only in whether they were supported by the governmental green industry policy, thereby reducing sample selection bias.
On this basis, a survival model was constructed for empirical research.We referenced Sun et al (2023) and He et al (2023) and used the number of enterprise green patents to measure enterprise green-tech innovation.The starting time is when an enterprise possesses a green patent in a certain year, and the ending time is when the enterprise possesses no patents in the following two years.In survival analysis, the survival function is commonly used to describe the distribution characteristics of the survival time.We define the survival function of an enterprise as the probability that the enterprise has continuously obtained green patents for more than t years as presented in equation (1).
where T represents the enterprise survival time in maintaining green-tech innovation.m t is the risk function that is the probability that an enterprise will no longer own a green patent in period t under the condition of normal operation in period t-1.Based on the above factors, we used the Kaplan-Meier product limit estimation formula to obtain the nonparametric survival function as presented in equation (2).


where N t represents the duration of the risk state in period t, and D t is the number of failed objects observed during the same period.Accordingly, we constructed a Kaplan-Meier survival curve and discrete-time cloglog model to study the impact of green industrial policy on the constancy of the green technological innovation levels of enterprises ( Hess and Persson 2012)   the enterprise.b 0 is the benchmark risk rate, green represents the virtual variable of green industrial policy, Control it represents the control variables that affect the probability of green-tech innovation, åYear, åInd, and åReg respectively represent time fixed effect, industry fixed effect, and regional fixed effect, and e it is a random disturbance term.The main variables used in this study are listed in table 1.

Sample selection and data sources
In this study, we include panel data detailing A-share listed companies in the Shanghai and Shenzhen stock markets from 2008 to 2019.To improve the reliability of the samples, we refer to the method of Brandt et al (2012) that identifies companies from different years based on their legal representative code, company name, phone number, address, and other information and merges the cross-sectional data into a panel data set.The samples were screened as follows: (1) excluding samples from the financial industry; (2) excluding samples with ST and * ST situations during the sample period; (3) excluding samples with missing, incomplete, and uncomplemented data on related variables.(4) To avoid the impact of enterprise entry or exit during the sample period, we retained only enterprises that continued to operate and possessed green-tech patent outputs from 2008 to 2019 as the analysis samples.
After screening, we obtained 1,391 enterprise samples.The author collected data detailing green industrial policies.The data regarding green patents were obtained from the Chinese Research Data Services database.The relevant financial data of the enterprises were collected from the China Stock Market & Accounting Research Database and the Wind database.

Propensity score match results
Based on the above analysis, we used the nearest-neighbor PSM method to identify a suitable control group for the treatment group.Through a matching balance test, we determined that the absolute values of the standard = å ( ) i Where X i is the sales volume of the ith enterprise in the industry.The larger the HHI, the lower the market competition.The smaller the HHI, the higher the market competition.b SA index(SA): SA = −0.737* Size + 0.043 * Size 2 −0.040 * Age, Where size represents the natural logarithm of the enterprise total asset scale, and Age represents the year of enterprise operation = the observation year (accounting period of the current year) -the time of establishment (year) (Hadlock and Pierce 2010).
deviations displayed were all less than 20%, thus indicating that the characteristics of the two groups were basically the same after matching.The t-test results reveal that each variable exhibits a significant deviation at the 5% level.The findings indicate that there is no significant difference in the enterprise matching variables between the treatment and control groups.Therefore, the matching variables and methods were used properly, and details are provided in the appendix.Furthermore, figure 2 presents the distribution of the nuclear density before and after PSM.We observed that the disposition score distribution of the treatment and control groups was similar, thus indicating that the matching effect was ideal.After matching, we obtained 548 green enterprises and 784 non-green enterprises.

Basic estimation results of survival analysis based on PSM samples
Based on the above PSM samples, a survival analysis was conducted to examine the impact of green industrial policy on the constancy of the green-tech innovation of enterprises.First, we constructed the Kaplan-Meier survival function to explore the relationship between green industrial policy and the constancy of green technological innovation.Figure 3(a) presents the Kaplan-Meier survival curve of the innovation constancy of green and non-green enterprises.
As presented in figure 3(a), the survival rates of both green and non-green enterprises decrease rapidly in the first year of green-tech innovation, thus indicating that the risk of enterprise innovation in the initial stage is the highest.The Kaplan-Meier survival curve exhibited a downward trend over time.Further, the survival curve of the supported enterprises is increases from the middle stage of green-tech innovation and gradually exceeds the survival curve of unsupported enterprises.The findings demonstrate that the innovation termination risk rate of the supported enterprises is significantly lower than the innovation termination risk rate of the unsupported enterprises.Thus, compared to unsupported enterprises, green enterprises exhibit longer green tech innovation constancy survival times.
Due to the different ownership structures in China, state-owned and private enterprises exhibit large differences in market environment and policy support.Accordingly, to explore the impact of green industrial policy on the constancy of enterprise green-tech innovation under different enterprise systems, we divided the sample into state-owned and private enterprises based on their different equity natures.Figure 3(b) reports the Kaplan-Meier survival curve of the green-tech innovation of state-owned and private enterprises.The survival time of state-owned enterprise green tech innovation is longer than that of private enterprises.In practice, although the green industrial policy can provide additional financial support for private enterprises and alleviate financial constraints, the policy of some command-control tools such as market access mechanisms and energy conservation and emission reduction requirements will increase the R&D investment cost and R&D pressure on private enterprises.However, the state-owned enterprise market mechanisms and standards are perfect.Due to their strategic position, state-owned enterprises are more vulnerable to the implicit guarantee and credit allocation of the government and are less vulnerable to financial discrimination.Therefore, under the effect of market competition, a green industrial policy plays a better role in promoting green technological innovation of state-owned enterprises.
In addition to the impact of the governmental green industrial policy, other factors can affect the consistency of enterprise innovation.Therefore, according to Model (3), we use a log risk model to further estimate if green industrial policy affects the survival time of enterprise green tech innovation.The estimated results are presented in table 2. The test results reveal that the coefficient of green industrial policy (green) is negative and significant at the 1% level, thus indicating that green industrial policy significantly reduces the termination risk rate of the innovation activities of supported enterprises.Thus, a green industrial policy is conducive to prolonging the survival of enterprise green-tech innovation, fully verifying Hypothesis 1.In terms of control variables, the quick ratio (qr), net profit rate of total assets (ROA), growth rate of total assets A (growth), and business nature (state) all exert a positive impact on the constancy of the green-tech innovation of enterprises, and this helps to reduce the risk of enterprises stopping innovation.Based on the research, the possible reasons are as follows.
(1) When the profit rate and total asset growth rate of enterprises are higher, the market arbitrage space is larger.The green industrial policy of market-oriented tools encourages enterprises to conduct technological innovation to increase innovation constancy and obtain more profits.(2) When the market arbitrage space is larger, it encourages enterprises to obtain more external financing, strengthens their willingness to carry out financing arbitrage, and enhances the constancy of enterprise innovation.

Test of action mechanism
These results indicate that a green industrial policy can promote the constancy of enterprises green technological innovations.Based on this, we further explored the essence of green technological innovation under different types of green industrial policies from a transmission path.As mentioned in the theoretical analysis above, green industrial policy in China is primarily divided into two categories that include command-control policy and market-oriented policy.Command-control policies primarily include market access mechanisms, project approval restrictions, energy conservation, and emission reduction requirements.This produces an industry competition effect under direct guidance.Market-oriented policies include financial subsidies, tax incentives, green credit, and other economic benefits.This produces short-term wealth, fiscal incentives, and financial investment effects under indirect guidance.The combined effect of these four market effects promotes enterprise green tech innovation.It is important to determine the effects of green industrial policies on promoting the constancy of enterprise green-tech innovation.Additionally, we must understand what motivates enterprises to innovate in the face of policy support.This study tests these four transmission mechanisms by constructing an intermediary effect model and examines the innovation motivation of enterprises.We refer to Hadlock and Pierce (2010) and Yu et al (2021) to measure the industry competition effect using the Herfindahl-Hirschman Index (HHI), the short-term wealth effect by tax incentives (Tax), the financial incentive effect of governmental financial subsidy intensity (Sub), and the financial investment effect of the enterprise financing constraint SA index (SA).Based on Model (3), we constructed a mediation effect model to analyze the action mechanism as presented in equations (4)-( 6).b b b e where Agg represents the action mechanism variables that are expressed by the market effect of the green industrial policy.It includes the industry competition (HHI), tax incentives (Tax), financial subsidy intensity (Sub), and enterprise financing constraint SA index (SA).e it is a random disturbance term.The other terms are the same as those presented in Model (1).Table 1 indicates the specific definitions of the control variables.The mediating effect is measured by b g * .
1 2 If both b 1 and g 2 are significant, the mediating effect is significant.

Industrial competition effect
In table 3, Columns (1) to (3) report the estimated results of the industry competition effects.The regression coefficient of green to green-tech innovation is significantly negative, and the regression coefficient of HHI to green-tech innovation is significantly positive.Some intermediary effects were established, thus indicating that the market competition effect promotes constancy in the enterprise innovation level.These findings indicate the establishment of a market competition mechanism.This is due to the observation that green industrial policy improves the market access mechanism of the industry and the requirements of energy conservation and increases project approval restrictions.Thus, command-control tools form a strong market competition effect, ultimately encouraging enterprises to enhance their market power through substantial green-tech innovation.
Command-controlled green industrial policy stimulates enterprises to carry out market-oriented substantial green-tech innovation constancy, thus verifying Hypothesis 2a.

Short-term wealth effect
In table 3, Columns (4) to (6) report the estimated results for the short-term wealth effect.The test results reveal that the regression coefficient of green on green technological innovation is significantly negative, the regression coefficient of tax incentives (tax) on enterprise green-tech innovation is significantly negative, and the regression coefficient of green on tax incentives (tax) is positive.However, this failed to pass the significance test, thus indicating the absence of an intermediary effect.These findings indicate that the path of the short-term wealth effect has not yet been established.Therefore, green industry policy cannot promote the constancy of enterprise green-tech innovation levels by increasing tax incentives.

Fiscal incentive effect
In table 3, Columns (7) to (9) report the estimated results for the fiscal incentive effect.The regression coefficient of green on enterprise green-tech innovation and the regression coefficient of financial subsidy intensity (Sub) on enterprise green-tech innovation are significantly negative.Some intermediary effects were established.These results indicate that the fiscal incentive mechanism path is established, but the fiscal incentive mechanism exerts a significant negative effect on the constancy of enterprise innovation levels.This is due to the observation that the incentive effect of financial subsidies encourages enterprises to carry out more subsidy-seeking activities to obtain more industrial policies.However, financial incentives are easily affected by incentive distortion and rent-seeking risks that may lead to insufficient innovation power for enterprises and an insensitive response to the direct intervention of the green industrial policy.Thus, the financial incentive mechanism in a marketoriented green industry policy will stimulate enterprises to carry out policy arbitrage green-tech innovation constancy, thus verifying Hypothesis 2b.

Financial investment effect
In table 3, Columns (10) to (12) report the estimated results of the financial investment effects.The regression coefficient of green technology on enterprise green technological innovation is significantly negative, and the regression coefficient of the financial investment effect (SA) on enterprise green technological innovation is significantly negative.Some intermediary effects were established.The research findings reveal that the transmission path can be achieved through the financial investment effect, but financial investment exerts a significant negative effect on the constancy of the enterprise innovation level.This is due to the observation that the incentive effect of green finance measures encourages enterprises to use the credit mechanism to obtain external financing on a large scale and at a low interest rate and encourages enterprises to carry out more technological innovation to finance arbitrage.As a result, a capital reservoir effect will be formed that may cause enterprises to carry out 'strategic' innovation.Accordingly, the financial investment mechanism in a marketoriented green industrial policy stimulates enterprises to carry out policy arbitrage innovation constancy, thus fully verifying Hypothesis 2b.
6. Robustness test 6.1.Re-estimation based on the semi-parametric model and the Logit model The basic model of this study is essentially a linear relationship between green industrial policy and the risk rate of green technological innovation based on theoretical analysis.To verify the reliability of the linear relationship, we referenced Bralower et al (1997) and used a semi-parametric model to fit the relationship between green industrial policy and the green-tech innovation termination risk rate as presented in equation (7).This model argues that a complex non-linear relationship exists.If the model is not established, the rationality of Model (3) can be verified in the reverse direction.
b e represent the smooth function part of Model (7).The spline smoothing method was used to fit the model.Table 4 and figure 4 present the estimation results of the model.The findings indicate that the fitting between the dependent and independent variables cannot be completed, and the regression results of all the control variables are not significant.The K-value of the spline was too low, thus indicating that a nonlinear relationship was not established, and this verified the reliability of the basic model in this study.
Furthermore, we refer to Yu et al (2021) who use the logit model to estimate the probability prediction value of enterprise green-tech innovation to investigate the enterprise green-tech innovation level for re-estimation.As presented in table 5, the coefficient of green is negative and significant at the 1% level, thus indicating that green industrial policy still significantly improves the innovation level.This further demonstrates the robustness of our findings.

Replacing variables and eliminating samples
In this study, we used the number of green patents to measure enterprise green-tech innovation.However, the level of enterprise green technological innovation may also be reflected in other innovation outputs.To ensure the robustness of the results, we refer to Li and Zhang (2019) who adopted the method of substitution variables using the total number of patents to measure enterprise green tech innovation and re-estimate the sample.The estimation results are presented in Columns (1)-( 3) of table A1.The test results reveal that the coefficient of green is negative and significant at the 1% level, and this indicates that the number of all patents is taken as the proxy variable for the green tech innovation level of enterprises, while green industrial policy still significantly reduces the termination risk rate of enterprise innovation activities.Therefore, the conclusions of this study are reliable.
The 2008 global financial crisis may have affected enterprise innovation.Therefore, we refer to Yu et al (2021) and exclude the 2008 sample for the re-estimation.The estimation results are presented in Columns (4)-(6) of table A1.After changing the sample size, Column (4) indicates that the coefficient of green is negative and significant at the 1% level.There was no significant change in the statistical significance, thus further confirming the robustness of the test results.

Conclusions and policy implications
This study empirically analyzes the effectiveness of the green industrial policy of China on enterprise green-tech innovation from the perspective of their green-tech innovation motivation.The findings demonstrate that the green industrial policy of China can significantly improve the constancy of enterprise green technological innovation capability and that the innovation constancy of state-owned enterprises is higher than that of private  enterprises.Furthermore, this study explored the essence of enterprise green technological innovation under different types of green industrial policies.Our results reveal that command-control policies can stimulate substantial innovation constancy in enterprises through the industrial competition effect, whereas marketoriented policies can induce enterprises to carry out strategic innovation constancy through fiscal incentives and financial investment effects.Based on the above research, we propose the following policy recommendations to improve the incentive efficiency of the green industrial policy of China and promote the constancy of enterprise innovation.(1) We must improve the construction of a green industrial policy system and promote long-term policy mechanisms.Policymakers should not only focus on the short-term effects of policies but should also clarify the long-term goals and directional planning for green industry development.(2) We must strengthen the coordination between policy systems and tools.In the short term, China should play a direct guiding role in commandcontrol policies and establish industrial norms and supervision mechanisms.In the long term, we should play an indirect guiding role in market-oriented policies, improve the efficiency of policy incentives, reduce the market arbitrage space, and promote the optimal allocation of market resources.(3)The government and policymakers should pay attention to the positive effects of industry competition and encourage fair competition within the industry.They should increase support for private enterprises and further elimination of differences between state-owned enterprises and private enterprises.Meanwhile, we should promote the construction of financial support and green financial systems and improve the efficiency of implementing green industrial policies.(4) We should promote the transformation of the enterprise green development concept and guide enterprises to actively fulfill their social responsibilities from an internal governance perspective.We should promote constant enterprise innovation and avoid the pursuit of distorted policy support and strategic innovation.
The limitations of this study are as follows.First, we consider the impact of the green industrial policy of China on technological innovation under different property rights.As basic statistical data are insufficient, we should not further explore the heterogeneity of the transmission paths between state-owned enterprises and private enterprises.Second, in terms of measurement indicators, this study uses the number of patents to measure green tech innovation; however, the number of patents primarily focuses on the measurement of innovation output and does not reflect the transformation of green technology innovation results.Third, the sample data and research duration should be improved in the future to ensure the universality of the conclusions.

Figure 1 .
Figure 1.The transmission path of green industrial policy affecting enterprises Green-tech innovation constancy.

Figure 2 .
Figure 2. Distribution of nuclear density before and after PSM.

Figure 3 .
Figure 3.The Kaplan-Meier survival curve of enterprise green-tech innovation.

Figure 4 .
Figure 4.The fitting results of semi-parametric model.
where h j represents the discrete-time risk rate, and -( ) Cloglog h 1 j is the explained variable that represents the termination risk rate of enterprises' green tech innovation.A larger value indicates a higher termination risk of

Table 1 .
Definition and description of main variables.

Table 2 .
The survival estimation results based on matched samples.

Table 3 .
Estimation result of action mechanism.