Heterogeneous environmental regulation and industrial structure 1 transformation and upgrading: Evidence from China

: There are obvious differences between different types of environmental regulations, which are 9 manifested in different environmental protection tendencies and goals, and there are also significant 10 differences in policy implementation. Therefore, it is an urgent empirical problem to quantitatively 11 evaluate the impact of heterogeneous environmental regulations on the transformation and upgrading of 12 industrial structure. Based on this, this paper empirically tests the impact of heterogeneous environmental 13 regulations on the transformation and upgrading of industrial structure on the basis of measuring the level 14 of industrial structure transformation and upgrading. The results show that the transformation and 15 upgrading level of China's industrial structure increased year by year from 2000 to 2017. The three types 16 of environmental regulatory measures have effectively promoted the transformation and upgrading of 17 the industrial structure, and the market incentive environmental regulation (MER) has a significantly 18 higher role in the advancement of the industrial structure than the command-controlled environmental 19 regulation (CER) and the voluntary public participation environmental regulation (VER). The three types 20 of environmental regulations have gradually highlighted the role of environmental regulations in 21 promoting the transformation and upgrading of industrial structure with the improvement of the level of 22 regional economic development. In the mechanism test, it is found that technological innovation has 23 partial mediating effect in the process of CER affecting the transformation and upgrading of industrial 24 structure, MER and public VER have complete mediating effect, while FDI has only partial mediating 25 effect in the process of three types of environmental regulation affecting the transformation and 26 upgrading of industrial structure.


Introduction 31
After more than 40 years of reform and opening up, China's economic growth has created a miracle 32 in the history of world economic development. According to data from the National Bureau of Statistics, 33 China's GDP grew at an average annual rate of 9.3% from 1979 to 2019, and its average annual 34 contribution to world economic growth was about 18%. However, in the early stage, China's excessive 35 reliance on extensive development with high investment, high energy consumption, and high pollution 36 caused excessive use of resources and severe environmental damage, which became the main obstacle 37 to economic growth (Lu, 2017;Liao, 2018), and resources increasingly depleted and deteriorating 38 ecological environment (Dai, 2015). Therefore, coordinating the relationship between ecological health 39 and economic growth is the key to implementing sustainable development strategies (Abu Seman et al., 40 2019). The "2018 Global Environmental Performance Index (EPI) Report" released by Yale University 41 and others pointed out that among the 180 countries participating in the evaluation, China ranked 120, 42 which shows that China's environment is in urgent need of improvement. China's economy has reached 43 a critical moment when it has to change its economic development mode, that is, from a stage of rapid 44 growth to a stage of high-quality development. The transformation and upgrading of industrial structure 45 is an important way to coordinate sustainable economic development and environmental protection. The 46 characteristic of industrial structure upgrading is that production factors enter high-value-added, high-47 efficiency and low-consumption industries from low-value-added, low-efficiency and high-consumption 48 industrial chain links. The process of chain links is one of the key drivers of economic growth. (Pipkin 49 and Fuentes, 2017; Zhu et al., 2019). Therefore, in order to alleviate the downward pressure on the 50 economy, the Chinese government is seeking to transform the economic development mode, optimize 51 the industrial economic structure, increase total factor productivity, and promote the national economy. 52 (Hu et al., 2020;Guo et al., 2020). The transformation of industrial structure can not only increase the 53 proportion of technology-intensive and knowledge-intensive industries, promote technological progress, 54 and support emerging industries, but also reduce the proportion of high-polluting and high-energy-55 consuming industries, encouraging environmental technology research and development and investment 56 in cleaner production equipment, and improve the level of cleaner production from the root cause. 57 However, the current problem of irrational and unbalanced industrial structure in China has 58 gradually emerged (Heinrich and Dai, 2016), which has attracted the attention of the government. In the 59 "13th Five-Year" development plan outline, it is clearly proposed to promote the transformation and industry's contribution to China's GDP exceeded 50% for the first time (Yu ang Wang, 2020). In 2019, 75 the three major industries accounted for 7.1%, 39% and 53.9% of GDP. 76 There are obvious differences between different types of environmental regulations, which are 77 reflected in different environmental protection tendencies and goals, and there are also significant 78 differences in policy implementation, so the impact on industrial structure transformation and upgrading 79 will also be different. The Chinese government has gradually established a command and control type, 80 market incentive type, and voluntary public participation type "three-dimensional one" environmental 81 policy tool system. The heterogeneity between the three will directly affect the cost constraints of 82 industrial structure transformation and upgrading, and thus affect "Porter effect" and "Compliance cost 83 effect" in environmental regulation. Ignoring the heterogeneity of environmental regulations or 84 conducting a total-level analysis of environmental regulations can easily lead to deviations in the 85 assessment of the effects of environmental regulations and policies (Li, 2020). Therefore, the research 86 on environmental regulations should not only involve the strength of environmental regulations, but also 87 include the reasonable mix and use of heterogeneous environmental regulatory policy tools. Based on 88 this, this paper studies the impact of heterogeneous environmental regulations on the transformation and 89 upgrading of industrial structure based on different types of environmental regulation perspectives. To 90 explore whether China's environmental regulations promote the transformation and upgrading of the 91 industrial structure while tackling environmental pollution? That is, can we achieve a win-win situation 92 between environmental protection and industrial structure transformation and upgrading? 93 The rest of this paper is organized as follows. The second part is a literature review. The third part 94 is research design. The fourth part is the discussion of the empirical results, and the final part is the 95 summary of the research conclusions and policy implications. 96

Literature review 97
Scholars focus on "Porter Hypothesis" and "Pollution Haven Hypothesis", and research shows that 98 environmental regulation policies can affect regional industrial structure in various ways ( on resources and the environment, promote the diversified development of the industrial structure of 120 resource-based countries, and reverse the disease problem in the Netherlands.

121
The second is that the "unfavorable theory" means that increasing the intensity of environmental 122 regulations will increase the company's "compliance cost", weaken its market price advantage, and then 123 put the company at a disadvantage in market competition, and ultimately inhibit the transformation and 124 upgrading of the industrial structure (Jaffe and Palmer, 1997). Specifically, in order to circumvent 125 environmental regulations or reduce environmental costs, regional differences in environmental 126 standards or environmental regulations will encourage polluting companies to transfer, thereby inhibiting 127 the adjustment of the industrial structure of the region or the country, that is, the "Pollution Refuge" effect. environmental regulations have increased the production costs of enterprises, reduced production 132 efficiency and profits, and affected the industry structure through enterprise entry or exit, production 133 scale adjustment and resource reallocation. Alpay et al. (2002) found that environmental regulation is a 134 negative hindrance on the food processing industry in the United States. 135 Third, the impact of environmental regulations on the transformation and upgrading of industrial 136 structure is uncertain, that is, the relationship between environmental regulations and industrial structure of economic development between regions in China, there may be differences in the "innovation 164 compensation effect" and the "compliance cost effect" produced by environmental regulations, leading 165 to differences in the impact of environmental regulations on the transformation and upgrading of regional 166 industrial structures. Moreover, the transformation and upgrading of the industrial structure is mainly 167 manifested in two aspects: the rationalization and advanced of the industrial structure. Only when 168 environmental regulations simultaneously promote the rationalization and advancement of the industrial 169 structure can the transformation and upgrading of the industrial structure be realized in a real sense.

170
Based on this, first of all, this research measures the development level of China's industrial structure 171 transformation and upgrading from two aspects, rationalization and advancement of the industrial 172 structure. Secondly, based on the perspective of different types of environmental regulation, this paper 173 uses panel data model and threshold model to study the impact of heterogeneous environmental 174 regulation on the transformation and upgrading of industrial structure, and discusses the impact 175 mechanism, so as to provide new empirical evidence for the relationship between environmental 176 regulation and the transformation and upgrading of industrial structure. 177

Model setting 179
This paper mainly explores the relationship between heterogeneous environmental regulations and 180 industrial structure transformation and upgrading. As the Chinese government has gradually established 181 a command-and-control, market-incentive and voluntary public participation "three-dimensional one" 182 environmental policy tool system. At the same time, considering that the three types of environmental 183 regulations may directly have complementary possibilities, the three types of environmental regulation 184 interaction items are introduced into the model to construct the following panel data model. 185 186 Where, TL and ES are industrial structure rationalization and advanced industrial structure, 189 respectively, CER , MER , and PER represent command-control, market incentive, and voluntary 190 public participation environmental regulations, X is a controlled variable, and subscripts i and t 191 represent province and time, it  and it  represent residual items.

192
Taking into account the large differences in the level of economic development between regions in 193 China, there may be differences in the "innovation compensation effect" and "compliance cost effect" 194 produced by environmental regulations, leading to differences in the impact of environmental regulations 195 on the transformation and upgrading of regional industrial structures. Furthermore, taking the level of 196 regional economic development as the threshold, this paper uses the threshold model to study the impact 197 of heterogeneous environmental regulation on the transformation and upgrading of industrial structure.

198
In the past, most studies on the "threshold effect" used methods such as grouping experience and adding 199 cross-terms. Grouping experience was restricted by the grouping criteria and could not accurately 200 estimate the threshold value. Although the cross-term was added to the model to estimate the threshold 201 value, it could not verify the correctness of the threshold. It was not until Hansen (1999) proposed the 202 non-dynamic panel threshold regression modeling idea that it solved the accurate measurement of the 203 "threshold effect". The choice of threshold variables can be determined exogenously by the theoretical 204 model. The threshold regression model originally proposed by Hansen (1999) is as follows. 205 is the threshold variable. The threshold variable i q serves to divide the sample into different groups, 209 and the regression coefficients in the model are determined based on whether the threshold variable 210 is greater than or less than a certain threshold value. 211 Therefore, the above model can be transformed into the following 213 model. 214 Where, This paper draws on the threshold regression model of Hansen (1999) and establishes a single 217 threshold regression model as follows.
(7) 226 (6) and (7) where ED is the threshold variable, which represents the level of regional economic 227 development, γ 1 , γ 2 , and γ 3 are the threshold values, and the meaning of other variables is consistent 228 with the benchmark model. 229

Dependent variable 231
Industrial structure refers to the composition of various industrial sectors and within each industrial 232 part in the national economy. The transformation and upgrading of industrial structure is the process of 233 industrial structure changing from a low-level form to a high-level form (Ernst, 1998). The dependent 234 variable of this paper is the transformation and upgrading of industrial structure, which is mainly 235 measured from two dimensions of industrial structure rationalization and industrial structure 236 advancement. The rationalization of industrial structure not only reflects the ability of inter-industry 237 structural transformation, but also reflects the effective use of resources. It is a measure of the degree of 238 coordination between factor input and industry. Existing studies generally use the degree of structural 239 deviation to measure the level of rationalization of industrial structure, but because this method ignores 240 the different roles of different industries in economic development, the measurement results may be 241 biased. Therefore, this paper selects the Theil Index, which takes into account the deviation of the output 242 value and employment structure of each industry, and the difference in the economic status of each 243 industry, as a measure of the rationalization of the regional industrial structure. The specific calculation 244 formula is as follows. 245 Where, TL stands for Theil Index, Y stands for output value, L stands for employment, and m stands 247 for three major industries. When TL=0, the economy is in an equilibrium state. The larger the TL value, 248 the more easily the economic development deviates from the equilibrium state and the more unreasonable 249 the industrial structure. 250 The advancement of industrial structure is an important part of the transformation and 251 upgrading of industrial structure. In general literature, the advancement of industrial structure is 252 defined as the increase in the proportion of non-agricultural output value, but this method cannot 253 accurately reflect the nature of industrial structure evolution. The advancement of industrial 254 structure involves the evolution of industrial proportional relations and the improvement of labor 255 productivity (Liu et al., 2008). Therefore, referring to the method of Liu et al. (2008), the product 256 of the inter-industry proportional relationship and labor productivity is used as a measure of 257 advanced industrial structure. Calculated as follows. 258 Where, ES stands for advanced industrial structure, and LP stands for labor productivity, which is 260 calculated by using the ratio of regional industrial added value to the number of employees at the end of 261 the same period. The larger the ES value, the higher the degree of advanced industrial structure. to select the employment index of regional environmental agencies instead. Market incentive 267 environmental regulation (MER) is stimulated by economic interests and influences the environmental 268 behavior of the parties through market forces, mainly including sewage charges, subsidies, taxes, etc. 269 Among them, sewage charges are the earliest market-incentive environmental regulation tool. And the 270 scope of implementation is wide, the time span is long, and it is more representative than policies such 271 as emissions trading and subsidies. Therefore, this paper draws on the practices of Levinson (1996) and  3.2.3 Threshold variable 280 The regional development level (ED) situation determines the environment in which the industrial 281 structure transformation and upgrading is located. Therefore, the regional economic development level 282 is very important to the industrial structure transformation and upgrading. This paper uses per capita 283 GDP as a measure of the regional economic development level, and takes it as a threshold variable. 284 3

.2.4 Control variables 285
According to the existing research, some provincial characteristics are added as control variables in 286 the regression analysis of the model to alleviate the bias of omitted variables as much as possible. Natural 287 resource endowment (NR). Natural resource endowment not only reflects the resource constraints of 288 various regions in the choice of industrial structure, but also reflects the principle of comparative 289 advantage of economic development. For the measurement of natural resource endowment, this paper 290 selects the sum of fixed asset investment of agriculture, forestry, animal husbandry, fishery and extractive 291 industries as the proportion of the total fixed asset investment of the whole society to reflect the 292 comprehensive endowment of natural resources in various regions to the greatest extent. Resident 293 consumption (RC). It is generally believed that with the improvement of living standards, people will 294 pay more attention to the quality of life, environment and other factors, which will have a compelling 295 effect on the transformation and upgrading of the industrial structure. This paper uses the ratio of the 296 consumption level of each region to the region's GDP to measure. Informatization (IT). Due to data 297 limitations, this paper uses per capita post and telecommunications business to measure the level of 298 informatization. Financial development level (FL). Levine (1997) believes that financial development is 299 accompanied by economic development. With the continuous development of financial markets, the 300 provision of financial services for economic entities has promoted the optimal allocation of resources, 301 and has a certain role in promoting the transformation and upgrading of industrial structure. This paper 302 draws on the common practice of existing literature and takes the proportion of the loan balance of each 303 region in the total output value of the region at the end of the year as a measure of the level of financial 304 development. are deflated by using the general production index and the consumer price index respectively. For a small 311 part of missing data, it is estimated by trend fitting.

The average change trend of industrial structure rationalization and advancedization 316
Based on the above methods, the degree of rationalization and advancement of the industrial 317 structure of 30 regions in China from 2000 to 2017 was measured, and the average change trend was 318 calculated. The details are shown in the figure. It can be seen from Figure 1 that, first of all, the Theil 319 Index is in a downward trend as a whole, indicating that China's industrial structure configuration level 320 is gradually increasing, and the industrial structure tends to be rationalized, but there is still a certain 321 distance from the optimal level. Secondly, judging from the magnitude of the change in Theil Index, most 322 years showed negative growth, especially after 2012, the decline was large, and only a small increase in 323 2017.

Benchmark regression 343
Firstly, before panel data regression, the model needs to be tested with Hausman statistics to 344 determine whether to use a random effects model or a fixed effects model. Secondly, in order to avoid 345 the bias of the estimation results caused by the multicollinearity of the model, this paper adopts the 346 stepwise regression method to perform regression estimation on the model. The results of Hausman test 347 show that the fixed effects model is better than the random effects model, so the fixed effects model is 348 selected. The regression results of the model are shown in Table 2 and Table 3. It can be seen from the 349 regression results of the model that the overall model has passed the significance test of the F statistic, 350 indicating that the measurement model is set reasonably. The following is a detailed analysis of the 351 regression structure of the model with the rationalization of industrial structure and advanced industrial 352 structure as dependent variables. 353 The model (1) in Table 2  structure has a certain negative effect on the rationalization of industries. In recent years, the income 366 level of residents has been increasing year by year, and the consumption structure has been upgraded. In 367 order to meet consumption needs, the industrial structure will inevitably change. The process of 368 informatization contains the mechanism of rapid industrial development, which directly promotes the 369 rationalization of industrial structure. With the improvement of the level of financial development, the 370 provision of financial services for industrial development has promoted the optimal allocation of 371 resources and reduced its financing costs, thereby conducive to the development of a rationalized 372 industrial structure. The natural resource endowment is positively correlated with the Theil Index, but it 373 has not passed the significance test, indicating that natural resource endowment has not played an 374 effective role in the rationalization of industrial structure. 375 Table 3

Robustness test 417
(1) Key independent variables lag by one period 418 Considering that the main body of research often has a time lag in response to environmental 419 regulations, that is, the effectiveness of environmental regulations will take a period of time to come into 420 play. Therefore, this paper will examine the impact of the lagging first-phase variables of three types of 421 environmental regulations on the transformation and upgrading of industrial structure to test the 422 robustness of the benchmark model estimation results. It can be seen from  F test is the adjoint probability of statistics.

439
(2) Replace the key independent variables 440 Since there are many indicators for environmental regulation, this paper will select an alternative 441 indicator for the three types of environmental regulations to test the robustness of the model. Among 442 them, the CER chooses "the number of environmental administrative punishment cases" as a substitute 443 indicator. MER select the "three-simultaneous deposit" as an alternative indicator, but based on the 444 availability of data, use the "implementation of the three-simultaneous project environmental investment 445 amount" for specific measurement in each region. VER select "the total number of environmental letters 446 and visits in the region" as a substitute indicator. It can be seen from  F test is the adjoint probability of statistics.

462
(3) Change the model estimation method 463 Considering that fixed effects regression may have endogenous problems, the generalized distance 464 estimation method is used to estimate the robustness of the benchmark model. System Generalized 465 Distance Estimation (System GMM) combines differential GMM and horizontal GMM into one, and 466 performs GMM estimation as a simultaneous equation system. In the system GMM, the endogenous 467 explanatory variable lag is used as its own instrumental variable, which not only improves the efficiency 468 and robustness of the estimation, but also estimates the regression coefficient that does not change with 469 time. Based on this, this paper adopts a two-step system GMM estimation method to test the robustness 470 of the above-mentioned benchmark model. It can be seen from

Non-linearity test 485
This paper selects the level of regional economic development as the threshold variable, and uses 486 the Bootstrap method to repeatedly sample 300 times for threshold self-sampling inspection. Table 7 and 487 9 show the results of the 300 self-sampling tests, and

510
Based on the above threshold effect test analysis, it can be seen that the three types of environmental 511 regulations have a threshold effect on the transformation and upgrading of the industrial structure with 512 the regional economic development level as the threshold variable. Therefore, the threshold model is 513 regressed. Table 11 and Table 12 are the regression results of the threshold model for the advanced and  514 rationalized industrial structure of the dependent variable. It can be seen from Table 11 that the three  515 types of environmental regulations have gradually highlighted the role of environmental regulations in 516 promoting the advancement of the industrial structure with the improvement of the level of regional 517 economic development. In particular, when the level of regional economic development is lower than 518 2.397, the VER does not play a significant positive role in the advancement of the industrial structure. 519 This shows that the level of regional economic development, as the material basis for the transformation 520 and upgrading of the industrial structure, is crucial to the realization of environmental regulation policies 521 to promote industrial transformation and upgrading. If the regional economic development level is high, 522 when the environmental regulation policy is implemented, it can use its own economic advantages to 523 achieve industrial transformation and upgrading smoothly. However, if the regional economic 524 development level is low, the local government may cause the effect of environmental regulation policies 525 to be greatly reduced in order to develop the economy. 526 It can be seen from Table 12 that MER only has a single threshold value of 0.347 when the regional 527 economic development level is used as the threshold variable. Similarly, the three types of environmental 528 regulations have gradually highlighted the role of environmental regulations in promoting the 529 rationalization of industrial structure with the improvement of the level of regional economic 530 development. And when the level of economic development is low, CER and VER have not effectively 531 promoted the improvement of the level of industrial structure rationalization. 532 Note: ***, **, * indicate significance at the level of 1%, 5%, and 10%, respectively. Values in parentheses 534 indicate standard errors. 535  Note: ***, **, * indicate significance at the level of 1%, 5%, and 10%, respectively. Values in parentheses 537 indicate standard errors. 538

Mechanism test 539
Since environmental resources are non-exclusive and competitive, the negative externalities caused 540 by environmental pollution need to be resolved through government forces. The theory of government 541 intervention believes that the externality of environmental pollution cannot be solved by the market 542 mechanism itself, while the government has complete information and can rely on environmental 543 regulations to internalize external effects. Porter et al (1995) expanded and deepened this theory and 544 proposed that strict environmental regulations will induce technological innovation in enterprises, 545 thereby achieving the goal of a win-win situation for environmental protection and economic 546 development, namely the "Porter Hypothesis". Specifically, on the one hand, enterprises affected by 547 environmental regulations may choose to improve production processes and pollution control 548 technologies in order to avoid being punished by environmental regulations. On the other hand, the green 549 subsidy policy may receive investment in technological innovation, thereby increasing the level of green 550 technological innovation, and ultimately promoting the transformation and upgrading of the industrial 551 structure. Those who hold the opposite view to Porter's hypothesis believe that because the cost of 552 complying with environmental regulations always exists, it will have a crowding-out effect on innovation 553 funds, thereby hindering the transformation and upgrading of industrial structure to a certain extent, that 554 is, "following the cost theory." Therefore, this paper further examines whether environmental regulation 555 can achieve the transformation and upgrading of industrial structure through the intermediary effect of 556 technological innovation. 557 Before foreign capital enters the host country, the strength of environmental regulations in a country 558 or region will inevitably affect the location of foreign investment. The "Pollution Paradise Hypothesis" 559 points out that in order to maximize profits, pollution-intensive companies will transfer their industries 560 to countries with lower levels of environmental regulation. On the one hand, strict environmental 561 regulations have become an obstacle to the capital entry of some polluting companies, increasing the cost 562 of foreign companies purchasing raw materials from the local area, and then squeezing out the investment 563 in production and operation links, which is not conducive to the spillover effect of foreign capital, so it 564 may have a negative impact on the transformation and upgrading of local industrial structure. On the 565 other hand, the establishment of environmental regulations as a screening threshold for foreign 566 investment can reduce the entry of pollution-intensive companies and leave room for other companies 567 with relatively clean production. Adom and Amuakwa-Mensah (2016) believe that the technology 568 spillover of foreign enterprises in the host country can improve the production efficiency of domestic 569 enterprises, thus promoting the transformation and upgrading of industrial structure to a certain extent. 570 Similarly, this paper examines whether environmental regulations can achieve industrial structural 571 transformation and upgrading through the intermediary role of foreign direct investment. 572 On the basis of existing literature, it is considered that environmental regulation may have an impact 573 on the transformation and upgrading of industrial structure through two paths: technological innovation 574 and foreign direct investment. Therefore, in order to test whether these two paths exist or not, this paper 575 uses the method of Baron and Kenny (1986)  include R&D investment, the number of scientific researchers, and the number of patents. We uses the 592 number of patent applications to measure. FDI is foreign direct investment, which is measured by the 593 ratio of the amount of regional foreign direct investment to local GDP. The meaning of other variables is 594 consistent with the benchmark model. 595 Table 13 reports the estimated results of technological innovation as an intermediary variable. It can 596 be seen that the three types of environmental regulations all have a significant positive role in promoting 597 technological innovation, which can be the next step of the mediation effect test. After the intermediary 598 variable PA is introduced into the model, the regression coefficients of CER are still significant, while 599 the regression coefficients of MER and VER have become insignificant. Part of the mediating effect 600 exists in the process of industrial structure transformation and upgrading, and there is a complete 601 mediating effect in the process of MER and VER on the process of industrial structure transformation 602 and upgrading. This also shows that China's MER and VER have the "Porter Hypothesis" in their 603 influence on the transformation and upgrading of industrial structure, while the CER is not. 604 Note: ***, **, * indicate significance at the level of 1%, 5%, and 10%, respectively. Values in parentheses 618 indicate standard errors. 619

Research conclusions and policy implications 620
Since the reform and opening up, China's economic development speed and overall scale have made 621 remarkable achievements, but at the same time, it is also facing problems such as overcapacity and 622 overall de-realization to virtuality, and the problem of irrational and unbalanced industrial structure has 623 gradually become prominent (Heinrich and Dai, 2016). To transform the mode of economic development 624 and lead the economy to high-quality development. The Chinese government has chosen a path that can 625 both improve the environment and promote sustainable economic development. For this reason, it has 626 issued a number of environmental policies to deal with environmental problems in the process of 627 industrial development. However, there are obvious differences between different types of environmental 628 regulations, which are reflected in different environmental protection tendencies and purposes, and there 629 are also significant differences in policy implementation. Therefore, quantitative assessment of 630 heterogeneous environmental regulations has become an empirical problem that needs to be solved 631 urgently for the transformation and upgrading of China's industrial structure. Therefore, this paper 632 describes the transformation and upgrading of industrial structure from two levels of industrial structure 633 rationalization and advancement. On this basis, it uses panel data models and threshold models to 634 empirically test the different effects of heterogeneous environmental regulations on industrial structure 635 transformation and upgrading, and further analyzes which influence mechanism. The main conclusions 636 are as follows: (1) The level of rationalization of China's industrial structure increased year by year from 637 2000 to 2017, but there is still a certain distance from the optimization level. The industrial structure is 638 changing from a low-level form of "primary, secondary, and tertiary industries", transformed into a high-639 level structure of "tertiary industry, secondary industry, and primary industry".
(2) The three types of 640 environmental regulation measures can effectively reduce the deviation degree of industrial structure and 641 improve the advanced level of industrial structure, and the effect of MER on the advanced level of 642 industrial structure is significantly higher than that of CER and PER. However, from the regression 643 results of the three interaction terms, it has no effect on the transformation of industrial structure 644 upgrading has a positive promoting effect.
(3) According to the results of non-linear testing, the three 645 types of environmental regulations have gradually highlighted the role of environmental regulations in 646 promoting the transformation and upgrading of the industrial structure as the level of regional economic 647 development increases. (4) In the examination of the mechanism of action, it is found that technological 648 innovation has a partial mediating effect in the process of the impact of CER on the transformation and 649 upgrading of industrial structure. MER and VER have a complete intermediary effect in the process of 650 industrial structure transformation and upgrading, while foreign direct investment has only a partial 651 intermediary effect in the process of the three types of environmental regulations affecting industrial 652 structure transformation and upgrading. 653 Based on the above research conclusions and combined with theoretical analysis, the following 654 policy implications are drawn. First, improve the environmental law enforcement system, continuously 655 innovate law enforcement methods, fundamentally solve the problems of extensive, random law 656 enforcement and unfair law enforcement, and effectively improve the reasonable level of command-and-657 control environmental regulations. The second is to build a market incentive-based environmental 658 regulation and guidance mechanism. MER has become the main regulatory force to promote industrial 659 transformation and upgrading. China should continue to carry out in-depth emissions trading and carbon 660 emissions trading pilot work, summarize and improve its regulatory methods, and finally promote it to 661 all provinces across the country to obtain environmental dividends. The third is to establish and improve 662 laws and regulations on public participation, improve the government information disclosure system, 663 strengthen corporate information disclosure systems, give full play to the role of media supervision and 664 reputation mechanisms, increase education and publicity, encourage people to participate in 665 environmental protection, and actively play VER promotes the transformation and upgrading of 666 industrial structure. Fourth, attach importance to the mediating effect of various types of environmental 667 regulations. First, increase investment in technological innovation in the manufacturing industry, 668 accelerate the transformation, utilization, promotion and application of research and development results, 669 and improve the intellectual property protection and other invention and creation incentive systems. 670