The role of environmental regulation, industrial upgrading and resource allocation on foreign direct investment: Evidence from 276 Chinese cities

11 Environmental pollution is becoming more and more prevalent in China, accompanied by the excessive 12 expansion of the country's foreign direct investment in the scale of resource-based industries. This 13 article uses the panel data of 276 prefecture-level cities in China from 2003 to 2016 to estimate the 14 impact of environmental regulation on foreign direct investment by employing the Spatial Durbin 15 model. The empirical results show that: firstly, environmental regulation, and foreign direct investment 16 have an obvious spatial correlation. Secondly, environmental regulation significantly inhibits foreign 17 direct investment and has significant negative space spillover. Thirdly, non-eastern cities' 18 environmental regulation has significantly greater inhibitory effects on foreign direct investment than 19 eastern cities, and the key cities' environmental regulation has greater inhibitory effects than ordinary 20 cities. Finally, from the perspective of industrial upgrading and resource configuration, environmental 21 regulation has significantly promoted foreign direct investment and have significant negative space 22 spillovers. Therefore, the reasonable use of environmental regulatory measures through industrial 23 upgrading and resource configuration to attract clean, capital-intensive and technology-intensive 24 enterprises and to achieve the effect of "decontamination and clean" for foreign-funded enterprises is 25 critical. 26


Introduction
At present, the research topics of environmental regulation and foreign direct investment are 48 considered by scholars at home and abroad to have the following two viewpoints. On the one hand, in 49 order to concentrate on the development of core industries in the country and keep the core industries 50 in the leading position in the global industrial chain, the "pollution paradise" hypothesis holds that 51 regulations. On the contrary, companies will be motivated by reasonable environmental regulations to 140 carry out technological innovation and other production activities. They will be prompted to optimize 141 resource allocation to improve economic performance so that companies can internalize pollution 142 control costs and maximize net profits (Borsatto et al., 2019). Secondly, from the perspective of the 143 competitive effect of foreign direct investment, with the inflow of foreign-funded enterprises with 144 higher production technology levels, domestic enterprises with relatively backward production 145 technology levels are forced by industry competition pressure to improve production efficiency by 146 optimizing resource allocation methods. As a result, the overall production efficiency of the industry 147 has been maintained at a relatively high level. Moreover, from the perspective of the technology 148 spillover effect of foreign direct investment, domestic companies can improve their own production 149 efficiency by learning the production technology and management models of clean and technological

Construction of the benchmark regression model 164
According to the hypothesis of "Pollution Heaven" and "Pollution Halo", there is a U-shaped 165 relationship between environmental regulation and foreign direct investment. In particular, foreign 166 direct investment is decreased with the strengthening of environmental regulation. When environmental 167 regulation develops to a certain stage, the foreign direct investment will reach the lowest value at the 168 turning point. Then, with the enhancement and optimization of environmental regulation, clean and 169 technology-intensive foreign-funded enterprises will continue to flow in. Therefore, this article 170 establishes an econometric model based on the hypotheses of "Pollution Heaven" and "Pollution Halo" 171 to study environmental regulation and foreign direct investment. 172 This paper introduces the performance-based environmental regulation index into the model, test 173 whether there is a u-shaped relationship between environmental regulation and foreign direct 174 investment. Furthermore, foreign direct investment is also affected by the rate of urbanization, society's 175 overall economic level, marketization level, the degree of openness to trade, and infrastructure 176 construction. Therefore, add urbanization rate, per capita GDP, marketization level, trade openness, and 177 infrastructure construction as control variables to the model to get the following basic econometric Next, on the one hand, reasonable environmental regulation will encourage companies to 208 internalize pollution control costs and maximize net profits by optimizing resource configuration. On 209 the other hand, according to the technology spillover effect of foreign direct investment, domestic 210 enterprises' resource configuration is optimized by learning the production technology of clean and 211 technological foreign-funded enterprises, ultimately realizing the improvement of the overall 212 production efficiency of the society. Therefore, to study the adjustment effect of resource configuration 213 on environmental regulation, in this paper, the interactive items of resource allocation and 214   Firstly, the spatial weight matrix is introduced to control the regional spatial geographic effect in the 225 spatial econometric model. Therefore, a correct and reasonable spatial weight matrix should be able to 226 accurately measure the spatial spillover effect. According to Tobler's first law of geography, everything 227 is related, but things nearby are more related than things far away (Tobler, 1970;Li et al., 2018).
Especially for air pollutants, the spread of air pollutants across regions is obvious under the influence 229 of atmospheric circulation. At present, spatial economists mainly use geographic distance matrix and 230 economic weight matrix to analyze spatial effects. Based on the influence of spatial location factors on 231 economic variables, this paper constructs a geographic distance matrix. Secondly, although with the 232 continuous development of the economy and the deepening of exchanges between different countries 233 and regions, environmental regulation between regions not only depends on the factors of spatial 234 distance but also on factors such as the level of economic development between regions(Wang et al., 235 2019) but in practice, for the spatial spillover effect of pollutant emissions between regions, 236 geographical distance is more important than economic distance. Finally, from an overall point of view, 237 this article chooses the geographic distance matrix as the reference matrix, calculates as follows: 238 In this formula, d indicates the distance between the centers of i and t in two regions or the 240 distance between two points (provincial capital cities). 241

Explained variable: foreign direct investment 243
In terms of foreign direct investment data, firstly, as China's trade opening continues to deepen, foreign 244 direct investment has become one of the important factors to promote China's economic development; 245 secondly, with the excessive expansion of foreign direct investment in the scale of resource-based 246 industries, the problem of environmental pollution is becoming more and more worrying. According to 247 previous literature, this article adopts the actual use amount of foreign direct investment in each region 248 and uses RMB's annual average rate against USD to convert into RMB, which represents foreign direct 249 investment (FDI). In addition, this paper uses the percentage of foreign direct investment in GDP by 250 region to replace the actual use of foreign direct investment (Huang et al., 2021) for the stability test. 251

The core explanatory variable: environmental regulation 252
Until now, the methods of measuring environmental regulation are mainly divided into cost-based 253 environmental regulation indicators and performance-based environmental regulation indicators. Next, 254 because of the lack of official data on air quality emissions fees and pollution investments, it may be 255 inaccurate to use pollution charges and pollution investments to measure environmental regulation. 256 Then, because the cost of pollution is closely related to the level of regional industrial development, the 257 gross industrial output value is closely related to the government's lowered threshold of the 258 environmental pollution to attract the inflow of foreign-funded enterprises; therefore, the payment cost 259 of pollution control is used as a measure of environmental regulation, may cause serious endogenous weights are assigned to different indicators to ensure that it accurately reflects the degree of pollution 276 control in each area. The specific calculation method is as follows: 277 In the formula, zi,t,j indicates the adjustment coefficient of j type indicator in i area in t year; eri,t,j, 279 gdpi,t respectively indicate the j type indicators and GDP of each region in t year. Third step, 280 according to the standardized value er * i,t,j and adjustment coefficient zi,t,j of SO2 removal rate and 281 industrial smoke dust removal rate, calculate the comprehensive index of environmental regulation: 282

Industrial upgrading 285
Firstly, from a macro perspective, industrial upgrading is manifested as the transformation of the 286 industrial structure from a low-level form to a high-level form. Secondly, from a micro perspective, 287 industrial upgrading is manifested in the overall transformation of domestic industries from 288 labor-intensive industries to capital-intensive and technology-intensive industries. Meanwhile, 289 industrial upgrading is also manifested in transferring production factors from industrial sectors with 290 lower production efficiency to industrial sectors with higher production efficiency. According to 291 previous literature, use the shares of the thrice industries and their comparative relationships to measure 292 industrial upgrading (Yuan et al., 2014). The specific calculation method is as follows: 293 In the formula, INDi,t indicates the industrial upgrading of area i in t year ; m indicates the serial 295 number of the thrice industries; indi,t,m respectively indicate the proportion of the primary industry, 296 secondary industry and tertiary industry in GDP. 297

Resource allocation 298
About resource configuration, this article uses the DEA-Malmquist index method to measure resource 299 allocation. According to previous literature, this article uses labor and capital as input factors, taking 300 the gross regional product as an output factor. The specific input and output elements are processed as 301 follows: output indicators, calculated using the gross regional product and converted the gross regional 302 product level to 2003. Input indicators, labor input is expressed by the number of employees in each 303 city over the years, capital input is measured by fixed capital stock, this data is not directly available 304 and needs to be calculated, this article uses the perpetual inventory method to calculate the stock of 305 fixed capital . The specific calculation method is as follows: 306 In the formula, Ki,t indicates the capital stock of area i in t year; Ii,t indicates the amount of fixed 308 asset investment in area i in t year; δ is depreciation rate, assign a value of 6%. The initial year capital 309 stock uses the depreciation of fixed assets of the year multiplied by 10%. 310

Control variables 311
Following existing literature, this article uses urbanization rate as the control variable, calculated as the 312 ratio of urban population to total population at the end of the year; per capita GDP represents the 313 overall economic level of a region; the ratio of fiscal expenditure to GDP represents the level of 314 marketization; the ratio of total import and export to GDP represents the degree of trade openness; per 315 million square meters of road area represent infrastructure construction.
come from EPS and CEIC databases.

Spatial autocorrelation test 326
First of all, this study uses the Moran I index to judge whether foreign direct investment and 327 environmental regulations between regions are spatially correlated. Specifically, when the index is 328 greater than 0, it indicates that a certain economic variable in each region is spatially positively 329 correlated; that is, there is spatial agglomeration. When the index is less than 0, it indicates that a 330 certain economic variable in each region is negatively correlated in space; that is, there is spatial 331 exclusion. When the index is equal to 0, it indicates that a certain economic variable is not related to 332 the regional distribution. Secondly, according to Table 2, foreign direct investment and environmental 333 regulations are significantly positive at the 1% level. It shows that foreign direct investment and 334 environmental regulations in various regions have an obvious positive autocorrelation in space, that is, 335 spatial agglomeration. Specifically, the Moran I index of foreign direct investment has shown a 336 downward trend as a whole. This shows that the spatial correlation of foreign direct investment is 337 gradually weakening; that is, the spatial distribution of foreign direct investment is becoming more and

Baseline estimation 356
In order to select a suitable spatial measurement model, this study conducted a series of tests as follows, 357 and the results are shown in Table 3. Firstly, the LM test is used to judge whether the model can be 358 simplified into a spatial autoregressive model or a spatial error model. This result shows that the null 359 hypothesis of no spatial error term and no spatial lag term is rejected at the 1% level. Secondly, LR and 360 Wald tests are further used to show that the spatial Dubin model is more suitable than the spatial 361 autoregressive model and the spatial error model. Thirdly, because prefecture-level city-level data is 362 used for empirical analysis in this study, and each city has its own characteristics, the Hausman test is 363 used to determine whether to choose a space and time double fixed effects model for estimation. 364 Finally, based on the above analysis, the spatial Dubin double fixed model is used in this study to 365 conduct an empirical analysis of the relationship between environmental regulations and foreign direct 366

investment. 367
This study uses the OLS basic regression model, spatial autoregressive model, spatial error model, 368 and spatial Dubin model to conduct empirical research on the relationship between foreign direct 369 investment and environmental regulation. The specific results are shown in Table 3. This result shows 370 that no matter which model is adopted, the coefficient of environmental regulation is significantly 371 negative at the 1% level. That is, foreign direct investment is significantly inhibited by environmental 372 regulations, and the research hypothesis H1 is initially verified. This may be the result of increasingly 373 stringent environmental regulations in various regions. First, under the call of "green water and green 374 mountains are golden mountains and silver mountains", increasingly stringent environmental 375 regulations have been proposed by the Chinese local government. This measure raised the marginal 376 cost of "environmentally unfriendly" and other foreign-funded enterprises, which in turn made them 377 withdraw from the market, and the inflow of high-polluting, high-energy-consuming foreign-funded 378 enterprises has been consciously restricted or prohibited by the Chinese local government. Second, 379 according to the "pollution halo" hypothesis, in order to achieve the effect of "decontamination and 380 cleaning", clean, capital-intensive and technology-intensive foreign-funded enterprises are actively 381

Decomposition effect 389
In order to judge whether the spatial spillover effect is significant, this study uses partial differentiation 390 to decompose the spatial effects of the spatial Dubin model into direct and indirect effects. After that, 391 the significance of the indirect effects of explanatory variables is used in this study to determine 392 whether the spatial spillover effects are significant, and the total effects are numerically equal to the 393 sum of the direct effects and the indirect effects. The specific results are shown in Table 4. First, from 394 the perspective of core explanatory variables, the coefficient of environmental regulation is 395 significantly negative at the 1% level. It shows that foreign direct investment is significantly inhibited 396 by environmental regulations and further validates the research hypothesis H1. In addition, the indirect 397 effects of environmental regulations are significantly negative at the 1% level. It shows that 398 environmental regulations will reduce local foreign direct investment and have significant negative 399 space spillover. This may be the result of environmental decentralization and competition between 400  On the other hand, the significant negative spatial spillover of environmental regulations may be 404 caused by the "top-to-top competition" of local governments for environmental governance from the 405 perspective of environmental decentralization. In order to release the ability signal to the higher-level government and increase its own promotion "weight", compared with neighboring regions, the intensity 407 of environmental governance has been further strengthened by the local government, resulting in 408 greater restrictions on the inflow of high-polluting and high-energy-consuming foreign-funded 409 enterprises. Secondly, in terms of control variables, on the one hand, the direct and indirect effects of 410 per capita GDP are both significantly positive at the level of 1%. It shows that foreign direct 411 investment is significantly promoted by per capita GDP, and per capita GDP has a significant positive 412 space overflow. This may be because the higher the per capita GDP, the more developed the overall 413 economic level of the society, and the more foreign companies are willing to invest in this area. On the 414 other hand, the direct effect of trade openness is significantly positive at the 1% level, and the indirect 415 effect is significantly negative at the 1% level. It shows that foreign direct investment is significantly 416 promoted by trade openness, and trade openness has a significant negative space spillover. This may be 417 because the higher the degree of local trade openness, the fewer restrictions on foreign companies 418 entering the local area, leading to more foreign companies' inflow. 419   Table 5, on the one hand, the 426 estimated results of eastern cities are relatively close to the national estimates; that is, the coefficient of 427 direct effect is significantly negative at the 1% level, and the coefficient of indirect effect is also 428 significantly negative. It shows that foreign direct investment is significantly suppressed by foreign 429 direct investment, and environmental regulations have a significant negative space overflow. On the 430 other hand, the coefficient of indirect effects in non-eastern cities is not significant, indicating that the 431 environmental regulations of non-eastern cities do not have spatial spillover effects. Secondly, it can be 432 seen from the above analysis that the difference between the estimation results of eastern cities and 433 non-eastern cities is more obvious. This may be the result of the overall economic difference between 434 eastern cities and non-eastern cities. On the one hand, cleaner and more technological foreign-funded 435 enterprises are attracted by eastern cities with more developed economies and complete infrastructure. 436 Based on its rich economic resources, optimized resource allocation and technology spillover effects of 437 foreign direct investment have been effectively used by local governments in eastern cities to take the 438 road to high-quality and sustainable development. Moreover, because the more demanding ecological 439 environment construction is proposed by residents of eastern cities with higher economic levels, the regional government has proposed more stringent environmental regulations than surrounding cities to 441 further restrict the inflow of high-polluting foreign-funded enterprises. On the other hand, the 442 technology spillover effect of foreign direct investment cannot be fully and effectively used by the 443 local governments of non-eastern cities with relatively backward economic development and education 444 to improve the economic and technological level of local cities. As a result, it is impossible to attract 445 high-quality foreign-invested enterprises, resulting in a substantial reduction in foreign direct 446 investment in non-eastern cities (Yang et al., 2019). Moreover, compared with surrounding cities, more 447 stringent environmental governance measures cannot be proposed by local governments in non-eastern 448 cities, resulting in environmental regulations that cannot generate spatial spillovers. 449 (2) In order to compare the differences in the impact of environmental regulations on foreign 450 direct investment between different levels of cities, the sample of prefecture-level cities is divided into 451 ordinary cities and key cities in this study. First of all, according to the results in Table 5, on the one 452 hand, the estimated results of ordinary cities and the whole country are relatively consistent, and 453 environmental regulations are significantly negative at the 1% level. It shows that foreign direct 454 investment is significantly inhibited by environmental regulations. On the other hand, the direct effect 455 of key cities is significantly negative at the 1% level and is significantly smaller than the direct effect 456 coefficients of national and ordinary cities. At the same time, the spatial coefficient of key cities is not 457 significant, and the indirect effect is only significantly negative at the 10% level, indicating that there is 458 no spatial spillover of environmental regulations in key cities. Secondly, it can be seen from the above 459 analysis that the difference between the estimation results of ordinary cities and key cities is more 460 obvious. This may be due to the difference in administrative hierarchy between ordinary cities and key 461 cities. Based on their own positioning and rich economic resources, key cities will look at the whole 462 country, improve their sustainable development capabilities through technological innovation and 463 industrial upgrading, and further restrict the inflow of high-polluting foreign-funded enterprises. On the 464 other hand, the administrative barriers between key cities are relatively strong, presenting a situation of 465 "fighting each other", resulting in no spatial spillover effect of environmental regulations. 466 Table 5. Results of heterogeneous effects. Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Mediation effect 469
(1) In order to explore the impact of environmental regulations on foreign direct investment from the 470 perspective of industrial upgrading, this study conducted the following research, and the specific results 471 are shown in Table 6. First, the direct effect of the interaction term between environmental regulation 472 and industrial upgrading is significantly positive at the 1% level, and the indirect effect is significantly 473 negative at the 1% level. It shows that from the perspective of industrial upgrading, foreign direct 474 investment is significantly promoted by environmental regulations; that is, the research hypothesis H2 475 is verified, and environmental regulations have significant negative spatial spillovers. Secondly, it can 476 be seen from the above analysis that this may be the result of the "innovation compensation" effect of 477 environmental regulations. On the one hand, based on the effect of "innovation compensation", 478 reasonable environmental regulations will appropriately increase the marginal cost of enterprises, 479 forcing enterprises to increase their R&D capital through financing and other methods and improve the 480 level of production technology. In the end, while guiding the overall green development of society (Du 481 et al., 2021), this initiative will transform from labor-intensive to capital-intensive and 482 technology-intensive through the "survival of the fittest" approach, which is industrial upgrading. On 483 the other hand, a higher level of product structure will bring a higher level of economic development. 484 Based on its own higher level of economic development, the local government will adopt more 485 stringent environmental governance measures than surrounding cities, resulting in further restrictions 486 on the inflow of high-polluting foreign-funded enterprises so that local cities can take an 487 environmentally friendly development path. 488 (2) In order to explore the impact of environmental regulations on foreign direct investment from 489 the perspective of resource allocation, this study conducted the following research, and the specific 490 results are shown in Table 6. Firstly, the direct effect of the interaction term between environmental 491 regulation and resource allocation is significantly positive at the 1% level, and the indirect effect is 492 significantly negative at the 1% level. It shows that from the perspective of resource allocation, foreign 493 direct investment is significantly promoted by environmental regulations; that is, the research 494 hypothesis H3 has been verified, and environmental regulations have a significant negative spatial 495 spillover effect. Secondly, it can be seen from the above analysis that this may be the result of the 496 "innovation compensation" effect of environmental regulations. On the one hand, in order to achieve 497 the optimization of resource allocation and the maximization of economic performance, the efficiency 498 of resource utilization and the degree of coupling and coordination between various economic 499 resources can be improved by reasonable environmental regulations through technological innovation. 500 On the other hand, foreign direct investment and the level of economic development are promoted by 501 higher resource allocation efficiency. Based on its own higher level of economic development, the 502 construction of an ecological civilization with higher requirements than surrounding cities was 503 proposed by the local government, thereby further restricting the inflow of foreign-funded enterprises. 504 In addition, it should be noted that the coefficient of resource allocation is significantly negative, 505 indicating that foreign direct investment is significantly suppressed by resource allocation. This may be 506 because the resource allocation is too low, that is, more than half of the samples with resource 507 allocation less than 0.5. The low resource allocation efficiency is not conducive to the inflow of 508 foreign-funded enterprises. 509

Robustness test 512
In order to verify whether the relationship between environmental regulations and foreign direct 513 investment is sound, the following methods are used. The specific results are shown in Table 7. First, 514 the calculation method of the explained variable was replaced by this study as the actual use of foreign 515 direct investment as a percentage of GDP. Secondly, considering that the spatial weight matrix is the 516 basis of the spatial measurement model, the spatial weight matrix is replaced by the spatial geographic 517 adjacency matrix in this study. Third, the calculation method of the explained variable and the spatial 518 weight matrix are replaced simultaneously. Finally, based on the above analysis, the robustness results 519 are basically consistent with the above results, which provides evidence for the reliability of the above 520 results. 521 Table 7. Robustness test results 522 Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.  negative spatial spillover effect. The environmental regulation of key cities does not have spatial 535 spillover. Finally, in the perspective of industrial upgrading, environmental regulation plays a 536 significant role in promoting foreign direct investment. The inflow of foreign-funded enterprises is 537 significantly inhibited by the improvement of local environmental regulation in neighboring regions; 538 that is to say, environmental regulation has a significantly negative spatial spillover effect. In the 539 perspective of resource configuration, environmental regulation also has a significant role in promoting 540 foreign direct investment and has a significantly negative spatial spillover. At the same time, too low 541 resource configuration efficiency will significantly inhibit foreign direct investment. Based on the 542 above conclusions, the following policy recommendations are put forward: 543 Firstly, from a national perspective, environmental regulation has a significant inhibitory effect 544 on foreign direct investment. On the one hand, reasonable environmental governance policies and 545 moderate environmental regulation measures can produce the "innovation compensation" effect to 546 achieve the effect of "decontamination and preservation" for foreign-funded enterprises. Domestic 547 enterprises can use the technology spillover effect of foreign direct investment to upgrade production 548 technology, introduce advanced equipment, attract clean, capital-intensive and technology-intensive 549 foreign-funded enterprises, and take the road of high-quality and sustainable development. On the other 550 hand, environmental regulation has a significantly negative spatial spillover effect. In the context of 551 environmental decentralization, the "top-to-top competition" of local government environmental 552 governance has been strengthened; that is, environmental decentralization has a positive role in 553 promoting local environmental governance. Environmental governance measures strengthened by local 554 government competition while significantly suppressing foreign direct investment are more likely to 555 appear a "one-size-fits-all approach" phenomenon. Therefore, it is necessary to formulate perfect and 556 reasonable environmental control measures and adopt different environmental control measures to deal 557 with foreign-funded enterprises with different pollution levels and development levels, but also to 558 further improve the local government competition mode under the context of environmental 559 decentralization. 560 Next, on the one hand, there are obvious differences between eastern cities and non-eastern cities 561 in China. First, foreign direct investment is significantly inhibited by the environmental regulation of 562 the eastern city; the environmental regulation also has significantly negative spatial spillover. Therefore, 563 the eastern cities with a relatively developed economy and comprehensive infrastructure should 564 continue to strengthen environmental regulation and establish a long-term mechanism for 565 environmental regulation to drive enterprise technological innovation through technological innovation 566 and other methods to make up for economic losses that restrict high pollution-related foreign 567 companies. Second, the non-eastern city environmental regulation has a significant inhibitory effect on 568 foreign direct investment, but its indirect effect is not significant. Therefore, non-eastern cities with 569 relatively backward economies should actively adopt higher education and talent attraction and other 570 methods to improve cities' overall technological innovation level to make full use of the technological 571 spillover effect of foreign direct investment to develop the economy.
On the other hand, there are obvious differences in the spatial spillover effects of environmental 573 regulation intensity between key cities and ordinary cities in China. The environmental regulation of 574 ordinary cities has significant negative spatial spillovers, but the environmental regulation of key cities 575 does not have spatial spillovers. Therefore, key cities should strengthen exchanges on the construction 576 of ecological civilization, break down administrative barriers between cities, form the integration of 577 environmental governance and avoid a "fragmented" pattern of environmental governance, ultimately, 578 it enables enterprises to carry out technological innovation，and avoid the "free ride" behavior of 579

enterprises. 580
At last, on the one hand, in the perspective of industrial upgrading, environmental regulation 581 significantly promotes foreign direct investment and has a significantly negative spatial spillover effect. 582 Therefore, firstly, more direct foreign investment will be directed into the technological innovation 583 sector, and the technological spillover effect of foreign direct investment shall be fully utilized to 584 complete the industrial upgrading of enterprises. Secondly, it is possible to appropriately lower the 585 environmental regulatory threshold for foreign-funded enterprises with lower pollution levels to better 586 exert the promotion effect of foreign direct investment on industrial upgrading. Finally, eastern cities 587 with relatively developed economies can make full use of foreign direct investment to upgrade their 588 industries; however, non-eastern cities with relatively backward economies are more suitable to force 589 enterprises to upgrade their industries through the "innovation compensation" effect of environmental 590 regulation. On the other hand, in resource configuration, environmental regulation also significantly 591 promotes foreign direct investment and has a significantly negative spatial spillover effect. Therefore, 592 local governments should make full use of the "innovation compensation" effect of environmental 593 regulation to force enterprises to upgrade production technology. Meanwhile, local governments can 594 also strengthen subsidies for enterprise technology innovation and provide low-interest loans to 595 enhance the enterprise's resource allocation efficiency, thus attracting environmentally friendly, 596 technical, and other foreign companies. 597 Declarations 598 • Ethics approval and consent to participate: Not applicable 599 • Consent for publication: Not applicable 600 • Availability of data and materials: All data generated or analyzed during this study are included 601 in this article. 602 • Competing interests: The authors declare that they have no competing interests. 603 • Funding: The authors acknowledge financial support from the Special project for national and 604 regional research of colleges and universities of the Ministry of Education (planning project)： 605 Research on high quality development of China-Central Asia-West Asia Economic Corridor in the 606 Context of the Belt and Road Initiative(2020-G70)and Xinjiang Social Science Fund Project 607 (20BJL062). 608